Power distribution equipment state monitoring system and monitoring method based on Internet of Things technology

Abstract

The invention discloses a power distribution equipment state monitoring system and monitoring method based on the Internet of Things technology, and relates to the technical field of power distribution equipment monitoring, the system comprises the following components: a sound acquisition module, a data transmission module, a data processing center and an early warning display module; by combining the Internet of Things technology, the expert system and the feature engineering technology, high-precision acquisition and analysis of operation sound signals of the power distribution equipment are realized, and by constructing the fault voiceprint feature library, the system can quickly recognize and match equipment fault features, so that early warning is given out at the initial stage of the fault, and the fault early warning efficiency is improved. In addition, due to the application of the cross-equipment associated fault prediction model and the fault propagation prediction model, the system can predict the propagation conditions of cascading faults and system-level faults, more comprehensive fault information is provided for operation and maintenance personnel, prevention measures can be taken in advance, and the fault monitoring accuracy and timeliness are improved. The influence of faults on a power distribution system is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution equipment monitoring, and in particular to a power distribution equipment status monitoring system and a monitoring method based on Internet of Things technology. Background Art

[0002] Distribution equipment plays a vital role in the power system. Its stable operation is the basis for ensuring the reliability and safety of power supply. With the continuous development of the power system and the improvement of its intelligence level, higher requirements are placed on the monitoring and maintenance of distribution equipment. In recent years, the rapid development of Internet of Things technology has provided new opportunities and technical means for the intelligent monitoring of distribution equipment. Through Internet of Things technology, real-time monitoring, data analysis and fault warning of the status of distribution equipment can be realized, thereby improving the operation and maintenance management level of the distribution system.

[0003] However, traditional distribution equipment monitoring methods have many limitations. On the one hand, traditional methods mainly rely on manual inspections and regular testing, which is not only inefficient, but also difficult to capture equipment fault information in real time, resulting in delays in fault discovery and processing. On the other hand, traditional methods lack accurate diagnostic means and can often only detect the macroscopic operating status of the equipment, but cannot deeply analyze the microscopic fault characteristics of the equipment. In addition, traditional methods have limited predictive capabilities for cross-device correlated faults and system-level fault propagation, and it is difficult to effectively respond to complex and changeable fault conditions. These problems limit the application effect of traditional methods in the field of distribution equipment monitoring, and it is difficult to meet the high-precision and high-adaptability requirements of modern power systems for distribution equipment monitoring.

[0004] In view of the above problems, it is necessary to optimize the existing distribution equipment status monitoring system and monitoring method. By integrating advanced technologies such as sound acquisition, data transmission, data processing and analysis, and combining expert systems and feature engineering, the accurate extraction and analysis of the voiceprint features of distribution equipment, as well as the in-depth prediction of cross-device related faults and system-level fault propagation paths can be achieved. Therefore, it is of great significance to develop a distribution equipment status monitoring system and monitoring method based on the Internet of Things technology that can comprehensively realize the above characteristics. Summary of the invention

[0005] The purpose of the present invention is to make up for the shortcomings of the prior art and to provide a distribution equipment status monitoring system and monitoring method based on the Internet of Things technology. It can realize real-time monitoring and accurate diagnosis of the status of distribution equipment through the comprehensive use of modules such as sound acquisition module, data transmission module, and data processing center. It uses expert system and feature engineering technology to extract and analyze voiceprint features of sound samples, establishes a fault voiceprint feature library, and improves the accuracy and adaptability of fault diagnosis. At the same time, through the correlation analysis and prediction unit, cross-device correlation fault prediction and system-level fault propagation path analysis are realized, providing comprehensive fault information and preventive measures for operation and maintenance personnel. The monitoring method of the present invention can not only accurately identify single equipment failures, but also predict cross-equipment chain failures and system-level fault propagation, effectively improving the operation and maintenance management level of the distribution system and ensuring the reliability and safety of power supply.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: On the one hand, a distribution equipment status monitoring system based on the Internet of Things technology includes the following components: a sound collection module, a data transmission module, a data processing center and an early warning display module;

[0007] The sound collection module uses a high-sensitivity microphone and is equipped with an electromagnetic shielding cover to reduce electromagnetic interference. It uses a low-noise and high-gain preamplifier to amplify the collected weak sound signal, and converts it into digital audio data by a high-precision analog-to-digital converter according to the corresponding sampling accuracy and a sampling frequency that is not less than twice the highest frequency of the sound signal. At the same time, a noise reduction algorithm is used to process the sound signal, so as to ensure the quality of the collected sound signal;

[0008] The data transmission module supports Wi-Fi, Bluetooth and 4G / 5G communication protocols, can automatically switch the communication mode according to the signal strength of the environment, and use the AES encryption algorithm to encrypt the transmitted data, combined with the CRC cyclic redundancy check to ensure data integrity, and use cache technology to ensure the continuity of data transmission when switching protocols, so as to achieve stable transmission of sound and various related data from the acquisition end to the data processing center;

[0009] The data processing center includes a storage unit, a feature analysis unit, a fault diagnosis unit, a correlation analysis and prediction unit, and a preventive measures generation unit;

[0010] The storage unit is used to store a large amount of sound sample data of power distribution equipment in normal operation and various known fault states, as well as processed fault voiceprint feature library data, association relationship data between various devices, cross-device association fault prediction model data, topological structure information of the power distribution system and electrical connection relationship model data;

[0011] The feature analysis unit extracts and analyzes the voiceprint features of the sound samples using an expert system combined with feature engineering technology. The expert system builds a rule base based on the knowledge and experience of experts in the power field. Feature engineering processes the frequency, amplitude, harmonic structure and time-frequency domain characteristic information of the sound. By analyzing a large amount of sample data and matching rules, a mapping relationship between different fault states and unique voiceprint patterns is established;

[0012] The fault diagnosis unit compares and matches the voiceprint features of the sound data of the power distribution equipment received in real time with the stored fault voiceprint feature library. If the match is successful, it is determined that the power distribution equipment has a corresponding fault and generates fault warning information;

[0013] The association analysis and prediction unit obtains the operation data of adjacent or associated equipment through the Internet of Things network, including various operation parameters of the switchgear in the distribution box, the connected cable lines and the upper-level substation, namely, temperature, current and voltage, and uses the association analysis algorithm to analyze the inherent connection and mutual influence between the equipment operation data, establish a cross-equipment association fault prediction model, and predict the scope and time node of the chain fault that will be caused in advance according to the current operation status change trend of each device. At the same time, based on the topological structure and electrical connection relationship of the distribution system, combined with the equipment fault information, analyze the propagation path of the fault in the system and the affected area, and build a fault propagation prediction model;

[0014] The preventive measures generating unit generates corresponding preventive measures plans for the possible chain failures and system-level failures according to the results of the correlation analysis and prediction unit;

[0015] The warning display module receives fault warning information from the data processing center, and displays the fault type, location, scope of the caused chain failure and estimated time node, fault propagation path and affected area to the operation and maintenance personnel through sound and light alarms and display screen prompts.

[0016] Furthermore, the sound collection module uses a noise reduction algorithm to process the sound signal to ensure the quality of the collected sound signal. The algorithm formula is: Among them, s(t) is the original collected sound signal, w(t) is the wavelet coefficient after s(t) is transformed by wavelet, and λ is the threshold.

[0017] Furthermore, the data transmission module uses the AES encryption algorithm to encrypt the transmitted data. Specifically, the key length is determined and the key is generated by a secure random number generator, which is stored in an area with strict access control and updated regularly. The block size is determined according to the requirements of the AES algorithm and the characteristics of the data, and the data is sequentially divided into blocks. When the data is less than 16 bytes, a specific padding method is used to fill it. The AES encryption algorithm is initialized with the stored key and the encryption mode is selected. The round function transformation is used to generate ciphertext blocks, and the ciphertext blocks are sequentially integrated to form an encrypted data stream for transmission, thereby ensuring the confidentiality and security of data transmission.

[0018] Furthermore, the feature analysis unit of the data processing center uses an expert system combined with feature engineering technology to extract and analyze the voiceprint features of the sound sample, and its feature extraction formula is: Where X represents the sound signal sequence, m is the subsequence length, r is the similarity tolerance, and N is the length of the sound signal sequence. is to satisfy the condition max 1≤j≤m |x i+j -x j The ratio of the number of subsequences with |≤r to the total number of subsequences.

[0019] Furthermore, the fault diagnosis unit of the data processing center compares and matches the voiceprint features of the power distribution equipment sound data received in real time with the stored fault voiceprint feature library, and the matching formula is: Among them, X is the feature vector of the sound data to be diagnosed, Y is a certain fault mode feature vector in the fault voiceprint feature library, μ Y is the mean of the eigenvectors of the fault mode Y, is its covariance matrix, It measures the statistical distance between the feature vector X and the fault mode Y in the feature space. DTW(X, Y) is the dynamic time warping distance, which is used to process the possible deformation or offset of the sound signal on the time axis. MDTW When (X, Y) is less than the set threshold δ, it is determined that the match is successful, that is, the device has a corresponding fault.

[0020] Furthermore, the correlation analysis and prediction unit of the data processing center uses a correlation analysis algorithm to analyze the inherent connection and mutual influence between the equipment operation data, and the algorithm formula is: Among them, k i and k j are the degrees of device i and device j in the complex network model of the power distribution system, i.e., the number of devices connected to them, and k ij is the number of common neighbors between device i and device j, C ij Indicates the association degree between device i and device j.

[0021] Furthermore, the correlation analysis and prediction unit of the data processing center analyzes the propagation path of the fault in the system and the affected area based on the topological structure and electrical connection relationship of the power distribution system and the equipment fault information. The calculation formula is: Among them, x (0) (k) is the original equipment operating parameter sequence, x (1) (k) is the value of x (0) (k) A new sequence is generated by first-order accumulation, where a is the development grayscale and u is the endogenous control grayscale. This formula is used to make short-term predictions on the changing trends of the equipment operating parameters. Based on the prediction results, it is determined whether a fault will occur and the speed at which the fault will develop, and preventive measures can be taken in advance.

[0022] Furthermore, the correlation analysis and prediction unit of the data processing center analyzes the propagation path of the fault in the system and the affected area based on the topological structure and electrical connection relationship of the power distribution system and the equipment fault information, and constructs a fault propagation prediction model, the model formula of which is: in are the planned values ​​of injected active power and reactive power of node i, V i is the voltage amplitude at node i, θ ij is the voltage phase difference between node i and node j, Y ij It is an element in the node admittance matrix. By iteratively solving the above equations, the voltage amplitude and phase angle of each node in the system under normal operating conditions are obtained. When a device fails, the iterative calculation is performed again to compare the changes in voltage and current of each node before and after the failure, and the scope and degree of the impact of the failure on the system are analyzed to determine the fault propagation path and the area that may be affected.

[0023] On the other hand, a distribution equipment status monitoring system method based on the Internet of Things technology includes the following specific steps:

[0024] Sound sample collection: During the installation and commissioning phase of the power distribution equipment and the daily operation and maintenance process, the sound collection module collects sound samples of the power distribution equipment in normal operation and various known fault simulation states, marks and classifies the samples, and records the corresponding equipment operation status information;

[0025] Feature library construction: The collected sound sample data is transmitted to the storage unit of the data processing center, and the sample data is processed by the expert system in the feature analysis unit combined with feature engineering technology. The feature information obtained after processing is constructed into a fault voiceprint feature library and stored in the storage unit;

[0026] Correlation data collection and model construction: The data transmission module continuously collects correlation data between each distribution device, including temperature, current and voltage parameter information, as well as topological structure data and electrical connection relationship data of the distribution system, and transmits it to the storage unit of the data processing center. Based on these correlation data, the correlation analysis and prediction unit uses machine learning technology to analyze the correlation relationship between devices and build a cross-device correlation fault prediction model. At the same time, a fault propagation prediction model is built based on the topological structure and electrical connection relationship to determine the propagation rules and impact range of equipment faults in the system. For example, by analyzing a large amount of historical data, the propagation path probability matrix of different types of equipment faults under a specific system structure is determined;

[0027] Real-time monitoring and diagnosis: During the normal operation of the power distribution equipment, the sound collection module continuously collects the equipment sound signals, which are transmitted to the data processing center in real time through the data transmission module. The fault diagnosis unit of the data processing center extracts the voiceprint features of the real-time sound data, and the feature analysis unit uses expert system rules and feature engineering technology to process and match it with the stored fault voiceprint feature library. If the match is successful, it is determined that the equipment has failed, and the early warning display module is immediately triggered to issue a fault warning message. If the match is not successful, the correlation analysis and prediction unit further analyzes the current equipment operation data and the operation data of the related equipment, and uses the cross-device correlation fault prediction model to predict whether there is a potential risk of chain failure. If the risk of chain failure is predicted, the early warning display module is also triggered to issue a warning containing information related to the chain failure.

[0028] Preventive measures execution: The preventive measures generation unit generates a preventive measures plan based on the prediction results of the correlation analysis and prediction unit and executes it automatically or prompts the operation and maintenance personnel to execute it;

[0029] Model update: Periodically or after a new fault type is discovered, repeat the sound sample collection steps and feature library construction steps to update and optimize the fault voiceprint feature library. At the same time, based on new equipment operation data, fault cases, and system structure change information, update the association rules and parameters in the cross-device association fault prediction model, as well as the propagation rules and probability matrix in the fault propagation prediction model to adapt to changes in the operating status of distribution equipment, the emergence of new fault conditions, and adjustments to the system structure.

[0030] Compared with the prior art, the power distribution equipment status monitoring system and monitoring method based on the Internet of Things technology have the following beneficial effects:

[0031] 1. The present invention combines Internet of Things technology, expert system and feature engineering technology to achieve high-precision collection and analysis of sound signals of power distribution equipment operation. By building a fault soundprint feature library, the system can quickly identify and match equipment fault characteristics, thereby issuing an early warning at the early stage of the fault, improving the accuracy and timeliness of fault monitoring. In addition, the application of cross-device association fault prediction model and fault propagation prediction model enables the system to predict the propagation of chain faults and system-level faults, providing operation and maintenance personnel with more comprehensive fault information, which helps to take preventive measures in advance and reduce the impact of faults on the power distribution system.

[0032] 2. By regularly collecting new sound samples and updating the fault voiceprint feature library, and updating the prediction model according to new equipment operation data, fault cases and system structure change information, the present invention can continuously adapt to the state changes of distribution equipment during long-term operation, the emergence of new fault types and the adjustment of system structure, thereby ensuring the long-term effectiveness and reliability of the monitoring system and providing strong support for the intelligent operation and maintenance management of the distribution system.

[0033] Other advantages, objectives and features of the present invention will be set forth in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be taught from the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 This is a flow chart of the power distribution equipment status monitoring method of the present invention;

[0035] Figure 2 This is a flow chart of the power distribution equipment status monitoring system of the present invention. DETAILED DESCRIPTION

[0036] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation mode, structure, characteristics and effects of the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments.

[0037] Embodiment 1

[0038] In the key power hub of the urban substation, the layout and design of the sound collection module fully consider the characteristics and operating environment of various distribution equipment in the substation. The high-sensitivity microphone uses a condenser microphone, whose frequency response accurately covers the sound spectrum range of the distribution equipment from 20Hz to 20kHz, and can keenly capture extremely subtle sound changes during the operation of the equipment. In order to resist the complex and strong electromagnetic interference in the substation, the microphone is encapsulated in a customized aluminum electromagnetic shielding cover. The shielding cover is a closed rectangular structure with only a specially designed circular channel facing the microphone pickup position. Multi-layer metal shielding nets and high-performance electromagnetic The filter forms an efficient electromagnetic shielding "barrier" to ensure that the sound signal collected by the microphone is pure and interference-free. The weak sound signal collected by the microphone first enters the low-noise, high-gain preamplifier, and its gain is precisely set to 30 times, which is enough to amplify the weak signal at the millivolt level to a level range suitable for subsequent analog-to-digital conversion. Subsequently, the high-precision analog-to-digital converter digitizes the amplified sound signal with a sampling frequency of 48kHz and a sampling accuracy of 16 bits. It can collect 48,000 sample data per second, thereby accurately restoring the detailed characteristics of the sound signal. The converted digital audio data is transmitted to the data transmission module via a high-speed data bus.

[0039] When the data transmission module is started, it quickly conducts a comprehensive scan and assessment of the network environment in the substation. Since a stable and high-speed Wi-Fi network is deployed in the substation, its signal strength has been detected to be -50dBm, which is much higher than the preset stable connection threshold of -70dBm. The data transmission module automatically prioritizes the Wi-Fi communication protocol for connection. During the connection process, according to the IEEE802.11 standard protocol, it performs identity authentication and parameter negotiation with the designated Wi-Fi access point. After the connection is successfully established, it obtains the dynamically allocated intranet IP address to ensure that the data can be accurately transmitted to the data processing center. In order to ensure the security and integrity of data transmission, the AES-128 encryption algorithm is used to encrypt sound data, device-related data, etc. For each data packet to be transmitted, it is first divided into a data block of 16 bytes. If the remaining data is less than 16 bytes, it is supplemented by PKCS#7 padding, and then , each data block is encrypted using a 128-bit key that is pre-stored in a secure storage area and updated regularly (updated every 30 days). During the encryption process, the AES algorithm runs in cipher block chaining (CBC) mode. Before encryption, each data block is first XORed with the ciphertext of the previous data block, and then undergoes a series of complex round function transformation operations such as byte substitution, row shift, and column confusion to generate the corresponding ciphertext data block. The encrypted ciphertext data block is then attached with a 4-byte checksum calculated by the CRC-32 algorithm to form the final transmission data packet and send it to the data processing center. During the data transmission process, if the Wi-Fi signal fluctuates or is interrupted briefly (such as due to equipment maintenance or temporary electromagnetic interference), the data transmission module automatically enables the cache technology and temporarily stores the data in a cache area with a capacity of 10MB. After the signal is restored to stability, the cached data is quickly transmitted to the data processing center at a transmission rate of 10Mbps to ensure the continuity and integrity of data transmission.

[0040] The data processing center is the core hub of the entire monitoring system. Its storage unit runs efficiently based on the distributed storage architecture built on Hadoop Distributed File System (HDFS). The storage nodes are composed of 5 high-performance servers. Each server is equipped with a 2TB large-capacity hard disk, which can store massive historical data and real-time data of power distribution equipment. When storing data, it is first classified and identified according to the type of data (such as sound sample data, equipment operation parameter data, topology structure data, etc.), and then the data is divided into 128MB data blocks and evenly distributed and stored on different storage nodes according to the hash value-based storage strategy. A multi-level index mechanism is also established. For example, for sound sample data, a first-level index is established according to the equipment type, such as transformer, switchgear, busbar, etc. Under the equipment type index, a second-level index is established according to the collection time, with the time accuracy accurate to seconds, so that the sound sample data of any time and any device can be quickly located and retrieved.

[0041] The feature analysis unit plays a key role in the data processing process. The construction of the expert system is based on the deep integration of the knowledge and experience of many senior experts in the power field. After many rounds of discussions and case analysis, the experts have compiled a set of rule bases covering the correspondence between more than 20 common distribution equipment fault types (such as transformer winding short circuit, insulation breakdown, switch cabinet contact overheating, loose busbar connection, etc.) and sound features. Each rule describes in detail the fault type, sound feature conditions (including specific frequency range, amplitude change mode, harmonic structure characteristics, etc.) and confidence (the value range is 0-1, indicating the reliability of the rule). For example, for a transformer winding short circuit fault, the rule It is described as "when the sound has a continuous high-frequency spike signal in the frequency range of 800Hz-2.5kHz, and the amplitude exceeds 5 times the normal operating value, the harmonic content increases significantly, and the confidence level is 0.9". The inference engine uses a forward reasoning algorithm. When the new sound data features are input, starting from the first rule in the rule library, it checks whether the sound data meets the sound feature conditions in the rule. In terms of feature engineering implementation, the sound data is first preprocessed, and the mean filter algorithm is used to remove high-frequency noise and effectively smooth the sound signal curve. Then, the amplitude range of the sound data is normalized to between 0 and 1 through linear transformation. Then, the voiceprint feature extraction formula based on multi-scale entropy is used. Extract time-frequency domain features, where the subsequence length m is preliminarily determined to be 30 based on the analysis and experiments of a large amount of historical sound data, and the similarity tolerance r is set to 0.25σ based on the standard deviation σ of the sound signal amplitude. In the feature optimization stage, the principal component analysis algorithm is used for dimensionality reduction processing. By calculating the eigenvalues ​​and eigenvectors of the feature covariance matrix, the first 8 eigenvectors with larger eigenvalues ​​are selected (experimentally verified to retain more than 95% of the information), and the original eigenvectors are projected onto these 8 eigenvectors to obtain the eigenvectors after dimensionality reduction, and the fault voiceprint feature library is constructed. After receiving the real-time sound data, the fault diagnosis unit starts the fault diagnosis process. First, the voiceprint features are extracted by the feature analysis unit to obtain the feature vector X. Then, the fault matching formula based on the fusion of Mahalanobis distance and dynamic time warping (DTW) is used. Perform fault matching. During the calculation process, for a certain fault mode Y in the fault voiceprint feature library, its feature vector mean μ Y and the covariance matrix The Mahalanobis distance part has been pre-calculated and stored in the feature library construction process. The statistical distance between the feature vector X and the fault mode Y in the feature space is accurately measured by matrix operation, and the correlation between the features is taken into account. At the same time, DTW (X, Y) efficiently finds the optimal matching path between X and Y on the time axis through a dynamic programming algorithm to obtain the DTW distance. In this embodiment, after a large number of experiments and data analysis, it is determined that when the Mahalanobis distance weight is 0.65 and the DTW distance weight is 0.35, the fault diagnosis accuracy is high. If the calculated D MDTW If the (X, Y) value exceeds the set threshold of 0.4, and other operating parameters of the equipment, such as temperature, current, and voltage, are also beyond the normal range after comprehensive judgment (for example, the transformer oil temperature exceeds 80°C, and the winding current exceeds 1.2 times the rated value), the equipment is judged to have a corresponding fault.

[0042] The association analysis and prediction unit deeply explores the potential association relationship between devices based on complex network theory. By constructing a complex network model of the distribution system, the transformers, switch cabinets, busbars, feeders and other equipment in the substation are abstracted as network nodes. The electrical connection relationship between the equipment is represented by directed edges. The weight of the edge is determined according to the electrical parameters of the connection line (such as impedance, conductivity, etc.). The equipment association degree calculation formula based on complex network theory is used Calculate the correlation between devices. For example, to calculate the correlation between a switch cabinet and the connected bus, first determine the degree k of the switch cabinet in the network. i =5 (that is, the number of devices connected to it is 5), the degree of the bus is k j =8, the number of common neighbors between them is k ij =3, then the correlation can be calculated The equipment association degree and the equipment's operating parameters such as temperature, current, and voltage are used as input feature vectors, and whether a cascading failure occurs, the fault range, and the time node are used as output labels. When the Newton-Raphson method in polar coordinate form is used for power flow calculation, for node i (i = 1, 2, ..., n), its power equation is: in are the planned values ​​of injected active power and reactive power of node i (given values, which can be obtained from the equipment operation parameters), V i is the voltage amplitude at node i, θ ij is the voltage phase difference between node i and node j, Y ij It is an element in the node admittance matrix. By iteratively solving the above equations, the voltage amplitude and phase angle of each node in the system under normal operating conditions can be obtained. When a device fails, the power flow calculation is re-performed to compare the changes in voltage and current of each node before and after the failure, and the scope and degree of the impact of the failure on the system are analyzed to determine the fault propagation path and the area that may be affected.

[0043] After receiving the results of the correlation analysis and prediction unit, the preventive measures generation unit quickly retrieves the matching preventive measures plan from the pre-established preventive measures knowledge base. The knowledge base is constructed based on the analysis and summary of a large number of historical failure cases, and detailed and specific preventive measures are formulated for different types of equipment failures and chain failure prediction results. For example, when it is predicted that a transformer failure may cause bus voltage fluctuations and affect downstream feeders, the plan generated by the preventive measures generation unit includes: first, automatically adjust the reactive compensation device of the substation to increase the reactive compensation capacity by 20% to stabilize the bus voltage; second, quickly transfer important downstream loads (such as hospitals, government agencies, etc.) to the backup power supply line through the intelligent switching device to ensure uninterrupted power supply to important areas; at the same time, arrange operation and maintenance personnel to conduct emergency inspections of the faulty transformer and related equipment, and prepare maintenance tools and spare parts so that maintenance work can be carried out quickly after the failure occurs, thereby minimizing the impact of the failure on the urban power supply system.

[0044] When the data processing center determines that the equipment fails or predicts a cascading failure, the early warning display module immediately activates the early warning mechanism. In the monitoring room of the substation, the sound and light alarm sends a strong alarm signal, and the high-brightness red LED indicator flashes at a frequency of 3 times per second to attract the attention of the operation and maintenance personnel. At the same time, the high-decibel speaker emits a continuous alarm sound with a volume of up to 90 decibels to ensure that it can be clearly heard in the noisy substation environment. The display screen displays the distribution system structure in the form of an intuitive topological diagram. The faulty equipment is prominently displayed on the topological diagram with a red flashing icon. For example, the icon of the faulty transformer is obviously flashing, and the fault propagation path is clearly depicted by orange lines, such as from the faulty transformer along a bus to multiple switch cabinets and feeders. The extended path is clear at a glance. Next to the topology map, the fault type, location, scope of cascading faults and estimated time nodes, fault propagation path and possible affected areas are displayed in detail in a list format, which is convenient for operation and maintenance personnel to quickly obtain key information. In addition, the early warning display module will also push the early warning information to the mobile phone application of the operation and maintenance personnel in a timely manner. The pushed text message content is concise and clear, including brief fault information and an exclusive link to the detailed information page of the early warning display module. After receiving the text message notification, the operation and maintenance personnel can click the link to conveniently view the complete early warning information on their mobile phones, including topology map display, detailed parameters and other content, so that they can respond quickly and take effective countermeasures in time to ensure the stability and reliability of the city's power supply.

[0045] Embodiment 2

[0046] In an industrial park environment with diverse electricity demands and complex distribution networks, the sound collection module is customized and deployed based on the unique operating conditions and sound characteristics of the industrial park's distribution equipment. The microphone selection takes into account the sound frequency range and intensity changes generated by the distribution boxes, cable connectors and other equipment in the park under different operating conditions. An electret microphone with a wide dynamic range and high signal-to-noise ratio is selected. Its frequency response range is 30Hz-18kHz, which can accurately capture the subtle differences in the operating sound of the equipment. In terms of installation, the microphone is fixed to the key parts of the inner wall of the distribution box near the switch equipment and cable connectors in view of the limited internal space and relatively complex electromagnetic environment of the distribution box. It is not only It is easy to install and can ensure stable and reliable sound collection. The electromagnetic shielding cover is made of stainless steel and has a compact box-shaped structure. A high-performance electromagnetic shielding film is installed at the sound collection port to effectively block external electromagnetic interference while having minimal effect on the attenuation of the sound signal. The gain of the preamplifier is finely adjusted according to the actual strength of the sound signal of the park equipment and is set to 25 times. It can fully amplify the weak sound signal to meet the analog-to-digital conversion requirements. The analog-to-digital converter digitizes the sound signal with a sampling frequency of 44.1kHz and a sampling accuracy of 16 bits to ensure high-quality collection and transmission of the sound signal. The converted digital audio data is transmitted to the data transmission module via the wireless Bluetooth module.

[0047] In the industrial park environment, the data transmission module is blocked by dense buildings and the Wi-Fi signal strength is weak (less than -70dBm). The data transmission module automatically switches to the 4G network for data transmission. During the switching process, the 4G network module is first initialized and configured, and the operator base station with the strongest signal is searched and connected to. After obtaining a valid IP address, a stable network connection channel is established. During data transmission, the AES-128 encryption algorithm is also used to encrypt and protect the data. The data segmentation and encryption process are similar to the first embodiment. Each 16-byte data block is encrypted by AES and attached with a CRC-16 checksum value to form a transmission data packet and send it to the data processing center. During the 4G network transmission process, the data transmission module monitors the network signal quality in real time. When the signal strength fluctuates or the transmission rate decreases, the transmission parameters are automatically adjusted, such as reducing the data transmission frequency or using data compression technology to ensure that the data can be transmitted stably. If the 4G network is briefly interrupted, the data transmission module caches the data in a cache area with a capacity of 5MB. After the network is restored, the cached data is transmitted to the data processing center at a transmission rate of 5Mbps to ensure the integrity and continuity of the data.

[0048] The storage unit of the data processing center adopts a distributed storage architecture. It consists of a storage cluster of three servers. Each server is equipped with a 1.5TB hard disk, which can meet the storage needs of the power distribution equipment data in the park. When storing data, an indexing mechanism is established according to multiple dimensions such as equipment area, equipment type, and collection time. For example, the first-level index is first divided according to different production areas in the park (such as production area A, production area B, etc.), and then a second-level index is established in each production area according to the equipment type (such as distribution box, transformer, cable, etc.). Finally, a third-level index is established according to the collection time accurate to milliseconds, so that the sound sample data and related information of any area and any device at any time can be quickly located and retrieved.

[0049] In view of the complexity and diversity of equipment operating conditions in industrial parks, the feature analysis unit has carried out targeted optimization in terms of expert system construction and feature engineering implementation. The expert system rule base covers the correspondence between 15 common equipment failure types and sound features in the park. The confidence of each rule has been repeatedly verified and adjusted to ensure its accuracy and reliability. For example, for the overheating failure of the switchgear contact in the distribution box, the rule is described as "when the sound has periodic amplitude fluctuations in the frequency range of 200Hz-800Hz, and the fluctuation amplitude gradually increases, the harmonic component increases slightly, and the confidence is 0.85". The reasoning engine adopts a hybrid reasoning algorithm combining forward reasoning and reverse reasoning. First, the possible fault types are preliminarily screened out through forward reasoning, and then the preliminary results are verified and refined through reverse reasoning to improve the accuracy of fault diagnosis. In feature engineering, the sound data preprocessing uses 7-point median filtering to remove noise interference, and then the sound data amplitude is normalized to between -1 and 1 to meet the requirements of the subsequent feature extraction algorithm. For the voiceprint feature extraction formula based on multi-scale entropy After a large number of experiments and data analysis, it was determined that when the subsequence length m = 25 and the similarity tolerance r = 0.2σ, the characteristic information of the sound of the park equipment can be better extracted. In the feature optimization stage, the independent component analysis (ICA) algorithm is used to reduce the dimension of the features, and 6 independent components are selected as the feature vectors after dimension reduction to construct the fault voiceprint feature library, which effectively removes feature redundancy and improves the efficiency of fault diagnosis.

[0050] Therefore, when processing the sound data of the power distribution equipment in the park, the diagnosis unit strictly follows the fault diagnosis process. After receiving the real-time sound data, the feature analysis unit extracts the voiceprint features to obtain the feature vector X, and then uses the fault matching formula based on the fusion of Mahalanobis distance and dynamic time warping (DTW) In this embodiment, multiple cross-validation experiments are performed on the historical sound sample data set of the park equipment to determine that the fault diagnosis accuracy is higher when the Mahalanobis distance weight is 0.6 and the DTW distance weight is 0.4. MDTWIf the (X, Y) value exceeds the set threshold of 0.35, and the operating parameters of the associated equipment also show an abnormal trend (for example, the current value of the connected cable fluctuates by more than 30% of the normal range in a short period of time), the equipment is judged to have a corresponding fault. The correlation analysis and prediction unit deeply analyzes the correlation between the equipment in the park distribution network based on complex network theory. When constructing the complex network model of the park distribution system, the distribution boxes, transformers, cables and other equipment are abstracted as network nodes. The directed edges are determined based on the electrical connection relationship between the equipment and the direction of power transmission. The weight of the edge comprehensively considers the electrical parameters such as the current carrying capacity and impedance of the connection line, and uses the equipment accessibility calculation formula based on complex network theory. Calculate the correlation between devices. For example, in the correlation calculation between a distribution box and its connected cables, the degree k of the distribution box in the network is i =4, the length of the connected cable k j =6, the number of common neighbors between them is k ij =2, then the correlation can be calculated The equipment association degree and the equipment's operating parameters such as temperature, current, and voltage are used as input feature vectors, and whether a cascading failure occurs, the fault range, and the time node are used as output labels. When the Newton-Raphson method in polar coordinate form is used for power flow calculation, for node i (i = 1, 2, ..., n), its power equation is: in are the planned values ​​of injected active power and reactive power of node i (given values, which can be obtained from the equipment operation parameters), V i is the voltage amplitude at node i, θ ij is the voltage phase difference between node i and node j, Y ij It is an element in the node admittance matrix. By iteratively solving the above equations, the voltage amplitude and phase angle of each node in the system under normal operation can be obtained. When a device fails, the power flow calculation is re-performed to compare the changes in voltage and current of each node before and after the failure, and the scope and degree of the impact of the failure on the system are analyzed to determine the fault propagation path and the area that may be affected.

[0051] After receiving the results of the correlation analysis and prediction unit, the preventive measures generation unit quickly matches and generates corresponding solutions from the preventive measures knowledge base. The knowledge base is constructed based on in-depth analysis and summary of various types of past failure cases in the park, and detailed response strategies are formulated for different equipment failures and chain failure prediction results. For example, when it is predicted that a failure in a distribution box may cause power outages in the connected production line equipment, the solution generated by the preventive measures generation unit includes: first, automatically triggering the orderly power-off protection program for the equipment on the production line, and shutting down the power supply of non-critical equipment in turn according to the importance of the equipment and the power-off sequence requirements to ensure the safe storage of key equipment data; second, quickly switching the backup power supply to the important equipment on the production line to ensure its continuous operation and avoid major losses caused by production interruptions; at the same time, arranging operation and maintenance personnel to carry professional testing equipment to the faulty distribution box and related equipment for detailed inspection, and preparing corresponding maintenance tools and spare parts according to the type of fault, so that maintenance work can be carried out quickly after the fault occurs, reducing the impact of the fault on the production activities of the park, and ensuring the reliable operation of the park's power distribution system.

[0052] When the data processing center determines that the equipment fails or predicts a cascading failure, the early warning display module immediately activates the early warning mechanism. Display screens are set up in the operation and maintenance office of the industrial park and key locations on site. When a fault occurs, the display screen displays the structure of the park's power distribution system in the form of a striking topology map. The faulty equipment is highlighted on the topology map with a red flashing icon. For example, the icon of the faulty distribution box flashes strongly, and the fault propagation path is clearly depicted by red lines. For example, the path extending from the faulty distribution box along the cable to the connected production line equipment is clear at a glance. Next to the topology map, the fault type, location, cascading failure range and estimated time node, fault propagation path and possible affected areas are displayed in detail in a list. At the same time, the sound and light alarm device is activated, the high-brightness red LED indicator flashes at a frequency of 2 times per second, and the high-decibel speaker emits a continuous alarm sound with a volume of up to decibels, ensuring that it can attract the attention of the operation and maintenance personnel even in the noisy environment of the park. In addition, the early warning display module also pushes the early warning information to the operation and maintenance personnel's mobile phone in the form of text messages. The content of the text message is concise and clear, including the location of the faulty distribution box and a link to view detailed information. The operation and maintenance personnel can click the link to view the complete early warning information on their mobile phones, including topology display, detailed parameters, etc., so that they can respond quickly and take effective countermeasures in time to ensure the normal progress of production activities in the park and the stable operation of the distribution system.

[0053] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A power distribution equipment status monitoring system based on Internet of Things technology, characterized in that: The system includes the following components: sound collection module, data transmission module, data processing center and early warning display module; The sound collection module uses a high-sensitivity microphone and is equipped with an electromagnetic shielding cover to reduce electromagnetic interference. It uses a low-noise and high-gain preamplifier to amplify the collected weak sound signal, and converts it into digital audio data by a high-precision analog-to-digital converter according to the corresponding sampling accuracy and a sampling frequency that is not less than twice the highest frequency of the sound signal. At the same time, a noise reduction algorithm is used to process the sound signal, so as to ensure the quality of the collected sound signal; The data transmission module supports Wi-Fi, Bluetooth and 4G / 5G communication protocols, can automatically switch the communication mode according to the signal strength of the environment, and use the AES encryption algorithm to encrypt the transmitted data, combined with the CRC cyclic redundancy check to ensure data integrity, and use cache technology to ensure the continuity of data transmission when switching protocols, so as to achieve stable transmission of sound and various related data from the acquisition end to the data processing center; The data processing center includes a storage unit, a feature analysis unit, a fault diagnosis unit, a correlation analysis and prediction unit, and a preventive measures generation unit; The storage unit is used to store a large amount of sound sample data of power distribution equipment in normal operation and various known fault states, as well as processed fault voiceprint feature library data, association relationship data between various devices, cross-device association fault prediction model data, topological structure information of the power distribution system and electrical connection relationship model data; The feature analysis unit extracts and analyzes the voiceprint features of the sound samples using an expert system combined with feature engineering technology. The expert system builds a rule base based on the knowledge and experience of experts in the power field. Feature engineering processes the frequency, amplitude, harmonic structure and time-frequency domain characteristic information of the sound. By analyzing a large amount of sample data and matching rules, a mapping relationship between different fault states and unique voiceprint patterns is established; The fault diagnosis unit compares and matches the voiceprint features of the sound data of the power distribution equipment received in real time with the stored fault voiceprint feature library. If the match is successful, it is determined that the power distribution equipment has a corresponding fault and generates fault warning information; The association analysis and prediction unit obtains the operation data of adjacent or associated equipment through the Internet of Things network, including various operation parameters of the switchgear in the distribution box, the connected cable lines and the upper-level substation, namely, temperature, current and voltage, and uses the association analysis algorithm to analyze the inherent connection and mutual influence between the equipment operation data, establish a cross-equipment association fault prediction model, and predict the scope and time node of the chain fault that will be caused in advance according to the current operation status change trend of each device. At the same time, based on the topological structure and electrical connection relationship of the distribution system, combined with the equipment fault information, analyze the propagation path of the fault in the system and the affected area, and build a fault propagation prediction model; The preventive measures generating unit generates corresponding preventive measures plans for the possible chain failures and system-level failures according to the results of the correlation analysis and prediction unit; The warning display module receives fault warning information from the data processing center, and displays the fault type, location, scope of the caused chain failure and estimated time node, fault propagation path and affected area to the operation and maintenance personnel through sound and light alarms and display screen prompts.

2. According to claim 1, a power distribution equipment status monitoring system based on Internet of Things technology is characterized in that: The sound collection module uses a noise reduction algorithm to process the sound signal to ensure the quality of the collected sound signal. The algorithm formula is: Among them, s(t) is the original collected sound signal, w(t) is the wavelet coefficient after s(t) is transformed by wavelet, and λ is the threshold.

3. The power distribution equipment status monitoring system based on Internet of Things technology according to claim 1 is characterized in that: The data transmission module uses the AES encryption algorithm to encrypt the transmitted data. Specifically, the key length is determined and the key is generated by a secure random number generator. The key is stored in an area with strict access control and updated regularly. The block size is determined according to the requirements of the AES algorithm and the characteristics of the data. The data is sequentially divided into blocks. When the data is less than 16 bytes, a specific padding method is used to fill it. The AES encryption algorithm is initialized with the stored key and the encryption mode is selected. The ciphertext block is generated by using the round function transformation. The ciphertext blocks are sequentially integrated to form an encrypted data stream for transmission, thereby ensuring the confidentiality and security of data transmission.

4. The power distribution equipment status monitoring system based on Internet of Things technology according to claim 1 is characterized in that: The feature analysis unit of the data processing center uses an expert system combined with feature engineering technology to extract and analyze the voiceprint features of the sound sample. The feature extraction formula is: Where X represents the sound signal sequence, m is the subsequence length, r is the similarity tolerance, and N is the length of the sound signal sequence. is to satisfy the condition max 1≤j≤m |x i+j -x j The ratio of the number of subsequences with |≤r to the total number of subsequences.

5. The power distribution equipment status monitoring system based on Internet of Things technology according to claim 1 is characterized in that: The fault diagnosis unit of the data processing center compares and matches the voiceprint features of the power distribution equipment sound data received in real time with the stored fault voiceprint feature library, and the matching formula is: Among them, X is the feature vector of the sound data to be diagnosed, Y is a certain fault mode feature vector in the fault voiceprint feature library, μ Y is the mean of the eigenvectors of the fault mode Y, is its covariance matrix, It measures the statistical distance between the feature vector X and the fault mode Y in the feature space. DTW(X, Y) is the dynamic time warping distance, which is used to process the possible deformation or offset of the sound signal on the time axis. MDTW When (X, Y) is less than the set threshold δ, it is determined that the match is successful, that is, the device has a corresponding fault.

6. The power distribution equipment status monitoring system based on Internet of Things technology according to claim 1 is characterized in that: The correlation analysis and prediction unit of the data processing center uses a correlation analysis algorithm to analyze the inherent connection and mutual influence between the equipment operation data. The algorithm formula is: Among them, k i and k j are the degrees of device i and device j in the complex network model of the power distribution system, i.e., the number of devices connected to them, and k ij is the number of common neighbors between device i and device j, C ij Indicates the association degree between device i and device j.

7. The power distribution equipment status monitoring system based on Internet of Things technology according to claim 1 is characterized in that: The correlation analysis and prediction unit of the data processing center analyzes the propagation path of the fault in the system and the affected area based on the topological structure and electrical connection relationship of the power distribution system and the equipment fault information. The calculation formula is: Among them, x (0) (k) is the original equipment operating parameter sequence, x (1) (k) is the value of x (0) (k) A new sequence is generated by first-order accumulation, where a is the development grayscale and u is the endogenous control grayscale. This formula is used to make short-term predictions on the changing trends of the equipment operating parameters. Based on the prediction results, it is determined whether a fault will occur and the speed at which the fault will develop, and preventive measures can be taken in advance.

8. The power distribution equipment status monitoring system based on Internet of Things technology according to claim 1 is characterized in that: The correlation analysis and prediction unit of the data processing center analyzes the propagation path of the fault in the system and the affected area based on the topological structure and electrical connection relationship of the power distribution system and the equipment fault information, and constructs a fault propagation prediction model, the model formula of which is: in are the planned values ​​of injected active power and reactive power of node i, V i is the voltage amplitude at node i, θ ij is the voltage phase difference between node i and node j, Y ij It is an element in the node admittance matrix. By iteratively solving the above equations, the voltage amplitude and phase angle of each node in the system under normal operating conditions are obtained. When a device fails, the iterative calculation is performed again to compare the changes in voltage and current of each node before and after the failure, and the scope and degree of the impact of the failure on the system are analyzed to determine the fault propagation path and the area that may be affected.

9. A method for monitoring the state of a power distribution equipment based on the Internet of Things technology, the method being applicable to a method for monitoring the state of a power distribution equipment based on the Internet of Things technology as claimed in any one of claims 1 to 8, characterized in that: The method comprises the following specific steps: Sound sample collection: During the installation and commissioning phase of the power distribution equipment and the daily operation and maintenance process, the sound collection module collects sound samples of the power distribution equipment in normal operation and various known fault simulation states, marks and classifies the samples, and records the corresponding equipment operation status information; Feature library construction: The collected sound sample data is transmitted to the storage unit of the data processing center, and the sample data is processed by the expert system in the feature analysis unit combined with feature engineering technology. The feature information obtained after processing is constructed into a fault voiceprint feature library and stored in the storage unit; Correlation data collection and model construction: The data transmission module continuously collects correlation data between each distribution device, including temperature, current and voltage parameter information, as well as topological structure data and electrical connection relationship data of the distribution system, and transmits it to the storage unit of the data processing center. Based on these correlation data, the correlation analysis and prediction unit uses machine learning technology to analyze the correlation relationship between devices and build a cross-device correlation fault prediction model. At the same time, a fault propagation prediction model is built based on the topological structure and electrical connection relationship to determine the propagation rules and impact range of equipment faults in the system. For example, by analyzing a large amount of historical data, the propagation path probability matrix of different types of equipment faults under a specific system structure is determined; Real-time monitoring and diagnosis: During the normal operation of the power distribution equipment, the sound collection module continuously collects the equipment sound signals, which are transmitted to the data processing center in real time through the data transmission module. The fault diagnosis unit of the data processing center extracts the voiceprint features of the real-time sound data, and the feature analysis unit uses expert system rules and feature engineering technology to process and match it with the stored fault voiceprint feature library. If the match is successful, it is determined that the equipment has failed, and the early warning display module is immediately triggered to issue a fault warning message. If the match is not successful, the correlation analysis and prediction unit further analyzes the current equipment operation data and the operation data of the related equipment, and uses the cross-device correlation fault prediction model to predict whether there is a potential risk of chain failure. If the risk of chain failure is predicted, the early warning display module is also triggered to issue a warning containing information related to the chain failure. Preventive measures execution: The preventive measures generation unit generates a preventive measures plan based on the prediction results of the correlation analysis and prediction unit and executes it automatically or prompts the operation and maintenance personnel to execute it; Model update: Periodically or after a new fault type is discovered, repeat the sound sample collection steps and feature library construction steps to update and optimize the fault voiceprint feature library. At the same time, based on new equipment operation data, fault cases, and system structure change information, update the association rules and parameters in the cross-device association fault prediction model, as well as the propagation rules and probability matrix in the fault propagation prediction model to adapt to changes in the operating status of distribution equipment, the emergence of new fault conditions, and adjustments to the system structure.

Images