A charging pile insulation performance testing system and method based on multimodal sensing

Abstract

This invention belongs to the field of power equipment testing technology and discloses a charging pile insulation performance testing system based on multimodal sensing. The system includes: a multimodal sensing unit for simultaneously collecting electrical parameters, environmental parameters, and physical state parameters of the charging pile; and a data acquisition and preprocessing unit for preprocessing the data collected by the multimodal sensing unit. By integrating multi-dimensional electrical, environmental, and physical information, it overcomes the limitations of single-parameter detection, more comprehensively and realistically reflecting the insulation status of the equipment, and significantly reducing false alarms and missed alarms. Secondly, based on early sign analysis such as partial discharge and temperature-humidity correlation trends, the system can issue early warnings before a significant decrease in insulation resistance, enabling predictive maintenance. It can also accurately distinguish different insulation fault types. Furthermore, combined with digital twins and cloud platform AI, the system can adaptively adjust diagnostic thresholds according to equipment aging and environmental changes.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power equipment detection, and particularly relates to a charging pile insulation performance detection system and method based on multi-modal sensing. BACKGROUND

[0002] The insulation performance of a charging pile refers to the ability of preventing current leakage between charged components and non-charged components (such as a shell and a grounding body) in the charging pile through an insulating material, which guarantees personal safety and avoids electric shock accidents caused by leakage when a person contacts the charging pile. Meanwhile, the insulation performance protects the stable operation of the equipment and prevents leakage from causing circuit short-circuit, component damage or fault shutdown.

[0003] With the popularity of electric vehicles, the operation safety of charging piles, as the core infrastructure, is crucial, and the insulation performance is a key indicator for measuring the safety of charging piles. Once the insulation fails, it may lead to leakage, short-circuit and even fire, which seriously threatens the safety of life and property of users. At present, the insulation detection of charging piles mainly relies on two types of methods. One type is periodic manual detection, in which professional personnel use megohmmeters and other tools for periodic detection. This method is inefficient and cannot realize real-time monitoring, and is prone to missed detection due to human factors. The other type is online detection by a single sensor. Although some charging piles are equipped with insulation monitoring devices, they usually only rely on the detection of a single parameter, i.e., insulation resistance. This method has obvious limitations. The decrease of insulation resistance is a slow process, and when an abnormality is detected, the insulation may have already deteriorated seriously. In addition, a single parameter is easily affected by humidity, condensation and other environmental factors, leading to false positives or false negatives. Furthermore, it is unable to effectively identify and warn the early signs of insulation deterioration, as well as the specific types such as moisture, damage and contamination. Therefore, the existing technology lacks a detection scheme that can evaluate the insulation state of a charging pile in real time, comprehensively and accurately, and perform early fault warning and type diagnosis. SUMMARY

[0004] The purpose of the present application is to provide a charging pile insulation performance detection system and method based on multi-modal sensing to solve the problems raised in the background.

[0005] In order to achieve the above-mentioned purpose, the present application provides the following technical scheme: a charging pile insulation performance detection system based on multi-modal sensing, comprising: a multi-modal sensing unit for synchronously collecting electrical parameters, environmental parameters and physical state parameters of a charging pile; a data acquisition and preprocessing unit for preprocessing the data collected by the multi-modal sensing unit; an edge computing and fusion analysis unit including a digital twin module and a fusion diagnosis module, for comprehensively evaluating the insulation state of the charging pile based on the preprocessed multi-modal data and the digital twin model through a data fusion algorithm, and outputting a diagnosis result; The cloud platform and user interaction unit are used to receive, store, and display the diagnostic results, and send alarm information to the user.

[0006] Preferably, the multimodal sensing unit includes at least an electrical sensing module, an environmental sensing module, a thermal sensing module, and a high-frequency current sensing module. The electrical sensing module is used to collect insulation resistance and leakage current, the environmental sensing module is used to collect ambient temperature and humidity, the thermal sensing module is used to collect cable joint temperature, and the high-frequency current sensing module is used to collect partial discharge signals.

[0007] Preferably, the fusion diagnostic module employs a machine learning model, which is trained using historical multimodal data and corresponding insulation status labels, and can output at least one of insulation health index, early warning signal, and fault type diagnosis.

[0008] Preferably, the alarm information is sent in multiple forms such as APP, SMS, and voice call to ensure that relevant personnel can receive the alarm information in a timely manner. The alarm information is sent every 3 minutes, and the sending will stop only after the relevant personnel confirm that they have received the alarm signal.

[0009] Preferably, the preprocessing process of the data acquisition and preprocessing unit first uses filtering technology to denoise the multimodal raw data to remove electromagnetic interference and environmental noise signals; secondly, it identifies and removes abnormal data points through the 3σ criterion to prevent extreme values ​​from affecting diagnostic accuracy; and finally, it uses the Z-score standardization method to convert heterogeneous data into feature data with uniform dimensions.

[0010] A method for detecting the insulation performance of charging piles based on multimodal sensing, the specific steps of which are as follows: Step 1: Synchronously collect multimodal sensor data from the charging pile; Step 2: Preprocess and extract features from the multimodal sensing data to obtain feature vectors; Step 3: Compare the feature vector with the output of the charging pile digital twin model; Step 4: Based on the comparison results, the feature vectors are comprehensively analyzed and diagnosed using a data fusion algorithm to generate insulation status assessment results; Step 5: Execute the corresponding decisions and provide feedback based on the evaluation results.

[0011] Preferably, the multimodal sensing data in step one includes transient ground voltage signal, insulation resistance value, pile temperature data, ambient humidity data, and partial discharge signal. The frequency range of the transient ground voltage signal covers 3MHz-100MHz and is synchronously acquired in a non-invasive manner through a capacitively coupled sensor.

[0012] Preferably, the data fusion algorithm in step four is one of the DS evidence theory algorithm and the Bayesian fusion algorithm, which achieves comprehensive diagnosis of multi-dimensional features through weighted calculation.

[0013] Preferably, the decision-making and feedback mechanism described in step five includes: issuing a timely warning when an early deterioration trend is diagnosed; and forcibly stopping the operation of the charging pile and issuing an emergency alarm immediately when a serious fault risk is diagnosed.

[0014] The beneficial effects of this invention are as follows: By integrating multi-dimensional information from electrical, environmental, and physical dimensions, the system overcomes the limitations of single-parameter detection, providing a more comprehensive and accurate reflection of equipment insulation status and significantly reducing false alarms and missed alarms. Secondly, based on early sign analysis such as partial discharge and temperature-humidity correlation trends, the system can issue warnings before insulation resistance drops significantly, enabling predictive maintenance. It can also accurately distinguish different types of insulation faults. Furthermore, by combining digital twins and cloud platform AI, the system can adaptively adjust diagnostic thresholds according to equipment aging and environmental changes, continuously improving diagnostic accuracy through continuous learning. Attached Figure Description

[0015] Fig. 1 This is a flowchart of the system of the present invention; Fig. 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 are within the scope of protection of the present invention.

[0017] like Figs. 1-2 As shown, this embodiment of the invention provides a charging pile insulation performance testing system based on multimodal sensing, comprising: Multimodal sensing unit, used to synchronously collect electrical parameters, environmental parameters and physical state parameters of charging pile; The data acquisition and preprocessing unit is used to preprocess the data acquired by the multimodal sensing unit; The edge computing and fusion analysis unit includes a digital twin module and a fusion diagnostic module, which are used to comprehensively evaluate the insulation status of the charging pile based on preprocessed multimodal data and digital twin models through data fusion algorithms, and output diagnostic results. The cloud platform and user interaction unit are used to receive, store, and display diagnostic results, and send alarm information to users.

[0018] Multimodal sensing simultaneously acquires three types of parameters, replacing single insulation resistance detection, avoiding false alarms and missed alarms caused by environmental interference, and providing more comprehensive detection.

[0019] It achieves real-time online monitoring, eliminating the limitations of manual periodic inspections, removing human error, and significantly improving efficiency; edge computing combined with digital twins and fusion algorithms can capture early signs of insulation degradation, accurately identify fault types such as moisture and damage, and provide early warnings; the cloud platform displays and stores diagnostic results in real time and issues timely alarms, allowing users to intuitively grasp the status and ensuring charging safety; overall, it fills the gap in existing technology, achieving real-time, comprehensive, and accurate assessment and early warning of the insulation status of charging piles.

[0020] The multimodal sensing unit includes at least an electrical sensing module, an environmental sensing module, a thermal sensing module, and a high-frequency current sensing module. The electrical sensing module is used to collect insulation resistance and leakage current, the environmental sensing module is used to collect ambient temperature and humidity, the thermal sensing module is used to collect cable joint temperature, and the high-frequency current sensing module is used to collect partial discharge signals.

[0021] The electrical sensing module simultaneously collects insulation resistance and leakage current, which more accurately captures subtle changes in insulation performance compared to single-parameter detection. The environmental sensing module collects temperature and humidity data, providing environmental compensation for the detection results and effectively avoiding false alarms and missed alarms caused by high humidity and other environmental factors. The thermal sensing module monitors the temperature of cable joints, accurately locating the risk points of insulation damage caused by local overheating. The high-frequency current sensing module captures partial discharge signals, which are key early warnings of insulation degradation and can detect potential insulation hazards in advance. The four modules work together to build a multi-dimensional monitoring system, realizing real-time and comprehensive perception of insulation status. This not only solves the problem of inefficiency in manual detection but also breaks through the limitations of single sensors, significantly improving detection accuracy and early warning capabilities.

[0022] The integrated diagnostic module employs a machine learning model, which is trained using historical multimodal data and corresponding insulation status labels. It can output at least one of the following: insulation health index, early warning signal, and fault type diagnosis.

[0023] The machine learning model, trained with historical multimodal data and insulation tags, can deeply mine the correlation patterns of parameters, overcome the limitations of single-parameter detection, and significantly reduce false alarms and missed alarms caused by environmental interference such as temperature and humidity, thus significantly improving diagnostic accuracy. The output insulation health index enables quantitative assessment of the state, replacing fuzzy judgments and allowing users to intuitively grasp the degree of insulation degradation. The model can capture early characteristics of insulation degradation and output early warning signals in advance, solving the problem of "severe aging by the time abnormalities appear" in traditional detection, allowing sufficient time for maintenance. It can accurately identify fault types such as moisture and damage, providing targeted maintenance basis and avoiding blind repairs. At the same time, the model has strong generalization ability and adapts to different working conditions, further ensuring the reliability of detection.

[0024] The system sends alarm messages via various means, including apps, text messages, and voice calls, to ensure that relevant personnel receive the alarm messages in a timely manner. The alarm message sending interval is set to remind the relevant personnel every 3 minutes until the relevant personnel confirm that they have received the alarm signal.

[0025] Multi-form alarm coverage across all scenarios, with APP, SMS, and voice calls working together to avoid missed calls from a single channel, ensuring that maintenance, management, and other relevant personnel can quickly receive warnings regardless of their work, commuting, or other scenarios, meeting the requirements for immediate reporting of power safety hazards; the 3-minute interval repetitive reminder mechanism forms an "alarm-confirmation" closed loop, which avoids both ignoring single notifications and wasting resources due to excessive frequency, effectively solving the management problem of "no response after alarm".

[0026] The data acquisition and preprocessing unit first uses filtering technology to denoise the multimodal raw data to remove electromagnetic interference and environmental noise signals; secondly, it identifies and removes abnormal data points using the 3σ criterion to prevent extreme values ​​from affecting diagnostic accuracy; and finally, it uses the Z-score standardization method to convert heterogeneous data into feature data with uniform dimensions.

[0027] Filtering technology specifically removes electromagnetic and environmental noise while retaining the core features of effective signals, preventing noise from masking subtle changes in insulation degradation and improving the signal-to-noise ratio. The 3σ criterion, based on statistical principles, objectively eliminates extreme outliers, removes distorted data caused by accidental interference, prevents misleading diagnostic results, and ensures data reliability. Z-score standardization unifies the dimensions of heterogeneous electrical and environmental data, transforming them into standardized data with a mean of 0 and a standard deviation of 1, eliminating analytical biases caused by parameter differences and providing a comparable and compatible feature basis for subsequent fusion diagnosis. The three-step process, from noise reduction and outlier removal to standardization, comprehensively improves data quality, providing solid data support for accurate insulation status assessment and early warning, and effectively avoiding false alarms and missed alarms caused by data problems.

[0028] A method for detecting the insulation performance of charging piles based on multimodal sensing, the specific steps of which are as follows: Step 1: Synchronously collect multimodal sensor data from the charging pile; Step 2: Preprocess and extract features from the multimodal sensing data to obtain feature vectors; Step 3: Compare the feature vector with the output of the charging pile digital twin model; Step 4: Based on the comparison results, the feature vectors are comprehensively analyzed and diagnosed using a data fusion algorithm to generate insulation status assessment results; Step 5: Implement corresponding decisions and provide feedback based on the evaluation results.

[0029] Simultaneous acquisition of multimodal data overcomes the limitations of single parameters, comprehensively capturing multidimensional information such as electrical and environmental data to provide complete data support for diagnosis; preprocessing and feature extraction processes purify data and extract key information, avoiding noise and redundant interference, and improving the efficiency and accuracy of subsequent analysis; combined with digital twin model comparison, it achieves state verification under virtual-real mapping, accurately identifies subtle signs of insulation degradation, and provides early warning of potential risks; data fusion algorithms deeply mine the correlation of multi-source data, breaking through the limitations of traditional detection, significantly reducing false alarm and missed alarm rates, while accurately distinguishing fault types; the decision feedback closed-loop design ensures that the evaluation results are quickly transformed into early warning, shutdown, and other actions, realizing full-process automation from monitoring to response, solving the problem of inefficiency in manual detection, and ensuring charging safety.

[0030] In step one, the multimodal sensing data includes transient ground voltage signals, insulation resistance values, pile temperature data, ambient humidity data, and partial discharge signals. The frequency range of the transient ground voltage signals covers 3MHz-100MHz and is synchronously acquired in a non-invasive manner through capacitively coupled sensors.

[0031] Multi-dimensional data collaboration covers core influencing factors of insulation status. Transient ground voltage signals and partial discharge signals complement each other, and combined with insulation resistance, temperature, and humidity data, a complete insulation degradation monitoring chain is constructed, breaking through the limitations of single parameters. The 3MHz-100MHz wide-spectrum design of the transient ground voltage signal accurately captures weak discharge pulses in the early stages of insulation defects, enabling early warning of degradation signs and solving the problem of "severe aging by the time abnormalities appear" in traditional detection. The capacitive coupling non-invasive acquisition does not require modification of the charging pile structure or affect normal charging, making installation and maintenance convenient, while avoiding potential damage to equipment caused by invasive detection. Multi-source data synchronous acquisition can eliminate interference from environmental humidity and other factors through cross-validation, significantly reducing false alarm and missed alarm rates, providing high-quality data support for subsequent accurate diagnosis, and comprehensively improving the reliability and timeliness of insulation status assessment.

[0032] In step four, the data fusion algorithm is one of the DS evidence theory algorithm and the Bayesian fusion algorithm, which achieves comprehensive diagnosis of multi-dimensional features through weighted calculation.

[0033] Both algorithms excel at handling uncertain information, effectively integrating multi-dimensional heterogeneous features, overcoming the limitations of single parameters, and significantly reducing false alarms and missed alarms caused by environmental interference. The weighted calculation mechanism can highlight the weights of key features, suppress redundant information, and improve diagnostic accuracy. The DS evidence theory is adapted to conflict data processing, while the Bayesian algorithm has the advantage of probabilistic reasoning, adapting to different working conditions and enhancing system robustness. It enables quantitative assessment of insulation status and accurate identification of fault types, providing a scientific basis for maintenance and solving the problems of fuzzy judgment and delayed response in traditional detection.

[0034] The decision-making and feedback mechanism in step five includes: issuing early warning information in a timely manner when an early deterioration trend is diagnosed; and forcibly stopping the operation of the charging pile and issuing an emergency alarm immediately when a serious fault risk is diagnosed.

[0035] Early warning of deterioration trends enables preventative measures, allowing ample time for maintenance and preventing further escalation of faults; in cases of severe risk, forced shutdown and emergency alarms quickly eliminate potential safety hazards, minimizing the risk of accidents such as electrical leakage and fire, and ensuring personal and property safety; a tiered handling mechanism adapts to different risk levels, ensuring that minor anomalies do not lead to excessive shutdowns affecting usage, while also addressing serious risks, balancing safety and practicality, and significantly improving the reliability of charging pile operation.

[0036] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0037] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A charging pile insulation performance testing system based on multimodal sensing, characterized in that, include: Multimodal sensing unit, used to synchronously collect electrical parameters, environmental parameters and physical state parameters of charging pile; The data acquisition and preprocessing unit is used to preprocess the data acquired by the multimodal sensing unit; The edge computing and fusion analysis unit includes a digital twin module and a fusion diagnostic module, which are used to comprehensively evaluate the insulation status of the charging pile based on preprocessed multimodal data and digital twin models through data fusion algorithms, and output diagnostic results. The cloud platform and user interaction unit are used to receive, store, and display the diagnostic results, and send alarm information to the user.

2. The charging pile insulation performance testing system based on multimodal sensing according to claim 1, characterized in that: The multimodal sensing unit includes at least an electrical sensing module, an environmental sensing module, a thermal sensing module, and a high-frequency current sensing module. The electrical sensing module is used to collect insulation resistance and leakage current, the environmental sensing module is used to collect ambient temperature and humidity, the thermal sensing module is used to collect cable joint temperature, and the high-frequency current sensing module is used to collect partial discharge signals.

3. The charging pile insulation performance testing system based on multimodal sensing according to claim 1, characterized in that: The fusion diagnostic module employs a machine learning model, which is trained using historical multimodal data and corresponding insulation status labels. It can output at least one of insulation health index, early warning signal, and fault type diagnosis.

4. The charging pile insulation performance testing system based on multimodal sensing according to claim 1, characterized in that: The alarm information is sent in multiple forms, including APP, SMS, and voice call, to ensure that relevant personnel can receive the alarm information in a timely manner. The alarm information is sent every 3 minutes, and the sending will stop only after the relevant personnel confirm that they have received the alarm signal.

5. The charging pile insulation performance testing system based on multimodal sensing according to claim 1, characterized in that: The data acquisition and preprocessing unit first uses filtering technology to denoise the multimodal raw data to remove electromagnetic interference and environmental noise signals; secondly, it identifies and removes abnormal data points using the 3σ criterion to prevent extreme values ​​from affecting diagnostic accuracy; and finally, it uses the Z-score standardization method to convert heterogeneous data into feature data with uniform dimensions.

6. A method for detecting the insulation performance of charging piles based on multimodal sensing, characterized in that, The specific steps are as follows: Step 1: Synchronously collect multimodal sensor data from the charging pile; Step 2: Preprocess and extract features from the multimodal sensing data to obtain feature vectors; Step 3: Compare the feature vector with the output of the charging pile digital twin model; Step 4: Based on the comparison results, the feature vectors are comprehensively analyzed and diagnosed using a data fusion algorithm to generate insulation status assessment results; Step 5: Execute the corresponding decisions and provide feedback based on the evaluation results.

7. The method for detecting the insulation performance of a charging pile based on multimodal sensing according to claim 6, characterized in that: The multimodal sensing data mentioned in step one includes transient ground voltage signals, insulation resistance values, pile temperature data, ambient humidity data, and partial discharge signals. The frequency range of the transient ground voltage signals covers 3MHz-100MHz and is synchronously acquired in a non-invasive manner through capacitive coupling sensors.

8. The method for detecting the insulation performance of a charging pile based on multimodal sensing according to claim 6, characterized in that: The data fusion algorithm described in step four is one of the DS evidence theory algorithm and the Bayesian fusion algorithm, which achieves comprehensive diagnosis of multi-dimensional features through weighted calculation.

9. The method for detecting the insulation performance of a charging pile based on multimodal sensing according to claim 6, characterized in that: The decision-making and feedback mechanism described in step five includes: issuing early warning information in a timely manner when an early deterioration trend is diagnosed; and forcibly stopping the operation of the charging pile and issuing an emergency alarm immediately when a serious fault risk is diagnosed.

Images