CN115693916B - Intelligent online monitoring method and system for direct-current power supply of transformer substation - Google Patents
Intelligent online monitoring method and system for direct-current power supply of transformer substation Download PDFInfo
- Publication number
- CN115693916B CN115693916B CN202211091761.3A CN202211091761A CN115693916B CN 115693916 B CN115693916 B CN 115693916B CN 202211091761 A CN202211091761 A CN 202211091761A CN 115693916 B CN115693916 B CN 115693916B
- Authority
- CN
- China
- Prior art keywords
- parameter
- model
- power supply
- characteristic
- current power
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Landscapes
- Remote Monitoring And Control Of Power-Distribution Networks (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses an intelligent online monitoring method and system for a direct-current power supply of a transformer substation, wherein the method comprises the following steps: receiving a first parameter and a second parameter of the direct current power supply, and acquiring a first analysis result representing whether the running performance of the direct current power supply is abnormal or not based on the first parameter as input of a first intelligent analysis model; based on the first parameter and the second parameter as input of a second intelligent analysis model, a second analysis result representing whether the running performance of the power supply is abnormal or not is obtained, and the running performance condition of the direct current power supply is determined based on the first analysis result and the second analysis result. According to the invention, the second analysis is performed by adopting the second intelligent analysis model matched with the direct current power supply to be analyzed, and the first analysis is performed by adopting the first intelligent analysis model matched with all the direct current power supplies, so that the rough monitoring and the accurate matching type monitoring of the direct current power supplies are simultaneously realized, and the accuracy of the running state analysis for all the direct current power supplies is effectively improved.
Description
Technical Field
The invention relates to the technical field of monitoring of a direct-current power supply system of a transformer substation, in particular to an intelligent online monitoring method and system of the direct-current power supply of the transformer substation.
Background
The direct current power supply system of the transformer substation provides reliable power supply for protection, control, automatic devices and the like and is an independent power supply. In normal operation, the charging device bears a constant load and simultaneously charges the storage battery pack to supplement self-discharge of the storage battery pack, so that the storage battery pack is in standby in a full-capacity state. Under the condition of interruption of the alternating current power supply, the storage battery pack continues to provide the direct current power supply for the load, so that the normal operation of the protection, control and automatic device can be ensured. As the running time of the direct current system of the transformer substation becomes longer, the power supply module and the storage battery in the direct current system may have abnormal running conditions. Defects in these dc systems may lead to more serious problems if not handled in time.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides an intelligent online monitoring method and system for a direct current power supply of a transformer substation, which realize accurate analysis of operation faults in a direct current system. The technical scheme is as follows:
in a first aspect, an embodiment of the application discloses an intelligent online monitoring method for a direct current power supply of a transformer substation, which includes the following steps:
receiving a first parameter and a second parameter of a direct current power supply, wherein the first parameter comprises: the internal resistance, voltage, current, temperature, humidity, operation working period and the second parameters of the direct current storage battery pack comprise: use duration, manufacturer, model and batch;
based on a first parameter as input of a first intelligent analysis model, acquiring a first analysis result representing whether the running performance of the direct current power supply is abnormal or not, wherein the first intelligent analysis model is obtained by training based on historical first parameters of all the direct current power supplies as training samples;
acquiring a second analysis result representing whether the running performance of the power supply is abnormal or not based on the first parameter and the second parameter as inputs of a second intelligent analysis model, wherein the second intelligent analysis model is an analysis model matched with the direct current power supply and obtained based on at least one historical first parameter and second parameter of the existing second intelligent analysis model in the direct current power supply as training samples;
and determining the running performance condition of the direct current power supply based on the first analysis result and the second analysis result.
As a further optimization of the above solution, the analysis process of the first intelligent analysis model includes:
extracting a first characteristic parameter and a second characteristic parameter based on the first parameter, wherein the first characteristic parameter comprises the duration of different operation working periods, the temperature and humidity of the running environment of the direct current power supply, the temperature and humidity of the power supply and the change characteristics of voltage and current during different operation working periods; the second characteristic parameter comprises a variation characteristic of internal resistance during different operation working periods;
comparing and analyzing based on the first characteristic parameter and a preset constraint rule to determine whether a first fault hidden danger exists in the running of the direct-current power supply;
under the condition that the first fault hidden danger does not exist, based on a first characteristic parameter and a second characteristic parameter, inputting the first characteristic parameter and the second characteristic parameter into a first intelligent analysis model, and analyzing whether the second characteristic parameter is abnormal under the first characteristic parameter;
and under the condition that the second characteristic parameter is determined to be normal, determining that the running performance of the direct current power supply is normal.
As a further optimization of the above solution, the running duty cycle includes: charging, floating charging, AC power-off discharging and AC recovery charging,
the preset constraint rule comprises the following steps: duration constraint rules of different operation working periods, constraint rules of temperature and humidity of a direct current power supply operation environment and temperature and humidity difference of a power supply under different operation working periods, and constraint rules of variation of voltage and current during different operation working periods.
As a further optimization of the above solution, the inputting the first and second characteristic parameters into the first intelligent analysis model to analyze whether the second characteristic parameter is abnormal under the first characteristic parameter includes:
performing depth feature analysis on the input first feature parameter and second feature parameter based on a first network, the first network performing depth feature analysis comprising:
obtaining a third feature through at least one feature extraction layer processing based on the first feature parameter and the second feature parameter;
combining based on the first characteristic parameter, the second characteristic parameter and the third characteristic parameter to obtain an initial characteristic vector, wherein the sum of the dimensions of the first characteristic parameter and the second characteristic parameter is d1, the dimension of the third characteristic is d2, and the dimension of the initial characteristic vector is d1+d2;
the first weight parameters are obtained through first preset activation function processing based on the first learnable parameter matrix acting on the initial feature vector, and the second weight parameters are obtained through first preset activation function processing based on the second learnable parameter matrix acting on the initial feature vector;
and based on the first weight parameter acting on the first characteristic parameter and the second characteristic parameter acting on the third characteristic, obtaining a fusion characteristic after weighted fusion based on the second weight parameter acting on the third characteristic, inputting the fusion characteristic into a preset first intelligent analysis model, and analyzing whether the second characteristic parameter is abnormal under the first characteristic parameter.
As a further optimization of the above scheme, the first network adopts 3-layer convolution, and includes 112 convolution kernels in total, where the convolution kernels are 3×3 in size, 1 in step size, and 1 in padding value.
As a further optimization of the above solution, the method for obtaining the second intelligent analysis model includes:
recording an existing second intelligent analysis model as an original model, and recording a second intelligent analysis model to be obtained aiming at the direct-current power supply as an updated model;
acquiring a second training sample set aiming at a second intelligent analysis model based on the historical first parameter and second parameter of the direct current power supply;
inputting a training sample based on an original model into the original model to obtain a first loss function of the original model and first model parameters of the original model updated based on the first loss function;
taking the first model parameter of the original model as the second model parameter of the updated model, and inputting the first model parameter into the updated model based on a training sample of the updated model to obtain an output result of the updated model;
correcting second model parameters of the updated model based on the difference between the training samples input by the original model and the training samples input by the updated model and the first loss function value;
repeating the steps, and carrying out iterative correction on the second model parameters of the updated model until the preset iterative stopping condition is reached, so as to obtain a trained second intelligent analysis model.
As a further optimization of the above solution, the correcting the second model parameter of the update model based on the difference between the training sample input by the original model and the training sample input by the update model and the first loss function value includes:
based on
Correcting a second model parameter of the updated model, wherein L is an accumulated error loss, L1 is a first loss function value, L2 is a difference between a training sample input by the original model and a training sample input by the updated model, and alpha i The weight of self-adaption and training precision in the original model is that i is the number of layers of the differential analysis network layer of the training sample input by the original model and the training sample input by the updated model; wherein (1)>
Wherein (1)>
R represents 1 -R 2 R1 and R2 are covariance matrices of training sample data input by the original model and the updated model respectively, and d is the dimensionality number of the training sample input by the original model and the updated model.
In a second aspect, an embodiment of the present application discloses an intelligent online monitoring system for a direct current power supply of a transformer substation, the system includes:
the data acquisition unit is used for receiving a first parameter and a second parameter of the direct current power supply, wherein the first parameter comprises: the internal resistance, voltage, current, temperature, humidity, operation working period and the second parameters of the direct current storage battery pack comprise: use duration, manufacturer, model and batch;
the first analysis unit is used for obtaining a first analysis result representing whether the running performance of the direct current power supply is abnormal or not based on the first parameter serving as input of a first intelligent analysis model, and the first intelligent analysis model is obtained by training based on historical first parameters of all the direct current power supplies serving as training samples;
the second analysis unit is used for obtaining a second analysis result representing whether the running performance of the power supply is abnormal or not based on the first parameter and the second parameter serving as input of a second intelligent analysis model, wherein the second intelligent analysis model is an analysis model matched with the direct current power supply and obtained based on at least one existing second intelligent analysis model by taking the historical first parameter and the second parameter of the direct current power supply as training samples;
and the online monitoring result acquisition unit is used for determining the running performance condition of the direct current power supply based on the first analysis result and the second analysis result.
In a third aspect, an embodiment of the present application discloses an electronic device, including:
a processor;
a memory for storing processor-executable instructions;
wherein the processor is configured to implement the intelligent on-line monitoring method of the substation direct current power supply according to any one of claims 1-7 by executing the executable instructions.
In a fourth aspect, embodiments of the present application disclose a computer readable storage medium having stored thereon computer instructions, wherein the instructions, when executed by a processor, implement the steps of the intelligent on-line monitoring method for a dc power supply of a substation according to any of claims 1-7.
The intelligent online monitoring method and system for the direct current power supply of the transformer substation have the following beneficial effects: acquiring a first analysis result representing whether the running performance of the direct-current power supply is abnormal or not by taking the first parameter as the input of a first intelligent analysis model; based on the first parameter and the second parameter as input of a second intelligent analysis model, obtaining a second analysis result representing whether the running performance of the power supply is abnormal; the running performance condition of the direct current power supply is determined based on the first analysis result and the second analysis result, the targeted analysis of different direct current power supplies is realized, the second analysis is performed by adopting a second intelligent analysis model matched with the direct current power supply to be analyzed, meanwhile, the first analysis is performed by adopting a first intelligent analysis model matched with all the direct current power supplies, the rough monitoring and the accurate matching type monitoring of the direct current power supplies are realized, and the accuracy of running state analysis for all the direct current power supplies is effectively improved.
Drawings
Fig. 1 is a schematic flow chart of an intelligent online monitoring method for a direct current power supply of a transformer substation according to an embodiment of the application;
FIG. 2 is a schematic illustration of an analysis process of a first smart analysis model in the present application;
FIG. 3 is a schematic diagram of a first network depth profile analysis process according to the present application;
FIG. 4 is a schematic diagram of a method of obtaining a second smart analytical model in the present application;
fig. 5 is a schematic structural diagram of an intelligent online monitoring system for a direct-current power supply of a transformer substation according to an embodiment of the application.
Detailed Description
In the running process of the direct current system of the transformer substation, the conversion between chemical energy and electric energy is involved, and when the running time is long, the direct current system of the transformer substation is influenced by factors such as human factors or equipment quality and the like, and faults are easy to occur. At this time, fault diagnosis and monitoring and early warning technologies are necessary to be applied, so that defects existing in the direct current system of the transformer substation can be found and mastered in time. By utilizing the fault diagnosis and monitoring and early warning technology of the direct current system of the transformer substation, the active isolation device for the state diagnosis and abnormal condition of the storage battery of the direct current system of the transformer substation can be developed, and the running state of the direct current system can be judged according to the acquired data information. When the abnormal condition of the direct current system is found, timely and actively isolating is carried out, so that the direct current system is ensured not to have a large range of faults.
Referring to fig. 1, an embodiment of the present application provides an intelligent online monitoring method for a dc power supply of a substation, where the method includes the following steps:
step 1, receiving a first parameter and a second parameter of a direct current power supply, wherein the first parameter comprises: the internal resistance, voltage, current, temperature, humidity, operation working period and the second parameters of the direct current storage battery pack comprise: use duration, manufacturer, model and batch;
step 2, based on a first parameter as input of a first intelligent analysis model, acquiring a first analysis result representing whether the running performance of the direct current power supply is abnormal, wherein the first intelligent analysis model is trained based on historical first parameters of all the direct current power supplies as training samples;
step 3, based on the first parameter and the second parameter as the input of a second intelligent analysis model, obtaining a second analysis result representing whether the running performance of the power supply is abnormal or not, wherein the second intelligent analysis model is an analysis model matched with the direct current power supply and obtained based on at least one historical first parameter and second parameter of the existing second intelligent analysis model in the direct current power supply as training samples;
and step 4, determining the running performance condition of the direct current power supply based on the first analysis result and the second analysis result.
The direct current power supply system provides stable and continuous working power for the protection and lighting equipment in the station when the alternating current power supply system for the station loses power, and serious accidents in the transformer substation are often caused by insufficient capacity of a storage battery pack in the direct current power supply system, open circuit inside or not connected with a direct current bus reliably and cannot supply power for the protection equipment when the alternating current system is abnormal. The alternating current load causing accidents cannot be timely removed from the power grid, and the fault loop can be removed only by means of override trip protection, so that the large-area load loses power, and immeasurable socioeconomic loss is caused.
According to the method, the operation data of the substation direct-current power supply system are comprehensively monitored, the operation state data of the direct-current power supply is comprehensively analyzed and predicted through the first acquisition parameters and the second acquisition parameters, the data are analyzed by combining the artificial intelligent neural network based on the deep neural network and the big data serving as training samples, and the accurate diagnosis of the operation state of the direct-current power supply is realized. In the application, the second analysis is performed by adopting the second intelligent analysis model matched with the direct current power supply to be analyzed, and the first analysis is performed by adopting the first intelligent analysis model matched with all the direct current power supplies, so that the rough type monitoring and the accurate matching type monitoring of the direct current power supplies are realized, and the accuracy of the running state analysis for all the direct current power supplies is effectively improved.
It can be understood that, for all dc power supplies, there are numerical constraint rules corresponding to the voltage, current, temperature, humidity, operation working period and operation internal resistance in a normal range during operation, and each parameter can represent different operation states of the dc power supply when different values are provided. It can be understood that the normal range of the state parameters of the storage battery can be influenced by different manufacturers, models and batches, and correspondingly, the second intelligent analysis model is adopted to analyze the running state of the direct current power supply by considering the using time length, the manufacturer, the model and the batch attribute of the direct current power supply, so as to obtain the running state analysis result which is accurately matched with the characteristics of the direct current power supply.
Referring to fig. 2, the analysis process of the first intelligent analysis model in the above step 2 includes:
step 21, extracting a first characteristic parameter and a second characteristic parameter based on the first parameter, wherein the first characteristic parameter comprises the duration of different operation working periods, the temperature and humidity of the DC power supply operation environment, the temperature and humidity of the power supply and the change characteristics of voltage and current during different operation working periods; the second characteristic parameter comprises a variation characteristic of internal resistance during different operation working periods;
step 22, comparing and analyzing based on the first characteristic parameter and a preset constraint rule to determine whether a first fault hidden danger exists in the operation of the direct current power supply;
step 23, under the condition that the first fault hidden danger does not exist, inputting the first and second characteristic parameters into a first intelligent analysis model based on the first and second characteristic parameters, and analyzing whether the second characteristic parameter is abnormal under the first characteristic parameter;
and step 24, determining that the running performance of the direct current power supply is normal under the condition that the second characteristic parameter is determined to be normal.
It can be understood that the temperature and humidity of the power supply have preset upper and lower threshold intervals, but under the condition of temperature and humidity change of the operating environment, the preset upper and lower threshold values can float, and of course, the floating upper and lower threshold value intervals cannot exceed the preset absolute threshold value interval range. When the running performance of the power supply is changed, the internal resistance characteristic of the power supply is abnormal, but the internal resistance of the power supply has relativity and variability, the real-time internal resistance of the power supply fluctuates in different real-time temperature and humidity environments and under the condition of normal running working cycles, in the embodiment of the application, under the condition that the first characteristic parameters are normal, the numerical characteristics of the internal resistance of the power supply under the first characteristic parameters in different running working cycles are analyzed through a first intelligent analysis model, the first intelligent analysis model takes a plurality of parameters in the first characteristic parameters and the second characteristic parameters as input, the proper theoretical second characteristic parameters are analyzed through the relation between the plurality of parameters in the first characteristic parameters and the second characteristic parameters which are fitted in advance through the model, and whether the input second characteristic parameters are normal is determined based on the deviation of the input second characteristic parameters and the theoretical second characteristic parameters.
As a further optimization of the above solution, the running duty cycle in the above step 21 includes: charging, floating charging, AC power-off discharging and AC recovery charging;
the preset constraint rules in the above step 22 include: duration constraint rules of different operation working periods, constraint rules of temperature and humidity of a direct current power supply operation environment and temperature and humidity difference of a power supply under different operation working periods, and constraint rules of variation of voltage and current during different operation working periods.
As a further optimization of the above solution, in the step 23, based on the first feature parameter and the second feature parameter input to the first intelligent analysis model, analyzing whether the second feature parameter is abnormal under the first feature parameter includes:
step 231, performing depth feature analysis on the input first feature parameter and second feature parameter based on the first network, specifically referring to fig. 3, the first network performing depth feature analysis, including:
step 2311, obtaining a third feature through at least one feature extraction layer process based on the first feature parameter and the second feature parameter;
step 2312, merging based on the first feature parameter, the second feature parameter and the third feature, thereby obtaining an initial feature vector, wherein the sum of the dimensions of the first feature parameter and the second feature parameter is d1, the dimension of the third feature is d2, and the dimension of the initial feature vector is d1+d2;
2313, applying the first matrix of learnable parameters to the initial feature vector and performing a first preset activation function process to obtain a first weight parameter, and applying the second matrix of learnable parameters to the initial feature vector and performing a first preset activation function process to obtain a second weight parameter;
step 2314, obtaining a fusion feature after weighted fusion based on the first weight parameter acting on the first feature parameter and the second feature parameter acting on the third feature and the second weight parameter acting on the third feature, inputting the fusion feature into a preset first intelligent analysis model, and analyzing whether the second feature parameter is abnormal under the first feature parameter.
Specifically, the first network in step 231 employs a 3-layer convolution, comprising a total of 112 convolution kernels, each of 3×3 in size, 1 in step size, and 1 in padding value.
In the embodiment of the application, the third feature is obtained through at least one layer of feature extraction layer processing based on the first feature parameter and the second feature parameter, the third feature is a depth feature, the first feature parameter and the second feature parameter are shallow features, and the richness and the high efficiency of feature data are effectively improved based on the combination of the depth feature and the shallow features, so that the accuracy of the analysis result of the first intelligent analysis model is improved. Specifically, in step 2313, v=h (f·j), where V is a weightThe re-parameters, H is a first preset activation function, F is an initial feature vector, M is a first or second matrix of learnable parameters, the first preset activation function may be
Further, referring to fig. 3, the method for obtaining the second intelligent analysis model in the step 3 includes:
step 31, recording the existing second intelligent analysis model as an original model, and recording the second intelligent analysis model to be obtained aiming at the direct current power supply as an updated model;
step 32, acquiring a second training sample set aiming at a second intelligent analysis model based on the historical first parameter and the second parameter of the direct current power supply;
step 33, inputting a training sample based on an original model into the original model to obtain a first loss function of the original model and first model parameters of the original model updated based on the first loss function;
step 34, taking the first model parameter of the original model as the second model parameter of the updated model, and inputting the second model parameter into the updated model based on the training sample of the updated model to obtain the output result of the updated model;
step 35, correcting the second model parameters of the updated model based on the difference between the training samples input by the original model and the training samples input by the updated model and the first loss function value;
and 36, repeating the steps 32-35, and carrying out iterative correction on the second model parameters of the updated model until the preset iterative stopping condition is reached, so as to obtain a trained second intelligent analysis model.
Further, in the step 35, the step of correcting the second model parameter of the updated model based on the difference between the training sample input by the original model and the training sample input by the updated model and the first loss function value includes:
based on
Correcting a second model parameter of the updated model, wherein L is an accumulated error loss, L1 is a first loss function value, L2 is a difference between a training sample input by the original model and a training sample input by the updated model, and alpha i The weight of self-adaption and training precision in the original model is that i is the number of layers of the differential analysis network layer of the training sample input by the original model and the training sample input by the updated model; wherein (1)>
Wherein (1)>
R represents 1 -R 2 R1 and R2 are covariance matrices of training sample data input by the original model and the updated model respectively, and d is the dimensionality number of the training sample input by the original model and the updated model.
In the embodiment of the application, the model parameters of the original model are used as references for training the updated model, so that the training time and the training efficiency of the updated model, namely the second intelligent analysis model, are effectively reduced, and in the training process, the second model parameters of the updated model are corrected based on the difference between the training samples input by the original model and the training samples input by the updated model and the first loss function value, so that the overfitting to the original model when the second model parameters of the updated model are corrected by using the first loss function value is avoided, the performance of the second intelligent analysis model is further reduced, and the problem that the accuracy of the second intelligent analysis model analysis is reduced when the second model parameters of the updated model are corrected by using the difference between the training samples input by the original model and the training samples input by the updated model is avoided.
Based on the intelligent online monitoring method of the direct current power supply of the transformer substation, the embodiment of the application also provides an intelligent online monitoring system of the direct current power supply of the transformer substation, which comprises the following steps:
the data acquisition unit is used for receiving a first parameter and a second parameter of the direct current power supply, wherein the first parameter comprises: the internal resistance, voltage, current, temperature, humidity, operation working period and the second parameters of the direct current storage battery pack comprise: use duration, manufacturer, model and batch;
the first analysis unit is used for obtaining a first analysis result representing whether the running performance of the direct current power supply is abnormal or not based on the first parameter serving as input of a first intelligent analysis model, and the first intelligent analysis model is obtained by training based on historical first parameters of all the direct current power supplies serving as training samples;
the second analysis unit is used for obtaining a second analysis result representing whether the running performance of the power supply is abnormal or not based on the first parameter and the second parameter serving as input of a second intelligent analysis model, wherein the second intelligent analysis model is an analysis model matched with the direct current power supply and obtained based on at least one existing second intelligent analysis model by taking the historical first parameter and the second parameter of the direct current power supply as training samples;
and the online monitoring result acquisition unit is used for determining the running performance condition of the direct current power supply based on the first analysis result and the second analysis result.
It can be understood that the intelligent online monitoring system for the dc power supply of the transformer substation provided in this embodiment belongs to the same concept as the intelligent online monitoring method embodiment for the dc power supply of the transformer substation provided in the foregoing embodiment, and detailed implementation processes of the intelligent online monitoring system are shown in the method embodiment, which is not described herein again. All or part of each unit in the intelligent online monitoring system for the direct-current power supply of the transformer substation can be realized by software, hardware and a combination thereof. The units can be embedded in hardware or independent of a processor in the computer equipment, and can also be stored in a memory in the computer equipment in a software mode, so that the processor can call and execute the operations corresponding to the units.
Based on the intelligent online monitoring method for the direct-current power supply of the transformer substation, the embodiment of the application also provides electronic equipment, which comprises:
a processor;
a memory for storing processor-executable instructions;
the processor runs the executable instructions to realize the intelligent online monitoring method for the direct-current power supply of the transformer substation.
It will be appreciated that the memory can be either volatile memory or nonvolatile memory, and can include both volatile and nonvolatile memory. The memory in the embodiment of the invention can store data to support the operation of the terminal. Examples of such data include: any computer program for operating on the terminal, such as an operating system and application programs. The operating system includes various system programs, such as a framework layer, a core library layer, a driver layer, and the like, for implementing various basic services and processing hardware-based tasks. The application may comprise various applications.
Based on the intelligent online monitoring method for the direct current power supply of the transformer substation, the embodiment of the application also provides a computer readable storage medium, wherein computer instructions are stored on the computer readable storage medium, and the instructions are executed by a processor to realize the steps of the intelligent online monitoring method for the direct current power supply of the transformer substation. The computer readable storage medium may be a read-only memory (ROM), a random access memory (random access memory, RAM), a compact disc read-only memory (CD-ROM), a magnetic tape, a floppy disk, an optical data storage node, etc.
The present invention is not limited to the above-described specific embodiments, and various modifications may be made by those skilled in the art without inventive effort from the above-described concepts, and are within the scope of the present invention.
Claims (8)
1. The intelligent online monitoring method for the direct-current power supply of the transformer substation is characterized by comprising the following steps of:
receiving a first parameter and a second parameter of a direct current power supply, wherein the first parameter comprises: the internal resistance, voltage, current, temperature, humidity, operation working period and the second parameters of the direct current storage battery pack comprise: use duration, manufacturer, model and batch;
based on a first parameter as input of a first intelligent analysis model, acquiring a first analysis result representing whether the running performance of the direct current power supply is abnormal or not, wherein the first intelligent analysis model is obtained by training based on historical first parameters of all the direct current power supplies as training samples;
acquiring a second analysis result representing whether the running performance of the power supply is abnormal or not based on the first parameter and the second parameter as inputs of a second intelligent analysis model, wherein the second intelligent analysis model is an analysis model matched with the direct current power supply and obtained based on at least one historical first parameter and second parameter of the existing second intelligent analysis model in the direct current power supply as training samples;
determining the running performance condition of the direct current power supply based on the first analysis result and the second analysis result;
the analysis process of the first intelligent analysis model comprises the following steps:
extracting a first characteristic parameter and a second characteristic parameter based on the first parameter, wherein the first characteristic parameter comprises the duration of different operation working periods, the temperature and humidity of the running environment of the direct current power supply, the temperature and humidity of the power supply and the change characteristics of voltage and current during different operation working periods; the second characteristic parameter comprises a variation characteristic of internal resistance during different operation working periods;
comparing and analyzing based on the first characteristic parameter and a preset constraint rule to determine whether a first fault hidden danger exists in the running of the direct-current power supply;
under the condition that the first fault hidden danger does not exist, based on a first characteristic parameter and a second characteristic parameter, inputting the first characteristic parameter and the second characteristic parameter into a first intelligent analysis model, and analyzing whether the second characteristic parameter is abnormal under the first characteristic parameter;
under the condition that the second characteristic parameter is determined to be normal, determining that the running performance of the direct current power supply is normal;
the first and second characteristic parameters are input into a first intelligent analysis model, and whether the second characteristic parameter is abnormal under the first characteristic parameter is analyzed, including:
performing depth feature analysis on the input first feature parameter and second feature parameter based on a first network, the first network performing depth feature analysis comprising:
obtaining a third feature through at least one feature extraction layer processing based on the first feature parameter and the second feature parameter;
combining based on the first characteristic parameter, the second characteristic parameter and the third characteristic parameter to obtain an initial characteristic vector, wherein the sum of the dimensions of the first characteristic parameter and the second characteristic parameter is d1, the dimension of the third characteristic is d2, and the dimension of the initial characteristic vector is d1+d2;
the first weight parameters are obtained through first preset activation function processing based on the first learnable parameter matrix acting on the initial feature vector, and the second weight parameters are obtained through first preset activation function processing based on the second learnable parameter matrix acting on the initial feature vector;
and based on the first weight parameter acting on the first characteristic parameter and the second characteristic parameter acting on the third characteristic, obtaining a fusion characteristic after weighted fusion based on the second weight parameter acting on the third characteristic, inputting the fusion characteristic into a preset first intelligent analysis model, and analyzing whether the second characteristic parameter is abnormal under the first characteristic parameter.
2. The intelligent online monitoring method for the direct-current power supply of the transformer substation according to claim 1, wherein,
the operational duty cycle includes: charging, floating charging, AC power-off discharging and AC recovery charging,
the preset constraint rule comprises the following steps: duration constraint rules of different operation working periods, constraint rules of temperature and humidity of a direct current power supply operation environment and temperature and humidity difference of a power supply under different operation working periods, and constraint rules of variation of voltage and current during different operation working periods.
3. The intelligent online monitoring method for the direct-current power supply of the transformer substation according to claim 1, wherein the first network adopts 3-layer convolution and totally comprises 112 convolution kernels, the convolution kernels are 3×3 in size, the step size is 1, and the filling value is 1.
4. The intelligent online monitoring method of the direct-current power supply of the transformer substation according to claim 1, wherein the acquiring method of the second intelligent analysis model comprises the following steps:
recording an existing second intelligent analysis model as an original model, and recording a second intelligent analysis model to be obtained aiming at the direct-current power supply as an updated model;
acquiring a second training sample set aiming at a second intelligent analysis model based on the historical first parameter and second parameter of the direct current power supply;
inputting a training sample based on an original model into the original model to obtain a first loss function of the original model and first model parameters of the original model updated based on the first loss function;
taking the first model parameter of the original model as the second model parameter of the updated model, and inputting the first model parameter into the updated model based on a training sample of the updated model to obtain an output result of the updated model;
correcting second model parameters of the updated model based on the difference between the training samples input by the original model and the training samples input by the updated model and the first loss function value;
repeating the steps, and carrying out iterative correction on the second model parameters of the updated model until the preset iterative stopping condition is reached, so as to obtain a trained second intelligent analysis model.
5. The intelligent online monitoring method of a dc power supply of a transformer substation according to claim 4, wherein the correcting the second model parameter of the update model based on the difference between the training sample input by the original model and the training sample input by the update model and the first loss function value comprises:
based on
Correcting a second model parameter of the updated model, wherein L is a cumulative error loss, L1 is a first loss function value, L2 is a training sample of the original model input and a training sample of the updated model inputDifference, alpha i The weight of self-adaption and training precision in the original model is that i is the number of layers of the differential analysis network layer of the training sample input by the original model and the training sample input by the updated model; wherein (1)>
Wherein (1)>
R represents 1 -R 2 R1 and R2 are covariance matrices of training sample data input by the original model and the updated model respectively, and d is the dimensionality number of the training sample input by the original model and the updated model.
6. The utility model provides a transformer substation direct current power supply intelligence on-line monitoring system which characterized in that includes:
the data acquisition unit is used for receiving a first parameter and a second parameter of the direct current power supply, wherein the first parameter comprises: the internal resistance, voltage, current, temperature, humidity, operation working period and the second parameters of the direct current storage battery pack comprise: use duration, manufacturer, model and batch;
the first analysis unit is used for obtaining a first analysis result representing whether the running performance of the direct current power supply is abnormal or not based on the first parameter serving as input of a first intelligent analysis model, and the first intelligent analysis model is obtained by training based on historical first parameters of all the direct current power supplies serving as training samples;
the second analysis unit is used for obtaining a second analysis result representing whether the running performance of the power supply is abnormal or not based on the first parameter and the second parameter serving as input of a second intelligent analysis model, wherein the second intelligent analysis model is an analysis model matched with the direct current power supply and obtained based on at least one existing second intelligent analysis model by taking the historical first parameter and the second parameter of the direct current power supply as training samples;
the online monitoring result acquisition unit is used for determining the running performance condition of the direct-current power supply based on the first analysis result and the second analysis result;
the analysis process of the first intelligent analysis model comprises the following steps:
extracting a first characteristic parameter and a second characteristic parameter based on the first parameter, wherein the first characteristic parameter comprises the duration of different operation working periods, the temperature and humidity of the running environment of the direct current power supply, the temperature and humidity of the power supply and the change characteristics of voltage and current during different operation working periods; the second characteristic parameter comprises a variation characteristic of internal resistance during different operation working periods;
comparing and analyzing based on the first characteristic parameter and a preset constraint rule to determine whether a first fault hidden danger exists in the running of the direct-current power supply;
under the condition that the first fault hidden danger does not exist, based on a first characteristic parameter and a second characteristic parameter, inputting the first characteristic parameter and the second characteristic parameter into a first intelligent analysis model, and analyzing whether the second characteristic parameter is abnormal under the first characteristic parameter;
under the condition that the second characteristic parameter is determined to be normal, determining that the running performance of the direct current power supply is normal;
the first and second characteristic parameters are input into a first intelligent analysis model, and whether the second characteristic parameter is abnormal under the first characteristic parameter is analyzed, including:
performing depth feature analysis on the input first feature parameter and second feature parameter based on a first network, the first network performing depth feature analysis comprising:
obtaining a third feature through at least one feature extraction layer processing based on the first feature parameter and the second feature parameter;
combining based on the first characteristic parameter, the second characteristic parameter and the third characteristic parameter to obtain an initial characteristic vector, wherein the sum of the dimensions of the first characteristic parameter and the second characteristic parameter is d1, the dimension of the third characteristic is d2, and the dimension of the initial characteristic vector is d1+d2;
the first weight parameters are obtained through first preset activation function processing based on the first learnable parameter matrix acting on the initial feature vector, and the second weight parameters are obtained through first preset activation function processing based on the second learnable parameter matrix acting on the initial feature vector;
and based on the first weight parameter acting on the first characteristic parameter and the second characteristic parameter acting on the third characteristic, obtaining a fusion characteristic after weighted fusion based on the second weight parameter acting on the third characteristic, inputting the fusion characteristic into a preset first intelligent analysis model, and analyzing whether the second characteristic parameter is abnormal under the first characteristic parameter.
7. An electronic device, the electronic device comprising:
a processor;
a memory for storing processor-executable instructions;
wherein the processor is configured to implement the intelligent on-line monitoring method of the substation direct current power supply according to any one of claims 1-5 by executing the executable instructions.
8. A computer readable storage medium having stored thereon computer instructions, which when executed by a processor, implement the steps of the intelligent on-line monitoring method of a substation dc power supply according to any of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211091761.3A CN115693916B (en) | 2022-09-07 | 2022-09-07 | Intelligent online monitoring method and system for direct-current power supply of transformer substation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211091761.3A CN115693916B (en) | 2022-09-07 | 2022-09-07 | Intelligent online monitoring method and system for direct-current power supply of transformer substation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115693916A CN115693916A (en) | 2023-02-03 |
| CN115693916B true CN115693916B (en) | 2023-05-30 |
Family
ID=85062622
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211091761.3A Active CN115693916B (en) | 2022-09-07 | 2022-09-07 | Intelligent online monitoring method and system for direct-current power supply of transformer substation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115693916B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117332361B (en) * | 2023-12-01 | 2024-03-01 | 赫芝特电气技术（广州）有限公司 | Online early warning method and detection system for uninterruptible power supply |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2010028883A (en) * | 2008-07-15 | 2010-02-04 | Nec Access Technica Ltd | Dc power supply for backup, and method of controlling the same |
| WO2022160705A1 (en) * | 2021-01-26 | 2022-08-04 | 中国电力科学研究院有限公司 | Method and apparatus for constructing dispatching model of integrated energy system, medium, and electronic device |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103701189B (en) * | 2013-12-25 | 2016-01-13 | 联方云天科技(北京)有限公司 | A kind of distributed dynamic balanced power supply method of networking |
| CN105158723B (en) * | 2015-07-30 | 2019-02-15 | 贵州电力试验研究院 | A kind of error evaluation system and method for digitalized electric energy metering system |
| CN107069959A (en) * | 2017-04-10 | 2017-08-18 | 华北电力大学 | A kind of intelligent substation power of alterating and direct current on-line monitoring system |
| CN110988559A (en) * | 2019-12-20 | 2020-04-10 | 国网湖北省电力公司咸宁供电公司 | Online monitoring method for full life cycle of transformer substation direct current system based on Internet of things |
| CN111308355A (en) * | 2020-03-14 | 2020-06-19 | 智洋创新科技股份有限公司 | Transformer substation storage battery state detection and analysis method based on deep learning |
| CN113872328B (en) * | 2021-09-26 | 2023-07-07 | 国网安徽省电力有限公司铜陵供电公司 | Remote intelligent substation inspection method and system based on neural network |
-
2022
- 2022-09-07 CN CN202211091761.3A patent/CN115693916B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2010028883A (en) * | 2008-07-15 | 2010-02-04 | Nec Access Technica Ltd | Dc power supply for backup, and method of controlling the same |
| WO2022160705A1 (en) * | 2021-01-26 | 2022-08-04 | 中国电力科学研究院有限公司 | Method and apparatus for constructing dispatching model of integrated energy system, medium, and electronic device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115693916A (en) | 2023-02-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112433907B (en) | Method and device for processing host operation parameter data of uninterruptible power supply and electronic device | |
| CN104316879B (en) | A kind of prediction technique in lead-acid batteries service life | |
| CN107121639B (en) | Multidimensional parameter direct current system storage battery management method and device | |
| CN114371409B (en) | Training method of battery state prediction model, battery state prediction method and device | |
| CN103378376A (en) | Battery controller with monitoring logic for model-based battery control | |
| CN114035098A (en) | Lithium battery health state prediction method integrating future working condition information and historical state information | |
| CN117318255B (en) | Battery state analysis system and method based on big data visualization | |
| CN115693916B (en) | Intelligent online monitoring method and system for direct-current power supply of transformer substation | |
| CN115951230B (en) | Abnormality detection method and system for lithium battery energy storage box | |
| CN112230146B (en) | Method, system and equipment for predicting battery charging remaining time | |
| Fendri et al. | Dynamic model to follow the state of charge of a lead-acid battery connected to photovoltaic panel | |
| CN116227637A (en) | Active power distribution network oriented refined load prediction method and system | |
| CN116953547A (en) | Energy storage battery health evaluation method, system, equipment and storage medium | |
| CN115980585A (en) | Battery fault detection method and device, computer equipment and storage medium | |
| CN111460656A (en) | Method and system for evaluating operation life of communication power supply of electric power machine room | |
| CN113406496B (en) | Battery capacity prediction method, system, device and medium based on model migration | |
| CN113093014B (en) | Online collaborative estimation method and system for SOH and SOC based on impedance parameters | |
| CN117169761A (en) | Battery state evaluation method, apparatus, device, storage medium, and program product | |
| CN116008846A (en) | Storage battery pack health state estimation method for space station power supply system | |
| CN116184214A (en) | Online rapid prediction method and device for service life of power battery for vehicle and storage medium | |
| CN114336969B (en) | Energy storage power station centralized monitoring method and system based on virtual power plant | |
| Uddin et al. | State of health estimation of lithium-ion batteries in vehicle-to-grid applications using recurrent neural networks for learning the impact of degradation stress factors | |
| Basia et al. | Comparison of data driven algorithms for SoH estimation of Lithium-ion batteries | |
| CN115389947B (en) | Lithium battery health state prediction method and device, electronic equipment and storage medium | |
| CN117970128B (en) | Battery comprehensive experiment debugging method and system based on real-time feedback control |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |