Davy work order element inspection method and device, electronic equipment and storage medium

Abstract

This invention provides a method, apparatus, electronic device, and storage medium for quality inspection of maintenance work orders. The method includes: acquiring maintenance work order data to be inspected, the maintenance work order data including work order text data and on-site image data; extracting text feature vectors from the work order text data and extracting image feature vectors from the on-site image data; inputting the text feature vectors and image feature vectors into a work order quality inspection model, whereby the work order quality inspection model, based on multiple preset quality inspection dimensions, fuses the text feature vectors and image feature vectors to obtain a fused feature vector corresponding to each quality inspection dimension; and based on the fused feature vectors corresponding to each quality inspection dimension, performs quality inspection on the maintenance work order data for each quality inspection dimension, and outputs the quality inspection result of the maintenance work order data. This invention improves the efficiency and accuracy of quality inspection by employing a work order quality inspection model, fusing text feature vectors and image feature vectors, and performing multi-dimensional quality inspection on maintenance work order data based on the fused feature vectors.

Description

Technical Field

[0001] This invention relates to the field of intelligent operation and maintenance technology, and in particular to a method, device, electronic device and storage medium for quality inspection of maintenance work orders. Background Technology

[0002] With the continuous development of communication technology, the scale of communication networks is growing rapidly, the number of network devices is increasing dramatically, and the network maintenance work involving outsourced maintenance is also constantly increasing. In order to better manage outsourced maintenance work, operators need to conduct quality inspections on outsourced maintenance work orders. At present, traditional outsourced maintenance work order quality inspection methods rely on manual inspection, which is inefficient and difficult to cope with the quality inspection needs of massive work orders; moreover, it is easily affected by the subjective factors of quality inspectors, and the accuracy and consistency of quality inspection results are difficult to guarantee. Summary of the Invention

[0003] This invention provides a method, apparatus, electronic device, and storage medium for quality inspection of outsourced maintenance work orders, in order to solve the shortcomings of low efficiency and poor accuracy in existing outsourced maintenance work order quality inspection methods.

[0004] This invention provides a method for quality inspection of maintenance work orders, comprising: Obtain the maintenance work order data to be inspected, the maintenance work order data including work order text data and on-site image data; Extract the text feature vector from the work order text data and extract the image feature vector from the on-site image data; The text feature vector and the image feature vector are input into the work order quality inspection model. The work order quality inspection model fuses the text feature vector and the image feature vector based on multiple preset quality inspection dimensions to obtain a fused feature vector corresponding to each quality inspection dimension. Based on the fused feature vector corresponding to each quality inspection dimension, the maintenance work order data is inspected for each quality inspection dimension, and the quality inspection result of the maintenance work order data is output. The work order quality inspection model is trained based on the text feature vector samples and image feature vector samples corresponding to the maintenance work order data samples, as well as the quality inspection result labels of the maintenance work order data samples.

[0005] In some embodiments, the step of fusing the text feature vector and the image feature vector based on multiple preset quality inspection dimensions to obtain a fused feature vector corresponding to each quality inspection dimension includes: For each quality inspection dimension, determine the first attention weight of the text feature vector and the second attention weight of the image feature vector; Based on the first attention weight and the second attention weight, the text feature vector and the image feature vector are fused to obtain the fused feature vector corresponding to each quality inspection dimension.

[0006] In some embodiments, the step of performing quality inspection on the maintenance work order data for each of the quality inspection dimensions based on the fused feature vector corresponding to each of the quality inspection dimensions includes: Based on the fusion feature vector corresponding to each of the quality inspection dimensions and each of the quality inspection dimensions, a subset of rules corresponding to each of the quality inspection dimensions is determined from a pre-built expert rule base; Based on the fusion feature vector corresponding to each of the quality inspection dimensions and the rule subset corresponding to each of the quality inspection dimensions, the maintenance work order data is inspected for each of the quality inspection dimensions to obtain the quality inspection results for each of the quality inspection dimensions. The quality inspection results of each of the aforementioned quality inspection dimensions are combined to obtain the quality inspection results of the maintenance work order data.

[0007] In some embodiments, the multiple quality inspection dimensions include: semantic consistency dimension, image repeatability dimension, and facility standardization dimension.

[0008] In some embodiments, obtaining the maintenance work order data to be inspected includes: Obtain the original maintenance work order data, which includes the original work order text data and the original on-site image data; The original work order text data and original site image data are preprocessed and correlated to obtain the maintenance work order data.

[0009] In some embodiments, after outputting the quality inspection results of the maintenance work order data, the method further includes: If the quality inspection results indicate that there is an anomaly in the maintenance work order data, a potential hazard warning message will be generated. The hazard warning information is sent to the client.

[0010] In some embodiments, the work order quality inspection model is trained based on the following steps: Obtain maintenance work order data samples, which include work order text data samples and on-site image data samples; Feature extraction is performed on the work order text data sample to obtain a text feature vector sample, and feature extraction is performed on the on-site image data sample to obtain an image feature vector sample; Determine the quality inspection result label of the maintenance work order data sample; Using the text feature vector samples and the image feature vector samples as training samples, and the quality inspection result labels as sample labels, an initial work order quality inspection model is trained. After training, the work order quality inspection model is obtained.

[0011] The present invention also provides a maintenance work order quality inspection device, comprising: The acquisition unit is used to acquire maintenance work order data to be inspected, the maintenance work order data including work order text data and on-site image data; The feature extraction unit is used to extract the text feature vector of the work order text data and the image feature vector of the on-site image data; The quality inspection unit is used to input the text feature vector and the image feature vector into the work order quality inspection model. The work order quality inspection model fuses the text feature vector and the image feature vector based on multiple preset quality inspection dimensions to obtain a fused feature vector corresponding to each quality inspection dimension. Based on the fused feature vector corresponding to each quality inspection dimension, the maintenance work order data is inspected for each quality inspection dimension, and the quality inspection result of the maintenance work order data is output. The work order quality inspection model is trained based on the text feature vector samples and image feature vector samples corresponding to the maintenance work order data samples, as well as the quality inspection result labels of the maintenance work order data samples.

[0012] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the maintenance work order quality inspection method as described above.

[0013] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the maintenance work order quality inspection method as described above.

[0014] The maintenance work order quality inspection method, device, electronic device, and storage medium provided by this invention acquire the maintenance work order data to be inspected, extract text feature vectors from the work order text data, and extract image feature vectors from the on-site image data; input the text feature vectors and image feature vectors into the work order quality inspection model, which fuses the text feature vectors and image feature vectors based on multiple preset quality inspection dimensions to obtain a fused feature vector corresponding to each quality inspection dimension; and perform quality inspection on the maintenance work order data for each quality inspection dimension based on the fused feature vector corresponding to each quality inspection dimension, outputting the quality inspection results of the maintenance work order data, thereby improving the efficiency and accuracy of maintenance work order quality inspection. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0016] Figure 1This is a flowchart illustrating the quality inspection method for maintenance work orders provided in this embodiment of the invention.

[0017] Figure 2 This is a flowchart illustrating the training process of the work order quality inspection model provided in this embodiment of the invention.

[0018] Figure 3 This is a schematic diagram of the structure of the maintenance work order quality inspection device provided in an embodiment of the present invention.

[0019] Figure 4 This is a schematic diagram of the structure of the electronic device provided in an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0021] The terms "first," "second," etc., used in this invention are used to distinguish similar objects, not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of the invention can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of the same class, not limited in number; for example, the first object can be one or more. Furthermore, in this invention, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0022] Figure 1 This is a flowchart illustrating the quality inspection method for maintenance work orders provided in an embodiment of the present invention. Figure 1 As shown, a method for quality inspection of maintenance work orders is provided, including the following steps: step 110, step 120, and step 130. This method's steps are merely one possible implementation of the present invention.

[0023] Step 110: Obtain the maintenance work order data to be inspected. The maintenance work order data includes work order text data and on-site image data. Optionally, the work order text data includes, but is not limited to: inspection results, equipment information, anomaly descriptions, and return work workload; the on-site image data can be image data of on-site equipment such as manholes, utility poles, and marker stones; the on-site image data can contain watermarks, such as timestamps, location information, and work order IDs.

[0024] In some embodiments, obtaining maintenance work order data to be inspected includes: Obtain the original maintenance work order data, which includes the original work order text data and the original site image data. Preprocessing and correlation processing are performed on the original work order text data and the original site image data to obtain the maintenance work order data.

[0025] Optionally, the original work order text data can be cleaned to remove invalid characters and extract key fields for logical verification; the original on-site image data can be standardized by adjusting the resolution, unifying the format, and automatically detecting the integrity of the image, such as blurriness and occlusion.

[0026] Step 120: Extract the text feature vector from the work order text data and extract the image feature vector from the on-site image data; Optionally, natural language processing techniques, such as BERT and other pre-trained models, can be used to segment and semantically encode the work order text data to generate high-dimensional text feature vectors. The text feature vectors contain semantic features related to each quality inspection dimension, such as keywords like "damage" and "detachment" in the anomaly description, and workload values.

[0027] Optionally, a convolutional neural network model is used to detect the equipment status, construction status, and environmental status based on on-site image data, and to identify image feature vectors; the image feature vectors include, but are not limited to: equipment hazard identification features, construction standard features, and environmental features.

[0028] Step 130: Input the text feature vector and image feature vector into the work order quality inspection model. The work order quality inspection model fuses the text feature vector and image feature vector based on multiple preset quality inspection dimensions to obtain the fused feature vector corresponding to each quality inspection dimension. Based on the fused feature vector corresponding to each quality inspection dimension, the maintenance work order data is inspected for each quality inspection dimension, and the quality inspection results of the maintenance work order data are output. The work order quality inspection model is trained based on the text feature vector samples and image feature vector samples corresponding to the maintenance work order data samples, as well as the quality inspection result labels of the maintenance work order data samples.

[0029] Optionally, quality inspection results include, but are not limited to, work order quality levels and issue markers for different quality inspection dimensions. Work order quality levels must include at least: High-quality work orders: all quality inspection dimensions meet the standards; Poor quality work orders: Work orders that have problems in one or more quality inspection dimensions, such as incomplete content or insufficient standardization.

[0030] Fake work orders: Work orders that are determined to be highly suspicious or fake through image repetition detection and spatiotemporal logic conflict detection.

[0031] Optionally, the work order quality inspection model includes a feature fusion layer and a quality inspection layer; the feature fusion layer is used to fuse text feature vectors and image feature vectors based on multiple preset quality inspection dimensions to obtain a fused feature vector corresponding to each quality inspection dimension; the quality inspection layer is used to perform quality inspection on the maintenance work order data for each quality inspection dimension based on the fused feature vector corresponding to each quality inspection dimension, and output the quality inspection results of the maintenance work order data.

[0032] In some embodiments, the multiple quality inspection dimensions include: semantic consistency dimension, image repeatability dimension, and facility standardization dimension.

[0033] Optionally, multiple quality inspection dimensions include completeness, authenticity, standardization, and timeliness.

[0034] For example, check if any required fields are missing; verify the authenticity and reliability of the work order content, such as identifying fake work orders through image duplication detection and text-image semantic consistency verification; check whether the construction or inspection process complies with safety and operation specifications, such as identifying the wearing of safety helmets and the setting of fences in the construction area through image recognition; and check whether the time difference between the work order submission time and the on-site image capture time is within a preset threshold range.

[0035] Optionally, for the semantic consistency dimension, the semantic matching degree between text and image is checked to obtain consistency quality inspection results; for the image repetition dimension, the repetition of images is detected to obtain repetition quality inspection results; for the facility standardization dimension, the construction standardization is detected to obtain standardization quality inspection results.

[0036] For example, if the text description device is working properly, but the image recognition cable is exposed, there is a logical conflict between the two, and the text and image are inconsistent; if there are multiple duplicate images, the authenticity is questionable.

[0037] In this embodiment of the invention, by acquiring the maintenance work order data to be inspected, text feature vectors are extracted from the work order text data, and image feature vectors are extracted from the on-site image data. The text feature vectors and image feature vectors are input into the work order quality inspection model. The work order quality inspection model fuses the text feature vectors and image feature vectors based on multiple preset quality inspection dimensions to obtain a fused feature vector corresponding to each quality inspection dimension. Based on the fused feature vector corresponding to each quality inspection dimension, the maintenance work order data is inspected for each quality inspection dimension, and the quality inspection result of the maintenance work order data is output, thereby improving the efficiency and accuracy of maintenance work order quality inspection.

[0038] In some embodiments, based on multiple preset quality inspection dimensions, text feature vectors and image feature vectors are fused to obtain a fused feature vector corresponding to each quality inspection dimension, including: For each quality inspection dimension, determine the first attention weight of the text feature vector and the second attention weight of the image feature vector; Based on the first attention weight and the second attention weight, the text feature vector and the image feature vector are fused to obtain the fused feature vector corresponding to each quality inspection dimension.

[0039] Among them, the fused feature vector is a new, more discriminative feature representation formed by effectively combining text feature vectors and image feature vectors for a specific quality inspection dimension. This vector can comprehensively reflect the correlation and contribution of text and image information in this dimension.

[0040] The first attention weight represents the importance of the text feature vector relative to a certain quality inspection dimension; the second attention weight represents the importance of the image feature vector for the same quality inspection dimension.

[0041] Understandably, by using attention mechanisms, text feature vectors and image feature vectors can be dynamically and selectively fused for different quality inspection dimensions, thereby improving the targeting of the fused feature vectors.

[0042] In some embodiments, step 130, based on the fused feature vector corresponding to each quality inspection dimension, performs quality inspection on the maintenance work order data for each quality inspection dimension, including: Step 131: Based on the fusion feature vector corresponding to each quality inspection dimension and each quality inspection dimension, determine the rule subset corresponding to each quality inspection dimension from the pre-built expert rule base; The expert rule base contains multiple rules for automated auditing of maintenance work orders. These rules are pre-built based on the knowledge and experience of domain experts, who can be senior operations engineers or quality inspectors.

[0043] For example, Rule 1: If the probability of image duplication is greater than 0.95, the authenticity dimension quality check fails; Rule 2: If the difference between the work order submission time and the on-site image capture time is greater than 2 hours, the timeliness dimension quality check fails.

[0044] Here, a rule subset refers to a set of rules most relevant to a specific quality inspection dimension, selected from the complete expert rule base. For example, the rule subset for judging authenticity includes image repeatability detection rules and time difference detection rules; while the rule subset for judging standardization includes safety helmet wearing recognition rules and construction area fence detection rules.

[0045] Step 132: Based on the fusion feature vector corresponding to each quality inspection dimension and the rule subset corresponding to each quality inspection dimension, perform quality inspection on the maintenance work order data for each quality inspection dimension to obtain the quality inspection result for each quality inspection dimension; Optionally, the fused feature vectors are passed through the model's classification / regression layer, outputting a probability value or state label, for example, the probability of image repetition is 0.97. The rule engine obtains this output value and compares it with the threshold or condition in the rule subset. For example, one rule in the rule subset is: if the probability of image repetition is greater than 0.95, then the authenticity dimension quality check fails. After executing all rules in the rule subset, the quality check result for that dimension is generated, such as authenticity: failed.

[0046] Step 133: Integrate the quality inspection results of each quality inspection dimension to obtain the quality inspection results of the maintenance work order data.

[0047] Optionally, if any dimension is judged to be false, the final quality inspection result will be directly marked as a false work order and the strictest handling will be triggered, such as freezing settlement and issuing a notice; if the quality inspection of the core dimension passes, but the secondary dimension fails, it can be marked as a poor quality work order and trigger manual review; if all dimensions pass, it will be marked as a high-quality work order.

[0048] In this embodiment of the invention, the maintenance work order data is inspected for each quality inspection dimension based on the fused feature vector corresponding to each quality inspection dimension and the rule subset corresponding to each quality inspection dimension, so as to obtain the quality inspection result of each quality inspection dimension; the quality inspection results of each quality inspection dimension are fused to obtain the quality inspection result of the maintenance work order data, thereby improving the accuracy, flexibility and interpretability of the quality inspection.

[0049] In some embodiments, after outputting the quality inspection results of the maintenance work order data, the method further includes: If the quality inspection results indicate that there are anomalies in the maintenance work order data, a potential hazard warning message will be generated. Send hazard warning information to the client.

[0050] Optionally, abnormal situations include, but are not limited to: Work order authenticity anomalies: such as identifying fake work orders with duplicate images or inconsistent spatiotemporal logic; Potential hazards in on-site facilities: such as equipment damage and exposed cables detected through image recognition; Abnormalities in construction compliance: such as not wearing a safety helmet or not setting up a fence in the construction area; Logical inconsistency anomaly: For example, the text describes the device as working properly, but the image recognition shows exposed cables.

[0051] Optionally, based on the quality inspection results and preset business rules, it can be determined whether there are any anomalies in the maintenance work order data. If so, a structured potential hazard warning message can be generated.

[0052] For example: Warning Title: High-Risk Hazard Alert; Notification: Work order ID [XXXXX] has a potential equipment hazard. AI has detected a missing manhole cover. Please check and handle it immediately. Related data: Simultaneously link the problematic image, the specific quality inspection rules that were triggered, and the basic information of the work order, providing sufficient context for subsequent processing.

[0053] In this embodiment of the invention, when anomalies are found in the maintenance work order data based on the quality inspection results, a potential hazard warning message is generated; the hazard warning message is sent to the client, which enables proactive operation and maintenance and real-time intervention, forming a closed-loop management of quality inspection and improving the efficiency and accuracy of hazard investigation.

[0054] Figure 2 This is a flowchart illustrating the training process of the work order quality inspection model provided in this embodiment of the invention, as shown below. Figure 2 As shown, in some embodiments, the work order quality inspection model is trained based on the following steps: Step 210: Obtain maintenance work order data samples, which include work order text data samples and on-site image data samples. Optionally, work order text data samples may include, but are not limited to: inspection results, equipment information, anomaly descriptions, return work workload, etc.; on-site image data samples may be image data of on-site equipment such as manholes, utility poles, and marker stones.

[0055] Step 220: Extract features from the work order text data sample to obtain text feature vector samples, and extract features from the on-site image data sample to obtain image feature vector samples; Optionally, natural language processing technology is used to extract features from work order text data samples to obtain text feature vector samples; a convolutional neural network model is used to extract features from on-site image data samples to obtain image feature vector samples.

[0056] Step 230: Determine the quality inspection result label for the maintenance work order data sample; Step 240: Using text feature vector samples and image feature vector samples as training samples, and quality inspection result labels as sample labels, train the initial work order quality inspection model. After training, the work order quality inspection model is obtained.

[0057] Optionally, the initial work order quality inspection model includes an initial feature fusion layer and an initial quality inspection layer. The initial feature fusion layer is used to fuse text feature vector samples and image feature vector samples based on multiple quality inspection dimensions to obtain the predicted fused feature vector corresponding to each quality inspection dimension. The initial quality inspection layer is used to perform quality inspection on the maintenance work order data samples for each quality inspection dimension based on the predicted fused feature vector corresponding to each quality inspection dimension, and output the predicted quality inspection result of the maintenance work order data samples.

[0058] Optionally, multiple quality inspection dimensions include: semantic consistency dimension, image repeatability dimension, and facility standardization dimension.

[0059] The following describes the maintenance work order quality inspection device provided in the embodiments of the present invention. The maintenance work order quality inspection device described below can be referred to in correspondence with the maintenance work order quality inspection method described above.

[0060] Figure 3 This is a schematic diagram of the structure of the maintenance work order quality inspection device provided in an embodiment of the present invention, as shown below. Figure 3 As shown, the maintenance work order quality inspection device 300 includes: The acquisition unit 310 is used to acquire the maintenance work order data to be inspected. The maintenance work order data includes work order text data and on-site image data. The feature extraction unit 320 is used to extract the text feature vector of the work order text data and the image feature vector of the on-site image data; The quality inspection unit 330 is used to input text feature vectors and image feature vectors into the work order quality inspection model. The work order quality inspection model fuses text feature vectors and image feature vectors based on multiple preset quality inspection dimensions to obtain a fused feature vector corresponding to each quality inspection dimension. Based on the fused feature vector corresponding to each quality inspection dimension, the work order data is inspected for each quality inspection dimension, and the quality inspection result of the work order data is output. The work order quality inspection model is trained based on the text feature vector samples and image feature vector samples corresponding to the maintenance work order data samples, as well as the quality inspection result labels of the maintenance work order data samples.

[0061] Optionally, based on multiple preset quality inspection dimensions, text feature vectors and image feature vectors are fused to obtain a fused feature vector corresponding to each quality inspection dimension, including: For each quality inspection dimension, determine the first attention weight of the text feature vector and the second attention weight of the image feature vector; Based on the first attention weight and the second attention weight, the text feature vector and the image feature vector are fused to obtain the fused feature vector corresponding to each quality inspection dimension.

[0062] Optionally, based on the fused feature vector corresponding to each quality inspection dimension, quality inspection is performed on the maintenance work order data for each quality inspection dimension, including: Based on the fusion feature vector corresponding to each quality inspection dimension and each quality inspection dimension, a subset of rules corresponding to each quality inspection dimension is determined from the pre-built expert rule base; Based on the fusion feature vector corresponding to each quality inspection dimension and the rule subset corresponding to each quality inspection dimension, the maintenance work order data is inspected for each quality inspection dimension to obtain the quality inspection result for each quality inspection dimension. By integrating the quality inspection results from each quality inspection dimension, the quality inspection results of the maintenance work order data are obtained.

[0063] Optionally, multiple quality inspection dimensions include: semantic consistency dimension, image repeatability dimension, and facility standardization dimension.

[0064] Optionally, obtain the maintenance work order data to be inspected, including: Obtain the original maintenance work order data, which includes the original work order text data and the original site image data. Preprocessing and correlation processing are performed on the original work order text data and the original site image data to obtain the maintenance work order data.

[0065] Optionally, the maintenance work order quality inspection device also includes: The generation unit is used to generate potential hazard warning information when the maintenance work order data is found to be abnormal based on the quality inspection results. The sending unit is used to send hazard warning information to the client.

[0066] Optionally, the work order quality inspection model is trained based on the following steps: Obtain maintenance work order data samples, which include work order text data samples and on-site image data samples. Feature extraction is performed on the work order text data samples to obtain text feature vector samples, and feature extraction is performed on the on-site image data samples to obtain image feature vector samples; Determine the quality inspection result labels for the maintenance work order data samples; Using text feature vector samples and image feature vector samples as training samples, and quality inspection result labels as sample labels, an initial work order quality inspection model is trained. After training, the work order quality inspection model is obtained.

[0067] It should be noted that the maintenance work order quality inspection device provided in this embodiment of the invention can realize all the method steps implemented in the above-mentioned maintenance work order quality inspection method embodiment, and can achieve the same technical effect. Here, the parts that are the same as those in the method embodiment and the beneficial effects will not be described in detail.

[0068] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention, such as... Figure 4As shown, the electronic device may include: a processor 410, a communications interface 420, a memory 430, and a communications bus 440, wherein the processor 410, the communications interface 420, and the memory 430 communicate with each other through the communications bus 440. The processor 410 can call logic instructions in the memory 430 to execute a maintenance work order quality inspection method. This method includes: acquiring maintenance work order data to be inspected, including work order text data and on-site image data; extracting text feature vectors from the work order text data and image feature vectors from the on-site image data; inputting the text feature vectors and image feature vectors into a work order quality inspection model, which, based on multiple preset quality inspection dimensions, fuses the text feature vectors and image feature vectors to obtain a fused feature vector corresponding to each quality inspection dimension; performing quality inspection on the maintenance work order data for each quality inspection dimension based on the fused feature vector corresponding to each quality inspection dimension, and outputting the quality inspection result of the maintenance work order data; wherein, the work order quality inspection model is trained based on text feature vector samples and image feature vector samples corresponding to the maintenance work order data samples, as well as the quality inspection result labels of the maintenance work order data samples.

[0069] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0070] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program performs the maintenance work order quality inspection method provided by the above methods. The method includes: acquiring maintenance work order data to be inspected, the maintenance work order data including work order text data and on-site image data; extracting text feature vectors from the work order text data and extracting image feature vectors from the on-site image data; inputting the text feature vectors and image feature vectors into a work order quality inspection model, whereby the work order quality inspection model, based on multiple preset quality inspection dimensions, fuses the text feature vectors and image feature vectors to obtain a fused feature vector corresponding to each quality inspection dimension; performing quality inspection on the maintenance work order data for each quality inspection dimension based on the fused feature vector corresponding to each quality inspection dimension, and outputting the quality inspection result of the maintenance work order data; wherein, the work order quality inspection model is trained based on text feature vector samples and image feature vector samples corresponding to the maintenance work order data samples, as well as the quality inspection result labels of the maintenance work order data samples.

[0071] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0072] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0073] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for quality inspection of maintenance work orders, characterized in that, include: Obtain the maintenance work order data to be inspected, the maintenance work order data including work order text data and on-site image data; Extract the text feature vector from the work order text data and extract the image feature vector from the on-site image data; The text feature vector and the image feature vector are input into the work order quality inspection model. The work order quality inspection model fuses the text feature vector and the image feature vector based on multiple preset quality inspection dimensions to obtain a fused feature vector corresponding to each quality inspection dimension. Based on the fused feature vector corresponding to each quality inspection dimension, the maintenance work order data is inspected for each quality inspection dimension, and the quality inspection result of the maintenance work order data is output. The work order quality inspection model is trained based on the text feature vector samples and image feature vector samples corresponding to the maintenance work order data samples, as well as the quality inspection result labels of the maintenance work order data samples.

2. The quality inspection method for maintenance work orders according to claim 1, characterized in that, The method involves fusing the text feature vector and the image feature vector based on multiple preset quality inspection dimensions to obtain a fused feature vector corresponding to each quality inspection dimension, including: For each quality inspection dimension, determine the first attention weight of the text feature vector and the second attention weight of the image feature vector; Based on the first attention weight and the second attention weight, the text feature vector and the image feature vector are fused to obtain the fused feature vector corresponding to each quality inspection dimension.

3. The quality inspection method for maintenance work orders according to claim 1, characterized in that, The step of performing quality inspection on the maintenance work order data for each of the quality inspection dimensions based on the fused feature vector corresponding to each of the quality inspection dimensions includes: Based on the fusion feature vector corresponding to each of the quality inspection dimensions and each of the quality inspection dimensions, a subset of rules corresponding to each of the quality inspection dimensions is determined from a pre-built expert rule base; Based on the fusion feature vector corresponding to each of the quality inspection dimensions and the rule subset corresponding to each of the quality inspection dimensions, the maintenance work order data is inspected for each of the quality inspection dimensions to obtain the quality inspection results for each of the quality inspection dimensions. The quality inspection results of each of the aforementioned quality inspection dimensions are combined to obtain the quality inspection results of the maintenance work order data.

4. The quality inspection method for maintenance work orders according to claim 1, characterized in that, The quality inspection dimensions include: semantic consistency, image repeatability, and facility standardization.

5. The quality inspection method for maintenance work orders according to claim 1, characterized in that, The process of obtaining the maintenance work order data to be inspected includes: Obtain the original maintenance work order data, which includes the original work order text data and the original on-site image data; The original work order text data and original site image data are preprocessed and correlated to obtain the maintenance work order data.

6. The quality inspection method for maintenance work orders according to claim 1, characterized in that, After outputting the quality inspection results of the maintenance work order data, the method further includes: If the quality inspection results indicate that there is an anomaly in the maintenance work order data, a potential hazard warning message will be generated. The hazard warning information is sent to the client.

7. The quality inspection method for maintenance work orders according to claim 1, characterized in that, The work order quality inspection model is trained based on the following steps: Obtain maintenance work order data samples, which include work order text data samples and on-site image data samples; Feature extraction is performed on the work order text data sample to obtain a text feature vector sample, and feature extraction is performed on the on-site image data sample to obtain an image feature vector sample; Determine the quality inspection result label of the maintenance work order data sample; Using the text feature vector samples and the image feature vector samples as training samples, and the quality inspection result labels as sample labels, an initial work order quality inspection model is trained. After training, the work order quality inspection model is obtained.

8. A quality inspection device for maintenance work orders, characterized in that, include: The acquisition unit is used to acquire maintenance work order data to be inspected, the maintenance work order data including work order text data and on-site image data; The feature extraction unit is used to extract the text feature vector of the work order text data and the image feature vector of the on-site image data; The quality inspection unit is used to input the text feature vector and the image feature vector into the work order quality inspection model. The work order quality inspection model fuses the text feature vector and the image feature vector based on multiple preset quality inspection dimensions to obtain a fused feature vector corresponding to each quality inspection dimension. Based on the fused feature vector corresponding to each quality inspection dimension, the maintenance work order data is inspected for each quality inspection dimension, and the quality inspection result of the maintenance work order data is output. The work order quality inspection model is trained based on the text feature vector samples and image feature vector samples corresponding to the maintenance work order data samples, as well as the quality inspection result labels of the maintenance work order data samples.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the maintenance work order quality inspection method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the maintenance work order quality inspection method as described in any one of claims 1 to 7.

Images