Method and system for identifying out-of-range operation of power system machine room
By constructing and training deep learning models based on convolutional layer, pooling layer and fully connected layer, processing video surveillance data of power system computer rooms, automatic identification of over-range operations in power system computer rooms is realized, solving the problems of low manual monitoring efficiency and low reliability in the existing technology, and improving the reliability and accuracy of identification.
Patent Information
- Application Number
- CN202411758115.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-03
- Publication Date
- 2025-03-07
AI Technical Summary
Extra-range operation monitoring of existing power system computer rooms relies on manual labor, and is low in efficiency, reliability and accuracy. Especially when multiple computer rooms are operating simultaneously, real-time and reliability will be greatly reduced.
Deep learning models based on convolutional layer, pooling layer and fully connected layer are adopted to obtain and process historical video surveillance data of power system computer room, and build and train the over-range operation recognition model of power system computer room to achieve automatic recognition.
It realizes automatic identification of over-range operations in the power system computer room, improves the reliability, accuracy and real-time identification, and reduces the dependence of manual monitoring.
Smart Images
Figure CN119580193A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of electrical automation, and in particular relates to a method and a system for identifying out-of-range operations in a power system room. Background Art
[0002] With the development of economy and technology and the improvement of people's living standards, electricity has become an indispensable secondary energy source in people's production and life, bringing endless convenience to people's production and life. Therefore, ensuring the stable and reliable supply of electricity has become one of the most important tasks of the power system.
[0003] The power system room is an important part of the power system; monitoring the operation process of the power system room is of great significance to ensure the safe and reliable operation of the power system. At present, the traditional power system room's out-of-range operation monitoring solution still adopts the manual monitoring solution, that is, the power system staff monitors the entire process of the power system room operation through monitoring or on-site methods. However, this manual monitoring method is not only time-consuming and labor-intensive, inefficient, but also has low reliability and accuracy. Especially for the situation where multiple rooms are operating at the same time, the reliability and real-time performance of the manual monitoring solution will be greatly reduced. Summary of the invention
[0004] One of the purposes of the present invention is to provide a method for identifying out-of-range operations in a power system room with high reliability, good accuracy and good real-time performance.
[0005] A second object of the present invention is to provide a system for implementing the method for identifying out-of-range operations in a power system room.
[0006] The method for identifying out-of-range operation in a power system room provided by the present invention comprises the following steps:
[0007] S1. Obtain historical video surveillance data information of the power system room;
[0008] S2. Process the data information obtained in step S1 to obtain a training data set;
[0009] S3. Based on the convolutional layer, pooling layer and fully connected layer, a primary model for identifying out-of-range operations in the power system room is constructed;
[0010] S4. Using the training data set obtained in step S2, the primary model for identifying out-of-range operation of the power system room constructed in step S3 is trained to obtain an out-of-range operation identification model of the power system room;
[0011] S5. Using the power system room out-of-range operation identification model obtained in step S5, actual power system room out-of-range operation identification is performed.
[0012] Step S2 processes the data information obtained in step S1 to obtain a training data set, which specifically includes the following steps:
[0013] Marking and position division of the data information obtained in step S1;
[0014] After the data is labeled and divided into positions, data cleaning and sample expansion are performed to obtain a training data set.
[0015] The marking and location division of the data information obtained in step S1 specifically includes the following steps:
[0016] Extract frames from the video surveillance picture acquired in step S1 to obtain key frame picture data, and mark them; the marking includes cabinet closed and cabinet opened;
[0017] The acquired video surveillance picture is divided according to the cabinet position to obtain the position of each cabinet in the video surveillance picture, and the number of each cabinet is associated with the cabinet position in the video surveillance picture.
[0018] The data after being marked and divided by position is cleaned and sample expanded, specifically including the following steps:
[0019] Perform data cleaning on the data after labeling and location division;
[0020] The cleaned data is subjected to sample expansion; the sample expansion includes image translation, image expansion, image rotation and image symmetry processing.
[0021] The primary model for identifying out-of-range operations in a power system room is constructed based on the convolutional layer, the pooling layer and the fully connected layer as described in step S3, and specifically includes the following steps:
[0022] The constructed primary model for identifying out-of-range operations in a power system room includes a first convolutional layer, a second convolutional layer, a first maximum pooling layer, a third convolutional layer, a fourth convolutional layer, a second maximum pooling layer, a fifth convolutional layer, a sixth convolutional layer, a seventh convolutional layer, a third maximum pooling layer, an eighth convolutional layer, a ninth convolutional layer, a tenth convolutional layer, a fourth maximum pooling layer, a first fully connected module, a second fully connected module, and a third fully connected module;
[0023] The first convolution layer, the second convolution layer, the first maximum pooling layer, the third convolution layer, the fourth convolution layer, the second maximum pooling layer, the fifth convolution layer, the sixth convolution layer, the seventh convolution layer, the third maximum pooling layer, the eighth convolution layer, the ninth convolution layer, the tenth convolution layer, the fourth maximum pooling layer, the first fully connected module, the second fully connected module and the third fully connected module are connected in series in sequence;
[0024] The first to tenth convolutional layers are used to extract feature information of the input image data;
[0025] The first to fourth maximum pooling layers are used to reduce the dimension of the input image data;
[0026] The full connection module is used to output the corresponding out-of-range operation judgment result according to the input data information; the out-of-range operation includes cabinet closing and cabinet opening.
[0027] The convolution kernels of the first and second convolution layers are both 3×3, with a stride of 1 and 64 output channels; the convolution kernels of the third and fourth convolution layers are both 3×3, with a stride of 1 and 128 output channels; the convolution kernels of the fifth, sixth and seventh convolution layers are all 3×3, with a stride of 1 and 256 output channels; the convolution kernels of the eighth, ninth and tenth convolution layers are all 3×3, with a stride of 1 and 512 output channels.
[0028] The pooling kernels of the first to fourth maximum pooling layers are all 2×2, and the step sizes are all 2.
[0029] The first fully connected module includes a first fully connected layer, a first ReLU activation function layer and a first Dropout layer; the first fully connected module includes 4096 neurons; the second fully connected module includes a second fully connected layer, a second ReLU activation function layer and a second Dropout layer; the second fully connected module includes 4096 neurons; the third fully connected module includes a third fully connected layer, a third softmax activation function layer and a third Dropout layer; the third fully connected module includes 2 neurons, corresponding to the judgment result of out-of-range operation; the ReLU activation function layer and the Dropout layer are both used to prevent model overfitting.
[0030] The training described in step S4 specifically includes the following steps:
[0031] During training, forward propagation is used to input the training data into the model, obtain the model output, and calculate the gap between the model output and the actual label;
[0032] Calculate the loss function, obtain the corresponding gradient information, and use back propagation to back propagate the gradient information from the third fully connected module to the first convolutional layer layer by layer;
[0033] Update the parameters of the model.
[0034] The present invention also provides a system for implementing the method for identifying out-of-range operations in a power system room, comprising a data acquisition module, a data processing module, a model construction module, a model training module and an operation identification module; the data acquisition module, the data processing module, the model construction module, the model training module and the operation identification module are connected in series in sequence; the data acquisition module is used to acquire historical video monitoring data information of the power system room, and upload the data information to the data processing module; the data processing module is used to process the acquired data information according to the received data information, obtain a training data set, and upload the data information to the model construction module; the model construction module is used to construct a primary model for identifying out-of-range operations in a power system room based on a convolutional layer, a pooling layer and a fully connected layer according to the received data information, and upload the data information to the model training module; the model training module is used to train the constructed primary model for identifying out-of-range operations in a power system room according to the received data information, using the obtained training data set, to obtain an out-of-range operation identification model for a power system room, and upload the data information to the operation identification module; the operation identification module is used to identify the actual out-of-range operations in a power system room according to the received data information and using the obtained out-of-range operation identification model for a power system room.
[0035] The method and system for identifying out-of-range operations in a power system computer room provided by the present invention, through the acquisition and processing of computer room operation video monitoring data, and the construction and training of an out-of-range operation identification model for the computer room, not only realizes the automatic identification of out-of-range operations in a power system computer room, but also has higher reliability, better accuracy, and better real-time performance. BRIEF DESCRIPTION OF THE DRAWINGS
[0036] Figure 1 The figure is a schematic diagram of the method flow of the present invention.
[0037] Figure 2 Schematic diagram of the functional modules of the system of the present invention. DETAILED DESCRIPTION
[0038] like Figure 1 The method flow chart of the method of the present invention is shown as follows: The method for identifying out-of-range operations in a power system room disclosed in the present invention comprises the following steps:
[0039] S1. Obtain historical video surveillance data information of the power system room;
[0040] S2. Process the data information obtained in step S1 to obtain a training data set; specifically, the steps include:
[0041] The data information obtained in step S1 is labeled and divided into positions; specifically, the steps include:
[0042] Extract frames from the video surveillance picture acquired in step S1 to obtain key frame picture data, and mark them; the marking includes cabinet closed and cabinet opened;
[0043] The acquired video surveillance images are divided according to the cabinet positions to obtain the position of each cabinet in the video surveillance images, and the number of each cabinet is associated with the cabinet position in the video surveillance images. At the same time, the association results can be stored in the database;
[0044] After the data is labeled and divided into positions, data cleaning and sample expansion are performed to obtain a training data set; the specific steps include the following:
[0045] Perform data cleaning on the data after labeling and location division;
[0046] The cleaned data is subjected to sample expansion; the sample expansion includes image translation, image expansion, image rotation and image symmetry processing;
[0047] Finally, the training data set is constructed;
[0048] S3. Based on the convolutional layer, pooling layer and fully connected layer, a primary model for identifying out-of-range operations in the power system room is constructed; specifically, the following steps are included:
[0049] The constructed primary model for identifying out-of-range operations in a power system room includes a first convolutional layer, a second convolutional layer, a first maximum pooling layer, a third convolutional layer, a fourth convolutional layer, a second maximum pooling layer, a fifth convolutional layer, a sixth convolutional layer, a seventh convolutional layer, a third maximum pooling layer, an eighth convolutional layer, a ninth convolutional layer, a tenth convolutional layer, a fourth maximum pooling layer, a first fully connected module, a second fully connected module, and a third fully connected module;
[0050] The first convolution layer, the second convolution layer, the first maximum pooling layer, the third convolution layer, the fourth convolution layer, the second maximum pooling layer, the fifth convolution layer, the sixth convolution layer, the seventh convolution layer, the third maximum pooling layer, the eighth convolution layer, the ninth convolution layer, the tenth convolution layer, the fourth maximum pooling layer, the first fully connected module, the second fully connected module and the third fully connected module are connected in series in sequence;
[0051] The first to tenth convolutional layers are used to extract feature information of the input image data;
[0052] The first to fourth maximum pooling layers are used to reduce the dimension of the input image data;
[0053] The full connection module is used to output the corresponding out-of-range operation judgment result according to the input data information; the out-of-range operation includes cabinet closing and cabinet opening.
[0054] When implementing:
[0055] The convolution kernels of the first and second convolution layers are both 3×3, with a stride of 1 and 64 output channels; the convolution kernels of the third and fourth convolution layers are both 3×3, with a stride of 1 and 128 output channels; the convolution kernels of the fifth, sixth and seventh convolution layers are all 3×3, with a stride of 1 and 256 output channels; the convolution kernels of the eighth, ninth and tenth convolution layers are all 3×3, with a stride of 1 and 512 output channels;
[0056] The pooling kernels of the first to fourth maximum pooling layers are all 2×2, and the step size is 2;
[0057] The first fully connected module includes a first fully connected layer, a first ReLU activation function layer and a first Dropout layer; the first fully connected module includes 4096 neurons; the second fully connected module includes a second fully connected layer, a second ReLU activation function layer and a second Dropout layer; the second fully connected module includes 4096 neurons; the third fully connected module includes a third fully connected layer, a third softmax activation function layer and a third Dropout layer; the third fully connected module includes 2 neurons, corresponding to the judgment result of the out-of-range operation; the ReLU activation function layer and the Dropout layer are both used to prevent the model from overfitting;
[0058] S4. Using the training data set obtained in step S2, the primary model for identifying out-of-range operation of the power system room constructed in step S3 is trained to obtain an out-of-range operation identification model of the power system room;
[0059] In specific implementation, the training process includes the following steps:
[0060] During training, forward propagation is used to input the training data into the model, obtain the model output, and calculate the gap between the model output and the actual label;
[0061] Calculate the loss function, obtain the corresponding gradient information, and use back propagation to back propagate the gradient information from the third fully connected module to the first convolutional layer layer by layer;
[0062] Update the parameters of the model, including the weights and biases of each neuron;
[0063] S5. Using the power system room out-of-range operation identification model obtained in step S5, actual power system room out-of-range operation identification is performed.
[0064] like Figure 2The functional module schematic diagram of the system of the present invention is shown as follows: the system disclosed in the present invention for realizing the method for identifying out-of-range operations in a power system room comprises a data acquisition module, a data processing module, a model construction module, a model training module and an operation identification module; the data acquisition module, the data processing module, the model construction module, the model training module and the operation identification module are connected in series in sequence; the data acquisition module is used to acquire historical video monitoring data information of the power system room, and upload the data information to the data processing module; the data processing module is used to process the acquired data information according to the received data information, obtain a training data set, and upload the data information to the model construction module; the model construction module is used to construct a primary model for identifying out-of-range operations in a power system room based on a convolutional layer, a pooling layer and a fully connected layer according to the received data information, and upload the data information to the model training module; the model training module is used to train the constructed primary model for identifying out-of-range operations in a power system room using the obtained training data set according to the received data information, obtain an out-of-range operation identification model for a power system room, and upload the data information to the operation identification module; the operation identification module is used to identify the actual out-of-range operations in a power system room using the obtained out-of-range operation identification model for a power system room according to the received data information.
Claims
1. A method for identifying out-of-range operations in a power system room, comprising the following steps: S1. Obtain historical video surveillance data information of the power system room; S2. Process the data information obtained in step S1 to obtain a training data set; S3. Based on the convolutional layer, pooling layer and fully connected layer, a primary model for identifying out-of-range operations in the power system room is constructed; S4. Using the training data set obtained in step S2, the primary model for identifying out-of-range operation of the power system room constructed in step S3 is trained to obtain an out-of-range operation identification model of the power system room; S5. Using the power system room out-of-range operation identification model obtained in step S5, actual power system room out-of-range operation identification is performed.
2. The method for identifying out-of-range operations in a power system room according to claim 1 is characterized in that Step S2 processes the data information obtained in step S1 to obtain a training data set, which specifically includes the following steps: Marking and position division of the data information obtained in step S1; After the data is labeled and divided into positions, data cleaning and sample expansion are performed to obtain a training data set.
3. The method for identifying out-of-range operations in a power system room according to claim 2 is characterized in that The marking and location division of the data information obtained in step S1 specifically includes the following steps: Extract frames from the video surveillance picture acquired in step S1 to obtain key frame picture data, and mark them; the marking includes cabinet closed and cabinet opened; The acquired video surveillance picture is divided according to the cabinet position to obtain the position of each cabinet in the video surveillance picture, and the number of each cabinet is associated with the cabinet position in the video surveillance picture.
4. The method for identifying out-of-range operations in a power system room according to claim 3 is characterized in that The data after being marked and divided by position is cleaned and sample expanded, specifically including the following steps: Perform data cleaning on the data after labeling and location division; The cleaned data is subjected to sample expansion; the sample expansion includes image translation, image expansion, image rotation and image symmetry processing.
5. The method for identifying out-of-range operations in a power system room according to claim 1 is characterized in that The primary model for identifying out-of-range operations in a power system room is constructed based on the convolutional layer, the pooling layer and the fully connected layer as described in step S3, and specifically includes the following steps: The constructed primary model for identifying out-of-range operations in a power system room includes a first convolutional layer, a second convolutional layer, a first maximum pooling layer, a third convolutional layer, a fourth convolutional layer, a second maximum pooling layer, a fifth convolutional layer, a sixth convolutional layer, a seventh convolutional layer, a third maximum pooling layer, an eighth convolutional layer, a ninth convolutional layer, a tenth convolutional layer, a fourth maximum pooling layer, a first fully connected module, a second fully connected module, and a third fully connected module; The first convolution layer, the second convolution layer, the first maximum pooling layer, the third convolution layer, the fourth convolution layer, the second maximum pooling layer, the fifth convolution layer, the sixth convolution layer, the seventh convolution layer, the third maximum pooling layer, the eighth convolution layer, the ninth convolution layer, the tenth convolution layer, the fourth maximum pooling layer, the first fully connected module, the second fully connected module and the third fully connected module are connected in series in sequence; The first to tenth convolutional layers are used to extract feature information of the input image data; The first to fourth maximum pooling layers are used to reduce the dimension of the input image data; The full connection module is used to output the corresponding out-of-range operation judgment result according to the input data information; the out-of-range operation includes cabinet closing and cabinet opening.
6. The method for identifying out-of-range operations in a power system room according to claim 5 is characterized in that The convolution kernels of the first and second convolution layers are both 3×3, with a stride of 1 and 64 output channels; the convolution kernels of the third and fourth convolution layers are both 3×3, with a stride of 1 and 128 output channels; the convolution kernels of the fifth, sixth and seventh convolution layers are all 3×3, with a stride of 1 and 256 output channels; the convolution kernels of the eighth, ninth and tenth convolution layers are all 3×3, with a stride of 1 and 512 output channels.
7. The method for identifying out-of-range operations in a power system room according to claim 5 is characterized in that The pooling kernels of the first to fourth maximum pooling layers are all 2×2, and the step sizes are all 2.
8. The method for identifying out-of-range operations in a power system room according to claim 5 is characterized in that The first fully connected module includes a first fully connected layer, a first ReLU activation function layer and a first Dropout layer; the first fully connected module includes 4096 neurons; the second fully connected module includes a second fully connected layer, a second ReLU activation function layer and a second Dropout layer; the second fully connected module includes 4096 neurons; the third fully connected module includes a third fully connected layer, a third softmax activation function layer and a third Dropout layer;; the third fully connected module includes 2 neurons, corresponding to the judgment results of out-of-range operations; the Dropout layers are all used to prevent model overfitting.
9. The method for identifying out-of-range operations in a power system room according to claim 5, characterized in that The training described in step S4 specifically includes the following steps: During training, forward propagation is used to input the training data into the model, obtain the model output, and calculate the gap between the model output and the actual label; Calculate the loss function, obtain the corresponding gradient information, and use back propagation to back propagate the gradient information from the third fully connected module to the first convolutional layer layer by layer; Update the parameters of the model.
10. A system for implementing the method for identifying out-of-range operations in a power system room as claimed in any one of claims 1 to 9, characterized in that It includes a data acquisition module, a data processing module, a model building module, a model training module and an operation identification module; the data acquisition module, the data processing module, the model building module, the model training module and the operation identification module are connected in series in sequence; the data acquisition module is used to obtain the historical video monitoring data information of the power system room, and upload the data information to the data processing module; The data processing module is used to process the acquired data information according to the received data information, obtain the training data set, and upload the data information to the model construction module; The model building module is used to build a primary model for identifying out-of-range operations in the power system room based on the received data information, based on the convolution layer, the pooling layer and the fully connected layer, and upload the data information to the model training module; The model training module is used to train the constructed primary model for identifying out-of-range operations in the power system room using the obtained training data set according to the received data information, obtain the out-of-range operation identification model for the power system room, and upload the data information to the operation identification module; The operation identification module is used to identify the actual out-of-range operation of the power system room based on the received data information and using the obtained out-of-range operation identification model of the power system room.