Power swap station equipment fault identification system fusing language and image information

Abstract

The invention, which belongs to the technical field of power equipment fault identification, discloses a power conversion station equipment fault identification system fusing language and image information, and the system comprises a data collection module which is used for collecting image data and language text data of power conversion station equipment; the image analysis module is used for carrying out image analysis on the image data by utilizing a preset image recognition algorithm to obtain an image analysis result; the language text analysis module is used for analyzing the language text data by utilizing the equipment fault tree of the battery swap station to obtain a text analysis result; the early warning module is used for triggering a grading treatment strategy according to the image analysis result and the text analysis result; through combination of image analysis and language text analysis, a more accurate equipment fault identification result of the battery swap station can be obtained.

Description

Technical Field

[0001] This invention relates to the field of power equipment fault identification technology, and in particular to a fault identification system for power swapping station equipment that integrates language and image information. Background Technology

[0002] Battery swapping stations are energy infrastructures that provide rapid battery replacement services for electric vehicles by centrally storing, charging, and distributing power batteries. Their core model separates the vehicle from the battery, allowing users to replenish energy simply by swapping in a fully charged battery, a process typically completed within 3-5 minutes.

[0003] Equipment failure at a battery swapping station can lead to battery swapping failures, requiring users to wait for repairs or transfer to other stations, directly impacting charging efficiency. However, equipment fault identification at battery swapping stations can detect potential faults in advance, reducing unplanned downtime. Existing equipment fault identification at battery swapping stations relies solely on image recognition, which has certain limitations and is not very accurate. Summary of the Invention

[0004] In view of the above-mentioned prior art, the present invention provides a fault identification system for battery swapping station equipment that integrates language and image information, mainly to solve the technical problems existing in the background art.

[0005] To achieve the above objectives, the technical solution of this invention is implemented as follows: A fault identification system for battery swapping station equipment that integrates language and image information includes: The data acquisition module is used to collect image data and voice text data from the battery swapping station equipment; The image analysis module performs image analysis on the image data using a preset image recognition algorithm to obtain image analysis results; The language text analysis module analyzes the language text data using the fault tree of the battery swapping station equipment to obtain text analysis results. The early warning module provides tiered early warnings based on the image analysis results and the text analysis results.

[0006] Optionally, the step of performing image analysis on the image data using a preset image recognition algorithm to obtain image analysis results includes: Visual analysis of the image data of the battery swapping station equipment was performed to obtain the results of the visual defect analysis of the battery swapping station equipment; The indicator light image data of the battery swapping station equipment is analyzed to obtain the indicator light status results; Thermal imaging analysis was performed on the image data of the battery swapping station equipment to obtain the temperature analysis results of the equipment.

[0007] Optionally, after analyzing the indicator light image data of the battery swapping station equipment to obtain the indicator light status result, the method further includes adjusting the indicator light status result to obtain an adjusted indicator light status result, specifically: The voltage and current values ​​of the battery swapping station equipment at each phase during a certain time period are obtained, and the current cycle data of the corresponding battery swapping station equipment is obtained based on the voltage and current values. A periodic feature vector library is constructed, and the current periodic data is matched with the periodic data in the periodic feature vector library to obtain the matching result; The matching result is used to correct the indicator light status result, and the adjusted indicator light status result is obtained based on the correction result.

[0008] Optionally, the construction of the periodic feature vector library includes: The equipment in the battery swapping station is classified according to its functional type. Multimodal data is collected from the classified battery swapping station equipment, and time-domain and frequency-domain analyses are performed on the multimodal data to obtain time-domain and frequency-domain features. The time-domain features and frequency-domain features are used to learn the time-series patterns of normal operation of the battery swapping station equipment based on the LSTM network, and a periodic feature vector library is generated.

[0009] Optionally, the step of analyzing the language text data using the fault tree of the battery swapping station equipment to obtain text analysis results includes: Stop words are removed from the language text, and word segmentation is performed to obtain multiple target words; A fault tree for the battery swapping station equipment is constructed, and text analysis results are obtained based on the target words and the fault tree for the battery swapping station equipment.

[0010] Optionally, constructing the fault tree for the battery swapping station equipment includes: Based on the functional type of the battery swapping station equipment, a fault tree for the sub-equipment of the battery swapping station is constructed, and each fault tree for the sub-equipment of the battery swapping station corresponds to a sub-equipment. Obtain historical fault information, perform semantic recognition based on the historical fault information, extract multiple keywords, and map the keywords to fault tree nodes; Based on the mapped fault nodes, causal analysis is used to determine the fault path.

[0011] Optionally, the causal analysis includes: The fault propagation path is identified by querying upwards from the faulty node to the root node. The root cause of the fault is located by querying downwards from the faulty node to the leaf node. Verify the rationality of the fault path by analyzing the correlation between events at the same level.

[0012] Optionally, the step of providing tiered early warnings based on the image analysis results and the text analysis results includes: Dynamic weight allocation is performed based on the image analysis results and the text analysis results, according to the image analysis results, the text analysis results, and the corresponding weights; A fault judgment criterion database is preset, and the confidence scores corresponding to the image analysis results and the text analysis results are determined based on the image analysis results, the text analysis results, and the fault judgment criterion database. Based on the confidence scores corresponding to the image analysis results and the text analysis results, as well as the weights corresponding to the image analysis results and the text analysis results, a comprehensive confidence score is calculated, and a graded early warning is issued based on the comprehensive confidence score.

[0013] The beneficial effects of this invention are as follows: This invention provides a fault identification system for battery swapping station equipment that integrates language and image information. By acquiring image data and language text data, analyzing the image data and language text data, it obtains image analysis results and language text analysis results. Based on the image analysis results and language text analysis results, it determines whether the battery swapping station equipment has malfunctioned. Compared with the traditional fault identification results obtained by using only image analysis or language text analysis, the combination of image analysis and language text analysis provided by this technical solution can obtain more accurate fault identification results for battery swapping station equipment. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of the structure of a fault identification system for a battery swapping station that integrates language and image information, provided in an embodiment of the present invention. Detailed Implementation

[0015] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. In the following description, the expression "some embodiments" refers to a subset of all possible embodiments; however, it should be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments and can be combined with each other without conflict.

[0016] In the following description, numerous specific details are set forth in order to provide a more thorough understanding of the invention. However, it will be apparent to those skilled in the art that the invention can be practiced without one or more of these details. In other instances, certain technical features well-known in the art have not been described in order to avoid obscuring the invention.

[0017] It should be understood that the present invention can be embodied in various forms and should not be construed as being limited to the embodiments set forth herein. Rather, providing these embodiments will make the disclosure thorough and complete, and will fully convey the scope of the invention to those skilled in the art. Furthermore, the terminology used herein is intended only to describe particular embodiments and is not intended to limit the invention. When used herein, the singular forms “a,” “an,” and “the” are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the terms “compose” and / or “comprising,” when used in this specification, identify the presence of the stated features, integers, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups. When used herein, the term “and / or” includes any and all combinations of the associated listed items.

[0018] It should also be noted that when an element is referred to as being "fixed to" another element, it can be directly attached to the other element or there may be an intervening element. When an element is referred to as being "connected to" another element, it can be directly connected to the other element or there may be an intervening element. The terms "vertical," "horizontal," "inner," "outer," "left," "right," and similar expressions used herein are for illustrative purposes only and do not represent the only possible implementation.

[0019] To fully understand this invention, a detailed structure will be presented in the following description to illustrate the technical solution proposed by this invention. Optional embodiments of the invention are described in detail below; however, in addition to these detailed descriptions, the invention may have other embodiments.

[0020] Example Please refer to the attached document. Figure 1 This application provides a fault identification system for battery swapping station equipment that integrates language and image information, including: The data acquisition module is used to collect image data and voice text data from the battery swapping station equipment; The image analysis module performs image analysis on the image data using a preset image recognition algorithm to obtain image analysis results; The language text analysis module analyzes the language text data using the fault tree of the battery swapping station equipment to obtain text analysis results. The early warning module provides tiered early warnings based on the image analysis results and the text analysis results.

[0021] Specifically, the battery swapping station equipment includes multiple components, such as a charging module, battery compartment, and battery management system, which work together to ensure the normal operation of the station. Cameras are deployed at multiple nodes of the equipment to collect image data, including both external and local images. An image analysis module analyzes the acquired images to obtain results. Text data, such as the station's operation logs, is also collected and analyzed to obtain text analysis results. The image analysis results reveal explicit faults, while the text analysis results reveal implicit faults. Combining both results provides a more comprehensive fault identification assessment, allowing for tiered warnings. Compared to methods relying solely on image or text analysis, this approach has limitations and may yield less accurate results.

[0022] As an optional implementation, the step of performing image analysis on the image data using a preset image recognition algorithm to obtain image analysis results includes: Visual analysis of the image data of the battery swapping station equipment was performed to obtain the results of the visual defect analysis of the battery swapping station equipment; The indicator light image data of the battery swapping station equipment is analyzed to obtain the indicator light status results; Thermal imaging analysis was performed on the image data of the battery swapping station equipment to obtain the temperature analysis results of the equipment.

[0023] Specifically, appearance analysis of the battery swapping station equipment's external image data can be performed using a YOLO-V8 inspection box to detect whether the equipment's appearance is complete or damaged, outputting an appearance label indicating normal or abnormal appearance. Since battery swapping stations involve multiple devices, each device's indicator lights will flash different colors and may have different on / off states in different locations within the same device when operating normally. Therefore, the indicator light states will differ under normal operating conditions and under fault conditions. Analysis of the indicator light image data can be achieved by inputting the data into a pre-trained classification model for identification, such as a neural network model, and finally outputting an indicator light status label indicating normal or abnormal. Because battery swapping stations are devices that enable rapid energy exchange for vehicles, they contain multiple battery compartments housing swapping batteries that charge and discharge. Heat is generated during the charging process. If the temperature is too high, it may lead to thermal runaway, which poses a significant threat to the safety of the battery swapping station. Temperature also affects the charging and discharging efficiency of the battery. High temperatures may accelerate battery aging and reduce capacity, while low temperatures may cause a decrease in battery activity and affect performance. Therefore, since multiple batteries are stored in a battery swapping station, if the temperature of one battery is abnormal, it will cause a chain reaction, thereby affecting the normal operation of the entire swapping station. Therefore, thermal imaging analysis is performed on the image data of the battery swapping station equipment, and a temperature threshold is set. If the temperature value obtained from the thermal imaging analysis does not exceed the temperature threshold, it indicates that the temperature of the battery swapping station equipment is normal. If the temperature value obtained from the thermal imaging analysis results exceeds the temperature threshold, it indicates that the temperature of the battery swapping station equipment is abnormal. Based on the judgment result of the temperature value and the temperature threshold, a temperature label indicating normal or abnormal temperature is output. Based on the corresponding labels obtained from the appearance defect analysis results, indicator light status results, and temperature analysis results, the labels of the three results are stitched together to obtain the image analysis result.

[0024] As an optional implementation, after analyzing the indicator light image data of the battery swapping station equipment to obtain the indicator light status result, the method further includes adjusting the indicator light status result to obtain an adjusted indicator light status result, specifically: The voltage and current values ​​of the battery swapping station equipment at each phase during a certain time period are obtained, and the current cycle data of the corresponding battery swapping station equipment is obtained based on the voltage and current values. A periodic feature vector library is constructed, and the current periodic data is matched with the periodic data in the periodic feature vector library to obtain the matching result; The matching result is used to correct the indicator light status result, and the adjusted indicator light status result is obtained based on the correction result.

[0025] Specifically, the indicator lights in different battery swapping station devices may have different states. For example, indicator lights may be continuously lit, flashing, or completely off. A single battery swapping station device may have all three states simultaneously, or only one or two states may exist at the same time. For the flashing state, there will be intermittent alternation between on and off, which may lead to misjudgments when analyzing indicator light state image data. To avoid inaccuracies in the indicator light image analysis results, corrections are made. Specifically, the voltage and current values ​​of each phase of the battery swapping station device are obtained over a period of time. Since different battery swapping station devices may have different numbers of indicator lights, and the on / off times of different indicator lights are different, the voltage and current values ​​of the indicator lights in different battery swapping station devices will also be different. Based on the voltage and current values ​​obtained over a period of time, the individual indicator lights of the battery swapping station device are determined. The system obtains the current cycle data; the cycle feature vector library stores the cycle data of each battery swapping station device under normal / abnormal conditions. The current cycle data of each indicator light on the battery swapping station device is matched with the cycle data in the cycle feature vector library to determine whether the working status of the battery swapping station device is normal or abnormal. The matching result is the indicator light status of the battery swapping station device as normal working state / abnormal working state. Then, the determined working state is compared with the indicator light status label obtained from the indicator light image data analysis to determine whether the indicator light working state obtained from the cycle feature vector library is consistent with the indicator light status label obtained from the image data analysis. If they are consistent, the indicator light status label obtained from the indicator light image data analysis result is used as the adjusted indicator light status result. If they are inconsistent, adjustments and corrections are made, and the indicator light working state obtained from the cycle feature vector library is used as the adjusted indicator light status result.

[0026] As an optional implementation, the construction of the periodic feature vector library includes: The equipment in the battery swapping station is classified according to its functional type. Multimodal data is collected from the classified battery swapping station equipment, and time-domain and frequency-domain analyses are performed on the multimodal data to obtain time-domain and frequency-domain features. The time-domain features and frequency-domain features are used to learn the time-series patterns of normal operation of the battery swapping station equipment based on the LSTM network, and a periodic feature vector library is generated.

[0027] Specifically, battery swapping station equipment is classified according to its function, for example, into charging units, battery storage units, and temperature monitoring units. Different swapping station equipment has different functions, and therefore requires different data collection. Thus, multimodal data is collected for different swapping station equipment, and time-domain and frequency-domain analyses are performed on this data. Time-domain analysis can analyze the changes in multimodal data over time, extracting time-domain features such as mean, variance, and peak value to capture the time-dimensional characteristics of equipment operation. Frequency-domain analysis uses Fourier transform to convert the time-domain signal into a frequency-domain signal, extracting frequency components, energy distribution, and other frequency-domain features to capture the frequency characteristics of equipment operation. Frequency dimension characteristics, through time-domain and frequency-domain analysis, can transform multimodal data into quantifiable feature vectors, providing structured input data for LSTM networks. LSTM networks can process time-series data and frequency-domain feature data, capturing long-term dependencies in equipment operating states. For example, they can capture the periodic patterns of start-up, operation, and shutdown of battery swapping station equipment, as well as the gradual changes in fault characteristics. By training the LSTM network, it can generate corresponding feature vectors (i.e., periodic feature vectors). The feature vectors learned by LSTM are then classified according to equipment type to form a periodic feature vector library for each battery swapping station.

[0028] As an optional implementation, the step of analyzing the language text data using the fault tree of the battery swapping station equipment to obtain text analysis results includes: Stop words are removed from the language text, and word segmentation is performed to obtain multiple target words; A fault tree for the battery swapping station equipment is constructed, and text analysis results are obtained based on the target words and the fault tree for the battery swapping station equipment.

[0029] Specifically, language text generally includes multiple words without actual meaning. These words are removed, and then word segmentation is performed to obtain multiple target words. The fault tree of the battery swapping station equipment has a multi-level structure, which allows the battery swapping station equipment to form multiple branches. Each branch includes multiple nodes. The extracted target words are associated with the nodes in the fault tree of the battery swapping station equipment, and the fault tree of the battery swapping station equipment is used for analysis to obtain the text analysis results.

[0030] As an optional implementation, the construction of the fault tree for the battery swapping station equipment includes: Based on the functional type of the battery swapping station equipment, a fault tree for the sub-equipment of the battery swapping station is constructed, and each fault tree for the sub-equipment of the battery swapping station corresponds to a sub-equipment. Obtain historical fault information, perform semantic recognition based on the historical fault information, extract multiple keywords, and map the keywords to fault tree nodes; Based on the mapped fault nodes, causal analysis is used to determine the fault path; Causal analysis includes: The fault propagation path is identified by querying upwards from the faulty node to the root node. The root cause of the fault is located by querying downwards from the faulty node to the leaf node. Verify the rationality of the fault path by analyzing the correlation between events at the same level.

[0031] Specifically, the equipment in a battery swapping station involves a variety of functional devices. According to the functional type of the equipment, it can be divided into various sub-devices such as charging modules and battery storage units. Since the functions of different sub-devices are different, the causes and symptoms of their failures may also be different. Therefore, by constructing a battery swapping station sub-device fault tree based on the sub-devices of different functional types, we can obtain the battery swapping station sub-device fault tree corresponding to different sub-devices. When it is determined that a certain functional sub-device has failed, we can analyze it according to its corresponding battery swapping station sub-device fault tree and quickly locate the fault problem of that functional sub-device. Historical fault information can be obtained from historical fault reports. Natural language processing is performed on these reports to extract keywords strongly related to the faults. For example, keywords such as equipment name, fault phenomenon, and fault cause can be extracted. These keywords are then mapped to corresponding nodes in the fault tree of the battery swapping station's sub-equipment. Keyword mapping transforms unstructured text fault information into structured fault tree node positions, providing input for subsequent causal analysis. Based on the mapped fault nodes, causal analysis is used to determine the fault path. Specifically, the mapped fault nodes are the initial fault nodes mapped to the fault tree of the battery swapping station's sub-equipment through semantic recognition. Based on these mapped fault nodes, the path is determined by tracing upwards. The complete fault path is determined by extending downwards. For example, tracing upwards involves: from the fault node (e.g., "charging module short circuit") along the parent node path to the root node (e.g., "swapping station overall failure shutdown"), recording the path nodes (e.g., "charging module fault" - "charging system abnormality"), and identifying sibling nodes at the same level (e.g., "battery storage unit fault" and "charging system abnormality" are sibling nodes that jointly cause "swapping station overall failure shutdown"); extending downwards involves: from the fault node along the child node path to the leaf node (e.g., "line aging" - "human error"), recording the path nodes and identifying sibling nodes (e.g., "line aging" and "poor contact" are sibling nodes that jointly cause "charging module short circuit").

[0032] The language text consists of textual data such as operation logs and operation records of the battery swapping station sub-equipment. The textual data is processed to extract multiple keywords, which are used as target words. The target words are then semantically matched with the nodes of the fault tree of the battery swapping station sub-equipment. For example, "voltage fluctuation" is mapped to the "charging module overvoltage" node, and causal analysis is performed to determine the fault path. The obtained fault path is used as the text analysis result.

[0033] As an optional implementation, the step of performing graded early warning based on the image analysis results and the text analysis results includes: Dynamic weight allocation is performed based on the image analysis results and the text analysis results, according to the image analysis results, the text analysis results, and the corresponding weights; A fault judgment criterion database is preset, and the confidence scores corresponding to the image analysis results and the text analysis results are determined based on the image analysis results, the text analysis results, and the fault judgment criterion database. Based on the confidence scores corresponding to the image analysis results and the text analysis results, as well as the weights corresponding to the image analysis results and the text analysis results, a comprehensive confidence score is calculated, and a graded early warning is issued based on the comprehensive confidence score.

[0034] Specifically, the fault content identified by image analysis and text analysis results may differ. For example, when image analysis identifies a fault but text analysis does not, the weight of the image analysis result can be set higher than that of the text analysis result; conversely, when image analysis does not identify a fault but text analysis does, the weight of the image analysis result can be set lower than that of the text analysis result. The fault criterion database stores all possible fault categories of the battery swapping station equipment. Each fault category has a corresponding confidence score. The image analysis results and text analysis results are retrieved from the fault criterion database to find the closest fault category and its corresponding confidence score, thereby obtaining the confidence scores corresponding to the image analysis results and text analysis results. A comprehensive confidence score is calculated based on the confidence score and weights. The corresponding level of warning signal is triggered based on the confidence score to remind staff to take action.

[0035] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A fault identification system for battery swapping station equipment that integrates language and image information, characterized in that, include: The data acquisition module is used to collect image data and voice text data from the battery swapping station equipment; The image analysis module performs image analysis on the image data using a preset image recognition algorithm to obtain image analysis results; The language text analysis module analyzes the language text data using the fault tree of the battery swapping station equipment to obtain text analysis results. The early warning module provides tiered early warnings based on the image analysis results and the text analysis results.

2. The fault identification system for battery swapping station equipment that integrates language and image information according to claim 1, characterized in that, The step of performing image analysis on the image data using a preset image recognition algorithm to obtain image analysis results includes: Visual analysis was performed on the image data of the battery swapping station equipment to obtain the results of the visual defect analysis of the battery swapping station equipment; The indicator light image data of the battery swapping station equipment is analyzed to obtain the indicator light status results; Thermal imaging analysis was performed on the image data of the battery swapping station equipment to obtain the temperature analysis results of the equipment.

3. The fault identification system for battery swapping station equipment that integrates language and image information according to claim 2, characterized in that, After analyzing the indicator light image data of the battery swapping station equipment to obtain the indicator light status results, the method further includes adjusting the indicator light status results to obtain adjusted indicator light status results, specifically as follows: The voltage and current values ​​of the battery swapping station equipment at each phase during a certain time period are obtained, and the current cycle data of the corresponding battery swapping station equipment is obtained based on the voltage and current values. A periodic feature vector library is constructed, and the current periodic data is matched with the periodic data in the periodic feature vector library to obtain the matching result; The matching result is used to correct the indicator light status result, and the adjusted indicator light status result is obtained based on the correction result.

4. The fault identification system for battery swapping station equipment that integrates language and image information according to claim 3, characterized in that, The construction of the periodic feature vector library includes: The equipment in the battery swapping station is classified according to its functional type. Multimodal data is collected from the classified battery swapping station equipment, and time-domain and frequency-domain analyses are performed on the multimodal data to obtain time-domain and frequency-domain features. The time-domain features and frequency-domain features are used to learn the time-series patterns of normal operation of the battery swapping station equipment based on the LSTM network, and a periodic feature vector library is generated.

5. A fault identification system for battery swapping station equipment that integrates language and image information according to claim 1, characterized in that, The analysis of the language text data using the fault tree of the battery swapping station equipment yields text analysis results, including: Stop words are removed from the language text, and word segmentation is performed to obtain multiple target words; A fault tree for the battery swapping station equipment is constructed, and text analysis results are obtained based on the target words and the fault tree for the battery swapping station equipment.

6. A fault identification system for battery swapping station equipment that integrates language and image information according to claim 5, characterized in that, The construction of the fault tree for the battery swapping station equipment includes: Based on the functional type of the battery swapping station equipment, a fault tree for the sub-equipment of the battery swapping station is constructed, and each fault tree for the sub-equipment of the battery swapping station corresponds to a sub-equipment. Obtain historical fault information, perform semantic recognition based on the historical fault information, extract multiple keywords, and map the keywords to fault tree nodes; Based on the mapped fault nodes, causal analysis is used to determine the fault path.

7. A fault identification system for battery swapping station equipment that integrates language and image information according to claim 6, characterized in that, The causal analysis includes: The fault propagation path is identified by querying upwards from the faulty node to the root node. The root cause of the fault is located by querying downwards from the faulty node to the leaf node. Verify the rationality of the fault path by analyzing the correlation between events at the same level.

8. A fault identification system for battery swapping station equipment that integrates language and image information according to claim 1, characterized in that, The step of providing tiered early warnings based on the image analysis results and the text analysis results includes: Dynamic weight allocation is performed based on the image analysis results and the text analysis results, according to the image analysis results, the text analysis results, and the corresponding weights; A fault judgment criterion database is preset, and the confidence scores corresponding to the image analysis results and the text analysis results are determined based on the image analysis results, the text analysis results, and the fault judgment criterion database. Based on the confidence scores corresponding to the image analysis results and the text analysis results, as well as the weights corresponding to the image analysis results and the text analysis results, a comprehensive confidence score is calculated, and a graded early warning is issued based on the comprehensive confidence score.

Images