Abnormal monitoring method and system for high-speed railway transformer, electronic equipment and medium

Abstract

The invention provides a high-speed railway transformer abnormity monitoring method and system, electronic equipment and a medium, and belongs to the technical field of power system monitoring. According to the method, a high-speed railway transformer anomaly monitoring model is constructed, input and output data of the transformer under no load and different loads are collected to form a data set, and the model is trained to achieve anomaly monitoring of the real-time operation state of the transformer. According to the system, a large amount of normal operation monitoring data can be effectively utilized to establish a model, and the operation condition of the transformer is judged according to the probability of actual data in probability distribution given by the model, so that the system does not depend on fault data. According to the method, the advantages of a mechanism model and a data driving method are combined, through the time-varying state space model and the recurrent neural network, the operation mechanism of the transformer is met, monitoring data can be fully utilized, and the monitoring accuracy and interpretability are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power system monitoring, and in particular to a method, system, electronic equipment and medium for monitoring abnormality of a high-speed railway transformer. Background Art

[0002] With the increase in high-speed railway capacity and the expansion of line scale, the operation safety of high-speed railways has become increasingly important. High-speed railway systems are complex and have many devices. Transformers are one of the important equipment in high-speed railway systems, playing an important role in power conversion and power transmission. Transformer operation failures will lead to serious problems such as damage to high-speed railway equipment and train suspension, so abnormal monitoring of high-speed railway transformers is particularly important.

[0003] The existing methods of monitoring abnormalities of high-speed railway transformers can be roughly divided into the following two categories: (1) using sensors to monitor the temperature, vibration and other variables inside the transformer, and comparing them with the set threshold to determine whether the operation is normal; (2) establishing a transformer mechanism model, and judging whether the operation is normal by comparing the size of the model output and the actual data; (3) training the neural network model with a large amount of data to directly give the transformer operation results; Among the above methods, the first type has a low degree of intelligence and is prone to false alarms and omissions, the second type has good interpretability but modeling is often difficult, and many parameters are difficult to calculate, and the third type of method makes full use of a large amount of transformer equipment operation monitoring data, but its interpretability is poor. Combining data with mechanism for abnormal monitoring of equipment can not only utilize the existing mechanism modeling information but also make full use of massive operation data. At present, there is still a lack of a method that combines the operation mechanism and operation data of high-speed railway transformers for abnormal monitoring. Summary of the invention

[0004] The purpose of the present invention is to provide a high-speed railway transformer abnormality monitoring method, system, electronic device and medium, which combines the advantages of mechanism model and data-driven method, and through time-varying state space model and recurrent neural network, it not only conforms to the operation mechanism of the transformer, but also can make full use of monitoring data, and improve the accuracy and interpretability of monitoring. In addition, the design of the present invention enables non-professionals to perform transformer abnormality monitoring, and with the increase of data volume and optimization of the model, the monitoring accuracy can be continuously improved.

[0005] To achieve the above object, the present invention provides a method for monitoring abnormality of a high-speed railway transformer, comprising the following steps:

[0006] Step S1, collecting monitoring data of high-speed railway transformers, including input and output data under no-load and different loads, to form a data set;

[0007] Step S2, constructing a high-speed railway transformer abnormality monitoring model;

[0008] Step S3, training a high-speed railway transformer abnormality monitoring model;

[0009] Step S4: acquiring real-time operation data of the high-speed railway transformer and performing abnormal operation status monitoring.

[0010] Preferably, in step S1, the data set is represents the input of the kth monitoring sequence at time t, represents the output of the kth monitoring sequence at time t, T k represents the length of the kth monitoring sequence, and L represents that there are L monitoring sequences in total.

[0011] Preferably, in step S2, the high-speed railway transformer abnormality monitoring model includes a time-varying state space model of the high-speed railway transformer model, a parameter prior matrix set, a parameter mask matrix set and a recurrent neural network, and the construction method of the high-speed railway transformer abnormality monitoring model is as follows:

[0012] Step S21, establishing a time-varying state space model of a high-speed railway transformer model according to the operating mechanism of the transformer;

[0013]

[0014] Among them, x t is an n-dimensional vector, representing the operating state of the transformer at time t; u t is a d-dimensional vector, representing the input vector of the transformer at time t; z t is an m-dimensional vector, representing the output vector of the transformer at time t; Φ t , B t and H t Respectively represent the state transfer matrix, input matrix and output matrix of the transformer at time t; ω t and v t is zero-mean Gaussian noise, whose covariance matrices are Q t and R t ;symbol represents Gaussian distribution;

[0015] Using the symbol Θ t represents the set of parameters in the time-varying state space model, Θ t ={Φ t , B t , H t , Q t , R t},Θ t is a time-varying parameter;

[0016] Step S22: construct a parameter prior matrix set θ pand the parameter mask matrix set Θ m ;

[0017] Step S23, combining the time-varying state space model in step S21 to construct a recurrent neural network;

[0018]

[0019] make represents the output set of the recurrent neural network, where and They are the state transfer matrix prediction value, input matrix prediction value, output matrix prediction value, ω given by the recurrent neural network. t The covariance matrix of the predicted value and v t The covariance matrix of the predicted value, The matrix dimensions are respectively t , B t , H t , Q t , R t same.

[0020] Preferably, in step S22, construct θ t The corresponding parameter prior matrix set Θ p ={Φ p , B p , H p , Q p , R p}, Φ p , B p , H p , Q p and R p They are Φ t , B t , H t , Q t and R t The corresponding prior matrix, Θ p and θ t With the following relationship:

[0021]

[0022]

[0023]

[0024]

[0025]

[0026] Among them, the symbol [] ijRepresents the element in the i-th row and j-th column of the matrix in [].

[0027] Preferably, in step S22, construct θ t The corresponding parameter prior matrix set Θ m ={Φ m , B m , H m , Q m , R m}, Φ m , B m , H m , Q m and R m They are Φ t , B t , H t , Q t and R t The corresponding mask matrix, Θ m and θ t With the following relationship:

[0028]

[0029]

[0030]

[0031]

[0032]

[0033] Preferably, in step S3, training a high-speed railway transformer abnormality monitoring model comprises the following steps:

[0034] Step S31: Input the input data in the data set into the recurrent neural network and calculate the network output

[0035] Step S32: Do the mask calculation and get The specific calculation formula is:

[0036]

[0037]

[0038]

[0039]

[0040]

[0041] in, and They are the state transfer matrix after mask calculation, input matrix, output matrix, ω t The covariance matrix and v t The covariance matrix of Indicates the multiplication of corresponding elements of the matrix;

[0042] Step S33: Use the Kalman filter to calculate the training loss of the network;

[0043]

[0044] in, and represents the best estimate of the state at time t and time t-1, represents the prior estimate of the state at time t, represents the prior estimate of the state covariance matrix at time t, and the superscript symbol T represents the matrix transpose operation. represents the prior estimate of the output vector at time t, S t represents the prior estimate of the covariance matrix of the output vector at time t, K t represents the Kalman gain at time t, ∑ t and∑ t-1 They represent the best state covariance matrix estimates at time t and time t-1, respectively. In particular, when t = 0, ∑0=I;

[0045] The training loss of the network is obtained as:

[0046]

[0047] Wherein, log represents the logarithmic function;

[0048] Step S34: Use the training loss to train the recurrent neural network.

[0049] Preferably, in step S4, obtaining real-time operation data of the high-speed railway transformer and performing abnormal operation status monitoring comprises the following steps:

[0050] Step S41: Transformer input vector u t Input into the trained recurrent neural network and calculate

[0051] Step S42: According to step S32, Do the mask calculation:

[0052]

[0053] Step S43: Given a threshold δ, calculate the and S t , we get the Gaussian probability distribution when If the voltage is 0, it indicates that the transformer is working normally; otherwise, it indicates that the transformer is working abnormally.

[0054] The present invention also provides a high-speed railway transformer abnormality monitoring system, comprising:

[0055] Data acquisition module, used to collect input and output data of high-speed railway transformers under no-load and different loads;

[0056] A model building module is used to build a high-speed railway transformer anomaly monitoring model, which includes a time-varying state space model, a parameter prior matrix, a parameter mask matrix and a recurrent neural network;

[0057] A training module, used to train the anomaly monitoring model, calculate the output of the recurrent neural network through input data, perform mask calculation and Kalman filter processing to calculate the training loss of the network, and use the training loss to train the recurrent neural network;

[0058] The real-time monitoring module is used to obtain the real-time operating data of the high-speed railway transformer and use the trained recurrent neural network to monitor the abnormal operating status.

[0059] The present invention also provides a computer device, comprising: a memory and a processor; the memory stores a computer program, and the processor implements the steps of the above-mentioned high-speed railway transformer abnormality monitoring method when executing the computer program.

[0060] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned high-speed railway transformer abnormality monitoring method are implemented.

[0061] Therefore, the present invention adopts the above-mentioned high-speed railway transformer abnormality monitoring method, system, electronic equipment and medium, and the technical effects are as follows:

[0062] (1) Although abnormal monitoring of high-speed railway transformer equipment is very important, its fault data is relatively small, while normal operation monitoring data is large, and the number of samples is unbalanced. The present invention uses a large amount of normal operation monitoring data of transformer equipment to establish a model, and judges the transformer operation status by the probability of actual data in the probability distribution given by the model, so it does not rely on fault data.

[0063] (2) In abnormal monitoring, each model has its own characteristics and focus. It is often difficult to establish an accurate mechanism model, while the data-driven method has poor interpretability. The present invention combines the advantages of both. The time-varying state space model makes the present invention basically consistent with the operating mechanism of the transformer, and the recurrent neural network enables the present invention to make full use of a large amount of monitoring data.

[0064] (3) Although there are many reference materials for modeling high-speed railway transformers, it is still difficult for non-professionals. Although the present invention combines the mechanism model of the transformer, as long as the parameters in the mechanism model are not fixed, the present invention is essentially a data-driven method. This allows non-professionals to use the present invention to monitor transformer abnormalities. Of course, the larger the amount of data and the more accurate the time-varying state space model is, the higher the accuracy of the present invention is. Therefore, in theory, the monitoring accuracy of the present invention can also be continuously improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 The present invention is a flowchart of a method for monitoring abnormality of a high-speed railway transformer. DETAILED DESCRIPTION

[0066] The technical solution of the present invention is further described below through the accompanying drawings and embodiments.

[0067] Unless otherwise defined, technical or scientific terms used in the present invention shall have the common meanings understood by one having ordinary skills in the field to which the present invention belongs.

[0068] Embodiment 1

[0069] The present invention provides a method for monitoring abnormality of a high-speed railway transformer, comprising the following steps:

[0070] Step S1, collecting monitoring data of high-speed railway transformers, including input and output data under no-load and different loads, to form a data set.

[0071] The dataset is represents the input of the kth monitoring sequence at time t, represents the output of the kth monitoring sequence at time t, T k represents the length of the kth monitoring sequence, and L represents that there are L monitoring sequences in total.

[0072] Step S2, constructing a high-speed railway transformer abnormality monitoring model;

[0073] Step S21, establishing a time-varying state space model of a high-speed railway transformer model according to the operating mechanism of the transformer;

[0074]

[0075] Among them, x t is an n-dimensional vector, representing the operating state of the transformer at time t; u t is a d-dimensional vector, representing the input vector of the transformer at time t; z t is an m-dimensional vector, which represents the output vector of the transformer at time t. Under no-load and different loads, the input, state and output of the transformer are all different; Φ t , B t and H t Respectively represent the state transfer matrix, input matrix and output matrix of the transformer at time t; ω t and v t is zero-mean Gaussian noise, whose covariance matrices are Q t and R t ;symbol represents Gaussian distribution;

[0076] Using the symbol Θ t represents the set of parameters in the time-varying state space model, Θ t ={Φ t , B t , H t , Q t , R t},Θ t is a time-varying parameter;

[0077] Step S22: construct a parameter prior matrix set θ p and the parameter mask matrix set Θ m ;

[0078] Construct Θ t The corresponding parameter prior matrix set Θ p ={Φ p , B p , H p , Q p , R p}, Φ p , B p , H p , Q p and R p They are Φ t , B t , H t , Q t and R t The corresponding prior matrix, Θ p and θ t With the following relationship:

[0079]

[0080]

[0081]

[0082]

[0083]

[0084] Among them, the symbol [] ij Represents the element in the i-th row and j-th column of the matrix in [].

[0085] Construct Θ t The corresponding parameter prior matrix set Θ m ={Φ m , B m , H m , Q m , R m}, Φ m , B m , H m , Q m and R m They are Φ t , B t , H t , Q t and R t The corresponding mask matrix, Θ m and θ t With the following relationship:

[0086]

[0087]

[0088]

[0089]

[0090]

[0091] Step S23, combining the time-varying state space model in step S21 to construct a recurrent neural network;

[0092]

[0093] make represents the output set of the recurrent neural network, where and They are the state transfer matrix prediction value, input matrix prediction value, output matrix prediction value, ω given by the recurrent neural network. t The covariance matrix of the predicted value and v t The covariance matrix of the predicted value, The matrix dimensions are respectively t , B t , H t , Q t , R t same.

[0094] Step S3, training a high-speed railway transformer abnormality monitoring model, comprises the following steps:

[0095] Step S31: Input the input data in the data set into the recurrent neural network and calculate the network output

[0096] Step S32: Do the mask calculation and get The specific calculation formula is:

[0097]

[0098]

[0099]

[0100]

[0101]

[0102] in, and They are the state transfer matrix after mask calculation, input matrix, output matrix, ω t The covariance matrix and v t The covariance matrix of Indicates the multiplication of corresponding elements of the matrix;

[0103] Step S33: Use the Kalman filter to calculate the training loss of the network;

[0104]

[0105] in, and represents the best estimate of the state at time t and time t-1, represents the prior estimate of the state at time t, represents the prior estimate of the state covariance matrix at time t, and the superscript symbol T represents the matrix transpose operation. represents the prior estimate of the output vector at time t, S t represents the prior estimate of the covariance matrix of the output vector at time t, K t represents the Kalman gain at time t, ∑ t and∑ t-1They represent the best state covariance matrix estimates at time t and time t-1, respectively. In particular, when t = 0, ∑0=I;

[0106] The training loss of the network is obtained as:

[0107]

[0108] Wherein, log represents the logarithmic function;

[0109] Step S34: Use the training loss to train the recurrent neural network.

[0110] Step S4, obtaining real-time operation data of the high-speed railway transformer and performing abnormal operation status monitoring, includes the following steps:

[0111] Step S41: Transformer input vector u t Input into the trained recurrent neural network and calculate

[0112] Step S42: According to step S32, Do the mask calculation:

[0113]

[0114] Step S43: Given a threshold δ, calculate the and S t , we get the Gaussian probability distribution when If the voltage is 0, it indicates that the transformer is working normally; otherwise, it indicates that the transformer is working abnormally.

[0115] Embodiment 2

[0116] A high-speed railway transformer abnormality monitoring system, comprising:

[0117] Data acquisition module, used to collect input and output data of high-speed railway transformers under no-load and different loads;

[0118] A model building module is used to build a high-speed railway transformer anomaly monitoring model, which includes a time-varying state space model, a parameter prior matrix, a parameter mask matrix and a recurrent neural network;

[0119] A training module, used to train the anomaly monitoring model, calculate the output of the recurrent neural network through input data, perform mask calculation and Kalman filter processing to calculate the training loss of the network, and use the training loss to train the recurrent neural network;

[0120] The real-time monitoring module is used to obtain the real-time operating data of the high-speed railway transformer and use the trained recurrent neural network to monitor the abnormal operating status.

[0121] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

[0122] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0123] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0124] Therefore, the present invention adopts the above-mentioned high-speed railway transformer abnormality monitoring method, system, electronic device and medium, combines the advantages of mechanism model and data-driven method, and through time-varying state space model and recurrent neural network, it not only conforms to the operation mechanism of the transformer, but also can make full use of monitoring data, and improves the accuracy and interpretability of monitoring. In addition, the design of the present invention enables non-professionals to perform transformer abnormality monitoring, and with the increase of data volume and optimization of the model, the monitoring accuracy can be continuously improved.

[0125] It is worth noting that the contents not elaborated in detail in the present invention are all prior art and are well known to those skilled in the art.

[0126] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for monitoring abnormality of a high-speed railway transformer, characterized in that: The following steps are involved: Step S1, collecting monitoring data of high-speed railway transformers, including input and output data under no-load and different loads, to form a data set; Step S2, constructing a high-speed railway transformer abnormality monitoring model; Step S3, training a high-speed railway transformer abnormality monitoring model; Step S4: acquiring real-time operation data of the high-speed railway transformer and performing abnormal operation status monitoring.

2. A high-speed railway transformer abnormality monitoring method according to claim 1, characterized in that: In step S1, the data set is represents the input of the kth monitoring sequence at time t, represents the output of the kth monitoring sequence at time t, T k represents the length of the kth monitoring sequence, and L represents that there are L monitoring sequences in total.

3. A high-speed railway transformer abnormality monitoring method according to claim 2, characterized in that: In step S2, the high-speed railway transformer abnormality monitoring model includes a time-varying state space model of the high-speed railway transformer model, a parameter prior matrix set, a parameter mask matrix set and a recurrent neural network. The construction method of the high-speed railway transformer abnormality monitoring model is as follows: Step S21, establishing a time-varying state space model of a high-speed railway transformer model according to the operating mechanism of the transformer; Among them, x t is an n-dimensional vector, representing the operating state of the transformer at time t; u t is a d-dimensional vector, representing the input vector of the transformer at time t; z t is an m-dimensional vector, representing the output vector of the transformer at time t; Φ t , B t and H t Respectively represent the state transfer matrix, input matrix and output matrix of the transformer at time t; ω t and v t is zero-mean Gaussian noise, whose covariance matrices are Q t and R t ;symbol represents Gaussian distribution; Using the symbol Θ t represents the set of parameters in the time-varying state space model, Θ t ={Φ t , B t , H t , Q t , R t },Θ t is a time-varying parameter; Step S22: construct a parameter prior matrix set θ p and the parameter mask matrix set Θ m ; Step S23, combining the time-varying state space model in step S21 to construct a recurrent neural network; make represents the output set of the recurrent neural network, where and They are the state transfer matrix prediction value, input matrix prediction value, output matrix prediction value, ω given by the recurrent neural network. t The covariance matrix of the predicted value and vt The covariance matrix of the predicted value, The matrix dimensions are respectively t , B t , H t , Q t , R t same.

4. A high-speed railway transformer abnormality monitoring method according to claim 3, characterized in that: In step S22, construct θ t The corresponding parameter prior matrix set Θ p ={Φ p , B p , H p , Q p , R p }, Φ p , B p , H p , Q p and R p They are Φ t , B t , H t , Q t and R t The corresponding prior matrix, Θ p and θ t With the following relationship: Among them, the symbol [ ] ij Represents the element in the i-th row and j-th column of the matrix in [ ].

5. A high-speed railway transformer abnormality monitoring method according to claim 4, characterized in that: In step S22, construct θ t The corresponding parameter prior matrix set Θ m ={Φ m , B m , H m , Q m , R m }, Φ m , B m , H m , Q m and R m They are Φ t , B t , H t , Q t and R t The corresponding mask matrix, Θ m and θ t With the following relationship:

6. A high-speed railway transformer abnormality monitoring method according to claim 5, characterized in that: In step S3, training a high-speed railway transformer abnormality monitoring model includes the following steps: Step S31: Input the input data in the data set into the recurrent neural network and calculate the network output Step S32: Do the mask calculation and get The specific calculation formula is: in, and They are the state transfer matrix after mask calculation, input matrix, output matrix, ω t The covariance matrix and v t The covariance matrix of Indicates the multiplication of corresponding elements of the matrix; Step S33: Use the Kalman filter to calculate the training loss of the network; in, and represents the best estimate of the state at time t and time t-1, represents the prior estimate of the state at time t, represents the prior estimate of the state covariance matrix at time t, and the superscript symbol T represents the matrix transpose operation. represents the prior estimate of the output vector at time t, S t represents the prior estimate of the covariance matrix of the output vector at time t, K t represents the Kalman gain at time t, ∑ t and Σ t-1 Represent the best state covariance matrix estimates at time t and time t-1 respectively; The training loss of the network is obtained as: Wherein, log represents the logarithmic function; Step S34: Use the training loss to train the recurrent neural network.

7. A high-speed railway transformer abnormality monitoring method according to claim 6, characterized in that: In step S4, real-time operation data of the high-speed railway transformer is obtained to monitor abnormal operation status, including the following steps: Step S41: Transformer input vector u t Input into the trained recurrent neural network and calculate Step S42: According to step S32, Do the mask calculation: Step S43: Given a threshold δ, calculate the and S t , we get the Gaussian probability distribution when If the voltage is 0, it indicates that the transformer is working normally; otherwise, it indicates that the transformer is working abnormally.

8. A high-speed railway transformer abnormality monitoring system, characterized in that: include: Data acquisition module, used to collect input and output data of high-speed railway transformers under no-load and different loads; A model building module is used to build a high-speed railway transformer anomaly monitoring model, which includes a time-varying state space model, a parameter prior matrix, a parameter mask matrix and a recurrent neural network; A training module, used to train the anomaly monitoring model, calculate the output of the recurrent neural network through input data, perform mask calculation and Kalman filter processing to calculate the training loss of the network, and use the training loss to train the recurrent neural network; The real-time monitoring module is used to obtain the real-time operating data of the high-speed railway transformer and use the trained recurrent neural network to monitor the abnormal operating status.

9. A computer device comprising: Memory and processor; The memory stores a computer program, characterized in that when the processor executes the computer program, the steps of the high-speed railway transformer abnormality monitoring method described in any one of claims 1-7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the high-speed railway transformer abnormality monitoring method described in any one of claims 1 to 7 are implemented.

Images