F5G all-optical substation real-time monitoring method and system fused with neural network, and medium

Abstract

The invention provides an F5G all-optical substation real-time monitoring method and system fused with a neural network, and a medium. The method comprises the following steps: S1, obtaining video data shot by cameras installed at the entrance and exit of a substation; s2, according to the video data in the step S1, judging whether an entering person is an illegal person; if the entering person is the illegal person, sending out warning information; and if the entering person is a worker, the warning information is not sent out. The method can achieve the real-time classification and judgment of workers for the transformer substation through the neural network.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent substations, and in particular to a real-time monitoring method, system and medium for an F5G all-optical substation integrating a neural network. Background Art

[0002] As an important part of the power system, smart substations integrate modern communication and automation technologies, use advanced digital control equipment, and have the ability of remote monitoring, remote operation and remote maintenance, achieving high intelligence and autonomous decision-making. The application of F5G all-optical network in smart substations is an important part of the digital and intelligent transformation of the power industry. F5G, the fifth generation fixed network, represents the latest progress in fixed network technology and aims to build a fixed network infrastructure with ultra-wide bandwidth, all-optical connection, ultimate experience, high security and intelligent management. Summary of the invention

[0003] The present invention aims to at least solve the technical problems existing in the prior art, and particularly innovatively proposes a F5G all-optical substation real-time monitoring method, system and medium integrating a neural network.

[0004] In order to achieve the above-mentioned object of the present invention, the present invention provides a F5G all-optical substation real-time monitoring method integrating a neural network, comprising the following steps:

[0005] S1, obtaining video data captured by cameras installed at the entrance and exit of the substation;

[0006] S2, judging whether the person entering is an illegal person according to the video data in step S1:

[0007] If the person entering is an illegal person, a warning message will be issued;

[0008] If the person entering is a staff member, no warning message will be issued.

[0009] S21, converting the video data into image data;

[0010] S22, preprocessing the image data;

[0011] S23, constructing a preprocessed face image to realize feature extraction;

[0012] S24, obtaining a facial feature sample according to a facial feature sample method;

[0013] S25, training the face image according to the neural network to learn and optimize the face recognition model;

[0014] S26, adjusting the convolutional neural network parameters according to the gradient algorithm;

[0015] S27, input the training results into the classifier for classification prediction to achieve the discrimination between illegal personnel and staff.

[0016] In a preferred embodiment of the present invention, the video data is obtained in step S1 through the F5G all-optical network.

[0017] In a preferred embodiment of the present invention, step S21 includes:

[0018] R r =T r ×s r ,

[0019] Among them, R r is the number of video data in the rth segment;

[0020] T r is the duration of the rth video data, in seconds;

[0021] s r Indicates the number of frames of the rth segment of video data; the unit is frames per second;

[0022] In a preferred embodiment of the present invention, in step S22, the method for preprocessing the facial images entering and exiting the smart substation acquired by using the F5G all-optical network includes one or any combination of grayscale, denoising, and normalization to improve image quality and reduce noise interference.

[0023] In a preferred embodiment of the present invention, the method for constructing the preprocessed face image to realize feature extraction in step S23 includes:

[0024] The integer value of the value where the face is not collected is 0, and the value where the face is collected is 1. The integers 0 and 1 are used for encoding. Two integer values ​​are assigned to the dataset to set the mapping of the binary vector length 2, that is, "face not collected" is marked as 0, and "face collected" is marked as 1. If the dataset sequence collects a face, its integer code is 1.

[0025] In a preferred embodiment of the present invention, the facial feature sample method in step S24 is a convolution and pooling method;

[0026] The fully connected layer uses the softmax function to classify the image, the pooling layer uses the RELU function to remove the negative values ​​of the grouped image, and the convolutional layer uses the loss function to regularize the weight of the image.

[0027] In a preferred embodiment of the present invention, step S25 includes:

[0028]

[0029] in, Represents the bias term of the nth feature map in the mth layer;

[0030] Represents the convolution kernel weight between the nth feature map in the mth layer and the jth feature map in the m-1th layer;

[0031] Represents the output of the jth feature map in the m-1th layer;

[0032] represents the output of the nth feature map in the mth layer.

[0033] In a preferred embodiment of the present invention, the loss function is used to learn the optimization model in step S25:

[0034]

[0035] Among them, U represents the loss function;

[0036] G represents the number of samples;

[0037] x i represents the true value;

[0038] Represents the predicted value.

[0039] In a preferred embodiment of the present invention, the method for adjusting the convolutional neural network parameters according to the gradient algorithm in step S26 includes:

[0040] m t =β 1 m t-1 +(1-β 1 ) t ,

[0041] Among them, m t Represents the moving average of the gradient of the current iteration;

[0042] β 1 Represents the first attenuation rate, which is 0.95 here;

[0043] m t-1 Represents the moving average of the gradient at the t-1 iteration time;

[0044] g t Represents the gradient of the current iteration;

[0045]

[0046] v t Represents the moving average of the square of the gradient of the current iteration;

[0047] β 2 represents the second attenuation rate, which is 0.996 here;

[0048] v t-1 Represents the moving average of the squared gradient at the t-1 iteration;

[0049] g t Represents the gradient of the current iteration;

[0050]

[0051] Represents the modified moving average of the gradient of the current iteration;

[0052] m t Represents the moving average of the gradient of the current iteration;

[0053] Indicates the gradient correction value;

[0054]

[0055] Represents the modified moving average of the squared gradient of the current iteration;

[0056] v t Represents the moving average of the squared gradient at t iterations;

[0057] Represents the squared gradient correction value;

[0058]

[0059] θ t+1 Network parameters indicating the current number of iterations;

[0060] θ t Network parameters representing the last iteration number

[0061] α represents the learning rate;

[0062] Represents the modified moving average of the gradient of the current iteration;

[0063] Represents the modified moving average of the squared gradient of the current iteration;

[0064] ε represents a very small value, which is used to prevent the denominator from being zero. Here, it is 10 -8 .

[0065] In a preferred embodiment of the present invention, step S27 includes:

[0066] The classifier used is the Bayesian classifier;

[0067] Assume that Q is a training set of samples, each sample has its class label, and there are q classes in total, namely D 1 , D 2 , D 3 ,……,D q , each sample is represented by an n-dimensional vector, H = {h 1 ,h 2 ,h 3 ,…,h n}, describing n measured values ​​of n attributes respectively, A 1 , A 2 , A 3 ,……,A n ;

[0068] Given a sample H, the classifier will predict that sample H belongs to the category with the highest posterior probability, conditional on H, that is, predict that sample H belongs to D i , and only when P(D i |H)>P(D j |H), 1≤j≤q, j≠i;

[0069] Thus we find the maximum P(D i |H) class, P(D i |H) The largest class D i It is called the maximum a posteriori hypothesis, and Bayes' theorem is:

[0070]

[0071] Since P(H) is the same for all classes, only P(H|D i )P(D i ) requires a maximum, if the class The prior probability P(D i ) is unknown, then it is usually assumed that these classes are equally likely, that is, P(D 1 )=P(D 2 )=P(D 3 )=…=P(D q ), so the maximum is P(H|D i ), otherwise the maximum value is P(H|D i )P(D i ).

[0072] The present invention also discloses a computer system, comprising:

[0073] processor;

[0074] a memory for storing processor-executable instructions;

[0075] Among them, the processor is configured to implement the F5G all-optical substation real-time monitoring method of the fused neural network when executing the executable instructions.

[0076] The present invention also discloses a computer-readable storage medium, comprising:

[0077] a memory having a computer program stored thereon;

[0078] A processor is used to execute the program in the memory to implement the F5G all-optical substation real-time monitoring method integrating the neural network.

[0079] In summary, due to the adoption of the above technical solution, the present invention can identify people entering and leaving the smart substation, thereby improving accuracy and efficiency.

[0080] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0082] Figure 1 It is a schematic block diagram of the process of the present invention. DETAILED DESCRIPTION

[0083] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0084] The present invention provides a real-time monitoring method for F5G all-optical substations integrating neural networks, such as Figure 1 As shown, the following steps are included:

[0085] S1, obtaining video data captured by cameras installed at the entrance and exit of the substation;

[0086] S2, judging whether the person entering is an illegal person according to the video data in step S1:

[0087] If the person entering is an illegal person, a warning message will be issued;

[0088] If the person entering is a staff member, no warning message will be issued.

[0089] S21, converting the video data into image data;

[0090] S22, preprocessing the image data; S3, constructing the preprocessed face image to realize feature extraction;

[0091] S4, obtaining a facial feature sample according to a facial feature sample method;

[0092] S5, training face images according to the neural network to learn and optimize the face recognition model;

[0093] S6, adjusting the convolutional neural network parameters according to the gradient algorithm;

[0094] S7, input the training results into the classifier for classification prediction to distinguish between illegal personnel and staff.

[0095] In a preferred embodiment of the present invention, the video data is obtained in step S1 through the F5G all-optical network.

[0096] In a preferred embodiment of the present invention, step S21 includes:

[0097] R r =T r ×s r ,

[0098] Among them, R r is the number of video data in the rth segment;

[0099] T r is the duration of the rth video data, in seconds;

[0100] s r Indicates the number of frames of the rth segment of video data; the unit is frames per second;

[0101] In a preferred embodiment of the present invention, in step S22, the method for preprocessing the facial images entering and exiting the smart substation acquired by using the F5G all-optical network includes one or any combination of grayscale, denoising, and normalization to improve image quality and reduce noise interference.

[0102] In a preferred embodiment of the present invention, the method for constructing the preprocessed face image to realize feature extraction in step S23 includes:

[0103] The integer value of the value where the face is not collected is 0, and the value where the face is collected is 1. The integers 0 and 1 are used for encoding. Two integer values ​​are assigned to the dataset to set the mapping of the binary vector length 2, that is, "face not collected" is marked as 0, and "face collected" is marked as 1. If the dataset sequence collects a face, its integer code is 1.

[0104] In a preferred embodiment of the present invention, the facial feature sample method in step S24 is a convolution and pooling method;

[0105] The fully connected layer uses the softmax function to classify the image, the pooling layer uses the RELU function to remove the negative values ​​of the grouped image, and the convolutional layer uses the loss function to regularize the weight of the image.

[0106] In a preferred embodiment of the present invention, step S25 includes:

[0107]

[0108] in, Represents the bias term of the nth feature map in the mth layer;

[0109] Represents the convolution kernel weight between the nth feature map in the mth layer and the jth feature map in the m-1th layer;

[0110] Represents the output of the jth feature map in the m-1th layer;

[0111] represents the output of the nth feature map in the mth layer.

[0112] In a preferred embodiment of the present invention, the loss function is used to learn the optimization model in step S25:

[0113]

[0114] Among them, U represents the loss function;

[0115] G represents the number of samples;

[0116] x i represents the true value;

[0117] Represents the predicted value.

[0118] In a preferred embodiment of the present invention, the method for adjusting the convolutional neural network parameters according to the gradient algorithm in step S26 includes:

[0119] m t =β 1 mt-1 +(1-β 1 ) t ,

[0120] Among them, m t Represents the moving average of the gradient of the current iteration;

[0121] β 1 Represents the first attenuation rate, which is 0.95 here;

[0122] m t-1 Represents the moving average of the gradient at the t-1 iteration time;

[0123] g t Represents the gradient of the current iteration;

[0124]

[0125] v t Represents the moving average of the square of the gradient of the current iteration;

[0126] β 2 represents the second attenuation rate, which is 0.996 here;

[0127] v t-1 Represents the moving average of the squared gradient at the t-1 iteration;

[0128] g t Represents the gradient of the current iteration;

[0129]

[0130] Represents the modified moving average of the gradient of the current iteration;

[0131] m t Represents the moving average of the gradient of the current iteration;

[0132] Indicates the gradient correction value;

[0133]

[0134] Represents the modified moving average of the squared gradient of the current iteration;

[0135] v t Represents the moving average of the squared gradient at t iterations;

[0136] Represents the squared gradient correction value;

[0137]

[0138] θ t+1 Network parameters indicating the current number of iterations;

[0139] θ t Network parameters representing the last iteration number

[0140] α represents the learning rate;

[0141] Represents the modified moving average of the gradient of the current iteration;

[0142] Represents the modified moving average of the squared gradient of the current iteration;

[0143] ε represents a very small value, which is used to prevent the denominator from being zero. Here, it is 10 -8 .

[0144] In a preferred embodiment of the present invention, step S27 includes:

[0145] The classifier used is the Bayesian classifier;

[0146] Assume that Q is a training set of samples, each sample has its class label, and there are q classes in total, namely D 1 , D 2 , D 3 ,……,D q , each sample is represented by an n-dimensional vector, H = {h 1 ,h 2 ,h 3 ,…,h n}, describing n measured values ​​of n attributes respectively, A 1 , A 2 , A 3 ,……,A n ;

[0147] Given a sample H, the classifier will predict that sample H belongs to the category with the highest posterior probability, conditional on H, that is, predicting that sample H belongs to Di, and only when P(D i |H)>P(D j |H), 1≤j≤q, j≠i;

[0148] Thus we find the maximum P(D i |H) class, P(D i |H) The largest class D i It is called the maximum a posteriori hypothesis, and Bayes' theorem is:

[0149]

[0150] Since P(H) is the same for all classes, only P(H|D i )P(D i ) requires a maximum, if the class The prior probability P(D i ) is unknown, then it is usually assumed that these classes are equally likely, that is, P(D 1 )=P(D 2 )=P(D 3 )=...=P(D q ), so the maximum is P(H|D i ), otherwise the maximum value is P(H | D i )P(D i ).

[0151] The present invention also discloses a computer system, comprising:

[0152] processor;

[0153] a memory for storing processor-executable instructions;

[0154] Among them, the processor is configured to implement the F5G all-optical substation real-time monitoring method of the fused neural network when executing the executable instructions.

[0155] The present invention also discloses a computer-readable storage medium, comprising:

[0156] a memory having a computer program stored thereon;

[0157] A processor is used to execute the program in the memory to implement the F5G all-optical substation real-time monitoring method integrating the neural network.

[0158] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A real-time monitoring method for F5G all-optical substation integrating neural network, characterized in that: The following steps are involved: S1, obtaining video data captured by cameras installed at the entrance and exit of the substation; S2, judging whether the person entering is an illegal person according to the video data in step S1: If the person entering is an illegal person, a warning message will be issued; If the person entering is a staff member, no warning message will be issued.

2. The F5G all-optical substation real-time monitoring method of fusion neural network according to claim 1 is characterized in that: Step S2 includes the following steps: S21, converting the video data into image data; S22, preprocessing the image data; S23, constructing a preprocessed face image to realize feature extraction; S24, obtaining a facial feature sample according to a facial feature sample method; S25, performing face image training according to a neural network to learn and optimize a face recognition model; S26, adjusting the convolutional neural network parameters according to the gradient algorithm; S27, input the training results into the classifier for classification prediction to achieve the discrimination between illegal personnel and staff.

3. The F5G all-optical substation real-time monitoring method of fusion neural network according to claim 1 is characterized in that: In step S1, the video data is obtained through the F5G all-optical network.

4. The F5G all-optical substation real-time monitoring method of fusion neural network according to claim 1 is characterized in that: Step S21 includes: R r =T r ×s r , Among them, R r is the number of video data in the rth segment; T r is the duration of the rth video data, in seconds; s r Indicates the number of frames in the rth segment of video data; the unit is frames per second.

5. The F5G all-optical substation real-time monitoring method of fusion neural network according to claim 1 is characterized in that: In step S22, the method for preprocessing the facial images entering and exiting the smart substation acquired by using the F5G all-optical network includes one or any combination of grayscale, denoising, and normalization to improve image quality and reduce noise interference.

6. The F5G all-optical substation real-time monitoring method of fusion neural network according to claim 1 is characterized in that: Step S25 includes: in, Represents the bias term of the nth feature map in the mth layer; Represents the convolution kernel weight between the nth feature map in the mth layer and the jth feature map in the m-1th layer; Represents the output of the jth feature map in the m-1th layer; represents the output of the nth feature map in the mth layer.

7. The F5G all-optical substation real-time monitoring method integrating neural network according to claim 1 is characterized in that: In step S25, the loss function is used to learn the optimization model: Among them, U represents the loss function; G represents the number of samples; x i represents the true value; Represents the predicted value.

8. The F5G all-optical substation real-time monitoring method of fusion neural network according to claim 1 is characterized in that: The method for adjusting the convolutional neural network parameters according to the gradient algorithm in step S26 includes: m t =β1m t-1 +(1-β1)g t , Among them, m t Represents the moving average of the gradient of the current iteration; β1 represents the first decay rate, which is 0.95 here; m t-1 Represents the moving average of the gradient at the t-1 iteration time; g t Represents the gradient of the current iteration; v t Represents the moving average of the square of the gradient of the current iteration; β2 represents the second decay rate, which is 0.996 here; v t-1 represents the moving average of the squared gradient at the t-1th iteration; g t Represents the gradient of the current iteration; Represents the modified moving average of the gradient of the current iteration; m t Represents the moving average of the gradient of the current iteration; Indicates the gradient correction value; represents the modified moving average of the square of the gradient of the current iteration; v t Represents the moving average of the squared gradient at t iterations; Represents the squared gradient correction value; θ t+1 Network parameters indicating the current number of iterations; θ t The network parameter α, which represents the number of previous iterations, represents the learning rate; Represents the modified moving average of the gradient of the current iteration; represents the modified moving average of the square of the gradient of the current iteration; ε represents a very small value to prevent the denominator from being zero, here it is 10 -8 .

9. A computer system, characterized in that: include: processor; a memory for storing processor-executable instructions; Wherein, the processor is configured to implement the F5G all-optical substation real-time monitoring method of the fused neural network described in one of claims 1 to 8 when executing the executable instructions.

10. A computer-readable storage medium, characterized in that: include: a memory having a computer program stored thereon; A processor is used to execute the program in the memory to implement the F5G all-optical substation real-time monitoring method integrating a neural network as described in any one of claims 1 to 8.

Images