Transformer global linkage digital monitoring safety protection system

Abstract

The invention discloses a transformer global linkage digital monitoring safety protection system which comprises a sensing layer, a data layer, an edge layer and an application layer. The sensing layer deploys a sensing device to comprehensively sense and monitor the operation state of the transformer; the data layer comprises aggregation node equipment and access node equipment and is used for realizing integration and specification of monitoring data; for the same type of data, the aggregation node measures the uncertainty contained in the data by adopting information entropy, and endows the data with a data weight according to the information reliability, so as to realize data fusion; and the edge layer is used for realizing the state differentiation defect diagnosis and evaluation of the transformer by embedding the trained fault diagnosis model into an edge calculation module, and the edge calculation module directly acts on a protection device to protect the safety of the protection system. According to the invention, a global linkage digital monitoring technology is adopted to analyze and determine the fault abnormity of the transformer, and a transformer diagnosis state evaluation index is given.

Description

Technical Field

[0001] The invention belongs to the technical field of transformer monitoring, and relates to a transformer full-area linkage digital monitoring safety protection system. Background Art

[0002] Due to the complex structure and operating environment of the transformer, there are problems such as surface dirt, moisture, insulation aging, etc. in the long-term operation process, which are easy to cause partial discharge, overheating and other faults. Therefore, it is necessary to establish a transformer status monitoring system to timely and accurately grasp the operating status of the transformer and provide a basis for the formulation of transformer maintenance decision-making plans.

[0003] The transformer condition monitoring system is a system that integrates transformer signal detection, signal analysis, and condition diagnosis. Existing technologies mainly monitor parameters such as dissolved gas in oil, online power, temperature, partial discharge, winding impedance, dielectric loss, etc., and use dissolved gas analysis technology in oil, fuzzy logic, expert system, artificial neural network and other technologies to diagnose transformer faults such as transformer insulation aging, oil gap breakdown discharge, winding deformation and looseness, on-load tap changer abnormalities, and bushing performance deterioration, evaluate the transformer status, and provide a basis for the formulation of its operation and maintenance strategy.

[0004] At present, a few manufacturers in China have realized the research of mobile intensive care system for main transformers, but most of them are limited to a few single functions such as online DGA of the main body and ultrasonic partial discharge, and have not realized the full-domain online monitoring function of the main transformer, and cannot realize the full-domain state quantity monitoring of the main transformer and give state decisions. At present, the domestic online DGA device can only reflect the multi-component dissolved gas in the oil, and has not realized the monitoring of oil micro-water and withstand voltage value; the online monitoring function of dissolved gas in the oil of the on-load voltage regulator has not been realized at home and abroad, and the key characteristic gas content such as acetylene in the vacuum switch or the oil micro-water and withstand voltage value monitoring in the oil switch cannot be realized, and the operating status of the on-load voltage regulator cannot be perceived in time. Summary of the invention

[0005] In view of the current working environment, development needs and insufficient operation and maintenance methods of heavy-load transformers, the present invention proposes a transformer full-domain linkage digital monitoring safety protection system, which adopts intelligent, miniature and diversified transformer full-domain linkage digital monitoring technology to monitor the vibration, partial discharge, hot spot imaging, power and other signals and working status of the transformer body and its affiliated on-load tap changers, bushings and other equipment. By using multi-state differentiated defect diagnosis and processing technologies such as monitoring signal change trends, event data, fault alarms, etc., the transformer fault anomaly is finally analyzed and determined, and the transformer diagnostic status evaluation index is given.

[0006] The present invention is implemented by the following technical solutions: A transformer global linkage digital monitoring safety protection system, including a perception layer, a data layer, an edge layer and an application layer;

[0007] The sensing layer deploys sensing devices to realize comprehensive sensing monitoring of the transformer operation status and obtain a transformer monitoring data set. The sensing devices include a multi-parameter sensor of gas in oil, oil temperature, oil pressure and trace water, a multi-parameter sensor of dielectric loss, capacitance and partial discharge, and an online vibration monitoring device.

[0008] The data layer includes aggregation nodes and access node devices to achieve the integration and regulation of monitoring data;

[0009] The aggregation node is used to receive and aggregate sensor device information, and then transmit the sensor device data to the access node; after realizing the standardized and intelligent access of multiple sensor devices, the access node sends the agreed data to the edge computing module;

[0010] For data of the same type, the aggregation node uses information entropy to measure the uncertainty contained in the data, assigns data weights according to their information reliability, and realizes data fusion; for data of different types, the principal component analysis method is used to reduce the data dimension through linear transformation to realize the fusion of heterogeneous data, and the fused data is used as the input data of the fault diagnosis model;

[0011] The edge layer embeds the trained fault diagnosis model into the edge computing module to realize the differentiated defect diagnosis and evaluation of the transformer status, and transmits the results to the visual monitoring platform. At the same time, when a serious fault occurs in the transformer, the edge computing module directly acts on the protection device to protect the safety of the protection system.

[0012] The application layer contains all the functional components of the system, including online monitoring module, data list module, fault diagnosis module and fault alarm module, and is displayed in the form of pages on the visual monitoring platform.

[0013] Further preferably, the multi-parameter sensor of gas in oil-oil temperature-oil pressure-trace water integrates the gas in oil, oil temperature, oil pressure and trace water sensing units on the sensor conditioning board, and the gas in oil sensing unit accurately converts the gas in oil and the resistance value through the signal processing circuit to characterize the gas in oil; the pressure sensing unit includes a pressure sensitive element and a signal conditioning circuit, the pressure sensitive element measures the pressure, and the pressure signal of the pressure sensitive element is zero-compensated and temperature-compensated through the signal conditioning circuit. The conversion relationship between pressure ε and resistance R is as follows:

[0014]

[0015] Among them, η is the piezoresistance coefficient, E is the elastic modulus, ε is the Poisson coefficient, and K is the sensitivity coefficient;

[0016] Further preferably, the water content in the oil is measured by a capacitive sensor method, and the relative dielectric constant when the oil contains water content is calculated according to the Maxwell-Garnetl theory:

[0017]

[0018] Among them, ε m is the relative dielectric constant of water in oil, ε w is the relative dielectric constant of water, ε is the relative dielectric constant of oil, and α is the water mass fraction. When the micro-water sensor unit senses that its dielectric constant changes, the capacitance also changes accordingly, and the water content in the oil is calculated based on the change in capacitance.

[0019] Further preferably, the dielectric loss-capacitance-partial discharge multi-parameter sensor integrates the high and low frequency signal coils, and the coils are sleeved on an extended grounding rod. One end of the grounding rod is connected to the bushing end screen lead wire in the form of crown spring crimping or spring top connection, and the other end is connected to the grounding end of the outer casing.

[0020] Further preferably, the fault diagnosis model uses the SMOTE algorithm to expand the minority class samples in the acquired transformer data set, reduces the imbalance of the original data set, constructs a CNN model based on the balanced data set, and optimizes the structure and hyperparameters of the CNN model through the SSA algorithm, and trains it to obtain an optimized SSA-CNN model for transformer fault diagnosis.

[0021] Further preferably, the online monitoring module monitors the vibration, partial discharge, hot spot imaging, power and other signals and working status of the transformer body and its attached on-load tap changers, bushings and other equipment in real time;

[0022] Further preferably, the data list module is used to display the change trend of the monitoring signal, event data, etc. in the form of a list;

[0023] Further preferably, the fault diagnosis module utilizes the equipment operation status monitoring information to realize the differentiated defect diagnosis and evaluation of the transformer status, analyzes and determines the fault anomaly of the transformer and evaluates its status;

[0024] Further preferably, after determining typical defects, abnormal noises, and fault information of the transformer, the fault alarm module promptly issues alarm information and formulates an operation and maintenance strategy.

[0025] Beneficial effects of the present invention:

[0026] (1) Use multiple sensors to monitor the transformer load, oil temperature, online DGA, load switch vibration and other equipment operating conditions, break through the limitation of the single monitoring parameter of the main transformer mobile intensive care system, and realize the full-area online monitoring of the main transformer. It can monitor and diagnose the transformer and its key equipment online, comprehensively and intelligently monitor the full-cycle operation status of the transformer, improve the operation reliability of the main transformer, reduce the operation and maintenance costs, extend the life of the transformer, improve the asset utilization rate of the transformer, timely warn and respond to power grid emergencies, improve the overall safe operation level of substation equipment, prevent the risks of power grid stability and power supply load loss caused by equipment failure, and greatly improve the power supply reliability; and effectively reduce the labor intensity of operation and maintenance inspection personnel, reduce the heavy work under the situation of rapid expansion of the power grid scale and the safety hazards of staff inspection under complex on-site conditions.

[0027] (2) Comprehensive online detection of transformer equipment operating status information is carried out through multi-sensor devices such as oil gas-oil temperature-oil pressure-micro-water multi-parameter sensors, dielectric loss-capacitance-partial discharge multi-parameter sensors, and online vibration monitoring devices. For data of the same type, information entropy is used to measure the uncertainty contained in the data, and data weights are assigned according to their information reliability to achieve data fusion. For data of different types, principal component analysis is used to reduce data dimensionality through linear transformation, simplify complex data, and achieve fusion of heterogeneous data. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a system schematic diagram of the present invention. DETAILED DESCRIPTION

[0029] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0030] The system includes primary equipment intelligent sensors, sensor interconnection equipment, intelligent embedded transformer digital monitoring and safety protection system, human-computer interaction interface and corresponding supporting software and hardware. The system can analyze and determine the operating status and fault evolution trend of the equipment by capturing the transformer equipment information.

[0031] like Figure 1 As shown, a transformer full-area linkage digital monitoring and safety protection system includes: a total of 4 layers, namely, a perception layer, a data layer, an edge layer and an application layer.

[0032] Among them, the perception layer deploys a large number of sensor devices to realize comprehensive perception and monitoring of the operating status of equipment such as transformer load monitoring, oil temperature, online DGA, and vibration of on-load switches, and obtains a transformer monitoring data set. The sensor devices include multi-parameter sensors of gas in oil-oil temperature-oil pressure-micro-water, multi-parameter sensors of dielectric loss-capacitance-partial discharge, and online vibration monitoring devices. After a transformer equipment fails, the physical properties of the equipment body and accessories change, such as abnormal changes in temperature, vibration, internal pressure, electromagnetic field, dissolved gas in oil, and other multi-physical quantities. The coupling relationship between different physical quantities acts on the transformer, which in turn affects the normal operation of the internal components of the transformer. It is necessary to use sensor devices to perform full-domain multi-state monitoring of the transformer.

[0033] Among them, the multi-parameter sensor of gas in oil, oil temperature, oil pressure and trace water is used to sense the changes of dissolved gas, temperature, internal pressure, trace water, etc. in the oil of the transformer and its accessories; the multi-parameter sensor of dielectric loss, capacitance and partial discharge is used for the synchronous monitoring of signal characteristics such as dielectric loss factor, capacitance and partial discharge of the transformer; the online vibration monitoring device realizes the status monitoring of the on-load switch.

[0034] The multi-parameter sensor of gas in oil-oil temperature-oil pressure-micro-water integrates the gas in oil, oil temperature, oil pressure and micro-water sensing units on the sensor conditioning board. The gas in oil sensing unit accurately converts the gas in oil and the resistance value through the signal processing circuit to characterize the gas in oil; the pressure sensing unit includes a pressure-sensitive element and a signal conditioning circuit. The pressure-sensitive element measures the pressure, and the pressure signal of the pressure-sensitive element is zero-compensated and temperature-compensated through the signal conditioning circuit. The conversion relationship between pressure ε and resistance R is as follows:

[0035]

[0036] Among them, η is the piezoresistance coefficient, E is the elastic modulus, ε is the Poisson coefficient, and K is the sensitivity coefficient.

[0037] The capacitive sensor method is used to measure the trace water in the oil. The relative dielectric constant when the oil contains trace water is calculated according to the Maxwell-Garnetl theory:

[0038]

[0039] Among them, ε m is the relative dielectric constant of water in oil, ε w is the relative dielectric constant of water, ε is the relative dielectric constant of oil, and α is the water mass fraction. When the micro-water sensor unit senses that its dielectric constant changes, the capacitance also changes accordingly, and the water content in the oil is calculated based on the change in capacitance.

[0040] The dielectric loss-capacitance-partial discharge multi-parameter sensor integrates high- and low-frequency signal coils. The coils are sleeved on the extended grounding rod. One end of the grounding rod is connected to the bushing end screen lead wire in the form of crown spring crimping or spring top connection, and the other end is connected to the grounding end of the shell. The high- and low-frequency signal coils are used to ensure the safety and reliability of grounding. The high- and low-frequency signal coils use the principle of electromagnetic induction to obtain the leakage current of the bushing and the high-frequency partial discharge current signal, realize the accurate measurement of the leakage current and partial discharge current signal, and combine the relative measurement method and the absolute measurement method to obtain the dielectric loss and capacitance of the transformer.

[0041] The data layer includes aggregation nodes and access node devices to achieve the integration and regulation of monitoring data.

[0042] The aggregation node is used to receive and aggregate information from sensor devices, and then transmit the sensor device data to the access node; after the access node realizes standardized and intelligent access to multiple sensor devices, it sends the protocolized data to the edge computing module.

[0043] For data of the same type, the aggregation node uses information entropy to measure the uncertainty contained in the data, assigns data weights according to their information reliability, and realizes data fusion.

[0044] Assuming that n sensors measure transformer data simultaneously, the data {X1, X2, ..., X j , ..., X n}, where X j (j=1, 2, ..., n) is the measurement data of the jth sensor. The weight value assigned to the jth sensor is ω j (j=1, 2, ..., n), the weighted average fusion model of the entropy weight method is:

[0045] X=[ω1,ω2,...,ω n ][X1,X2,...,X n ] T

[0046] Where T represents transposition, X represents the weighted average fusion data of the entropy weight method; ω1, ω2, ..., ω n Respectively represent the weights of the measurement data of the 1st, 2nd, ..., nth sensors, X1, X2, ..., X n Represent the measurement data of the 1st, 2nd,…, nth sensors respectively.

[0047] For different types of data, principal component analysis is used to reduce data dimension through linear transformation, simplify complex data, realize the fusion of heterogeneous data, and use the fused data as input data of the fault diagnosis model.

[0048] Assume that there are P monitoring indicators in the transformer safety protection system, T1, T2, ..., T are the time series values ​​of the 1st, 2nd, ..., pth monitoring indicators, respectively, forming a vector with p dimensions (T1, T2, ..., T p ).

[0049] Since heterogeneous monitoring data have dimensional differences, in order to achieve the standardization of multiple sensor devices and eliminate the impact of dimensional differences on subsequent data processing. Assume that in T, the jth sensor device, the i-th monitoring value indexed is t ij (i=1, 2, ..., M; j=1, 2, ..., p), M is the number of monitoring values, and standardization is achieved by the following formula:

[0050]

[0051] Z is the obtained standardized matrix, μ j is the average value of the monitoring value of the jth sensor device, σ j is the standard deviation of the monitoring value of the jth sensor device;

[0052] The correlation coefficient R of the standardized matrix Z is:

[0053]

[0054] Calculate the cumulative contribution rate a and determine the principal components of the overall data:

[0055]

[0056] Among them, λ i is the i-th eigenvalue, and K is the number of principal components containing the elements.

[0057] The edge layer embeds the trained fault diagnosis model into the edge computing module to realize differentiated defect diagnosis and evaluation of the transformer status, and transmits the results to the visual monitoring platform, changing the current working situation of resource consumption and manual analysis energy occupied by large-scale data concentrated in the main station calculation; at the same time, when a serious fault occurs in the transformer, the edge computing module can directly act on the protection device to protect the safety of the protection system.

[0058] Fault diagnosis model: The SMOTE algorithm is used to expand the minority class samples in the obtained transformer monitoring data set, reduce the imbalance of the original data set, and obtain a balanced data set. The balanced data set is then fused through the aggregation node, and the fused data is used as the data set for training the SSA-CNN model; a CNN model is constructed, and the structure and hyperparameters of the CNN model are optimized and trained through the SSA algorithm to obtain an optimized SSA-CNN model. The SSA-CNN model is then trained with the constructed data set to obtain a fault diagnosis model for transformer fault diagnosis.

[0059] Of course, in this application, in addition to using the fused data for training the SSA-CNN model data set, the signal data such as dissolved gas in oil, temperature, internal pressure, micro-water, dielectric loss factor, capacitance, partial discharge, etc. can also be used as feature variables based on the transformer historical data to establish a transformer multi-parameter state evaluation index for training the SSA-CNN model. The single state quantity threshold is shown in the following table, and the transformer fault is divided into low-temperature overheating, high-temperature overheating, low-energy discharge, and high-energy discharge to achieve transformer fault diagnosis.

[0060] index Normal value Note value <![CDATA[H2 content]]> 0 500 <![CDATA[C2H2 content]]> 0 1 <![CDATA[CH4 content]]> 0 100 <![CDATA[C2H4 content]]> 0 1.45 Total hydrocarbon content 0 150 Initial value difference of capacitance / % 0 5 Bushing partial discharge / ρC 0 20 Dielectric loss factor / % 0 0.8

[0061] The hyperparameters of the CNN model are as follows: the kernel sizes of the convolutional layers are 3×3 and 2×2, and the numbers are 32 and 64 respectively; a pooling layer and a normalization layer are set after each convolutional layer, the size of the pooling kernel is 2×2, and the step size is 1; the activation function uses the ReLu function; the fully connected layer uses 2 layers, and the number of neurons is 64 and 32 respectively; after activation by the Softmax function, the output is 5; the optimizer is Adam; and the learning rate is 0.001.

[0062] The parameter settings of the SSA-CNN classification model: the convolutional layer, fully connected layer, optimizer, etc. are the same as the basic CNN model, but the hyperparameters such as the size and number of convolution kernels, the number of neurons in the fully connected layer, and the initial learning rate are optimized using the SSA algorithm.

[0063] The application layer contains all the functional components of the system, including online monitoring module, data list module, fault diagnosis module and fault alarm module, and is displayed in the form of a page on the visual monitoring platform, which is helpful for operation and maintenance personnel to understand the recent operating status of the transformer and has certain guiding significance for the next operation and maintenance decision of the transformer.

[0064] Among them, the online monitoring module monitors the vibration, partial discharge, hot spot imaging, power and other signals and working status of the transformer body and its affiliated on-load tap changers, bushings and other equipment in real time; the data list module is used to display the changing trend of the monitoring signal, event data, etc. in the form of a list; the fault diagnosis module uses load monitoring, oil temperature, online DGA, on-load switch vibration and other equipment operation status monitoring information to realize differentiated defect diagnosis and evaluation of the transformer status, analyze and determine the transformer fault anomaly and evaluate its status; the fault alarm module promptly issues alarm information and formulates operation and maintenance strategies after determining the typical defects, abnormal noises and fault information of the transformer.

[0065] The present invention is not limited to what is described in the specification and implementation modes, and therefore additional advantages and modifications can be easily realized by those skilled in the art. Therefore, without departing from the spirit and scope of the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details, representative devices, and illustrative examples shown and described herein.

Claims

1. A transformer global linkage digital monitoring safety protection system, comprising a perception layer, a data layer, an edge layer and an application layer; characterized in that: The sensing layer deploys sensing devices to realize comprehensive sensing monitoring of the transformer operation status and obtain a transformer monitoring data set. The sensing devices include a multi-parameter sensor of gas in oil, oil temperature, oil pressure and trace water, a multi-parameter sensor of dielectric loss, capacitance and partial discharge, and an online vibration monitoring device. The data layer includes aggregation nodes and access node devices to achieve the integration and regulation of monitoring data; The aggregation node is used to receive and aggregate sensor device information, and then transmit the sensor device data to the access node; after realizing the standardized and intelligent access of multiple sensor devices, the access node sends the agreed data to the edge computing module; For data of the same type, the aggregation node uses information entropy to measure the uncertainty contained in the data, assigns data weights according to their information reliability, and realizes data fusion. For data of different types, the principal component analysis method is used to reduce the data dimension through linear transformation to realize the fusion of heterogeneous data, and the fused data is used as the input data of the fault diagnosis model. The edge layer embeds the trained fault diagnosis model into the edge computing module to realize the differentiated defect diagnosis and evaluation of the transformer status, and transmits the results to the visual monitoring platform. At the same time, when a serious fault occurs in the transformer, the edge computing module directly acts on the protection device to protect the safety of the protection system. The application layer contains all the functional components of the system, including online monitoring module, data list module, fault diagnosis module and fault alarm module, and is displayed in the form of pages on the visual monitoring platform.

2. The transformer global linkage digital monitoring safety protection system according to claim 1 is characterized in that: The multi-parameter sensor of gas in oil-oil temperature-oil pressure-micro-water integrates the gas in oil, oil temperature, oil pressure and micro-water sensing units on the sensor conditioning board. The gas in oil sensing unit accurately converts the gas in oil and the resistance value through the signal processing circuit to characterize the gas in oil; the pressure sensing unit includes a pressure sensitive element and a signal conditioning circuit. The pressure sensitive element measures the pressure, and the pressure signal of the pressure sensitive element is zero-compensated and temperature-compensated through the signal conditioning circuit. The conversion relationship between pressure ε and resistance R is as follows: Among them, η is the piezoresistance coefficient, E is the elastic modulus, ε is the Poisson coefficient, and K is the sensitivity coefficient; The capacitive sensor method is used to measure the trace water in the oil, and the relative dielectric constant when the oil contains trace water is calculated according to the Maxwell-Garnetl theory: Among them, ε m is the relative dielectric constant of water in oil, ε w is the relative dielectric constant of water, ε is the relative dielectric constant of oil, and α is the water mass fraction. When the micro-water sensor unit senses that its dielectric constant changes, the capacitance also changes accordingly, and the water content in the oil is calculated based on the change in capacitance.

3. The transformer global linkage digital monitoring safety protection system according to claim 1 is characterized in that: The dielectric loss-capacitance-partial discharge multi-parameter sensor integrates high and low frequency signal coils. The coils are sleeved on an extended grounding rod. One end of the grounding rod is connected to the bushing end screen lead wire in the form of crown spring crimping or spring top connection, and the other end is connected to the grounding end of the shell.

4. The transformer global linkage digital monitoring safety protection system according to claim 1 is characterized in that: The SMOTE algorithm is used to expand the minority class samples in the obtained transformer monitoring data set to reduce the imbalance of the original data set and obtain a balanced data set. The balanced data set is then fused through the aggregation node, and the fused data is used as the data set for training the SSA-CNN model. A CNN model is constructed, and the structure and hyperparameters of the CNN model are optimized and trained using the SSA algorithm to obtain an optimized SSA-CNN model. The SSA-CNN model is then trained using the constructed data set to obtain a fault diagnosis model for transformer fault diagnosis.

5. The transformer global linkage digital monitoring safety protection system according to claim 1 is characterized in that: The online monitoring module monitors the vibration, partial discharge, hot spot imaging, power and other signals and working status of the transformer body and its attached on-load tap-changers, bushings and other equipment in real time.

6. The transformer global linkage digital monitoring safety protection system according to claim 1 is characterized in that: The data list module is used to display the changing trends of monitoring signals, event data, etc. in the form of a list.

7. The transformer global linkage digital monitoring safety protection system according to claim 1 is characterized in that: The fault diagnosis module uses the equipment operation status monitoring information to realize the differentiated defect diagnosis and evaluation of the transformer status, analyze and determine the transformer fault anomaly and evaluate its status.

8. The transformer global linkage digital monitoring safety protection system according to claim 1 is characterized in that: After determining the typical defects, abnormal noises and fault information of the transformer, the fault alarm module will promptly issue alarm information and formulate operation and maintenance strategies.

Images