Partial discharge monitoring and early warning system and method based on deep learning

Abstract

The application discloses a partial discharge monitoring and early warning system and method based on deep learning, relates to the technical field of partial discharge monitoring, and discloses a partial discharge monitoring module, a discharge early warning step starting module and a partial discharge deep early warning module. The application not only effectively reduces the number of partial discharge monitoring sensors installed in electrical equipment, but also facilitates the maintenance convenience and maintenance cost of subsequent partial discharge monitoring sensors. Through low-cost deployment, multidimensional monitoring and intelligent analysis, the application solves the problems of high misjudgment rate, poor real-time performance and complex maintenance of traditional partial discharge monitoring, and provides an efficient solution for the whole life cycle management of electrical equipment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of partial discharge monitoring, more particularly, it relates to a partial discharge monitoring and early warning system and method based on deep learning. BACKGROUND

[0002] During the operation of electrical equipment, partial discharge is one of the important hidden dangers that leads to the degradation of equipment insulation performance and further causes faults. Partial discharge often starts from the failure of internal components of the equipment. With the passage of time, these partial discharges will gradually develop and expand, eventually causing serious faults such as insulation breakdown and short circuit of the equipment, affecting the safe and stable operation of the power system.

[0003] In order to monitor the partial discharge as comprehensively as possible, the traditional method often needs to install a large number of sensors of various types at each key position and different components of the electrical equipment. Through the discharge monitoring sensor corresponding to each component, the partial discharge of each component in the electrical equipment is monitored. Although this method can effectively monitor and locate the partial discharge, it will significantly increase the cost of purchasing sensors, and arranging numerous sensors in the limited space inside the electrical equipment will face the problems of difficult installation and subsequent maintenance inconvenience.

[0004] In summary, the present application proposes a partial discharge monitoring and early warning system and method based on deep learning, which utilizes the rational layout of multiple types of sensors, the fusion of multi-dimensional monitoring data, and intelligent analysis and processing means to achieve low-cost deployment, high-accuracy monitoring, real-time dynamic analysis, and accurate decision-making, effectively reducing the misjudgment rate of partial discharge monitoring, improving real-time performance, and simplifying the maintenance process. It provides a reliable and efficient solution for the whole life cycle management of electrical equipment, ensures the safe and stable operation of electrical equipment, and meets the increasing demand for equipment reliability of modern power systems. SUMMARY

[0005] In view of the deficiencies of the prior art, the present application aims to provide a partial discharge monitoring and early warning system and method based on deep learning.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0007] The partial discharge monitoring and early warning system based on deep learning comprises a partial discharge monitoring module, a discharge early warning step starting module, and a partial discharge deep early warning module.

[0008] The partial discharge monitoring module determines the electrical equipment to which the partial discharge monitoring is directed, installs various types of partial discharge monitoring sensors in the electrical equipment, and constructs a device partial discharge monitoring model for the electrical equipment.

[0009] The discharge early warning step starts a module that periodically determines the partial discharge image index of the electrical equipment, and determines whether to start the partial discharge early warning step based on the partial discharge image index.

[0010] The partial discharge deep early warning module, after starting the partial discharge early warning step, controls the equipment partial discharge monitoring model to complete multiple partial discharge monitoring of the same image simulation, and further determines the partial discharge early warning index of the electrical equipment, determines the partial discharge early warning level of the electrical equipment based on the partial discharge early warning index, and generates the partial discharge screening sequence of the components in the electrical equipment.

[0011] Further, the partial discharge image index of the electrical equipment is periodically determined as follows: obtaining the monitoring data features of each type of partial discharge monitoring sensor, combining the monitoring data features of each type of partial discharge monitoring sensor into a monitoring data feature set in the form of a feature set, and taking the monitoring data feature set as the input data of the partial discharge image deep learning model, and outputting the partial discharge image index of the electrical equipment.

[0012] Further, the monitoring data features of a type of partial discharge monitoring sensor are obtained as follows: selecting a partial discharge monitoring sensor, collecting all partial discharge monitoring data collected by the partial discharge monitoring sensor in the last partial discharge monitoring and early warning cycle, preprocessing and feature extraction of all collected partial discharge monitoring data, and obtaining the monitoring data features of the type of partial discharge monitoring sensor.

[0013] Further, the partial discharge screening sequence of the components in the electrical equipment is generated: all components are sorted in descending order according to the values of the comprehensive partial discharge index, and then the partial discharge screening sequence of the components is generated.

[0014] Further, the partial discharge early warning index of the electrical equipment is determined: the refined component partial discharge index of each component in the equipment partial discharge monitoring model after each partial discharge monitoring of the same image simulation is determined, the comprehensive partial discharge index of each component is obtained, the comprehensive partial discharge upper index and the comprehensive partial discharge lower index are set, when the comprehensive partial discharge index is greater than or equal to the comprehensive partial discharge upper index, the number of partial discharge unexpected components is increased by one, the total number of partial discharge unexpected components is marked as n, when the comprehensive partial discharge index is less than or equal to the comprehensive partial discharge lower index, the number of partial discharge stable components is increased by one, the total number of partial discharge stable components is marked as m, when the comprehensive partial discharge index is between the comprehensive partial discharge upper index and the comprehensive partial discharge lower index, the corresponding component is marked as an unexpected stable wandering component, and the partial discharge early warning index of the electrical equipment is determined based on the total number of partial discharge unexpected components, the total number of partial discharge stable components, and the total number of unexpected stable wandering components.

[0015] Further, the comprehensive partial discharge index of the component is obtained in the following specific process: selecting a component, obtaining the detailed component partial discharge index of the component in the equipment partial discharge monitoring model after the same visual simulation of each partial discharge monitoring, performing sum average calculation on all the detailed component partial discharge indexes, and calculating the average detailed component partial discharge index performing pairwise comparison on all the detailed component partial discharge indexes, performing absolute difference calculation on the two compared detailed component partial discharge indexes, calculating the component partial discharge swing index, performing sum average calculation on all the component partial discharge swing indexes, and calculating the average component partial discharge swing index , the comprehensive partial discharge index of the component is calculated wherein x1 is the first coefficient, and x2 is the second coefficient.

[0016] Further, the same visual simulation of one-time partial discharge monitoring of the equipment partial discharge monitoring model is completed in the following specific process: randomly adjusting the key operating parameters of the components in the equipment partial discharge monitoring model, adjusting the equipment partial discharge monitoring model to perform simulation operation for one partial discharge monitoring early warning period after the adjustment, collecting the real-time partial discharge monitoring data of each virtual monitoring point in the simulation process, obtaining the monitoring data features of each virtual monitoring point after the simulation, and then determining the monitoring approximation index of each virtual monitoring point, performing sum average calculation on the monitoring approximation indexes of all the virtual monitoring points, calculating the average monitoring approximation index, setting the average monitoring approximation threshold index, and determining that the equipment partial discharge monitoring model completes the same visual simulation of one-time partial discharge monitoring when the average monitoring approximation index is greater than or equal to the average monitoring approximation threshold index.

[0017] Further, the monitoring approximation index of the virtual monitoring point is determined in the following specific process: selecting a virtual monitoring point, obtaining the monitoring data features of the partial discharge monitoring sensor corresponding to the virtual monitoring point, combining the monitoring data features of the virtual monitoring point and the monitoring data features of the partial discharge monitoring sensor into a monitoring feature comparison set, determining the type of the partial discharge monitoring sensor, obtaining the monitoring feature comparison model of the partial discharge monitoring sensor of the type, inputting the monitoring feature comparison set into the monitoring feature comparison model, and outputting the monitoring approximation index of the virtual monitoring point from the monitoring feature comparison model.

[0018] ​Further, the refined part partial discharge index of each part in the equipment partial discharge monitoring model after one-time partial discharge monitoring same image simulation is determined, and the specific determination process is as follows: after the equipment partial discharge monitoring model completes one-time partial discharge monitoring same image simulation, the operation parameter set of each part is obtained, the part partial discharge deep learning model corresponding to each part is obtained, the operation parameter set of each part is taken as the input data of the corresponding part partial discharge deep learning model, and the refined part partial discharge index of each part is output.

[0019] The operation parameter set of the part is obtained in the following specific process: one part in the equipment partial discharge monitoring model is selected, all key operation parameters of the part in the simulation operation process are obtained, and all the key operation parameters are combined into the operation parameter set in the form of a data set.

[0020] Further, the deep learning-based partial discharge monitoring and early warning method comprises the following steps:

[0021] Step one: determine the electrical equipment to which the partial discharge monitoring is directed, and install various types of partial discharge monitoring sensors in the electrical equipment;

[0022] Step two: build an equipment partial discharge monitoring model for the electrical equipment;

[0023] Step three: periodically determine the partial discharge image index of the electrical equipment, and determine whether to start the partial discharge early warning step based on the partial discharge image index;

[0024] Step four: after starting the partial discharge early warning step, control the equipment partial discharge monitoring model to complete multiple partial discharge monitoring same image simulations, and then determine the partial discharge early warning index of the electrical equipment;

[0025] Step five: determine the partial discharge early warning level of the electrical equipment based on the partial discharge early warning index, and generate the partial discharge screening sequence of the parts in the electrical equipment.

[0026] Compared with the prior art, the present application has the following beneficial effects:

[0027] The system of the present application sets up a partial discharge monitoring module, a discharge early warning step starting module and a partial discharge deep early warning module, installs various types of partial discharge monitoring sensors in the electrical equipment, carries out multi-dimensional partial discharge monitoring on the electrical equipment, quickly locks whether there is a partial discharge hidden danger according to the monitoring data, judges the possible existence of each component in the electrical equipment through multiple partial discharge monitoring of the same apparent image simulation when it is determined that the electrical equipment has a partial discharge hidden danger, and then comprehensively judges and analyzes the partial discharge condition of the entire electrical equipment, accurately determines the current partial discharge complexity and criticality of the electrical equipment, and formulates a partial discharge checking sequence for the components in the electrical equipment, which not only effectively reduces the installation number of partial discharge monitoring sensors in the electrical equipment, but also facilitates the maintenance convenience and maintenance cost of the subsequent partial discharge monitoring sensors. The method of the present application solves the problems of high misjudgment rate, poor real-time performance and complex maintenance of traditional partial discharge monitoring through low-cost deployment, multi-dimensional monitoring and intelligent analysis, and provides an efficient solution for the whole life cycle management of electrical equipment. BRIEF DESCRIPTION OF DRAWINGS

[0028] Figure 1 The system of the present application is a schematic diagram of the principle;

[0029] Figure 2 The device partial discharge monitoring model of the present application is a flow chart;

[0030] Figure 3 The flow chart of the method of the present application. DETAILED DESCRIPTION

[0031] Embodiment one: refer to Figures 1 to 2 The partial discharge monitoring and early warning system based on deep learning includes a partial discharge monitoring module, a discharge early warning step starting module and a partial discharge deep early warning module.

[0032] The partial discharge monitoring module determines the electrical equipment to which the partial discharge monitoring is directed, installs various types of partial discharge monitoring sensors (the types of partial discharge monitoring sensors include ultrahigh frequency sensors, pulse current sensors, ultrasonic sensors, etc., different types of partial discharge monitoring sensors are used to monitor different types of partial discharge monitoring data, the installation number of each type of partial discharge monitoring sensor is one, which reduces the installation number of partial discharge monitoring sensors, i.e. installation cost, without the need to install multiple types of partial discharge monitoring sensors for each component, avoiding the difficulty of installing multiple partial discharge monitoring sensors due to the small internal structure of the electrical equipment, and facilitating the subsequent maintenance of the partial discharge monitoring sensors) in the electrical equipment, and constructs a device partial discharge monitoring model of the electrical equipment.

[0033] The device partial discharge monitoring model about the electrical equipment is constructed: according to the actual drawing of the electrical equipment, a three-dimensional geometric model of the electrical equipment is constructed in a general finite element analysis software, structural features that may cause partial discharge are reserved, such as conductor edges, insulation air gaps, cable terminal heads, insulator surface defects and the like, corresponding material parameters are given to each component, according to the type and installation position of the partial discharge monitoring sensor, corresponding virtual monitoring points are added to the three-dimensional geometric model of the electrical equipment, the virtual monitoring points can collect the partial discharge monitoring data of the corresponding type in real time during simulation, and finally, the device partial discharge monitoring model about the electrical equipment is constructed.

[0034] The discharge early warning step starting module sets a partial discharge monitoring early warning period (the length of the period is set and adjusted according to requirements), determines the partial discharge image index of the electrical equipment based on the partial discharge monitoring early warning period, sets a partial discharge image threshold index (the partial discharge image threshold index is a preset index, which is used for comparison with the partial discharge image index), and starts the partial discharge early warning step when the partial discharge image index of the electrical equipment is greater than or equal to the partial discharge image threshold index.

[0035] The partial discharge image index of the electrical equipment is determined periodically as follows: the monitoring data features of each type of partial discharge monitoring sensor are obtained, the monitoring data features of each type of partial discharge monitoring sensor are combined into a monitoring data feature set in the form of a feature set, the monitoring data feature set is used as input data of the partial discharge image deep learning model, and the partial discharge image index of the electrical equipment is obtained.

[0036] The monitoring data features of one type of partial discharge monitoring sensor are obtained as follows: a partial discharge monitoring sensor is selected, all partial discharge monitoring data collected by the partial discharge monitoring sensor in the last partial discharge monitoring early warning period are collected, the collected all partial discharge monitoring data are preprocessed and feature extracted (the preprocessing methods include data cleaning, data normalization and the like, and the feature extraction methods include time domain feature extraction, frequency domain feature extraction and the like), and the monitoring data features of the partial discharge monitoring sensor of the type are obtained.

[0037] The partial discharge image deep learning model is constructed based on a deep learning model, and the specific construction process is: constructing a deep learning model, collecting a plurality of monitoring data feature sets, training the deep learning model with the monitoring data feature sets, assigning a partial discharge image index to each monitoring data feature set, the value range of the partial discharge image index is (0.1~15.0), the larger the partial discharge image index, the greater the possibility of partial discharge in the electrical equipment, dividing the training data into a training set, a validation set and a test set, the proportion is 70%:15%:15%, training the training set, the validation set and the validation set, training is completed, and the partial discharge image deep learning model is constructed.

[0038] The partial discharge deep early warning module, when starting the partial discharge early warning step, controls the equipment partial discharge monitoring model to complete a plurality of partial discharge monitoring same image simulations (the components, key operation parameters and adjustment amplitudes of the key operation parameters of each partial discharge monitoring same image simulation are different), and then determines the partial discharge early warning index of the electrical equipment, determines the partial discharge early warning level of the electrical equipment based on the partial discharge early warning index, and generates the partial discharge screening sequence of the components in the electrical equipment.

[0039] The control equipment partial discharge monitoring model completes the same image simulation of partial discharge monitoring once. Specifically, the key operating parameters of the components in the equipment partial discharge monitoring model are randomly adjusted (randomly means that one key operating parameter of one component can be adjusted, or multiple key operating parameters of multiple components can be adjusted, such as adjusting the contact resistance of the bus connection in the power equipment to 60 μΩ, or adjusting the pulse current amplitude of the bus connection in the power equipment to 120 mA, adjusting the repetition frequency to 11 Hz, and adjusting the effective value of the ultrasonic wave of the cable terminal head to 25 dB, and adjusting the main frequency to 110 kHz). After the adjustment is completed, the control equipment partial discharge monitoring model is simulated for a partial discharge monitoring warning period. During the simulation, each virtual monitoring point collects real-time partial discharge monitoring data of the corresponding type. After the simulation is completed, the monitoring data characteristics of each virtual monitoring point are obtained (the monitoring data characteristics of the virtual monitoring point are obtained in the same way as the monitoring data characteristics of the corresponding type of partial discharge monitoring sensor), and the monitoring approximation index of each virtual monitoring point is determined. The monitoring approximation indexes of the virtual monitoring points are summed and averaged to calculate the average monitoring approximation index. The average monitoring approximation threshold index (the average monitoring approximation threshold index is a preset index, which is used for comparison with the average monitoring approximation index) is set. When the average monitoring approximation index is greater than or equal to the average monitoring approximation threshold index, it indicates that the equipment partial discharge monitoring model completes the same image simulation of partial discharge monitoring once (when the average monitoring approximation index is less than the average monitoring approximation threshold index, it indicates that the same image simulation of partial discharge monitoring once is not completed).

[0040] The monitoring approximation index of the virtual monitoring point is determined as follows: select a virtual monitoring point, obtain the monitoring data characteristics of the corresponding partial discharge monitoring sensor of the virtual monitoring point, combine the monitoring data characteristics of the virtual monitoring point and the monitoring data characteristics of the partial discharge monitoring sensor into a monitoring feature comparison set, determine the type of the partial discharge monitoring sensor, obtain the monitoring feature comparison model of the partial discharge monitoring sensor of the same type, and input the monitoring feature comparison set into the monitoring feature comparison model. The monitoring feature comparison model outputs the monitoring approximation index of the virtual monitoring point.

[0041] Each type of partial discharge monitoring sensor corresponds to a monitoring feature comparison model, and each monitoring feature comparison model is constructed based on a neural network model. In this embodiment, taking the ultra-high frequency sensor as an example, the monitoring feature comparison model for the ultra-high frequency sensor is constructed as follows: a neural network model is constructed, a plurality of monitoring feature comparison sets about the ultra-high frequency sensor are collected (if a monitoring feature comparison model for the pulse current sensor is constructed, a plurality of monitoring feature comparison sets about the pulse current sensor are collected), each monitoring feature comparison set contains a monitoring data feature of an ultra-high frequency sensor and a monitoring data feature of a corresponding virtual monitoring point, the neural network model is trained by using the monitoring feature comparison set of the ultra-high frequency sensor, a monitoring approximation index is given to each monitoring feature comparison set of the ultra-high frequency sensor, the value range of the monitoring approximation index is (1.0~5.0), the greater the monitoring approximation index, the more similar the two monitoring data features in the monitoring feature comparison set, the training data is divided into a training set, a validation set and a test set, the proportion is 60%:20%:20%, the neural network is iteratively trained on the training set, the validation set and the validation set, and finally, the monitoring feature comparison model of the ultra-high frequency sensor is constructed.

[0042] After one-time partial discharge monitoring of the same simulation of the equipment, the refined component partial discharge index of each component in the equipment partial discharge monitoring model is obtained, and the specific determination process is as follows: after one-time partial discharge monitoring of the same simulation of the equipment, the operating parameter set of each component is obtained, the component partial discharge deep learning model corresponding to each component is obtained, the operating parameter set of each component is taken as the input data of the corresponding component partial discharge deep learning model, and the refined component partial discharge index of each component is obtained by output.

[0043] The operating parameter set of the component is obtained as follows: select a component in the equipment partial discharge monitoring model, obtain all key operating parameters of the component during simulation operation (different components need to obtain different key operating parameters, for example, the key operating parameters of the bus connection include pulse current, contact resistance, UHF signal, contact resistance, etc., and the key operating parameters of the cable terminal include ultrasonic wave, infrared temperature rise, insulation resistance, UHF signal, etc.), and all the key operating parameters are combined into an operating parameter set in the form of a data set.

[0044] Each component corresponds to a component partial discharge deep learning model, and each component partial discharge deep learning model is constructed based on a neural network model. In this embodiment, the bus connection is taken as an example to construct a component partial discharge deep learning model about the bus connection: a neural network model is constructed, and a plurality of operating parameter sets of the bus connection are collected (if a component partial discharge deep learning model of a cable terminal head is constructed, a plurality of operating parameter sets of the cable terminal head are collected), the operating parameter sets of the bus connection are used to train the neural network model, and a refined component partial discharge index is given to each operating parameter set of the bus connection. The value range of the refined component partial discharge index is (10.0-20.0), and the greater the refined component partial discharge index, the greater the possibility of partial discharge of the bus connection (if a component partial discharge deep learning model of a cable terminal head is constructed, the greater the refined component partial discharge index, the greater the possibility of partial discharge of the cable terminal head). The training data is divided into a training set, a validation set and a test set, and the proportion is 60%:25%:15%. The training set, the validation set and the validation set are iteratively trained, and finally, the component partial discharge deep learning model of the bus connection is constructed.

[0045] The partial discharge screening sequence of the components in the electrical equipment is generated: all the components are sequentially sorted according to the values of the comprehensive partial discharge indexes from large to small, and then the partial discharge screening sequence of the components is generated (the components can be sequentially screened for partial discharge according to the sorting).

[0046] The comprehensive partial discharge index of the component is obtained as follows: select a component, obtain the refined component partial discharge index of the component in the equipment partial discharge monitoring model after each partial discharge monitoring and the same image simulation, sum all the refined component partial discharge indexes, and calculate the average to obtain the average refined component partial discharge index , all the refined component partial discharge indexes are compared with each other, the absolute difference of the two compared refined component partial discharge indexes is calculated to obtain the component partial discharge swing index, and the average component partial discharge swing index is calculated by summing all the component partial discharge swing indexes , the comprehensive partial discharge index of the component is calculated by , wherein x1 is the first coefficient, x2 is the second coefficient, the value of x1 is 0.39, and the value of x2 is 1.14.

[0047] ​determine the refined part partial discharge index of each component in the partial discharge monitoring model of the equipment after the same partial discharge monitoring simulation each time, obtain the comprehensive partial discharge index of each component, set the upper comprehensive partial discharge index and the lower comprehensive partial discharge index (the upper comprehensive partial discharge index is greater than the lower comprehensive partial discharge index, and both the upper comprehensive partial discharge index and the lower comprehensive partial discharge index are preset indexes, which are used for comparison with the comprehensive partial discharge index), when the comprehensive partial discharge index is greater than or equal to the upper comprehensive partial discharge index, the number of the partial discharge unexpected components is increased by one, and the total number of the partial discharge unexpected components is marked as Lwbte, when the comprehensive partial discharge index is less than or equal to the lower comprehensive partial discharge index, the number of the partial discharge stable components is increased by one, and the total number of the partial discharge stable components is marked as Pedsz, when the comprehensive partial discharge index is between the upper comprehensive partial discharge index and the lower comprehensive partial discharge index, the corresponding component is marked as an unexpected stable wandering component, and the total number of the unexpected stable wandering components is marked as Kert, and the partial discharge early warning index of the electrical equipment is calculated by . .

[0048] determine the partial discharge early warning level of the electrical equipment based on the partial discharge early warning index: obtain the partial discharge early warning index of the electrical equipment , set the range of each partial discharge early warning index corresponding to one partial discharge early warning level, the range of the partial discharge early warning index includes [0, 1], (( 1, 2], …, (( B-1, B], the partial discharge early warning level includes the partial discharge early warning level 1, the partial discharge early warning level 2, …, the partial discharge early warning level B-1, and the partial discharge early warning level B, the partial discharge risk of the partial discharge early warning level 1 is lower than that of the partial discharge early warning level 2, and so on.

[0049] set the partial discharge monitoring module, the partial discharge early warning step starting module, and the partial discharge deep early warning module, perform multi-dimensional partial discharge monitoring on the electrical equipment by installing various types of partial discharge monitoring sensors in the electrical equipment, quickly lock whether there is a partial discharge hidden danger according to the monitoring data, when it is determined that the power equipment has a partial discharge hidden danger, judge the possible situation of each component in the power equipment by multiple partial discharge monitoring of the same apparent image simulation, and then comprehensively determine and analyze the partial discharge situation of the entire power equipment, accurately determine the current partial discharge complexity and criticality of the power equipment, and develop a partial discharge checking sequence for the components in the power equipment, which not only effectively reduces the number of partial discharge monitoring sensors installed in the electrical equipment, but also facilitates the maintenance convenience and maintenance cost of the subsequent partial discharge monitoring sensors.

[0050] Embodiment two: reference Figure 3 , the deep learning-based partial discharge monitoring and early warning method, the steps are as follows:

[0051] Step one: determine the electrical equipment to which the partial discharge monitoring is directed, and install various types of partial discharge monitoring sensors in the electrical equipment;

[0052] Step two: build a device partial discharge monitoring model for the electrical equipment;

[0053] Step three: periodically determine the partial discharge image index of the electrical equipment, and determine whether to start the partial discharge early warning step based on the partial discharge image index;

[0054] Step four: after starting the partial discharge early warning step, control the device partial discharge monitoring model to complete multiple partial discharge monitoring of the same image simulation, and then determine the partial discharge early warning index of the electrical equipment;

[0055] Step five: determine the partial discharge early warning level of the electrical equipment based on the partial discharge early warning index, and generate the partial discharge screening sequence of the components in the electrical equipment.

[0056] The above method solves the problems of high misjudgment rate, poor real-time performance, and complex maintenance of traditional partial discharge monitoring through low-cost deployment, multi-dimensional monitoring, and intelligent analysis, and provides an efficient solution for the whole life cycle management of electrical equipment.

[0057] The above formulas are dimensionless values, and the formulas are obtained by collecting a large amount of data to simulate the latest real situation, and the preset parameters in the formula are set by a person skilled in the art according to the actual situation.

[0058] The above-described embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, the above-described embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are wholly or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center through a wired or wireless (such as infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. containing one or more available medium collections. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state disk.

[0059] It should be understood that the size of the sequence number of each process described above in various embodiments of the present application does not mean the order of execution, and the execution order of each process should be determined by its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0060] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be realized in electronic hardware, or a combination of computer software and electronic hardware. Whether the functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0061] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described system, device and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be described here.

[0062] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other manners. For example, the division of the above-described device embodiments is only a logical function division, and there can be another division manner for actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different units, or the among different units, can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0063] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, and the computer software product is stored in a storage medium, and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, and various other media that can store program codes.

[0064] The above describes only the specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. The partial discharge monitoring and early warning system based on deep learning is characterized by: It includes partial discharge monitoring module, discharge warning step start module, and partial discharge depth warning module; The partial discharge monitoring module determines the electrical equipment to be monitored for partial discharge, installs various types of partial discharge monitoring sensors in the electrical equipment, and constructs an equipment partial discharge monitoring model for the electrical equipment; The discharge warning step starting module regularly determines the partial discharge image index of the electrical equipment and determines whether to start the partial discharge warning step based on the partial discharge image index; The partial discharge visibility index of electrical equipment is regularly determined as follows: monitoring data features of various types of partial discharge monitoring sensors are obtained, the monitoring data features of various types of partial discharge monitoring sensors are combined into a monitoring data feature set in the form of a feature set, the monitoring data feature set is used as input data of a partial discharge visibility deep learning model, and the partial discharge visibility index of the electrical equipment is obtained as output; The partial discharge depth warning module, after initiating the partial discharge warning step, controls the equipment partial discharge monitoring model to complete multiple partial discharge monitoring simulations with the same image, thereby determining a partial discharge warning index for the electrical equipment, determining a partial discharge warning level for the electrical equipment based on the partial discharge warning index, and generating a partial discharge screening order for components in the electrical equipment; Determining a partial discharge warning index for electrical equipment: determining a detailed component partial discharge index for each component in a partial discharge monitoring model of the equipment after the same display simulation for each partial discharge monitoring, obtaining a comprehensive partial discharge index for each component, setting a comprehensive partial discharge upper index and a comprehensive partial discharge lower index; when the comprehensive partial discharge index is greater than or equal to the comprehensive partial discharge upper index, increasing the number of partial discharge accidental components by one, and marking the total number of partial discharge accidental components as ; when the comprehensive partial discharge index is less than or equal to the comprehensive partial discharge lower index, increasing the number of partial discharge stable components by one, and marking the total number of partial discharge stable components as ; when the comprehensive partial discharge index is between the comprehensive partial discharge upper index and the comprehensive partial discharge lower index, marking the corresponding component as an accidental stable wandering component; and determining a partial discharge warning index for the electrical equipment based on the total number of partial discharge accidental components, the total number of partial discharge stable components, and the total number of accidental stable wandering components; The comprehensive partial discharge index of a component is specifically obtained by the following process: selecting a component, obtaining the detailed component partial discharge index of the component in the equipment partial discharge monitoring model after the same display simulation of each partial discharge monitoring, calculating the average of all detailed component partial discharge indices to obtain an average detailed component partial discharge index, comparing all detailed component partial discharge indices pairwise, calculating the absolute difference between the two compared detailed component partial discharge indices to obtain a component partial discharge swing index, calculating the average of all component partial discharge swing indices to obtain an average component partial discharge swing index, and determining the comprehensive partial discharge index of the component based on the average detailed component partial discharge index and the average component partial discharge swing index; Controlling the equipment partial discharge monitoring model to complete a partial discharge monitoring same image simulation, specifically: randomly adjusting the key operating parameters of the components in the equipment partial discharge monitoring model, after the adjustment is completed, controlling the equipment partial discharge monitoring model to perform a simulation operation of a partial discharge monitoring early warning cycle, during the simulation process, each virtual monitoring point collects the corresponding type of partial discharge monitoring data in real time, after the simulation is completed, obtaining the monitoring data characteristics of each virtual monitoring point, and then determining the monitoring approximation index of each virtual monitoring point, summing and averaging the monitoring approximation indexes of each virtual monitoring point to obtain an average monitoring approximation index, setting an average monitoring approximation threshold index, and when the average monitoring approximation index is greater than or equal to the average monitoring approximation threshold index, it indicates that the equipment partial discharge monitoring model has completed a partial discharge monitoring same image simulation; The monitoring approximation index of the virtual monitoring point is determined as follows: select a virtual monitoring point, obtain the monitoring data characteristics of the local discharge monitoring sensor corresponding to the virtual monitoring point, combine the monitoring data characteristics of the virtual monitoring point with the monitoring data characteristics of the local discharge monitoring sensor into a monitoring feature comparison set, determine the type of local discharge monitoring sensor, obtain the monitoring feature comparison model of this type of local discharge monitoring sensor, use the monitoring feature comparison set as input data of the monitoring feature comparison model, and the monitoring feature comparison model outputs the monitoring approximation index of the virtual monitoring point.

2. The partial discharge monitoring and early warning system based on deep learning according to claim 1 is characterized in that: The monitoring data characteristics of a type of partial discharge monitoring sensor are specifically obtained through the following process: selecting a partial discharge monitoring sensor, collecting all partial discharge monitoring data collected by the partial discharge monitoring sensor in the previous partial discharge monitoring and early warning cycle, preprocessing and feature extraction of all collected partial discharge monitoring data, and obtaining the monitoring data characteristics of the type of partial discharge monitoring sensor.

3. The partial discharge monitoring and early warning system based on deep learning according to claim 1 is characterized in that: Generate a partial discharge screening order for components in electrical equipment: sort all components in descending order according to the values ​​of the comprehensive partial discharge index, and then generate a partial discharge screening order for the components.

4. The partial discharge monitoring and early warning system based on deep learning according to claim 1 is characterized in that: The detailed component partial discharge index of each component in the equipment partial discharge monitoring model after a partial discharge monitoring simulation with the same image is determined as follows: after the equipment partial discharge monitoring model completes a partial discharge monitoring simulation with the same image, an operating parameter set of each component is obtained, and a component partial discharge deep learning model corresponding to each component is obtained. The operating parameter set of each component is used as input data of the corresponding component partial discharge deep learning model, and the refined component partial discharge index of each component is obtained as output; The specific acquisition process of the component operating parameter set is as follows: select a component in the equipment partial discharge monitoring model, obtain all key operating parameters of the component during the simulated operation process, and combine all key operating parameters into an operating parameter set in the form of a data set.

5. A partial discharge monitoring and early warning method based on deep learning, applied to a partial discharge monitoring and early warning system based on deep learning according to any one of claims 1 to 4, characterized in that: Here are the steps: Step 1: Determine the electrical equipment for partial discharge monitoring and install various types of partial discharge monitoring sensors in the electrical equipment; Step 2: Construct a partial discharge monitoring model for electrical equipment; Step 3: Regularly determine the partial discharge image index of the electrical equipment and determine whether to initiate the partial discharge warning step based on the partial discharge image index; Step 4: After the partial discharge warning step is initiated, the partial discharge monitoring model of the control equipment completes multiple partial discharge monitoring simulations with the same image, thereby determining the partial discharge warning index of the electrical equipment; Step 5: Determine the partial discharge warning level of the electrical equipment based on the partial discharge warning index, and generate a partial discharge screening order for components in the electrical equipment.

Images