Power service data analysis method, device, equipment, medium and program

Abstract

The invention discloses a power service data analysis method, device and equipment, a medium and a program. The method comprises the following steps: processing acquired first video data and first work order data related to power service to obtain second video data and second work order data; performing feature extraction on the second video data and the second work order data to obtain a visual feature vector of the second video data and a semantic feature vector of the second work order data; performing feature fusion on the visual feature vector and the semantic feature vector to obtain a target fusion feature; and analyzing the target fusion feature according to an expected fusion feature, and determining an association result between the first video data and the first work order data according to an analysis result. The relevance between the video data and the work order text in the electric power service can be automatically determined, and the intelligent level and the response efficiency of the electric power service are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data analysis, and in particular to a power service data analysis method, device, equipment, medium and program. BACKGROUND

[0002] In the field of power service, video data and work order text are two types of resources that are crucial and contain rich associated information. Analyzing and establishing the association between video data and work order text can quickly locate the cause of the fault, provide support for intelligent work order allocation, and assist in service quality evaluation, which has great application value.

[0003] However, current association analysis of video data and work order text in power service mostly relies on manual operation. Manual analysis is not only inefficient and difficult to cope with massive data size, but also highly subjective, with different personnel having different standards for judging association, and it is difficult to achieve real-time response to data, which to some extent restricts the improvement of the intelligent level of power service.

[0004] Therefore, there is an urgent need for a technical solution that can automatically analyze the association of video data and work order text in power service. SUMMARY

[0005] The present application provides a power service data analysis method, device, equipment, medium and program, which can automatically determine the association between video data and work order text in power service, improve the intelligent level and response efficiency of power service, and promote the development of power service in a more efficient and intelligent direction.

[0006] According to a first aspect of the present application, a power service data analysis method is provided, comprising:

[0007] processing the obtained first video data and first work order data related to power service to obtain second video data and second work order data;

[0008] respectively extracting features from the second video data and the second work order data to obtain a visual feature vector of the second video data and a semantic feature vector of the second work order data;

[0009] performing feature fusion on the visual feature vector and the semantic feature vector to obtain a target fusion feature;

[0010] analyzing the target fusion feature according to an expected fusion feature, and determining an association result between the first video data and the first work order data according to the analysis result.

[0011] According to a second aspect of the present application, a power service data analysis device is provided, comprising:

[0012] a processing module, configured to process the acquired first video data and first work order data related to the power service to obtain second video data and second work order data;

[0013] an extraction module, configured to perform feature extraction on the second video data and the second work order data respectively to obtain a visual feature vector of the second video data and a semantic feature vector of the second work order data;

[0014] a fusion module, configured to perform feature fusion on the visual feature vector and the semantic feature vector to obtain target fusion features;

[0015] a determination module, configured to analyze the target fusion features according to expected fusion features, and determine an association result between the first video data and the first work order data according to an analysis result.

[0016] According to a third aspect of the present application, an electronic device is provided, which comprises:

[0017] at least one processor; and a memory connected with the at least one processor in communication; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to perform the power service data analysis method according to any one of the embodiments of the present application.

[0018] According to a fourth aspect of the present application, a computer readable storage medium is provided, which stores computer instructions for enabling a processor to perform the power service data analysis method according to any one of the embodiments of the present application when executed by the processor.

[0019] According to a fifth aspect of the present application, a computer program product is provided, which comprises a computer program for enabling a processor to perform the power service data analysis method according to any one of the embodiments of the present application when executed by the processor.

[0020] The technical scheme of the present application can process the obtained first video data and first work order data related to power service to obtain second video data and second work order data, and respectively extract features of the second video data and the second work order data to obtain a visual feature vector of the second video data and a semantic feature vector of the second work order data, then fuse the visual feature vector and the semantic feature vector to obtain a target fusion feature, finally analyze the target fusion feature according to an expected fusion feature, and determine the association result between the first video data and the first work order data according to the analysis result, thereby realizing automatic determination of the association between video data and work order text in power service, solving the problem of low intelligent level and response efficiency in manual analysis of the association between video data and work order text in power service, improving the intelligent level and response efficiency of power service, and promoting the development of power service in a more efficient and intelligent direction.

[0021] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained according to these drawings without creative labor for those skilled in the art.

[0023] Figure 1 is a flow chart of a power service data analysis method provided by the first embodiment of the present application;

[0024] Figure 2 is a flow chart of a power service data analysis method provided by the second embodiment of the present application;

[0025] Figure 3 is a structural schematic diagram of a power service data analysis device provided by the third embodiment of the present application;

[0026] Figure 4 is a structural schematic diagram of an electronic device for implementing a power service data analysis method of the present application. DETAILED DESCRIPTION

[0027] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0029] Example 1

[0030] Figure 1 This is a flowchart of a power service data analysis method provided in Embodiment 1 of the present invention. This embodiment is applicable to the automatic analysis of the correlation between video data and work order text in power services. This method can be executed by a power service data analysis device, which can be implemented in hardware and / or software and can be configured in a power service data processing server or management platform. Figure 1 As shown, the method includes:

[0031] S101. Process the acquired first video data and first work order data related to power services to obtain second video data and second work order data.

[0032] It should be noted that power services encompass a series of power operation and maintenance, power repair, power management, and power customer service activities undertaken to ensure the safe and stable operation of the power system. These service scenarios may include, but are not limited to, power equipment inspection, fault reporting and handling, on-site work safety monitoring, electricity consumption inspection, and customer service response. Power service systems can record a large amount of video data of on-site power service situations and work order text data describing the tasks. Analyzing and establishing the correlation between video data and work order text can quickly locate the cause of faults, support intelligent work order allocation, and assist in service quality assessment.

[0033] The first video data can be raw video data collected from the power service site by image acquisition devices (such as cameras or inspection robots), covering dynamic visual content of scenarios such as on-site inspection, fault handling, and equipment operation. For example, the appearance of the equipment, the environmental conditions on site, and the operation process can all be presented intuitively through the video data.

[0034] The first work order data can be the original work order text data obtained from the records of the power service system, which carries semantic information such as fault reporting, service requests, and task assignment, and includes key content such as fault description, service requests, and processing requirements.

[0035] The second video data can be standardized video data generated after preprocessing. The second work order data can be standardized work order text data generated after preprocessing.

[0036] For example, this embodiment can acquire raw video data from the power service site and raw work order text data recorded in the power service system through communication and transmission with image acquisition equipment and power service system. Then, the raw video data and raw work order text data can be standardized to obtain standardized video data and standardized work order text data, thereby providing standardized data for subsequent feature extraction and correlation analysis.

[0037] Optionally, the process of processing the acquired first video data and first work order data related to power services to obtain second video data and second work order data includes:

[0038] Extract video frames from the first video data according to a preset time interval to obtain a target video frame set;

[0039] Normalize each video frame in the target video frame set to obtain the second video data;

[0040] The first work order data is cleaned, segmented, and encoded to obtain the second work order data.

[0041] It should be noted that the first video data can be a video dataset, including more than one video. The first work order data can be a work order dataset, including more than one work order text. This embodiment will use the processing of one video and one work order text as an example to illustrate the preprocessing process of the first video data and the first work order data.

[0042] For example, a video dataset can be represented by V: V = {v1, v2, v3, ..., v...} N}, where N can represent the total number of video data, V i It can represent the i-th video.

[0043] Based on this, for the i-th video V i Video frames can be extracted according to a preset fixed time interval Δt to obtain the i-th video V. i Target video frame set F i :F i ={f i1 f i2 , ..., f ik}, where k is the total number of video frames extracted, f ij It can represent the j-th video frame of the i-th video obtained from extraction.

[0044] Furthermore, to eliminate the influence of differences in frame size and pixel value range, each pixel in each video frame can be normalized to obtain normalized video frames. All normalized video frames are then combined into a video frame sequence, serving as the second video data. The normalization process for each pixel in the video frame can be as follows:

[0045] ;

[0046] in, It can be the normalized pixel matrix of the j-th video frame in the i-th video; (x, y) can represent pixel coordinates; It can be the pixel matrix of the j-th video frame in the i-th video; μ can be the mean of the pixel values ​​of all video frames in the i-th video; σ can be the standard deviation of the pixel values ​​of all video frames in the i-th video.

[0047] On the other hand, the work order dataset can be represented by W: W = {w1, w2, ..., w...} M}, where M can be the total number of work order data, w i It can represent the text of the i-th work order.

[0048] Based on this, for the i-th work order text w i First, the text content can be cleaned to remove special characters, punctuation marks, and meaningless words, resulting in the cleaned text T corresponding to the i-th work order text. i .

[0049] Then, the cleaned text T can be processed. i Perform word segmentation to obtain the word sequence S corresponding to the i-th work order text. i :S i ={s i1 s i2 , ..., s il}, where l can represent the total number of words after text segmentation. s ijIt can represent the j-th word in the i-th work order text.

[0050] Finally, each word segment can be uniformly mapped to a fixed dimension to obtain the word vector sequence E corresponding to the i-th work order text. i The word vector sequence is then used as the second work order data. The word vector sequence E... i ={e i1 e i2 , ..., e il}, e ij It can represent the word vector corresponding to the j-th word in the i-th work order text.

[0051] S102. Perform feature extraction on the second video data and the second work order data respectively to obtain the visual feature vector of the second video data and the semantic feature vector of the second work order data.

[0052] The visual feature vector can be a feature vector that represents the overall visual content of the video data, and can represent key visual information in the video data, such as device status, environmental scene, operation behavior and other visual information.

[0053] The semantic feature vector can be a feature vector that represents the overall semantic content of the work order text data, and can represent key semantic information in the text, such as descriptive content information such as fault type, location of occurrence, and processing requirements.

[0054] For example, a pre-defined deep learning model can automatically identify the spatial and temporal features in each video frame of the second video data, thereby obtaining a visual feature vector containing the overall visual content of the second video data. Simultaneously, a natural language processing model can automatically determine the semantic features in the text content of the second work order data, thereby generating a semantic feature vector containing the overall semantic information of the second work order data. It is understood that the entire feature extraction process requires no manual intervention, achieving automation and intelligence in feature extraction.

[0055] S103. The visual feature vector and the semantic feature vector are fused to obtain the target fused feature.

[0056] The target fusion features can simultaneously contain key information from both visual and semantic modalities, providing a data foundation for subsequent correlation analysis.

[0057] For example, this embodiment can combine visual feature vectors and semantic feature vectors from different modalities through vector concatenation operations to obtain a high-dimensional fusion feature to be processed. Then, dimensionality reduction and feature recombination can be performed on the fusion feature to be processed to eliminate redundant information and enhance key information between multimodal features, thereby generating a target fusion feature that simultaneously contains key information from both visual and semantic modalities.

[0058] S104. Analyze the target fusion features according to the expected fusion features, and determine the correlation result between the first video data and the first work order data based on the analysis results.

[0059] The expected fusion feature can be an ideal feature vector used to characterize the correlation between video and work order.

[0060] For example, when analyzing the target fusion features based on the expected fusion features, the probability of a correlation between the first video data and the first work order data can be determined by calculating the similarity between the two. The higher the similarity, the higher the probability that the first video data and the first work order data are associated. Furthermore, the association result between the first video data and the first work order data can be determined according to preset decision rules (such as threshold comparison), thereby achieving automatic identification of the correlation between video and work order.

[0061] The technical solution of this embodiment can process the acquired first video data and first work order data related to power services to obtain second video data and second work order data. Features are then extracted from the second video data and the second work order data to obtain visual feature vectors and semantic feature vectors for the second video data and the second work order data, respectively. These visual feature vectors and semantic feature vectors are then fused to obtain target fused features. Finally, the target fused features are analyzed based on the expected fused features, and the association between the first video data and the first work order data is determined based on the analysis results. This achieves automatic determination of the association between video data and work order text in power services, solving the problem of low intelligence and response efficiency in manual analysis of the association between video data and work order text in power services. It improves the intelligence level and response efficiency of power services, promoting the development of power services towards a more efficient and intelligent direction.

[0062] Based on the above embodiments, the present invention also provides an optional embodiment. This optional embodiment can further optimize the analysis of the target fusion features based on the expected fusion features and the determination of the association result between the first video data and the first work order data based on the analysis results, and may include:

[0063] Calculate the two-dimensional similarity between the expected fusion feature and the target fusion feature to obtain the similarity result;

[0064] If the similarity result meets the preset conditions, then the correlation between the first video data and the first work order data is taken as the correlation result.

[0065] It should be noted that relying solely on the similarity between the expected fusion features and the target fusion features may result in low accuracy in determining the correlation.

[0066] To improve the accuracy of correlation judgment, this optional embodiment can enhance the similarity judgment between the expected fusion feature and the target fusion feature by calculating the two-dimensional similarity between the expected fusion feature and the target fusion feature, thereby realizing two-dimensional similarity calculation and improving the accuracy of judging the correlation between video data and work order text.

[0067] For example, after determining the target fusion feature and the expected fusion feature, the cosine similarity between the two can be calculated to measure directional consistency, thereby obtaining a first similarity score between them. The first similarity score can be the cosine similarity value between the target fusion feature and the expected fusion feature, used to represent the directional consistency between the target fusion feature and the expected fusion feature.

[0068] Simultaneously, the numerical similarity can be measured by calculating the normalized Euclidean distance between the two features, thus obtaining the first similarity score. The second similarity score can be the normalized value of the Euclidean distance between the target fused feature and the expected fused feature, used to represent the numerical similarity between the target fused feature and the expected fused feature.

[0069] Finally, considering that the directional consistency and numerical similarity of the target fusion feature and the expected fusion feature contribute differently to the correlation, an adjustable weight can be introduced to weight the first similarity score and the second similarity score to obtain the final similarity result. The correlation result can be determined according to preset conditions (such as threshold comparison).

[0070] Example 2

[0071] Figure 2 This is a flowchart of a power service data analysis method provided in Embodiment 2 of the present invention. This embodiment can further optimize step S102 of Embodiment 1, which involves extracting features from the second video data and the second work order data to obtain the visual feature vector of the second video data and the semantic feature vector of the second work order data. Figure 2 As shown, the method includes:

[0072] S201. Process the acquired first video data and first work order data related to power services to obtain second video data and second work order data.

[0073] S202. Perform depth feature extraction on each video frame in the second video data to obtain the frame-level features corresponding to the video frame.

[0074] The frame-level features can be feature maps extracted from a single video frame that can characterize the local visual information of that frame.

[0075] For example, after obtaining the preprocessed second video data, a pre-trained deep convolutional neural network (DCNN) model can be used to extract features from each video frame in the second video data. This DCNN model can extract features from each video frame layer by layer through stacked convolutional layers and activation functions, thereby outputting high-level frame-level feature maps with strong representational capabilities. The DCNN model can be a ResNet50 model.

[0076] This embodiment can be illustrated by taking the extraction of features from a single video frame to obtain the corresponding frame-level features as an example.

[0077] Specifically, the second video data can be represented as F i ={f i1 f i2 , ..., f ik}, where k can be the total number of video frames, f ij It can be represented as the j-th video frame in the i-th second video data.

[0078] For a single video frame f ij By using the first 17 convolutional layers and activation functions in a deep convolutional neural network model to extract features from video frames layer by layer, the following can be achieved:

[0079] ;

[0080] in, It can be used for video frame f ij The feature map output by the l-th layer in a deep convolutional neural network model, and the initial input. It can be a video frame f ij ReLU() can be an activation function; K l It can be the convolution kernel of the l-th layer; b l It can be the bias term of the l-th layer; It can be used for video frame f ij The final output of the deep convolutional neural network model is a frame-level feature map.

[0081] S203. Determine the temporal correlation between each of the frame-level features, and generate the visual feature vector of the second video data based on the temporal correlation.

[0082] The temporal correlation can be the temporal relationship between frame-level features of consecutive video frames in the second video data. The visual feature vector can be the global feature vector of the second video data.

[0083] It should be noted that after obtaining the frame-level features corresponding to each video frame in the second video data, the frame-level features corresponding to each video frame can be further correlated to obtain the global feature vector of the second video data.

[0084] For example, in this embodiment, a global feature vector of the second video data can be constructed by temporal correlation between features at each frame level.

[0085] Specifically, for each video frame corresponding to a given frame-level feature map, global average pooling can be used to transform the two-dimensional frame-level feature map into a one-dimensional frame-level feature vector to eliminate spatial size differences. Then, the frame-level feature vectors corresponding to each video frame are summarized to obtain a feature vector sequence.

[0086] Then, the temporal correlation between frame-level feature vectors can be captured by inputting the feature vector sequence into a preset recurrent neural network model (which can be a bidirectional LSTM), thereby performing feature concatenation on each frame-level feature vector to obtain the global feature vector of the second video data, which serves as the visual feature vector of the second video data.

[0087] S204. Perform multi-scale feature extraction on the second work order data to obtain the local semantic feature matrix of the second work order data at each scale.

[0088] The local semantic feature matrix can be a semantic feature vector matrix that reflects the combined words of the second work order data at various scales.

[0089] For example, when performing multi-scale feature extraction on the second work order data, various convolutional kernels of different widths can be used to extract features from the word vector sequence in the second work order data in parallel. Each convolutional kernel of a certain width can be responsible for capturing the semantic feature vectors of each word combination (such as a combination of two adjacent words or a combination of three adjacent words) of a specific length, and outputting the local semantic feature matrix at the corresponding scale.

[0090] Understandably, this method can simultaneously capture the local semantic feature matrix of phrase combinations (such as combinations of two adjacent words) and the local semantic feature matrix of long phrase combinations (such as combinations of four adjacent words) in the work order text.

[0091] S205. By pre-setting attention weights, the local semantic feature matrix is ​​weighted and fused to generate key sentence-level features.

[0092] The preset attention weights can represent the importance of the local semantic feature matrices at various scales. These preset attention weights can be pre-set by relevant personnel based on experience, or they can be determined in real-time based on the local semantic feature matrices. This embodiment does not impose specific limitations on this.

[0093] The key sentence-level features can be feature vectors that, after attention filtering and enhancement, can highlight the core semantic information of the work order text at the corresponding scale.

[0094] For example, after determining the local semantic feature matrix of the second work order data at the corresponding scale, the feature vectors of each combined word in the local feature matrix can be weighted and combined based on the preset attention weight to highlight key semantics (such as "fault location" and "handling solution"), and obtain the core semantic information of the work order text at the corresponding scale as key sentence-level features, so as to automatically ignore redundant or secondary information and focus on core descriptions such as "fault location" and "equipment model".

[0095] S206. Concatenate the key sentence-level features at each scale to obtain the semantic feature vector of the second work order data.

[0096] The semantic feature vector can be the global semantic feature of the second work order data, which can comprehensively represent the overall semantic content of the work order text.

[0097] For example, by connecting the key sentence-level features at various scales along the vector dimension, a global feature vector of the second work order data can be formed, thus obtaining the semantic feature vector of the second work order data.

[0098] S207. The visual feature vector and the semantic feature vector are fused to obtain the target fused feature.

[0099] S208. Analyze the target fusion features according to the expected fusion features, and determine the correlation result between the first video data and the first work order data based on the analysis results.

[0100] The technical solution of this embodiment can extract deep features from each video frame in the second video data to obtain frame-level features corresponding to the video frames, determine the temporal correlation between the frame-level features, and generate a visual feature vector of the second video data based on the temporal correlation. At the same time, multi-scale feature extraction is performed on the second work order data to obtain local semantic feature matrices of the second work order data at each scale. The local semantic feature matrices are then weighted and fused using preset attention weights to generate key sentence-level features. Finally, the key sentence-level features at each scale are concatenated to obtain the semantic feature vector of the second work order data. This approach of first extracting local features and then constructing global features can achieve accurate feature extraction of video data and work order data, thereby laying a more reliable feature foundation for subsequent accurate correlation determination and further improving the accuracy of power service video-work order correlation analysis.

[0101] Based on the above embodiments, the present invention also provides an optional embodiment. This optional embodiment can further optimize the feature fusion of the visual feature vector and the semantic feature vector to obtain the target fused feature, and may include:

[0102] Determine a first weight of the visual feature vector relative to the semantic feature vector, and a second weight of the semantic feature vector relative to the visual feature vector;

[0103] Based on the first weight and the second weight, the target fusion feature is obtained by weighted summation of the visual feature vector and the semantic feature vector.

[0104] The first weight represents the importance of the visual feature vector to the current semantic feature vector. The second weight represents the importance of the semantic feature vector to the current visual feature vector.

[0105] For example, in this optional embodiment, the visual feature vector and the semantic feature vector can first be mapped to a common feature space, so that the features of the two modalities have the same dimension, which facilitates subsequent weight calculation and feature interaction. Then, the first weight of the visual feature vector relative to the semantic feature vector and the second weight of the semantic feature vector relative to the visual feature vector can be determined by calculating the proportion of the similarity between the current video and the current work order feature pair to the total similarity between all video and current work order feature pairs, and the proportion of the similarity between the current work order and the current video feature pair to the total similarity between all work order and current video feature pairs. Finally, based on the first and second weights, the visual feature vector and the semantic feature vector mapped to the same feature space are weighted and summed to obtain the target fusion feature.

[0106] Specifically, mapping the current visual feature vector and semantic feature vector to a common feature space can be achieved as follows:

[0107] ;

[0108] in, It can be the visual feature vector of the second video data. It can be the visual feature vector mapped to the feature space; It can be the semantic feature vector of the second work order data. It can be a semantic feature vector mapped to the feature space; It can be a weight matrix used for visual feature vector mapping. It can be a weight matrix used for semantic feature vector mapping; It can be a bias vector used for visual feature vector mapping. It can be a bias vector used for semantic feature vector mapping.

[0109] Furthermore, the similarity between the current video and the current work order feature pair is calculated as a proportion of the total similarity between all video and current work order feature pairs. This determines the first weight of the visual feature vector relative to the semantic feature vector, which can be:

[0110] ;

[0111] Where, α i It can be the first weight of the current visual feature vector relative to the current semantic feature vector; It can represent the similarity score (dot product result) of the feature pairs between the current video and the current work order. Let represent the similarity score between the feature pair of the m-th video and the current work order, where m ∈ (1, N), and N is the total number of videos in the second video data; () T It can be a transpose function, used to transpose the semantic feature vector mapped to the same feature space; exp() can be an exponential function, used to map the similarity score (dot product result) to the positive number space.

[0112] Furthermore, the similarity between the current work order and the current video feature pair is calculated as a proportion of the total similarity between all work orders and the current video feature pairs to determine the second weight of the semantic feature vector relative to the visual feature vector. Specifically, this can be done as follows:

[0113] ;

[0114] Where, β i It can be the second weight of the current semantic feature vector relative to the visual feature vector; It can represent the similarity score (dot product result) of the feature pairs between the current work order and the current video. Let represent the similarity score between the feature pairs of the nth work order and the current video, where n ∈ (1, M), and M is the total number of work order texts in the second work order data; ( ) T It can be a transpose function, used to transpose the visual feature vector mapped to the same feature space; exp() can be an exponential function, used to map the similarity score (dot product result) to the positive number space.

[0115] Example 3

[0116] Figure 3 This is a schematic diagram of the structure of a power service data analysis device provided in Embodiment 3 of the present invention. Figure 3 As shown, the device includes:

[0117] Processing module 31 can be used to process the acquired first video data and first work order data related to power services to obtain second video data and second work order data.

[0118] The extraction module 32 can be used to extract features from the second video data and the second work order data respectively, to obtain the visual feature vector of the second video data and the semantic feature vector of the second work order data;

[0119] The fusion module 33 can be used to fuse the visual feature vector and the semantic feature vector to obtain the target fused feature;

[0120] The determination module 34 can be used to analyze the target fusion features according to the expected fusion features, and determine the correlation result between the first video data and the first work order data based on the analysis results.

[0121] The technical solution of this embodiment can process the acquired first video data and first work order data related to power services to obtain second video data and second work order data. Features are then extracted from the second video data and the second work order data to obtain visual feature vectors and semantic feature vectors for the second video data and the second work order data, respectively. These visual feature vectors and semantic feature vectors are then fused to obtain target fused features. Finally, the target fused features are analyzed based on the expected fused features, and the association between the first video data and the first work order data is determined based on the analysis results. This achieves automatic determination of the association between video data and work order text in power services, solving the problem of low intelligence and response efficiency in manual analysis of the association between video data and work order text in power services. It improves the intelligence level and response efficiency of power services, promoting the development of power services towards a more efficient and intelligent direction.

[0122] Optionally, the processing module 31 can be specifically used to extract video frames of the first video data according to a preset time interval to obtain a target video frame set;

[0123] Normalize each video frame in the target video frame set to obtain the second video data;

[0124] The first work order data is cleaned, segmented, and encoded to obtain the second work order data.

[0125] Optionally, the extraction module 32 can be specifically used to extract depth features from each video frame in the second video data to obtain frame-level features corresponding to the video frame;

[0126] Determine the temporal correlation between each of the frame-level features, and generate a visual feature vector of the second video data based on the temporal correlation.

[0127] Optionally, the extraction module 32 can also be used to perform multi-scale feature extraction on the second work order data to obtain the local semantic feature matrix of the second work order data at each scale.

[0128] By pre-setting attention weights, the local semantic feature matrix is ​​weighted and fused to generate key sentence-level features;

[0129] The key sentence-level features at each scale are concatenated to obtain the semantic feature vector of the second work order data.

[0130] Optionally, the fusion module 33 can be specifically used to determine a first weight of the visual feature vector relative to the semantic feature vector, and a second weight of the semantic feature vector relative to the visual feature vector;

[0131] Based on the first weight and the second weight, the target fusion feature is obtained by weighted summation of the visual feature vector and the semantic feature vector.

[0132] Optionally, the determining module 34 can be specifically used to calculate the two-dimensional similarity between the expected fusion feature and the target fusion feature to obtain a similarity result;

[0133] If the similarity result meets the preset conditions, then the correlation between the first video data and the first work order data is taken as the correlation result.

[0134] The power service data analysis device provided in this embodiment of the invention can execute the power service data analysis method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of executing the method.

[0135] Example 4

[0136] Figure 4 A schematic diagram of an electronic device 40 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0137] like Figure 4 As shown, the electronic device 40 includes at least one processor 41 and a memory, such as a read-only memory (ROM) 42 or a random access memory (RAM) 43, communicatively connected to the at least one processor 41. The memory stores computer programs executable by the at least one processor. The processor 41 can perform various appropriate actions and processes based on the computer program stored in the ROM 42 or loaded from storage unit 48 into the RAM 43. The RAM 43 may also store various programs and data required for the operation of the electronic device 40. The processor 41, ROM 42, and RAM 43 are interconnected via a bus 44. An input / output (I / O) interface 45 is also connected to the bus 44.

[0138] Multiple components in electronic device 40 are connected to I / O interface 45, including: input unit 46, such as keyboard, mouse, etc.; output unit 47, such as various types of monitors, speakers, etc.; storage unit 48, such as disk, optical disk, etc.; and communication unit 49, such as network card, modem, wireless transceiver, etc. Communication unit 49 allows electronic device 40 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0139] Processor 41 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 41 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 41 performs the various methods and processes described above, such as power service data analysis methods.

[0140] In some embodiments, the power service data analysis method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 48. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 40 via ROM 42 and / or communication unit 49. When the computer program is loaded into RAM 43 and executed by processor 41, one or more steps of the power service data analysis method described above may be performed. Alternatively, in other embodiments, processor 41 may be configured to perform the power service data analysis method by any other suitable means (e.g., by means of firmware).

[0141] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0142] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0143] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0144] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0145] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0146] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0147] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the expected results of the technical solution of this invention can be achieved, and this is not limited herein.

[0148] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method of analyzing electric service data, the method comprising: The method comprises the following steps: processing the obtained first video data and first work order data related to power service to obtain second video data and second work order data; extracting features from the second video data and the second work order data respectively to obtain a visual feature vector of the second video data and a semantic feature vector of the second work order data; performing feature fusion on the visual feature vector and the semantic feature vector to obtain a target fusion feature; analyzing the target fusion feature according to an expected fusion feature, and determining an association result between the first video data and the first work order data according to an analysis result.

2. The method of claim 1, wherein, The processing of the obtained first video data and first work order data related to power service to obtain second video data and second work order data comprises the following steps: extracting video frames of the first video data according to a preset time interval to obtain a target video frame set; performing normalization processing on each video frame in the target video frame set to obtain the second video data; performing text cleaning, word segmentation and coding on the first work order data to obtain the second work order data.

3. The method of claim 1, wherein, The feature extraction from the second video data to obtain the visual feature vector of the second video data comprises the following steps: performing deep feature extraction on each video frame in the second video data to obtain frame-level features corresponding to the video frames; determining the time sequence association between the frame-level features, and generating a visual feature vector of the second video data based on the time sequence association.

4. The method of claim 1, wherein, The feature extraction from the second work order data to obtain the semantic feature vector of the second work order data comprises the following steps: performing multi-scale feature extraction on the second work order data to obtain local semantic feature matrices of the second work order data at each scale; performing weighted fusion on the local semantic feature matrices through a preset attention weight to generate key sentence-level features; splicing the key sentence-level features at each scale to obtain a semantic feature vector of the second work order data.

5. The method of claim 1, wherein, The feature fusion on the visual feature vector and the semantic feature vector to obtain a target fusion feature comprises the following steps: determining a first weight of the visual feature vector relative to the semantic feature vector, and a second weight of the semantic feature vector relative to the visual feature vector; based on the first weight and the second weight, performing weighted summation on the visual feature vector and the semantic feature vector to obtain the target fusion feature.

6. The method of claim 1, wherein, The analysis of the target fusion feature according to an expected fusion feature, and the determination of an association result between the first video data and the first work order data according to an analysis result comprises the following steps: calculating a two-dimensional similarity between the expected fusion feature and the target fusion feature to obtain a similarity result; if the similarity result meets a preset condition, it is determined that there is an association between the first video data and the first work order data as the association result.

7. An electric service data analysis apparatus characterized by comprising: The method comprises the following steps: a processing module for processing the obtained first video data and first work order data related to power service to obtain second video data and second work order data; The extraction module is configured to perform feature extraction on the second video data and the second work order data respectively to obtain a visual feature vector of the second video data and a semantic feature vector of the second work order data. The fusion module is configured to perform feature fusion on the visual feature vector and the semantic feature vector to obtain a target fusion feature. The determination module is configured to analyze the target fusion feature according to an expected fusion feature, and determine an association result between the first video data and the first work order data according to an analysis result.

8. An electronic device, comprising: The electronic device comprises: at least one processor; and a memory connected to the at least one processor in communication, wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to perform the power service data analysis method of any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions for enabling the processor to perform the power service data analysis method of any one of claims 1-6 when executed.

10. A computer program product, characterised in that, The computer program product comprises a computer program, and the computer program implements the power service data analysis method of any one of claims 1-6 when executed by the processor.

Images