Electric power mobile terminal fault detection system
By combining multimodal sensors and edge computing, the power mobile terminal fault detection system has achieved high precision, rapid response, and enhanced security, solving the problems of limited sensing capabilities and weak communication security in existing technologies.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-11
- Publication Date
- 2026-03-10
AI Technical Summary
Existing mobile power terminal fault detection systems suffer from limitations in sensing capabilities, high response delays, and weak communication protocol security.
It employs a multimodal sensing unit combined with an edge computing unit, including vibration sensors, infrared thermal imagers, and electromagnetic sensors, to perform real-time data processing and fault diagnosis. It also achieves end-to-end dynamic key encryption through an encrypted communication unit and performs long-term predictive maintenance in conjunction with a cloud collaboration unit.
It improves the accuracy and response speed of fault detection, reduces the false alarm rate, and enhances the security and stability of the system.
Smart Images
Figure CN121633648A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The application belongs to the technical field of power detection, and particularly relates to a power mobile terminal fault detection system. BACKGROUND
[0002] With the development of power business, the demand of customers is continuously improved, and users put forward higher and higher requirements in reliability and rapidity of power supply. The existing common power mobile terminal such as a patrol robot and a handheld detector has the following shortcomings in fault detection work.
[0003] 1. The existing system mainly relies on a single type of sensor for fault monitoring, such as only using a vibration sensor for monitoring, but this method is easily interfered by a strong electromagnetic field, resulting in a high false alarm rate.
[0004] 2. The traditional architecture mostly uploads original detection data to a cloud server for centralized analysis and judgment, and there is a large network transmission delay and calculation waiting time in this process, and the average response time is more than 2 seconds, which is much higher than the standard requirement of 200 milliseconds.
[0005] 3. The traditional communication mostly adopts a static encryption method, and lacks effective semantic recognition and behavior discrimination mechanism for communication contents of Modbus and other protocols, and specific instructions such as FLT are easily injected to induce misoperation or false alarm, which seriously threatens the stability of the system and the safety of the equipment. SUMMARY
[0006] The technical problem to be solved by the application is that the existing fault detection process has great limitations in sensing ability, high response delay and weak communication protocol security.
[0007] To solve the above problems, the technical scheme adopted by the application is as follows:
[0008] The power mobile terminal fault detection system provided by the application comprises:
[0009] The multi-modal sensing unit comprises:
[0010] a vibration sensor for capturing mechanical structure fault signals, an infrared thermal imager for detecting local temperature anomalies and an electromagnetic sensor for monitoring electromagnetic disturbance signals, and outputs multi-modal data;
[0011] The edge computing unit is connected with the multi-modal sensing unit, carries a microprocessor and has local computing capability, and comprises:
[0012] a protocol feature extraction module for extracting Modbus protocol keywords and behavior features;
[0013] Real-time diagnosis engine: used for performing wavelet packet decomposition, extracting energy entropy and other features, fusing multi-modal data and outputting fault confidence level, etc.
[0014] Encryption communication unit: used for realizing end-to-end dynamic key encryption.
[0015] Cloud collaborative unit: connected with the edge computing unit through the encryption communication unit, configured with a deep prediction model, used for analyzing the feature vectors uploaded from the edge compression, and realizing long-term predictive maintenance.
[0016] Further, a dynamic trust evaluation module is further included: connected with the protocol feature extraction module, used for monitoring the frequency of communication protocol keywords and triggering a security alarm.
[0017] Further, a behavior baseline modeling module is further included: connected with the real-time diagnosis engine, used for constructing a baseline model of the normal operation state of the device, and triggering an abnormal alarm when the state deviates from the baseline standard by n times of standard deviation, and n is preferably 3.
[0018] Preferably, the wavelet packet decomposition of the real-time diagnosis engine adopts a db4 wavelet base function, and the number of decomposition layers is 4.
[0019] Preferably, the keywords monitored by the dynamic trust evaluation module include FLT, OVR and RST, and the alarm threshold is 5 times per minute.
[0020] Further, the encryption communication unit adopts an AES-256 dynamic key encryption technology, and the key update period is less than or equal to 1 hour.
[0021] Further, the deep prediction model adopts an LSTM-Attention deep prediction model, and the input dimension of the LSTM-Attention model is a 24-dimensional feature vector, and the number of hidden layer units is 128.
[0022] Further, the fault confidence level includes normal, early warning and fault.
[0023] By adopting the above scheme, the application has the following beneficial effects:
[0024] 1. By fusing multi-modal data, the false alarm rate is reduced, and the detection accuracy is improved. In combination with the model prediction of the cloud collaborative unit, the accuracy of fault prediction is improved.
[0025] 2. Real-time diagnosis of 20ms level is realized by using edge computing, and the response speed is improved.
[0026] 3. The dynamic trust evaluation module is used to improve the security protection performance and reduce the security vulnerabilities of the protocol. BRIEF DESCRIPTION OF DRAWINGS
[0027] Figure 1 A configuration diagram of a power mobile terminal fault detection system provided by the present application;
[0028] Figure 2 A workflow diagram of the cloud collaboration unit in the present application.
[0029] Among them, 1, multi-modal sensing unit, 2, edge computing unit, 3, encryption communication unit, 4, cloud collaboration unit, 5, dynamic trust evaluation module, 6, behavior baseline modeling module.
[0030] The accompanying drawings are used to provide a further understanding of the present application, and constitute a part of the specification, together with the embodiments of the present application, to explain the present application, and do not constitute a limitation on the present application. DETAILED DESCRIPTION
[0031] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.
[0032] In the description of the present application, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, therefore cannot be understood as a limitation on the present application.
[0033] Embodiment 1
[0034] As Figure 1 shown, the power mobile terminal fault detection system proposed by the present application comprises:
[0035] For data acquisition:
[0036] The multi-modal sensing unit 1 comprises:
[0037] The vibration sensor is used to capture mechanical structure fault signals;
[0038] The infrared thermal imager is used to detect local temperature anomalies;
[0039] The electromagnetic sensor is used to monitor electromagnetic disturbance signals.
[0040] In data collection, the vibration sensor sampling rate is 10 kHz, capturing 0-5 kHz mechanical fault frequency band; the infrared thermal imager detects temperature anomalies at 30 fps frame rate.
[0041] For edge computing:
[0042] Edge computing unit 2, which communicates with multi-modal sensing unit 1 through RS485 connection, carries microprocessor, has local computing power, and contains:
[0043] Protocol feature extraction module: for extracting Modbus protocol keywords and behavior characteristics;
[0044] Real-time diagnosis engine: for performing wavelet packet decomposition, extracting energy entropy and other features, fusing multi-modal data and outputting fault confidence level.
[0045] Among them, the wavelet packet decomposition of the real-time diagnosis engine adopts db4 wavelet basis function, and the decomposition layer number is 4. The fault confidence level includes normal, warning and fault.
[0046] In edge processing, the real-time diagnosis engine performs 4-layer db4 wavelet packet decomposition on the vibration signal, extracts 8 sub-band energy entropies, and fuses temperature / electromagnetic features to generate fault levels: normal (confidence <0.3), warning (0.3-0.7), and fault (>0.7).
[0047] For security monitoring:
[0048] Dynamic trust evaluation module 5, which is connected with the protocol feature extraction module, is used for monitoring the frequency of communication protocol keywords and triggering security alarm.
[0049] In the workflow of security monitoring, the frequency of protocol keywords is counted, and the keywords monitored by the dynamic trust evaluation module 5 include FLT, OVR and RST, and the alarm threshold is 5 times / minute.
[0050] For cloud prediction:
[0051] Encryption communication unit 3: for realizing end-to-end dynamic key encryption;
[0052] Among them, the encryption communication unit 3 adopts AES-256 dynamic key encryption technology;
[0053] As Figure 2 shown, cloud collaborative unit 4: connected with edge computing unit 2 through encryption communication unit 3, configured with LSTM-Attention deep prediction model, its input dimension is 24-dimensional feature vector, hidden layer unit number is 128, used for analyzing the feature vector uploaded from the edge, realizing long-term predictive maintenance.
[0054] In cloud prediction, the encryption communication unit 3 updates the key every hour, compresses the 24-dimensional feature vector uploaded, and the LSTM-Attention deep prediction model outputs the 30-day fault probability.
[0055] For baseline comparison:
[0056] The behavior baseline modeling module 6 is connected with the real-time diagnosis engine, which is used to construct the baseline model of the normal operation state of the device, and triggers an abnormal alarm when the state deviates from the baseline standard n times standard deviation.
[0057] In this process, the behavior baseline modeling module 6 calculates the Mahalanobis distance between the real-time state and the baseline, and triggers an alarm when it exceeds 3σ.
[0058] Embodiment 2
[0059] Based on the system of embodiment 1, the application further discloses a power mobile terminal fault detection method, comprising the following steps:
[0060] S1, the multi-modal sensing unit 1 acquires vibration, temperature and electromagnetic signals in real time;
[0061] S2, the edge computing unit 2 performs wavelet packet decomposition and energy entropy calculation on the data, and generates a fault confidence level by fusing multi-modal data;
[0062] S3, the dynamic trust evaluation module 5 synchronously analyzes the frequency of communication protocol keywords and triggers a security alarm;
[0063] S4, the encryption communication module compresses and uploads the feature vector to the cloud collaboration unit 4;
[0064] S5, the cloud performs long-term fault prediction through the LSTM-Attention model;
[0065] S6, the behavior baseline modeling module 6 compares the real-time state with the baseline model and triggers an abnormal alarm.
[0066] The above describes the present application and its embodiments, which are not limited, and the drawings only show one of the embodiments of the present application, and the actual structure is not limited thereto. In general, if a person skilled in the art is inspired thereby, without departing from the purpose of the present application, without creative design, similar structural methods and embodiments of the technical solutions should belong to the protection scope of the present application.
Claims
1. A power mobile terminal failure detection system, characterized by, Comprise: Multi-modal sensing unit: contains vibration sensors for capturing mechanical structure failure signals, infrared thermal imagers for detecting local temperature anomalies, and electromagnetic sensors for monitoring electromagnetic disturbance signals, outputting multi-modal data; Edge computing unit connected with multi-modal sensing unit, equipped with microprocessor, with local computing power, containing protocol feature extraction module and real-time diagnosis engine; Encryption communication unit: for realizing end-to-end dynamic key encryption; Cloud collaborative unit: connected with edge computing unit through encryption communication unit, configured with deep prediction model, used for analyzing feature vectors uploaded from edge compression, realizing long-term predictive maintenance.
2. The power mobile terminal failure detection system of claim 1, wherein: The protocol feature extraction module is used to extract Modbus protocol keywords and behavior features; The real-time diagnosis engine is used to perform wavelet packet decomposition, extract energy entropy features, fuse multi-modal data and output fault confidence level; The fault confidence level includes normal, warning and fault.
3. The power mobile terminal failure detection system of claim 2, wherein: It also includes a dynamic trust evaluation module connected with the protocol feature extraction module, which is used to monitor the frequency of communication protocol keywords and trigger safety alarms.
4. The power mobile terminal failure detection system of claim 2, wherein: It also includes a behavior baseline modeling module connected with the real-time diagnosis engine, which is used to construct a baseline model of the normal operation state of the device, and trigger an abnormal alarm when the state deviates from the baseline standard by n times standard deviation.
5. The power mobile terminal fault detection system of claim 1, wherein: The deep prediction model adopts LSTM-Attention deep prediction model, the input dimension of LSTM-Attention model is 24-dimensional feature vector, and the number of hidden layer units is 128.
6. The power mobile terminal failure detection system of claim 2, wherein: The wavelet packet decomposition of the real-time diagnosis engine uses db4 wavelet basis function, and the number of decomposition layers is 4.
7. The power mobile terminal fault detection system of claim 3, wherein: The keywords monitored by the dynamic trust evaluation module include FLT, OVR and RST, and the alarm threshold is 5 times per minute.
8. The power mobile terminal fault detection system of claim 1, wherein: The encryption communication unit adopts AES-256 dynamic key encryption technology, and the key update period is less than or equal to 1 hour.