A method and system for early warning of electric energy with automatic electric energy measurement

Abstract

The present invention discloses an electric energy early warning method and system for automatic electric energy metering, the method comprising photographing a radio frequency electronic tag on an electric energy meter by high-definition photography equipment. The present invention obtains a label image, performs noise reduction and compression on it, on the one hand, improves the visual quality of the label image, makes it clearer, easier to understand and analyze, on the other hand, reduces the space occupied by the image storage, and speeds up the loading speed of the image; secondly, the text in the label is framed, and a BLSTM network is used for recognition, so as to obtain the type, product name, specification, batch number, production date, validity period, storage time, and outbound time data of the electric energy meter. There are usually multiple electric energy meters in the factory, all of which have standard information. The above-mentioned extracted parameters are compared with the standard information. If they can match, it means that the electric energy meter information is correct, otherwise an early warning is issued to remind the staff to correct the data on the label.

Description

Technical Field

[0001] The present invention generally relates to the technical field of electric energy meter detection, and in particular to an electric energy early warning method and system for automatic electric energy metering. Background Art

[0002] As the country continues to strengthen the construction of smart grids, my country's electricity metering meters (single-phase electricity meters, three-phase electricity meters, collection terminals, small transformers, etc.) have made great progress in quantity and quality, thus putting forward new requirements for the electricity metering asset management system.

[0003] Electricity meters are usually affixed with radio frequency electronic tags, also known as RFID tags. During the production process, the information on the RFID tags needs to be inspected on the assembly line to ensure that the RFID tags are correct. The existing inspection method is usually to set up high-definition photography equipment on the assembly line to collect RFID tag images, but the system lacks the step of compressing images. There are many electricity meters on the assembly line, which can easily cause slow image opening speed and system crash, making it difficult to meet the needs of staff. Summary of the invention

[0004] In order to solve the above technical problems, a method and system for electric energy early warning with automated electric energy metering are provided. This technical solution solves the problem that the existing detection method proposed in the above background technology is usually to set up high-definition photography equipment on the assembly line to collect RFID tag images, but the system lacks the step of compressing images, and there are a large number of electric energy meters on the assembly line, which easily causes the image opening speed to be slow and the system to crash.

[0005] In a first aspect of the present invention, a pipeline fault early warning method based on artificial intelligence is provided. The method comprises:

[0006] Use high-definition photography equipment to photograph the radio frequency electronic tag on the electric energy meter to obtain the tag image;

[0007] Preprocessing the label image to obtain a preprocessed label image, wherein the preprocessing includes image denoising and image compression;

[0008] Detect text boxes in pre-processed label images and correct the pre-processed label images;

[0009] BLSTM is used to identify text regions, extract the type, product name, specification, batch number, production date, expiration date, storage time, and delivery time data of the electric energy meter, and import them into the information system for storage;

[0010] Retrieve standard data information of several groups of electric energy meters from the database, compare the standard data information with the type, product name, specification, batch number, production date, validity period, storage time, and storage time data of the extracted electric energy meters, and obtain the comparison results;

[0011] Based on the comparison results, determine whether to issue an early warning.

[0012] Preferably, the image noise reduction includes:

[0013] Adaptive wavelet threshold denoising is performed on label images;

[0014] By constructing a probability model of wavelet coefficients, the optimal wavelet threshold is determined using Bayesian estimation.

[0015] The wavelet coefficients are processed with adaptive threshold and then reconstructed by wavelet to obtain a denoised image.

[0016] Preferably, the image compression includes:

[0017] Analyze the denoised image to obtain the length, width and color data of the image, and cut the image into several blocks of equal size;

[0018] According to the background color and non-background color of the image, all display areas with non-background colors are obtained in several blocks, the coordinates of each display area are established, and the arithmetic mean of the color data of all pixels in each display area is calculated;

[0019] Divide the display areas with equal arithmetic mean values ​​into an area group, and record the number of display areas in the area group, to obtain multiple area groups and the number of display areas in each area group;

[0020] Assign the low pixel value preset in the system to each display area to obtain a compressed image.

[0021] Preferably, the cut image is a plurality of blocks of equal size, including:

[0022] The length and width of the tile are selected according to the length and width of the image;

[0023] The tile length is selected so that the length of the image is exactly cut;

[0024] The tile width is chosen so that the width of the image is completely cut;

[0025] Based on the cutting results, an integer number of tiles are obtained.

[0026] Preferably, establishing the coordinates of each display area includes:

[0027] Get the background color data and non-background color data in the image;

[0028] According to the background color data and non-background color data of the image, all the cut blocks are screened to obtain all the display areas with non-background colors;

[0029] According to the number of pixels in each display area and the length and width of the display area, the coordinates of each display area are constructed and saved.

[0030] Preferably, the correcting the pre-processed label image includes:

[0031] Perform fitting processing on each text line in the preprocessed label image to obtain a fitting curve corresponding to each text line;

[0032] Obtaining a spacing, a first offset, and a second offset, wherein the spacing is the spacing between any two adjacent fitting curves, the first offset is the offset between the starting points of any two fitting curves, and the second offset is the offset of each fitting curve from the center point;

[0033] determining a mapping matrix according to the spacing, the first offset and the second offset;

[0034] The mapping matrix is ​​used to perform pixel mapping on the preprocessed label image to obtain the corrected label image.

[0035] Preferably, the detecting of the text box in the preprocessing label image includes:

[0036] Input the corrected label image into the VGG16 network;

[0037] Get the mapping of the convolutional layer in the VGG16 network and get the features of size W×H×C, where W is the image width, H is the image height, and C is the number of image channels;

[0038] Extracting feature vectors from the convolutional layer using a sliding window;

[0039] The extracted feature vector is input into the BLSTM network for processing to obtain the text box in the corrected label image.

[0040] Preferably, the adopting BLSTM to identify the text region includes:

[0041] Input the text box into the VGG16 network to generate several convolution feature matrices;

[0042] Each convolution feature matrix corresponds to a rectangular area, and the convolution feature matrix is ​​input into the BLSTM layer;

[0043] Set the maximum time length of the BLSTM layer, classify the results of the BLSTM output using the softmax function, and obtain the type, product name, specification, batch number, production date, validity period, storage time, and delivery time data of the electric energy meter.

[0044] Preferably, judging whether to issue an early warning based on the comparison result includes:

[0045] Compare the type, product name, specification, batch number, production date, validity period, storage time, and storage time of the electric energy meter with the standard data information of several groups of electric energy meters in the database;

[0046] If the type, product name, specification, batch number, production date, validity period, storage time, and storage time of the electric energy meter match the standard data information of several groups of electric energy meters in the database, it means that the electric energy meter information is correct and no warning is needed;

[0047] If any of the meter's type, product name, specification, batch number, production date, expiration date, storage time, and delivery time data cannot match the standard data information of several groups of meter sets, it means that the meter information is incorrect and an early warning is issued.

[0048] In a second aspect of the present invention, an electric energy early warning system for automatic electric energy metering is provided. The system comprises:

[0049] An acquisition module is used to photograph the radio frequency electronic tag on the electric energy meter through a high-definition photographic device to obtain a tag image;

[0050] A preprocessing module is used to preprocess the label image to obtain a preprocessed label image, wherein the preprocessing includes image denoising and image compression;

[0051] A detection and correction module is used to detect text boxes in pre-processed label images and correct the pre-processed label images;

[0052] The recognition module is used to use BLSTM to recognize the text area, extract the type, product name, specification, batch number, production date, validity period, storage time, and delivery time data of the electric energy meter, and import it into the information system for storage;

[0053] A comparison module is used to retrieve standard data information of several groups of electric energy meters in the database, compare the standard data information with the type, product name, specification, batch number, production date, validity period, storage time, and storage time data of the extracted electric energy meters, and obtain the comparison results;

[0054] The early warning module is used to determine whether to issue an early warning based on the comparison results.

[0055] Compared with the prior art, the present invention provides an electric energy early warning method and system for automatic electric energy metering, which has the following beneficial effects:

[0056] The present invention obtains the label image, performs noise reduction and compression on it. On the one hand, it improves the visual quality of the label image to make it clearer, easier to understand and analyze. On the other hand, it reduces the space occupied by the image storage and speeds up the loading speed of the image. Secondly, the text in the label is framed and recognized by the BLSTM network, so as to obtain the type, product name, specification, batch number, production date, validity period, storage time, and outbound time data of the electric energy meter. There are usually multiple electric energy meters in the factory, all with standard information. The above-mentioned extracted parameters are compared with the standard information. If they can match, it means that the electric energy meter information is correct. Otherwise, an early warning is issued to remind the staff to correct the data on the label, thereby meeting the needs of the staff. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The above and other features, advantages and aspects of the embodiments of the present invention will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, wherein:

[0058] Figure 1 It is a schematic diagram of the electric energy early warning method for automatic electric energy metering in the present invention;

[0059] Figure 2 Schematic diagram of the method for image noise reduction in the present invention;

[0060] Figure 3 It is a schematic diagram of the method for image compression in the present invention;

[0061] Figure 4 A schematic diagram of a method for cutting an image into a plurality of blocks of equal size in the present invention;

[0062] Figure 5 A schematic diagram of a method for establishing coordinates of each display area in the present invention;

[0063] Figure 6 A schematic diagram of a method for correcting a pre-processed label image in the present invention;

[0064] Figure 7 A schematic diagram of a method for detecting text boxes in a preprocessed label image in the present invention;

[0065] Figure 8 A schematic diagram of a method for identifying text regions using BLSTM in the present invention;

[0066] Fig. 9 A schematic diagram of a method for determining whether to issue an early warning based on a comparison result in the present invention;

[0067] Fig.10 The block diagram of the electric energy early warning system for automatic electric energy metering according to the present invention is shown. DETAILED DESCRIPTION

[0068] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0069] In addition, the term "and / or" in this article is only a description of the association relationship between the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0070] Example 1

[0071] Please refer to Figure 1 As shown, an electric energy early warning method for automatic electric energy metering includes:

[0072] S100, photographing the radio frequency electronic tag on the electric energy meter using a high-definition photographic device to obtain a tag image;

[0073] S200, preprocessing the label image to obtain a preprocessed label image, wherein the preprocessing includes image noise reduction and image compression;

[0074] S300, detecting a text box in a preprocessed label image, and correcting the preprocessed label image;

[0075] S400, using BLSTM to identify the text area, extract the type, product name, specification, batch number, production date, validity period, storage time, and delivery time data of the electric energy meter, and import it into the information system for storage;

[0076] S500, retrieve standard data information of several groups of electric energy meters in the database, compare the standard data information with the type, product name, specification, batch number, production date, validity period, storage time, and storage time data of the extracted electric energy meters, and obtain the comparison result;

[0077] S600: Based on the comparison result, determine whether to issue an early warning.

[0078] Those skilled in the art can understand that, during the production and inspection of electric energy meters, the parameters on the radio frequency electronic tags are important objects for inspection. By obtaining the label image, reducing noise and compressing it, on the one hand, the visual quality of the label image is improved, making it clearer, easier to understand and analyze, and on the other hand, the space occupied by the image storage is reduced and the loading speed of the image is accelerated; secondly, the text in the label is framed and recognized by the BLSTM network to obtain the type, product name, specification, batch number, production date, validity period, storage time, and outbound time data of the electric energy meter. There are usually multiple types of electric energy meters in the factory, all with standard information. The above-mentioned extracted parameters are compared with the standard information. If they all match, it means that the electric energy meter information is correct. Otherwise, an early warning is issued to remind the staff to correct the data on the label, thereby meeting the needs of the staff.

[0079] like Figure 2 As shown, image noise reduction includes:

[0080] S211, performing adaptive wavelet threshold denoising on the label image;

[0081] S212, determining the optimal wavelet threshold by constructing a probability model of wavelet coefficients and using Bayesian estimation;

[0082] S213, performing adaptive threshold processing on the wavelet coefficients and performing wavelet reconstruction to obtain a denoised image.

[0083] It can be understood by those skilled in the art that wavelet transform is an effective time-frequency analysis tool that can decompose an image into components of different frequencies and scales for subsequent processing; find the optimal wavelet threshold so as to retain as much important information of the image as possible while removing noise. This usually involves constructing a probability model of the wavelet coefficients, that is, assuming that the wavelet coefficients obey a certain distribution (such as Gaussian distribution, Laplace distribution, etc.). Then, the Bayesian estimation method is used to estimate the optimal threshold based on known image statistical characteristics and prior knowledge. Bayesian estimation is a statistical method that combines the likelihood of data and the prior distribution of parameters to obtain the posterior distribution of parameters, thereby making the best decision, setting the wavelet coefficients less than the threshold to zero (or a value close to zero) to remove noise, and retaining the wavelet coefficients greater than the threshold to retain important information of the image.

[0084] like Figure 3 As shown, image compression includes:

[0085] S221, parsing the denoised image to obtain the length, width and color data of the image, and cutting the image into a number of blocks of equal size;

[0086] S222, according to the background color and non-background color of the image, obtain all display areas with non-background colors in a plurality of image blocks, establish the coordinates of each display area, and calculate the arithmetic mean of the color data of all pixels in each display area;

[0087] S223, dividing the display areas with equal arithmetic means into an area group, and recording the number of display areas in the area group, to obtain multiple area groups and the number of display areas in each area group;

[0088] S224: assign a low pixel value preset in the system to each display area to obtain a compressed image.

[0089] Those skilled in the art can understand that, by using the method of data matching and multiplexing symmetrical display areas, the size of the space occupied by the display area storage and the time for reading the display area can be reduced, thereby reducing the size of the space occupied by the image storage and speeding up the loading of the image; the data processing of the display area becomes smaller, thereby reducing the time for the amount of data, making the image interface display smoother.

[0090] like Figure 4 As shown, the cut image is divided into several blocks of equal size, including:

[0091] S2211, selecting values ​​for the length and width of the block according to the length and width of the image;

[0092] S2212, selecting a value for the length of the image block so that the length of the image is just completely cut;

[0093] S2213, selecting a value for the width of the image block so that the width of the image is just completely cut;

[0094] S2214. Based on the cutting result, an integer number of blocks is obtained.

[0095] It can be understood by those skilled in the art that the length and width of the tile are set according to the length and width of the picture so that the length of the picture is just completely cut and the width of the picture is just completely cut to obtain an integer number of tiles, and the length and width of the tile may be equal or unequal, that is, the length of the picture is just divisible by the length of the tile, and the width of the picture is just divisible by the width of the tile, or the length of the picture is just divisible by the width of the tile, and the width of the picture is just divisible by the length of the tile, to obtain an integer number of tiles, so that the picture is just completely cut, and the number of tiles after cutting is an integer. The tile can be rectangular or square, and this cutting method can avoid the different sizes of the tiles obtained by cutting during the picture cutting process, thereby avoiding affecting the subsequent steps.

[0096] like Figure 5As shown, establish the coordinates of each display area, including:

[0097] S2221, obtaining background color data and non-background color data in the image;

[0098] S2222, screening all the cut blocks according to the background color data and non-background color data of the image to obtain all display areas with non-background colors;

[0099] S2223. Construct and save the coordinates of each display area according to the number of pixels in each display area and the length and width of the display area.

[0100] It can be understood by those skilled in the art that the color data of the picture is obtained by parsing the picture information, and the color data is the color data information of each pixel in the picture, and the color data information includes at least background color and non-background color, the background color is the initial color of the picture, and the non-background color is the color rendered on the background color. The background color of the picture is parsed and stored. The stored background color is used to fill the part of the picture other than the non-background color part when the picture is loaded, thereby reducing the total data size of the picture, that is, reducing the storage space of the picture, and can also reduce the processing time of the picture loading, thereby reducing the memory of the electronic device consumed by the picture loading, that is, increasing the speed of picture loading.

[0101] like Figure 6 As shown, the pre-processed label image is corrected, including:

[0102] S311, performing fitting processing on each text line in the preprocessed label image to obtain a fitting curve corresponding to each text line;

[0103] S312, obtaining a spacing, a first offset, and a second offset, where the spacing is the spacing between any two adjacent fitting curves, the first offset is the offset between the starting points of any two fitting curves, and the second offset is the offset of each fitting curve from the center point;

[0104] S313, determining a mapping matrix according to the interval, the first offset and the second offset;

[0105] S314: Use a mapping matrix to perform pixel mapping on the preprocessed label image to obtain a corrected label image.

[0106] Those skilled in the art can understand that, based on the spacing between any two adjacent curves, the offset between the starting points of any two curves, and the offset of each curve from the center point, the mapping relationship between the pixel points in the target image and the pixel points in the intermediate target image can be solved to obtain an optimal solution matrix, and then the target image can be pixel mapped using the optimal solution matrix, thereby achieving flattening of the target image. By flattening the target image, the text of the intermediate target image is made straighter. In this way, on the intermediate target image with straight text, the text in the area selected by the text selection operation can be recognized, and a more accurate text recognition result can be obtained.

[0107] like Figure 7 As shown, the text box in the preprocessed label image is detected, including:

[0108] S321, input the corrected label image into the VGG16 network;

[0109] S322, obtain the mapping of the convolutional layer in the VGG16 network, and obtain features of size W×H×C, where W is the image width, H is the image height, and C is the number of image channels;

[0110] S323, extracting feature vectors from the convolutional layer using a sliding window;

[0111] S324: Input the extracted feature vector into the BLSTM network for processing to obtain the text box in the corrected label image.

[0112] like Figure 8 As shown, BLSTM is used to identify text areas, including:

[0113] S401, input the text box into the VGG16 network to generate several convolution feature matrices;

[0114] S402, each convolution feature matrix corresponds to a rectangular area, and the convolution feature matrix is ​​input into the BLSTM layer;

[0115] S403. Set the maximum time length of the BLSTM layer, classify the results of the BLSTM output using the softmax function, and obtain the type, product name, specification, batch number, production date, validity period, storage time, and delivery time data of the electric energy meter.

[0116] like Fig. 9 As shown, based on the comparison results, it is determined whether to issue an early warning, including:

[0117] S601, comparing the type, product name, specification, batch number, production date, validity period, storage time, and storage time of the electric energy meter with the standard data information of several groups of electric energy meters in the database;

[0118] S602: If the type, product name, specification, batch number, production date, validity period, storage time, and storage time of the electric energy meter match the standard data information of several groups of electric energy meters in the database, it means that the electric energy meter information is correct and no warning is required;

[0119] S603. If any of the type, product name, specification, batch number, production date, validity period, storage time, and storage time of the electric energy meter cannot match the standard data information of several groups of electric energy meters, it means that the electric energy meter information is incorrect and an early warning is issued.

[0120] The above is an introduction to a method embodiment. The following is a further explanation of the present invention through a device embodiment having the same inventive concept as the method in the aforementioned embodiment.

[0121] like Fig.10 As shown, an electric energy early warning system for automatic electric energy metering, the system 700 includes:

[0122] The acquisition module 710 is used to photograph the radio frequency electronic tag on the electric energy meter by using a high-definition photography device to obtain a tag image;

[0123] A preprocessing module 720 is used to preprocess the label image to obtain a preprocessed label image, wherein the preprocessing includes image noise reduction and image compression;

[0124] A detection and correction module 730 is used to detect the text box in the pre-processed label image and correct the pre-processed label image;

[0125] The recognition module 740 is used to recognize the text area using BLSTM, extract the type, product name, specification, batch number, production date, validity period, storage time, and storage time data of the electric energy meter, and import it into the information system for storage;

[0126] The comparison module 750 is used to retrieve standard data information of several groups of electric energy meters in the database, compare the standard data information with the type, product name, specification, batch number, production date, validity period, storage time, and storage time data of the extracted electric energy meters, and obtain the comparison result;

[0127] The warning module 760 is used to determine whether to issue a warning based on the comparison result.

[0128] In summary, the present invention obtains the label image, performs noise reduction and compression on it. On the one hand, it improves the visual quality of the label image to make it clearer, easier to understand and analyze. On the other hand, it reduces the space occupied by the image storage and speeds up the loading speed of the image. Secondly, the text in the label is framed and recognized by the BLSTM network to obtain the type, product name, specification, batch number, production date, validity period, storage time, and outbound time data of the electric energy meter. There are usually multiple electric energy meters in the factory, all with standard information. The above-mentioned extracted parameters are compared with the standard information. If they all match, it means that the electric energy meter information is correct. Otherwise, an early warning is issued to remind the staff to correct the data on the label, thereby meeting the needs of the staff.

[0129] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An electric energy early warning method for automatic electric energy metering, characterized in that: include: Use high-definition photography equipment to photograph the radio frequency electronic tag on the electric energy meter to obtain the tag image; Preprocessing the label image to obtain a preprocessed label image, wherein the preprocessing includes image denoising and image compression; Detect text boxes in pre-processed label images and correct the pre-processed label images; BLSTM is used to identify text regions, extract the type, product name, specification, batch number, production date, expiration date, storage time, and delivery time data of the electric energy meter, and import them into the information system for storage; Retrieve standard data information of several groups of electric energy meters from the database, compare the standard data information with the type, product name, specification, batch number, production date, validity period, storage time, and storage time data of the extracted electric energy meters, and obtain the comparison results; Based on the comparison results, determine whether to issue an early warning.

2. The electric energy early warning method for automatic electric energy measurement according to claim 1 is characterized in that: The image noise reduction comprises: Adaptive wavelet threshold denoising is performed on label images; By constructing a probability model of wavelet coefficients, the optimal wavelet threshold is determined using Bayesian estimation. The wavelet coefficients are processed with adaptive threshold and then reconstructed by wavelet to obtain a denoised image.

3. The electric energy early warning method for automatic electric energy measurement according to claim 2 is characterized in that: The image compression comprises: Analyze the denoised image to obtain the length, width and color data of the image, and cut the image into several blocks of equal size; According to the background color and non-background color of the image, all display areas with non-background colors are obtained in several blocks, the coordinates of each display area are established, and the arithmetic mean of the color data of all pixels in each display area is calculated; Divide the display areas with equal arithmetic mean values ​​into an area group, and record the number of display areas in the area group, to obtain multiple area groups and the number of display areas in each area group; Assign the low pixel value preset in the system to each display area to obtain a compressed image.

4. The electric energy early warning method for automatic electric energy measurement according to claim 3 is characterized in that: The cut image is a number of blocks of equal size, including: The length and width of the tile are selected according to the length and width of the image; The tile length is selected so that the length of the image is exactly cut; The tile width is chosen so that the width of the image is completely cut; Based on the cutting results, an integer number of tiles are obtained.

5. The electric energy early warning method for automatic electric energy measurement according to claim 4 is characterized in that: The establishing of the coordinates of each display area includes: Get the background color data and non-background color data in the image; According to the background color data and non-background color data of the image, all the cut blocks are screened to obtain all the display areas with non-background colors; According to the number of pixels in each display area and the length and width of the display area, the coordinates of each display area are constructed and saved.

6. The electric energy early warning method for automatic electric energy measurement according to claim 5 is characterized in that: The correcting of the pre-processed label image includes: Perform fitting processing on each text line in the preprocessed label image to obtain a fitting curve corresponding to each text line; Obtaining a spacing, a first offset, and a second offset, wherein the spacing is the spacing between any two adjacent fitting curves, the first offset is the offset between the starting points of any two fitting curves, and the second offset is the offset of each fitting curve from the center point; determining a mapping matrix according to the spacing, the first offset and the second offset; The mapping matrix is ​​used to perform pixel mapping on the preprocessed label image to obtain the corrected label image.

7. The electric energy early warning method for automatic electric energy measurement according to claim 6 is characterized in that: The detecting of the text box in the preprocessing label image includes: Input the corrected label image into the VGG16 network; Get the mapping of the convolutional layer in the VGG16 network and get the features of size W×H×C, where W is the image width, H is the image height, and C is the number of image channels; Extracting feature vectors from the convolutional layer using a sliding window; The extracted feature vector is input into the BLSTM network for processing to obtain the text box in the corrected label image.

8. The electric energy early warning method for automatic electric energy measurement according to claim 7 is characterized in that: The method of using BLSTM to identify the text region includes: Input the text box into the VGG16 network to generate several convolution feature matrices; Each convolution feature matrix corresponds to a rectangular area, and the convolution feature matrix is ​​input into the BLSTM layer; Set the maximum time length of the BLSTM layer, classify the results of the BLSTM output using the softmax function, and obtain the type, product name, specification, batch number, production date, validity period, storage time, and delivery time data of the electric energy meter.

9. The electric energy early warning method for automatic electric energy measurement according to claim 8, characterized in that: The determining whether to issue an early warning based on the comparison result includes: Compare the type, product name, specification, batch number, production date, validity period, storage time, and storage time of the electric energy meter with the standard data information of several groups of electric energy meters in the database; If the type, product name, specification, batch number, production date, validity period, storage time, and storage time of the electric energy meter match the standard data information of several groups of electric energy meters in the database, it means that the electric energy meter information is correct and no warning is needed; If any of the meter's type, product name, specification, batch number, production date, expiration date, storage time, and delivery time data cannot match the standard data information of several groups of meter sets, it means that the meter information is incorrect and an early warning is issued.

10. An electric energy early warning system for automatic electric energy metering, used to implement an electric energy early warning method for automatic electric energy metering as claimed in any one of claims 1 to 9, characterized in that: The system comprises: An acquisition module is used to photograph the radio frequency electronic tag on the electric energy meter through a high-definition photographic device to obtain a tag image; A preprocessing module is used to preprocess the label image to obtain a preprocessed label image, wherein the preprocessing includes image denoising and image compression; A detection and correction module is used to detect text boxes in pre-processed label images and correct the pre-processed label images; The recognition module is used to use BLSTM to recognize the text area, extract the type, product name, specification, batch number, production date, validity period, storage time, and delivery time data of the electric energy meter, and import it into the information system for storage; A comparison module is used to retrieve standard data information of several groups of electric energy meters in the database, compare the standard data information with the type, product name, specification, batch number, production date, validity period, storage time, and storage time data of the extracted electric energy meters, and obtain the comparison results; The early warning module is used to determine whether to issue an early warning based on the comparison results.

Images