CN117912023A - A method and terminal for automatically identifying electrical drawings of a hydropower plant - Google Patents

Abstract

The invention discloses an automatic identification method and terminal for an electrical drawing of a hydropower plant, which are characterized in that a segmentation and identification model is used for segmenting and identifying the electrical drawing of the hydropower plant to be identified to obtain a primitive, a text and a category of the electrical drawing of the hydropower plant to be identified, the primitive is identified based on the category to obtain primitive information, the text is subjected to semantic identification to obtain text information, the categories of the primitive, the text and the drawing in the electrical drawing of the hydropower plant are automatically identified and separated by the segmentation and identification model, then the primitive identification and the semantic identification are respectively carried out, automatic, comprehensive and accurate electrical drawing identification can be realized, and the output identification result can be used for further analysis and processing, so that informatization and intellectualization of the electrical drawing are effectively realized.

Description

A method and terminal for automatically identifying electrical drawings of a hydropower plant

Technical Field

The present invention relates to the technical field of image recognition, and in particular to an automatic recognition method and terminal for electrical drawings of a hydropower plant.

Background technique

Hydropower enterprises have accumulated a large number of drawings in decades of development, but a large number of electrical drawings of hydropower plants have also caused difficulties in data processing, especially the consistency, accuracy and timeliness of electrical drawing data. Therefore, it has become an urgent technical challenge to realize the digitization and informatization of electrical drawings of hydropower plants and facilitate the retrieval and extraction of drawings by staff during use.

At present, the main method for digitizing electrical drawings in hydropower plants is drawing scanning. Although the drawings are transferred from physical form to digital form, this form change is only in terms of carrier rather than information. Although it is conducive to storage, reference and transmission, it does not realize the informatization and intelligence of drawing data, and cannot realize further processing and handling of drawing information.

Summary of the invention

The technical problem to be solved by the present invention is to provide a method and a terminal for automatically identifying electrical drawings of a hydropower plant, which can effectively realize the informatization and intelligence of the electrical drawings.

In order to solve the above technical problems, the technical solution adopted by the present invention is:

A method for automatically identifying electrical drawings of a hydropower plant, comprising the steps of:

Obtain electrical drawings of the hydropower plant to be identified;

Use the segmentation and recognition model to segment and recognize the electrical drawings of the hydropower plant to be recognized, and obtain the graphic elements, texts and categories of the electrical drawings of the hydropower plant to be recognized;

Identify the graphic element based on the category to obtain graphic element information, and perform semantic recognition on the text to obtain text information;

The recognition result is output according to the graphic element information and the text information.

In order to solve the above technical problems, another technical solution adopted by the present invention is:

An automatic identification terminal for electrical drawings of a hydropower plant comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are implemented:

Obtain electrical drawings of the hydropower plant to be identified;

Use the segmentation and recognition model to segment and recognize the electrical drawings of the hydropower plant to be recognized, and obtain the graphic elements, texts and categories of the electrical drawings of the hydropower plant to be recognized;

Identify the graphic element based on the category to obtain graphic element information, and perform semantic recognition on the text to obtain text information;

The recognition result is output according to the graphic element information and the text information.

The beneficial effects of the present invention are as follows: a segmentation and recognition model is used to segment and recognize the electrical drawings of a hydropower plant to be recognized, and the graphics, text and categories of the electrical drawings of the hydropower plant to be recognized are obtained; graphics are recognized based on the categories to obtain graphics information; and text is semantically recognized to obtain text information. In this way, the graphics, text and categories of the electrical drawings of the hydropower plant are automatically recognized and separated by using the segmentation and recognition model, and then graphics and semantic recognition are performed respectively, so that automatic, comprehensive and accurate electrical drawing recognition can be realized, and the output recognition results can be used for further analysis and processing, thereby effectively realizing the informatization and intelligence of electrical drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a flowchart of a method for automatically identifying electrical drawings of a hydropower plant according to an embodiment of the present invention;

FIG2 is a schematic diagram of the structure of an automatic identification terminal for electrical drawings of a hydropower plant according to an embodiment of the present invention;

FIG3 is a recognition flow chart of the method for automatic recognition of electrical drawings of a hydropower plant according to an embodiment of the present invention.

Detailed ways

In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in conjunction with the implementation modes and the accompanying drawings.

Referring to FIG. 1 , a method for automatically identifying electrical drawings of a hydropower plant includes the following steps:

Obtain electrical drawings of the hydropower plant to be identified;

Use the segmentation and recognition model to segment and recognize the electrical drawings of the hydropower plant to be recognized, and obtain the graphic elements, texts and categories of the electrical drawings of the hydropower plant to be recognized;

Identify the graphic element based on the category to obtain graphic element information, and perform semantic recognition on the text to obtain text information;

The recognition result is output according to the graphic element information and the text information.

From the above description, it can be seen that the beneficial effects of the present invention are: using a segmentation and recognition model to segment and recognize the electrical drawings of the hydropower plant to be recognized, obtaining the graphics, text and categories of the electrical drawings of the hydropower plant to be recognized, recognizing the graphics based on the categories to obtain the graphics information, and performing semantic recognition on the text to obtain the text information, thereby utilizing the segmentation and recognition model to automatically recognize and separate the graphics, text and categories of the electrical drawings of the hydropower plant, and then performing graphics recognition and semantic recognition respectively, which can realize automatic, comprehensive and accurate electrical drawing recognition, and the output recognition results can be used for further analysis and processing, thereby effectively realizing the informatization and intelligence of electrical drawings.

Furthermore, the step of obtaining the electrical drawings of the hydropower plant to be identified includes:

Obtain electrical drawings of different types of hydropower plants;

Performing expansion processing on the electrical drawings of the hydropower plant of different categories to obtain expanded electrical drawings of the hydropower plant, and dividing the expanded electrical drawings of the hydropower plant to obtain a first training set;

Using category labels to label the first training set to obtain a segmentation and discrimination sample set;

Constructing an improved YOLOv5 model and a segmentation discrimination loss function corresponding to the improved YOLOv5 model;

The segmentation recognition sample set is input into the improved YOLOv5 model, and the network parameters of the improved YOLOv5 model are trained using the segmentation recognition loss function to obtain a segmentation recognition model.

From the above description, it can be seen that the improved YOLOv5 model is trained using the first training set marked with category labels to obtain a segmentation recognition model, which can accurately and quickly identify the graphics elements, texts and drawing categories in the electrical drawings of the hydropower plant.

Further, the identifying the graphic element based on the category to obtain the graphic element information includes:

Determine whether the category is an electrical primary drawing. If so, use an electrical primary drawing recognition model to recognize the graphic element to obtain graphic element information. If not, use an electrical secondary drawing recognition model to recognize the graphic element to obtain graphic element information.

From the above description, it can be seen that different recognition models are used to recognize graphic elements according to the types of drawings to improve the recognition accuracy.

Furthermore, the electrical primary drawing recognition model is used to recognize the graphic element, and the graphic element information obtained includes:

The electrical primary drawing recognition model is used to identify the graphic element according to the hydropower plant drawing to be identified, and the electrical connection relationship of the position of the graphic element relative to other graphic elements and the actual physical name of the graphic element are obtained.

As can be seen from the above description, the identified primitive information includes the electrical connection relationship of the primitive's location relative to other primitives and the actual physical name of the primitive, so that the user can directly and quickly understand the primitive information.

Furthermore, the step of obtaining the electrical drawings of the hydropower plant to be identified includes:

Obtain the primary electrical drawing set of the hydropower plant;

Expanding the primary electrical drawing set of the hydropower plant to obtain an expanded primary electrical drawing set of the hydropower plant;

Annotate the graphic element information of each hydropower plant electrical primary drawing in the expanded hydropower plant electrical primary drawing set to obtain an annotated hydropower plant electrical primary drawing set;

Dividing the annotated primary electrical drawing set of the hydropower plant to obtain a second training set;

Constructing an improved YOLOv5 model and an electrical primary loss function corresponding to the improved YOLOv5 model;

The second training set is input into the improved YOLOv5 model, and the network parameters of the improved YOLOv5 model are trained according to the electrical primary loss function to obtain an electrical primary drawing recognition model.

From the above description, it can be seen that the second training set is a set of electrical primary drawings of hydropower plants with annotated graphic element information. By using this training set to train the improved YOLOv5 model, an electrical primary drawing recognition model with high recognition accuracy can be obtained.

Furthermore, the performing semantic recognition on the text to obtain text information includes:

A text recognition model is used to perform semantic recognition on the text to obtain text information.

From the above description, it can be seen that the use of text recognition models for semantic recognition improves the automation level and accuracy of electrical drawing recognition.

Furthermore, the step of obtaining the electrical drawings of the hydropower plant to be identified includes:

Get the text box area image set of electrical terms;

Performing text area segmentation on the text box area picture set to obtain a segmented text box area picture set, and dividing the segmented text box area picture set to obtain a third training set;

Constructing an optimized FCOS model and an electrical term loss function corresponding to the optimized FCOS model;

The third training set is input into the optimized FCOS model, and the network parameters of the optimized FCOS model are trained according to the electrical terminology loss function to obtain a text recognition model.

From the above description, we can see that the text recognition model is trained by the optimized FCOS model. The FCOS model is a fully convolutional single-stage target detector, which is designed based on anchor-free and proposal free, and can solve the text recognition problem at a low computational cost.

Furthermore, the step of obtaining the electrical drawings of the hydropower plant to be identified includes:

The Backbone feature extraction part in the original YOLOv5 model is modified into a ShuffleNet module that comprehensively uses grouped convolution and Channel Shuffle operations, and a CBAM mechanism is added to the Neck feature aggregation module in the original YOLOv5 model to obtain an improved YOLOv5 model.

From the above description, it can be seen that the Backbone feature extraction part in the original YOLOv5 model is modified to the ShuffleNet module that comprehensively uses group convolution and Channel Shuffle operations, and the CBAM mechanism is added to the Neck feature aggregation module in the original YOLOv5 model to obtain the improved YOLOv5 model, which strengthens the algorithm's drawing feature extraction capability and improves the algorithm's operation efficiency to ensure accurate recognition results.

Furthermore, the electrical term loss function is a cross entropy loss function.

From the above description, it can be seen that using the cross entropy loss function as the electrical term loss function can avoid the gradient vanishing problem, accelerate the optimization process, and also avoid model overfitting, ensuring the best model training effect.

Please refer to Figure 2. Another embodiment of the present invention provides a hydropower plant electrical drawing automatic recognition terminal, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, each step in the above-mentioned hydropower plant electrical drawing automatic recognition method is implemented.

The above-mentioned hydropower plant electrical drawing automatic recognition method and terminal of the present invention can be applied to the hydropower plant electrical drawing recognition scenario, which is described below through specific implementation methods:

Please refer to Figures 1 and 3, the first embodiment of the present invention is:

A method for automatically identifying electrical drawings of a hydropower plant, comprising the steps of:

S1. Obtain the electrical drawings of the hydropower plant to be identified.

Specifically, the user end converts the electrical drawings of the hydropower plant to be identified from DWG, DXF or PDF format into PNG format, and uploads the electrical drawings of the hydropower plant to be identified in PNG format to the server end, and the server end obtains the electrical drawings of the hydropower plant to be identified.

S2. Segment and identify the electrical drawings of the hydropower plant to be identified using a segmentation and identification model to obtain the graphic elements, texts and categories of the electrical drawings of the hydropower plant to be identified, as shown in FIG3 ;

Among them, the categories include electrical primary drawings and electrical secondary drawings.

In an optional implementation, before S1, the steps include:

Obtain electrical drawings of different types of hydropower plants;

Performing expansion processing on the electrical drawings of the hydropower plant of different categories to obtain expanded electrical drawings of the hydropower plant, and dividing the expanded electrical drawings of the hydropower plant to obtain a first training set;

The first training set is labeled with a category label to obtain a segmentation and discrimination sample set, so that the category output after the network is trained can be compared with the category label, the loss value is calculated, and the parameters of the network model are adjusted;

Constructing an improved YOLOv5 model and a segmentation discrimination loss function corresponding to the improved YOLOv5 model;

The segmentation recognition sample set is input into the improved YOLOv5 model, and the network parameters of the improved YOLOv5 model are trained using the segmentation recognition loss function to obtain a segmentation recognition model.

The expansion process includes image cropping, rotation, data enhancement, etc. When the expanded electrical drawings of the hydropower plant are divided, a first test set is also obtained. In an optional implementation, the ratio of the first training set to the first test set is 8:2.

That is, first use the first training set with category labels as the segmentation and discrimination set to train the improved YOLOv5 model, and the label is the category. After the segmentation and discrimination model is trained, an electrical drawing in the first test set is input to output the corresponding category; the principle of the segmentation and discrimination loss function is to calculate the mean square error between the predicted category label and the true category label, and adjust the network parameters of the model according to the mean square error until the mean square error reaches a certain threshold, then the training can be stopped. The threshold can be set according to the actual situation. In this embodiment, the threshold is set to 0.001.

S3, identifying the graphic element based on the category to obtain graphic element information, and performing semantic recognition on the text to obtain text information, as shown in FIG3, specifically including S31-S32:

S31. Determine whether the category is an electrical primary drawing. If so, execute S311; if not, execute S312.

S311, using an electrical primary drawing recognition model to recognize the graphic element to obtain graphic element information.

Specifically, the electrical primary drawing recognition model is used to identify the graphic element according to the hydropower plant drawing to be identified, and the electrical connection relationship of the position of the graphic element relative to other graphic elements and the actual physical name of the graphic element are obtained.

S312: Use an electrical secondary drawing recognition model to recognize the graphic element to obtain graphic element information.

Specifically, the electrical secondary drawing recognition model is used to identify the graphic element according to the hydropower plant drawing to be identified, and the electrical connection relationship of the position of the graphic element relative to other graphic elements and the actual physical name of the graphic element are obtained.

The graphic elements include physical equipment graphic elements such as transformers and current transformers.

S32: Use a text recognition model to perform semantic recognition on the text to obtain text information.

In an optional implementation, before S1, the steps include:

Obtain the primary electrical drawing set of the hydropower plant;

Expanding the primary electrical drawing set of the hydropower plant to obtain an expanded primary electrical drawing set of the hydropower plant;

Annotate the graphic element information of each hydropower plant electrical primary drawing in the expanded hydropower plant electrical primary drawing set to obtain an annotated hydropower plant electrical primary drawing set;

Dividing the annotated primary electrical drawing set of the hydropower plant to obtain a second training set;

Constructing an improved YOLOv5 model and an electrical primary loss function corresponding to the improved YOLOv5 model;

The second training set is input into the improved YOLOv5 model, and the network parameters of the improved YOLOv5 model are trained according to the electrical primary loss function to obtain an electrical primary drawing recognition model.

Among them, the annotated content includes canvas, basic plant information, primitives and connecting lines; the canvas includes canvas width and height, the basic plant information includes name and type, the primitives include primitive category, attribute, rotation angle and related primitive information, and the connecting lines include a list of all connected primitive IDs and the coordinates of the beginning and end. The electrical primary loss function is a mean square error function. In an optional embodiment, the deep learning annotation tool labelimg is used to annotate the primitive information of each hydropower plant electrical primary drawing in the expanded hydropower plant electrical primary drawing set to obtain the annotated hydropower plant electrical primary drawing set, so that the length and width of the text box area, the center point coordinates and the tilt angle can be automatically calculated during the annotation process; the annotated hydropower plant electrical primary drawing set is converted into XML format to facilitate model training; when the annotated hydropower plant electrical primary drawing set is divided, a second test set is also obtained. After the electrical primary drawing recognition model is trained, the corresponding primitive information can be output by inputting a hydropower plant electrical primary drawing in the second test set.

The training process of the electrical secondary drawing recognition model is similar to the above, except that the above-mentioned hydropower plant electrical primary drawing set is replaced by the hydropower plant electrical secondary drawing set, which will not be repeated here.

In an optional implementation, before S1, the steps include:

The Backbone feature extraction part in the original YOLOv5 model is modified into a ShuffleNet module that comprehensively uses grouped convolution and Channel Shuffle operations, and a CBAM mechanism is added to the Neck feature aggregation module in the original YOLOv5 model to obtain an improved YOLOv5 model; the target detection problem is converted into a regression prediction problem, and the algorithm uses a CNN network to extract image features and perform feature fusion by directly obtaining bounding boxes and category information from image features, and then classifies and regresses based on these features to obtain the bounding box and category confidence of the target; during the execution of the algorithm, the image to be detected is divided into s×s panes, and each pane can detect whether the center point of the object is located therein; in this way, target detection can be performed on multiple locations in the image at the same time, and each pane will have multiple bounding boxes and target category confidences, thereby improving the detection speed and accuracy of the algorithm.

In an optional implementation, before S1, the steps include:

Get the text box area image set of electrical terms;

Performing text area segmentation on the text box area picture set to obtain a segmented text box area picture set, and dividing the segmented text box area picture set to obtain a third training set;

Constructing an optimized FCOS model and an electrical term loss function corresponding to the optimized FCOS model;

The third training set is input into the optimized FCOS model, and the network parameters of the optimized FCOS model are trained according to the electrical terminology loss function to obtain a text recognition model.

Among them, the electrical term loss function is a cross entropy loss function, the principle of which is to calculate the learning rate of the cross entropy loss function of the predicted character label and the real character label, and adjust the network parameters of the above FCOS model according to the learning rate, until the learning rate reaches a certain threshold, the training can be stopped, the threshold can be flexibly set according to the actual situation, in this embodiment, the threshold is set to 0.0001. The optimized FCOS model has 50 convolutional layers, each layer uses a deep residual network, and the last layer of the network sets the relevant training evaluation index: the test set accuracy, and the training is stopped when the test set accuracy reaches the expected value, and finally the text recognition model is obtained.

FCOS has some unique advantages over traditional neural network target detection algorithms: Traditional neural network target detection algorithms are based on anchor points, that is, before training the network model, multiple rectangular anchor boxes are generated for the image, and the predicted category and offset are marked for each anchor box as a training sample. When training the model, multiple predicted anchor box positions will be obtained, and then the maximum intersection-over-union ratio between the predicted bounding box and the true rectangular box will be calculated, and the best predicted bounding box and its category probability will be output through the non-maximum suppression method. FCOS does not rely on anchor boxes or proposed areas obtained in advance through big data analysis. By using the anchor-free mode, FCOS completely removes the operations on the anchor box part, and the memory usage during the training process is also greatly reduced, which makes it more convenient to train and deploy the algorithm. Compared with existing technologies, it achieves better or more competitive performance in text scenarios, namely character recognition.

S4. Output a recognition result according to the graphic element information and the text information.

Please refer to FIG. 2 , the second embodiment of the present invention is:

An automatic identification terminal for electrical drawings of a hydropower plant comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, each step of the method for intelligently organizing cable information of a hydropower plant in the first embodiment is implemented.

In summary, the present invention provides an automatic recognition method and terminal for electrical drawings of hydropower plants, which use a segmentation and recognition model to segment and recognize the electrical drawings of hydropower plants to be recognized, obtain the graphics, text and categories of the electrical drawings of hydropower plants to be recognized, recognize the graphics based on the categories to obtain graphics information, and perform semantic recognition on the text to obtain text information, thereby using the segmentation and recognition model to automatically recognize and separate the graphics, text and categories of the electrical drawings of hydropower plants, and then perform graphics and semantic recognition respectively, which can realize automatic, comprehensive and accurate electrical drawing recognition, and the output recognition results can be used for further analysis and processing, thereby effectively realizing the informatization and intelligence of electrical drawings; and, using the first training set marked with category labels to train the improved YOLOv5 model, a segmentation and recognition model is obtained, so that the graphics, text and drawing categories in the electrical drawings of hydropower plants can be accurately and quickly recognized; in addition, the Backbone feature extraction part in the original YOLOv5 model is modified to comprehensively use grouped convolution and Channel The ShuffleNet module of the Shuffle operation is used, and the CBAM mechanism is added to the Neck feature aggregation module in the original YOLOv5 model to obtain the improved YOLOv5 model, which strengthens the algorithm's drawing feature extraction capability and improves the algorithm's operation efficiency to ensure accurate recognition results.

The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims (10)

1. A method for automatic identification of electrical drawings of a hydropower plant, characterized by comprising the steps of: Obtain electrical drawings of the hydropower plant to be identified; Use the segmentation and recognition model to segment and recognize the electrical drawings of the hydropower plant to be recognized, and obtain the graphic elements, texts and categories of the electrical drawings of the hydropower plant to be recognized; Identify the graphic element based on the category to obtain graphic element information, and perform semantic recognition on the text to obtain text information; The recognition result is output according to the graphic element information and the text information. 2. The method for automatic identification of electrical drawings of a hydropower plant according to claim 1, characterized in that before obtaining the electrical drawings of the hydropower plant to be identified, the method comprises: Obtain electrical drawings of different types of hydropower plants; Performing expansion processing on the electrical drawings of the hydropower plant of different categories to obtain expanded electrical drawings of the hydropower plant, and dividing the expanded electrical drawings of the hydropower plant to obtain a first training set; Using category labels to label the first training set to obtain a segmentation and discrimination sample set; Constructing an improved YOLOv5 model and a segmentation discrimination loss function corresponding to the improved YOLOv5 model; The segmentation recognition sample set is input into the improved YOLOv5 model, and the network parameters of the improved YOLOv5 model are trained using the segmentation recognition loss function to obtain a segmentation recognition model. 3. The method for automatic recognition of electrical drawings of a hydropower plant according to claim 1, characterized in that the step of identifying the graphic element based on the category to obtain the graphic element information comprises: Determine whether the category is an electrical primary drawing. If so, use an electrical primary drawing recognition model to recognize the graphic element to obtain graphic element information. If not, use an electrical secondary drawing recognition model to recognize the graphic element to obtain graphic element information. 4. A method for automatic recognition of electrical drawings of a hydropower plant according to claim 3, characterized in that the step of using an electrical primary drawing recognition model to recognize the graphic element to obtain graphic element information includes: The electrical primary drawing recognition model is used to identify the graphic element according to the hydropower plant drawing to be identified, and the electrical connection relationship of the position of the graphic element relative to other graphic elements and the actual physical name of the graphic element are obtained. 5. The method for automatic identification of electrical drawings of a hydropower plant according to claim 1, characterized in that before obtaining the electrical drawings of the hydropower plant to be identified, the method comprises: Obtain the primary electrical drawing set of the hydropower plant; Expanding the primary electrical drawing set of the hydropower plant to obtain an expanded primary electrical drawing set of the hydropower plant; Annotate the graphic element information of each hydropower plant electrical primary drawing in the expanded hydropower plant electrical primary drawing set to obtain an annotated hydropower plant electrical primary drawing set; Dividing the annotated primary electrical drawing set of the hydropower plant to obtain a second training set; Constructing an improved YOLOv5 model and an electrical primary loss function corresponding to the improved YOLOv5 model; The second training set is input into the improved YOLOv5 model, and the network parameters of the improved YOLOv5 model are trained according to the electrical primary loss function to obtain an electrical primary drawing recognition model. 6. The method for automatic recognition of electrical drawings of a hydropower plant according to claim 1, characterized in that the semantic recognition of the text to obtain text information comprises: A text recognition model is used to perform semantic recognition on the text to obtain text information. 7. The method for automatic identification of electrical drawings of a hydropower plant according to claim 6, characterized in that before obtaining the electrical drawings of the hydropower plant to be identified, the method comprises: Get the text box area image set of electrical terms; Performing text area segmentation on the text box area picture set to obtain a segmented text box area picture set, and dividing the segmented text box area picture set to obtain a third training set; Constructing an optimized FCOS model and an electrical term loss function corresponding to the optimized FCOS model; The third training set is input into the optimized FCOS model, and the network parameters of the optimized FCOS model are trained according to the electrical terminology loss function to obtain a text recognition model. 8. A method for automatic identification of electrical drawings of a hydropower plant according to claim 2 or 5, characterized in that before obtaining the electrical drawings of the hydropower plant to be identified, the method comprises: The Backbone feature extraction part in the original YOLOv5 model is modified into a ShuffleNet module that comprehensively uses grouped convolution and ChannelShuffle operations, and a CBAM mechanism is added to the Neck feature aggregation module in the original YOLOv5 model to obtain an improved YOLOv5 model. 9. A method for automatic recognition of electrical drawings of a hydropower plant according to claim 7, characterized in that the electrical terminology loss function is a cross entropy loss function. 10. An automatic identification terminal for electrical drawings of a hydropower plant, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements each step of an automatic identification method for electrical drawings of a hydropower plant as described in any one of claims 1 to 9 when executing the computer program.