An emergency portable power generation apparatus with a fault self-detection module

Abstract

The application provides an emergency portable power generation equipment with a fault self-detection module, and relates to the technical field of portable power generation.The equipment collects operation data such as voltage, current, vibration, acoustics and temperature in real time through an integrated multi-modal sensor module.A built-in edge detection and analysis module in the core controller adopts a lightweight digital twin model to perform fusion analysis on multi-dimensional data, can accurately identify early fault signs and diagnose fault types, and forms a fault fingerprint.According to a preset fault grading logic table, the system drives an execution control module to realize intelligent grading response from early warning, automatic load reduction, output isolation to emergency shutdown.The application also has the functions of environmental compensation and fault information reporting after power failure, greatly improves the reliability, safety and maintenance efficiency of the equipment in key scenes such as emergency rescue and field operation through active sensing, prediction and self-healing control, and realizes predictive maintenance.

Description

Technical Field

[0001] This invention relates to the field of portable power generation technology, specifically to an emergency portable power generation device with a fault self-detection module. Background Technology

[0002] Existing portable emergency power generation equipment, especially traditional gasoline generators, is quite mature in structure and function, but it still has significant shortcomings in terms of intelligence and reliability. The main problems are as follows: Problem one: the lack of and passivity of condition monitoring. Traditional power generation equipment is typically equipped with only the most basic instruments, such as voltmeters or simple indicator lights, which cannot provide real-time, detailed monitoring of key internal parameters such as vibration, winding temperature, exhaust temperature, operating noise, and power quality. Users are completely unaware of the equipment's health status and can only perform passive, reactive maintenance. The second problem is the lack of fault early warning capabilities. Due to the lack of in-depth monitoring, the equipment often only shuts down suddenly when a serious fault occurs, such as a bearing seizure, winding burnout, or engine cylinder scoring. Such sudden failures not only lead to power outages, severely impacting critical tasks, but also incur high repair costs and may even result in the direct scrapping of the equipment. Thirdly, the protection mechanism is simplistic and crude. Traditional protection measures are usually limited to overload protection switches. When an overload occurs, the protection switch directly cuts off all outputs, resulting in a complete power outage. It cannot perform intelligent load management, such as prioritizing core loads or implementing flexible handling like load reduction. Its response to electrical faults such as short circuits and overvoltages, or mechanical faults such as severe impacts, is also not fast or intelligent enough. Fourthly, standard power generation equipment suffers from poor environmental adaptability and lacks remote management. Its performance is affected by environmental factors such as altitude and temperature, which it cannot compensate for or adjust for, potentially leading to decreased operating efficiency or misjudgments. Furthermore, it generally lacks remote communication capabilities, making remote monitoring, diagnosis, and management impossible in unattended scenarios. Summary of the Invention

[0003] Technical problems to be solved To address the shortcomings of existing technologies, this invention provides an emergency portable power generation device with a fault self-detection module, solving the following problems: 1. Traditional power generation equipment only has basic instrument displays and cannot monitor key operating parameters such as vibration, temperature, noise, and power quality in real time, resulting in the equipment's health status being invisible and maintenance being carried out only after a failure occurs; 2. Due to insufficient monitoring data, traditional power generation equipment usually only shuts down after serious faults such as bearing seizure, winding burnout, and engine cylinder scoring occur, making it impossible to detect abnormalities in advance, resulting in sudden shutdowns and high maintenance costs. 3. Traditional power generation equipment relies on simple overload protection switches, which cannot respond intelligently according to fault type or load priority. The protection against complex faults such as short circuits, overvoltages or mechanical shocks is not timely or accurate enough.

[0004] 4. Traditional power generation equipment lacks the ability to adapt to changes in environment such as altitude and temperature, limiting its operational stability; at the same time, it lacks remote communication and management functions, making it unable to support unattended or centralized monitoring application scenarios.

[0005] Technical solution To achieve the above objectives, the present invention provides the following technical solution: an emergency portable generator with a fault self-detection module, comprising a frame, an internal frame, a fuel tank on the top of the frame, a control panel on one side of the frame, and a controller on the other side of the frame. An output interface panel is located on one side of the control panel, and the control panel is electrically connected to the controller. Both the control panel and the output interface panel are mounted on the side of the frame. A gasoline generator is located at the bottom of the fuel tank, a battery is located on one side of the gasoline generator, and a muffler is located on the other side of the gasoline generator. An air filter is located on one side of the muffler. The controller contains a fault self-detection module, including a sensor module for collecting and analyzing operating status and an edge detection and analysis unit for fault identification. The surfaces of the frame, frame, and gasoline generator are provided with a sensor module, an execution control module, and an environmental monitoring module. The controller is electrically connected to the sensor module, the execution control module, and the environmental monitoring module via signal lines.

[0006] Preferably, the frame is L-shaped, the fuel tank is installed on the top surface of the frame and is bolted to the frame, the gasoline generator and battery are both installed inside the frame, the frame, gasoline generator and battery are all bolted to the frame, the air filter and muffler are both connected to the gasoline generator pipeline, and the surfaces of the gasoline generator and frame are provided with sensor modules, execution control modules and environmental monitoring modules.

[0007] Preferably, the sensor module collects the operating status parameters of the gasoline generator in real time, including at least two of voltage, current, vibration, acoustic signals, and temperature.

[0008] Preferably, the sensor module includes a voltage sensor, a current sensor, a harmonic acquisition sensor, a triaxial accelerometer, a microphone sensor, and a temperature sensor. The voltage sensor, current sensor, and harmonic acquisition sensor are all electrically connected to the power output line between the control panel and the output interface panel. The triaxial accelerometer is installed on the surface of the frame near the engine end. The microphone sensor is installed near the ventilation opening of the frame housing. The temperature sensor is installed on the surface of the gasoline generator near the generator windings, the exhaust port, and the controller.

[0009] Preferably, the controller includes an edge detection and analysis module that is electrically connected to the sensor module to perform fusion processing on the collected multimodal data. The edge detection and analysis module consists of a microprocessor, a storage chip, and an algorithm program. The embedded microprocessor runs a preset lightweight digital twin algorithm program to achieve fusion analysis of multimodal signals, thereby identifying abnormal operation of the gasoline generator.

[0010] Preferably, the edge detection and analysis module includes a signal acquisition circuit, a data preprocessing unit, and an embedded microprocessor. The embedded microprocessor stores a lightweight model for operational status evaluation to perform fusion analysis on multimodal signals, thereby identifying operational anomalies.

[0011] Preferably, the controller further includes a fault classification and self-healing control module. The fault classification and self-healing control module has a built-in fault classification logic table, outputs control signals to the execution control module according to the input parameter range, and outputs corresponding control commands to the execution control module. The execution control module is connected to the power output terminal and fuel supply port of the gasoline generator, and realizes load switching, output isolation, load reduction operation and emergency shutdown according to the control commands.

[0012] Preferably, the execution control module includes a power relay, a solid-state relay, and a fuel control solenoid valve, and has redundant hardware protection circuits, including parallel fuses and overvoltage protection units, to prevent equipment damage and safety risks caused by malfunctions. The control panel also includes a human-machine interaction and communication unit, which displays the health status of the equipment, stores fault evidence packages, and reports fault summaries to a remote terminal via wired or wireless means.

[0013] Preferably, the controller is further communicatively connected to an environmental monitoring module, which includes a temperature sensor, a humidity sensor, an atmospheric pressure sensor, and a microphone sensor to detect external temperature, humidity, air pressure, and noise levels. The edge detection and analysis module includes an environmental compensation subunit, which corrects the sensor acquisition values ​​based on the output parameters of the environmental monitoring module to improve fault identification accuracy and stability. The power generation equipment further includes an independently powered backup battery that automatically switches to power supply mode when the main circuit is powered off, providing power to the edge detection and analysis module and the communication unit.

[0014] Beneficial effects This invention provides an emergency portable power generation device with a fault self-detection module. It has the following beneficial effects: 1. In this solution, multiple types of high-precision sensor modules are deployed in key parts of the rack, including electrical parameter sensors, vibration sensors, acoustic sensors, and temperature sensors, enabling real-time acquisition of key operating parameters such as voltage, current, vibration, noise, winding and exhaust temperatures. The acquired signals are amplified, filtered, and converted from digital to digital (A / D) before entering the edge analysis module for feature extraction and fusion. Combined with a digital twin status assessment model, continuous monitoring and trend judgment of the equipment's operating status are achieved, allowing users to understand the equipment's health status at any time, thus transforming the traditional "post-event response" into "real-time perception and proactive control."

[0015] 2. This solution uses an embedded variational autoencoder model to reconstruct the multimodal feature vectors and analyze the error. It then uses the comparison between the anomaly score and an adaptive dynamic threshold to achieve anomaly detection. When the equipment status begins to deviate from the normal distribution, the model can capture subtle anomaly trends in advance and analyze the error contribution of different feature dimensions using a fault fingerprint database. This allows for accurate identification of early potential problems such as bearing wear, partial short circuits in windings, and abnormal combustion, thus shifting from "sudden shutdown" to "early warning."

[0016] 3. This solution employs a tiered self-healing control strategy, automatically executing different responses based on the fault level: minor anomalies are only indicated and documented; moderate anomalies automatically reduce load while retaining core load output; and severe faults result in output isolation and fuel cutoff, achieving intelligent protection throughout the entire process, from flexible load reduction to emergency shutdown. The system's internal solid-state relays and power relays work together to independently control different loads, avoiding the crude protection problem of traditional equipment that simply "completely shuts down power."

[0017] 4. This solution incorporates an environmental monitoring module, which acquires environmental data in real time through temperature, humidity, and atmospheric pressure sensors. An environmental compensation subunit dynamically adjusts the judgment thresholds of the digital twin model, ensuring the equipment maintains judgment accuracy and operational stability under various operating conditions, including high altitudes, high temperatures, and low temperatures. Simultaneously, the system integrates 4G / Wi-Fi / Bluetooth communication modules, enabling real-time reporting of equipment status, fault levels, and diagnostic results to a remote monitoring center or mobile terminal, achieving unattended operation, remote diagnostics, and maintenance scheduling. Attached Figure Description

[0018] Figure 1 This is an isometric drawing of the present invention; Figure 2 This is a front view of the present invention; Figure 3 This is a partial isometric view of the present invention; Figure 4 This is a partial first perspective view of the present invention; Figure 5 This is a partial second perspective view of the present invention; Figure 6 This is a flowchart of the process of the present invention; Figure 7 This is a system flowchart of the present invention.

[0019] The components include: 1. Rack; 2. Frame; 3. Control panel; 4. Output interface panel; 5. Fuel tank; 6. Gasoline generator; 7. Battery; 8. Air filter; 9. Muffler; 10. Controller; 11. Execution control module; 12. Sensor module; 13. Environmental monitoring module. Detailed Implementation

[0020] 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. Specific Implementation Example 1

[0022] like Figures 1 to 7 As shown, this invention proposes an emergency portable power generation device with a fault self-detection module, the structure and operation of which are as follows: This invention provides an emergency portable generator with a fault self-detection module. The main structure of the emergency portable generator is fixed to a high-strength metal frame 1 to ensure operational stability and durability during transport. An L-shaped frame 2 is located inside the frame 1. A fuel tank 5 is mounted on the top horizontal surface of the frame 2 using high-strength anti-loosening bolts. The bottom of the fuel tank is connected to the carburetor of the gasoline generator 6 via a hose clamp-fixed fuel-resistant pipeline. On the vertical plane of the L-shaped frame 2, a control panel 3 and an output interface panel 4 are integrated on the front, forming the interface for human-machine interaction and power output. A controller 10 is installed on the other side of the frame 2 or in a concealed internal location. The base of the frame 1 is used to mount the core power and generator unit, including the gasoline generator 6 and a battery 7 that provides starting power and system backup power. These heavy components are bolted to the frame 1 using shock-absorbing pads to absorb and isolate vibrations generated during operation. To reduce operating noise and treat exhaust gases, the exhaust port of the gasoline generator 6 is connected to a muffler 9 via a metal pipe, while the intake port is connected to an air filter 8. The entire device can be covered with a protective shell with ventilation openings, which protects the internal components and ensures good heat dissipation.

[0023] The core of this equipment lies in its highly integrated fault self-detection and control system. This system, coordinated by the controller 10, mainly includes a sensor module 12, an edge detection and analysis module, a fault classification and self-healing control module, an execution control module 11, an environmental monitoring module 13, and a human-machine interaction and communication unit. To comprehensively and accurately grasp the operating status of the power generation equipment, the system deploys a rich array of sensor modules 12 for real-time acquisition of multi-dimensional and multi-modal data. This module specifically includes: Electrical parameter sensors: A high-precision Hall current sensor is connected in series and a voltage divider sensor is connected in parallel on the main power output line between control panel 3 and output interface panel 4 to monitor the effective values ​​of output voltage and current in real time. Simultaneously, a harmonic acquisition sensor is integrated into the line. This sensor, based on the fast Fourier transform principle, can acquire total harmonic distortion and key harmonic components such as the 3rd, 5th, and 7th harmonics. These parameters are important indicators for judging the health status of the generator windings and the power quality.

[0024] Mechanical condition sensor: A MEMS triaxial accelerometer is securely mounted on the surface of frame 1 near the end of the gasoline generator 6 engine block to capture vibration signals of the equipment in the X, Y, and Z dimensions. Spectral analysis of the vibration signals can effectively identify mechanical faults such as bearing wear, rotor imbalance, and loose screws.

[0025] Acoustic Status Sensor: A wideband microphone sensor is installed near the ventilation opening on the rack 1 housing. This location effectively captures internal operating sounds while avoiding interference from direct airflow. By collecting and analyzing acoustic signals, abnormal sound characteristics such as knocking, unusual noises, and excessive valve clearance can be identified.

[0026] Temperature sensors: Multiple high-precision NTC thermistors and PT100 platinum resistance temperature sensors are strategically placed at key temperature measurement points. Specifically, one sensor is attached close to the stator winding housing of the gasoline generator 6 to monitor the temperature rise of the generator unit; another is installed near the exhaust port to monitor engine combustion conditions and exhaust temperature; and yet another is installed near the housing of the controller 10 to monitor the operating temperature of the electronic control unit and prevent it from malfunctioning due to overheating.

[0027] The raw analog signals collected by all sensors first enter the signal acquisition circuit inside the edge detection and analysis module. This circuit includes signal amplification, filtering, and A / D conversion functions to convert the analog signals into digital signal streams. Subsequently, the data enters the data preprocessing unit for normalization, noise reduction, Kalman filtering of current and voltage signals, wavelet transform of vibration signals, and timestamp alignment to form standardized multimodal data frames.

[0028] The processed data is then fed into an embedded microprocessor for core analysis. This microprocessor uses an ARM Cortex-M series processor, but higher-level processors can also be used. Internally, the microprocessor runs a pre-defined, lightweight digital twin state assessment model optimized for embedded environments. This model consists of two parts: Baseline Model: This is a hybrid model combining a simplified physical model of the equipment and a data-driven model. The physical model estimates the voltage, current, and temperature ranges under normal operating conditions based on the generator's rated parameters. The data-driven component is an autoencoder neural network trained using a large amount of normal operating data collected before the equipment leaves the factory. This network learns a compressed representation of multimodal data under normal operating conditions, namely, a fused feature vector of voltage, current, vibration spectrum, acoustic characteristics, and temperature.

[0029] Anomaly Detection Algorithm: During device operation, preprocessed real-time multimodal data feature vectors are input into a pre-trained autoencoder. The model attempts to reconstruct the input data and then calculates the reconstruction error between the input vector and the reconstructed vector. Under normal operation, the reconstruction error is small because the input data and training data have the same distribution. Once the device malfunctions, its real-time data features deviate from the normal distribution, leading to a significant increase in the reconstruction error. The system sets a dynamic threshold; when the reconstruction error exceeds this threshold, it is considered an operational anomaly.

[0030] In addition, this module includes an environmental compensation subunit. It receives data from the environmental monitoring module 13, including temperature, humidity, and atmospheric pressure sensors, and corrects the expected parameter range of the baseline model according to a preset compensation algorithm or lookup table method. Specifically, in high-altitude areas, reduced atmospheric pressure affects engine intake efficiency and combustion, leading to decreased output power and changes in exhaust temperature. The environmental compensation subunit adjusts the criteria for determining normal operating conditions accordingly, thereby avoiding false alarms caused by environmental changes and greatly improving the accuracy and robustness of fault identification.

[0031] Once the edge detection and analysis module identifies an operational anomaly, it transmits the anomaly type—specifically, which sensor data deviated from the normal range and the degree of deviation, i.e., the magnitude of the reconstruction error—to the fault classification and self-healing control module. This module incorporates a detailed fault classification logic table, which is a rule base built upon expert knowledge and extensive experimental data.

[0032] The fault levels are as follows: Level 1: Warning: Generator winding temperature exceeds normal range by 5%, vibration signal energy slightly increases in a specific frequency band, and output voltage and current fluctuate slightly (<5%) continuously. System response and control command output: Display "Warning" information and fault code on the control panel 3 screen; store fault evidence package, sensor data for 30 seconds before and after the event; send low-priority alarms to remote terminals via communication unit.

[0033] Level 2: Load Reduction and Correction: Output total harmonic distortion >8%, generator winding temperature exceeds normal range by 15%, continuous overload (current exceeds rated value by 110% for more than 1 minute), vibration signal amplitude significantly increased. System Response and Control Command Output: Output control commands to execution control module 11, disconnect non-core load interfaces through solid-state relays to achieve load reduction operation - display "load reduction operation" status on the control panel, and report medium-priority faults.

[0034] Level 3: Critical Fault: Output voltage deviates significantly by >±15%, critical vibration frequency amplitude exceeds the danger threshold, and exhaust port temperature and winding temperature rise rapidly. System Response and Control Command Output: Immediately output a high-level signal to the power relay of the execution control module 11 to achieve complete output isolation and protect electrical equipment. At the same time, the output signal controls the fuel control solenoid valve to close, cutting off the fuel supply and achieving an emergency shutdown.

[0035] Level 4: Catastrophic Failure: The current sensor detects a short circuit, i.e., the instantaneous current value exceeds 5 times the rated value; the triaxial accelerometer detects a severe impact and overturning; the hardware redundant protection circuit, i.e., the parallel fuse and the overvoltage protection unit, directly activate, physically breaking the circuit; the execution control module executes an emergency stop command, reports the highest priority fault alarm, and activates the buzzer.

[0036] The execution control module 11 interacts with the human-machine interface: The execution control module 11 is the final executor of the system control logic. It receives instructions from the fault classification module and completes specific operations through its internal components. This module includes: Power relays: used to control the on / off state of the main power output, and to cut off the main circuit during "output isolation" and "emergency stop". Solid-state relays: used to control multiple parallel output interfaces, and can cut off parts of the load individually and in groups according to instructions, realizing "load switching" and "load reduction operation". Fuel control solenoid valve: installed in the fuel line from the fuel tank to the engine, it closes immediately upon receiving a stop command, providing the most fundamental emergency stop safety guarantee. Redundant hardware protection circuit: including an overvoltage protection unit consisting of a fast-acting fuse and a TVS diode connected in parallel to the main output line, serving as the last line of defense for the software control system to prevent extreme electrical faults caused by controller failure or malfunction.

[0037] The human-machine interface and communication unit on control panel 3 is responsible for information display and data uploading. It typically includes an LCD screen and multiple status indicator lights. The screen can display real-time information such as generator voltage, current, power, temperature, and remaining operating time. When a fault occurs, it clearly displays the fault level, code, and suggested actions. This unit has a built-in storage chip to store "fault evidence packages" generated by the edge computing module. This data can be exported via USB or reported to the remote monitoring center and user mobile app via integrated 4G / Wi-Fi / Bluetooth wireless communication modules. This information includes the device ID, fault time, fault level, and a summary of key parameters, facilitating remote diagnosis and maintenance scheduling.

[0038] The working principle of this power generation equipment is as follows: When the user presses the start button, battery 7 powers the main circuit and controller 10. Controller 10 first performs a power-on self-test (POST) to confirm that all sensors, actuators, and communication modules are functioning correctly. Then, the controller controls the starter motor to drive the gasoline generator 6 to start. After the generator enters stable operation, sensor module 12 and environmental monitoring module 13 begin continuously collecting data at a high frequency. The raw data stream is input to the edge detection and analysis module of controller 10. After signal conditioning, preprocessing, and feature extraction, a unified format multimodal feature vector is formed. This feature vector is input in real-time to the lightweight digital twin state assessment model. The model calculates the reconstruction error and compares it with a dynamic threshold. Simultaneously, the environmental compensation subunit adjusts the judgment criteria based on environmental data. This module determines the fault level according to the fault classification logic table and generates corresponding control commands, which are output to the execution control module 11. The execution control module 11 drives relays or solenoid valves according to the commands to complete operations such as load reduction, isolation, and shutdown. At the same time, the human-machine interface unit updates the equipment status on the local display screen and packages and stores fault information for remote reporting. During operation, if the main circuit loses power due to a fault or emergency shutdown, the system will automatically and seamlessly switch to independent backup battery 7 for power. This backup battery ensures that the edge detection and analysis module and communication unit of controller 10 can continue to work, complete the final fault data storage and reporting tasks, and ensure the integrity of fault information.

[0039] This emergency portable generator's self-healing control module includes an automatic health recovery algorithm and optimized battery backup circuitry to maintain equipment continuity and data integrity during minor and severe faults, and to achieve automatic repair as much as possible. Upon detecting a fault, the self-healing algorithm first classifies it: minor faults (Level 1) trigger an automatic recovery process upon warning; moderate faults (Level 2) attempt controlled recovery and report the fault; severe and catastrophic faults (Levels 3 and 4) are immediately locked, requiring manual reset and model retraining. Automatic recovery for minor faults begins with the edge detection and analysis module storing the fault evidence package and reconstruction error vector in non-volatile memory and marking them as "candidates for automatic recovery." Subsequently, under conditions such as safety requirements, fuel valve status, load isolation, and stable controller power supply, the controller performs a controlled soft restart: shutting down unnecessary outputs, isolating parallel loads, waiting for the generator and motor speeds to drop back to a safe range before re-energizing the drive subsystem, and immediately initiating a power-on self-test program to quickly verify sensor responses, ADC bias, communication links, and key electrical parameters. If the self-test passes, the system enters a short-term observation window. A multi-modal consistency check—cross-validation of vibration, acoustic, temperature, and electrical parameters—confirms whether the anomaly has disappeared. If the anomaly disappears, it is marked as "automatic recovery successful," and the fault evidence package and recovery timestamp are reported and recorded with high priority. During successful automatic recovery and when the equipment is in steady-state operation, the system executes a self-calibration procedure for the sensor zero-point and temperature compensation coefficients. It calculates the bias and proportional error using redundant sensors or reference channels, updates the zero-point and temperature coefficients through weighted filtering, and performs regression consistency tests before and after the update to ensure calibration effectiveness and support rollback operations, avoiding miscalibration under abnormal conditions.

[0040] To ensure the self-healing algorithm operates safely during power outages or failures, the system optimizes the battery backup circuit and introduces a supercapacitor + diode OR-ing structure for seamless switching. When the main power supply is disconnected or the voltage drops, the supercapacitor rapidly releases energy, while the backup battery provides continuous current, preventing the controller and critical edge detection and communication modules from resetting. A power-down detection interrupt is added to the circuit; when the voltage falls below a set threshold, an interrupt program is triggered to immediately save critical data, ensuring information integrity. Subsequently, the system, powered by the supercapacitor and backup battery, continues to perform health recovery operations, data reporting, and self-tests, completing automatic repair of minor faults. In the event of a severe fault, the system enters a "pending reset" state, locking the automatic recovery and calibration functions. All critical data is saved and uploaded to the remote maintenance center. After manual reset and retraining of the certified model, the device enters a controlled verification mode. Once the new model is confirmed to be stable, normal operation resumes.

[0041] In this invention, the fault self-detection function inputs multimodal data collected by the sensor module into a lightweight variational autoencoder model. The model calculates the reconstruction error and compares it with a dynamic threshold. If the error exceeds the threshold, it is determined to be a fault. The fault classification and fault fingerprint database further analyze the error contribution dimension to determine the fault type.

[0042] The self-healing function selects the corresponding strategy according to the fault level through the execution control module: for minor faults, it performs controlled soft restart and output isolation; for moderate faults, it performs load reduction and load switching; and for severe faults, it performs emergency shutdown and fuel cut-off, thereby realizing the 'fault self-detection and self-healing' function as described in the claims.

[0043] Through the aforementioned series of closely collaborative modules and processes, this portable emergency power generation device has achieved a leap from passive use to proactive sensing, and from post-fault repair to pre-fault prediction and self-healing, greatly improving its reliability and intelligence in key scenarios such as emergency rescue and field operations. Specific Implementation Example 2: like Figures 1 to 7 As shown, based on the content of the above specific embodiments, the following content is further disclosed: Based on Embodiment 1, this embodiment performs more refined preprocessing and feature extraction on the multimodal data collected from sensor module 12 to generate feature vectors that can accurately reflect the health status of the equipment.

[0045] Vibration Signal Processing: For the raw time-domain vibration signals acquired from the triaxial accelerometer, the data preprocessing unit in the edge detection and analysis module first performs frame processing, with each frame consisting of 1024 sampling points. Then, a Fast Fourier Transform is applied to each frame to convert it from the time domain to the frequency domain, obtaining the vibration spectrum. Key features are extracted from the spectrum, including: Fundamental frequency amplitude: the frequency corresponding to the engine speed and the energy amplitude of its harmonics, used to diagnose rotor imbalance and misalignment faults. High-frequency energy: the root mean square of energy in a specific high-frequency band, used to characterize early signs such as bearing wear and gear failures. Sideband features: analyzing the sidebands near the fundamental frequency to identify modulation phenomena related to electrical faults. Time-domain statistical features: such as kurtosis and peak factor, used to capture impact events.

[0046] Acoustic signal processing: For the acoustic signals acquired by the microphone, feature extraction is performed using MFCC, a mature technology in audio processing. MFCC can effectively simulate the auditory characteristics of the human ear and is highly sensitive to abnormal sounds emitted by the device, such as popping, friction sounds, and air leakage sounds. The preprocessing unit calculates 13-dimensional and higher-dimensional MFCCs for each frame of audio signal as acoustic features.

[0047] Electrical signal processing: For high-frequency sampling data from voltage and current sensors, in addition to calculating the effective value, the peak-to-peak value and waveform factor are also calculated. For harmonic acquisition sensor data, the total harmonic distortion and the amplitude percentage of each harmonic are directly obtained.

[0048] This power generation equipment employs a lightweight digital twin state assessment model with a variational autoencoder as its core. Compared to ordinary autoencoders, the variational autoencoder can learn the probability distribution of normal operating data, making anomaly detection more robust.

[0049] Offline training phase: Before the equipment leaves the factory, it is run in a controlled environment, covering all normal load ranges (0%, 25%, 50%, 75%, 100% load) and typical environmental conditions. During this period, a large amount of multimodal sensor data is collected, and the aforementioned feature vectors are extracted, forming a large "normal operating condition dataset." This dataset is used to train the variational autoencoder model. The model maps the high-dimensional input feature vector to a probability distribution in a low-dimensional latent space through an encoder network, and then samples from the latent space through a decoder network to attempt to reconstruct the original feature vector. The training objective is to minimize the reconstruction loss, i.e., the MSE between the input and output vectors and the KL divergence between the latent space distribution and the standard normal distribution.

[0050] Online Inference and Anomaly Scoring: The trained variational autoencoder model is deployed to the embedded microprocessor of controller 10. During device operation, the real-time generated "device state feature vector" is fed into the variational autoencoder model. The model calculates the reconstructed vector and uses the MSE between the original vector and the reconstructed vector as the "anomaly score".

[0051] Dynamic threshold setting: The system does not use a fixed threshold, but rather an adaptive dynamic threshold. The system calculates the moving average of outlier scores over the past five minutes. and standard deviation The real-time threshold is set to... ,in This is the moving average of outlier scores over a recent period. The standard deviation of outlier scores over a recent period; `threshold` is an adjustable parameter; it is a dynamic threshold calculated in real time by the system. When the real-time anomaly score exceeds this dynamic threshold at a certain moment... When this happens, the system determines that an anomaly has occurred.

[0052] Fault fingerprint diagnosis: Once an anomaly is detected, the system enters the fault diagnosis phase. At this stage, it's not just about the magnitude of the anomaly score, but more importantly, analyzing the "contribution of reconstruction error." That is, identifying the dimensions where the original feature vector and the reconstructed feature vector differ most significantly. For example, if the fundamental frequency and second harmonic of the vibration signal contribute the majority of the reconstruction error, while other feature errors are small, the system will initially diagnose "rotor imbalance." If the reconstruction error of the temperature sensor, especially near the windings, and the current harmonic characteristics significantly increases, the system will diagnose "generator overheating or risk of inter-turn short circuit in the windings." If the reconstruction error of the MFCC acoustic characteristics is large and accompanied by slight vibration anomalies, it may be diagnosed as "combustion anomaly or internal foreign object." This error contribution-based analysis method creates a "fault fingerprint" database, enabling the system not only to identify a fault but also its specific type. Specific Implementation Example 3: like Figures 1 to 7 As shown, based on the content of the above specific embodiments, the following content is further disclosed: The specific hardware model of this power generation equipment is as follows: In the emergency portable power generation device of this invention, all hardware components are selected from industrial-grade or high-reliability parts to ensure stable operation in harsh environments. The foundation of the overall device is a high-strength metal frame 1 and an internal L-shaped frame 2, which is typically made of Q235 steel but can also be made of lighter aluminum alloy profiles, and is constructed by welding and bolting. The core power unit, a gasoline generator 6, can be a mature and reliable commercial engine and generator set such as the Honda GX200 or Yamaha MZ175, and is securely mounted on the base of the frame 1 with shock-absorbing pads. The battery 7, which provides starting power and system backup power, can be a 12V valve-regulated lead-acid battery or a lithium iron phosphate (LiFePO4) battery with a capacity of 7Ah to 12Ah, model Tianneng 6-DZM-12, and is mounted on the base of the frame 1 adjacent to the gasoline generator 6. Fuel is supplied from a fuel tank 5 mounted on the top of the frame 2, and is connected to the engine carburetor via an oil-resistant rubber hose. An air filter 8 is connected to the engine's air intake, while the exhaust port is connected to a muffler 9 mounted on the other side of the frame 1 via a metal bellows. The core of this intelligent device is the controller 10, whose hardware core is a custom-designed PCB board, installed in a concealed location inside the frame 2 for dust and moisture protection. The central processing unit (CPU) of this PCB board, i.e., the embedded microprocessor, uses a high-performance Cortex-M series chip. The controller 10 is electrically connected to the control panel 3 and output interface panel 4 mounted on the vertical plane of the frame 2. The human-machine interface unit on the control panel 3 includes an industrial-grade LCD display for data display, as well as several status indicator LEDs and physical buttons.

[0054] The specific models and locations of the sensor networks are as follows: Electrical parameter sensors: A Hall effect current sensor, using an Allegro ACS758LCB-100U, is connected in series on the main power output line to accurately measure output current up to 100A. A voltage divider circuit composed of precision resistors is connected in parallel as a voltage sensor, and its output signal is sampled by the high-precision ADC built into the STM32 processor. Harmonic acquisition is not performed by a separate hardware sensor, but rather by the processor performing high-speed synchronous sampling of voltage and current signals. The total harmonic distortion and individual harmonic components are then calculated in real-time by a Fast Fourier Transform algorithm running on the processor. Mechanical state sensor: A high-precision, low-power MEMS triaxial accelerometer, model AnalogDevices ADXL355, is firmly bonded with epoxy resin or screwed to the end position on the surface of frame 1 closest to the engine block of the gasoline generator 6 to maximize the capture of equipment operating vibrations. Acoustic Status Sensor: A wideband MEMS microphone sensor, using the TDK InvenSense ICS-40720, is installed near the inside of the ventilation opening of the rack 1 housing. This location effectively picks up internal operating noise while avoiding wind noise caused by direct airflow. Temperature Sensor: The system deploys two types of temperature sensors in different locations. A fast-response NTC thermistor is installed near the key chips on the stator winding housing of the gasoline generator 6 and the PCB board of the controller 10; while in the high-temperature area near the exhaust port, a K-type thermocouple or a packaged PT100 platinum resistance thermometer is installed to monitor combustion conditions. Environmental Monitoring Module 13 can be a Bosch BME280, which can simultaneously measure temperature, humidity, and atmospheric pressure. This module is mounted on the PCB board of the controller 10 and communicates with the external environment through a small opening in the housing to accurately sense the environmental conditions of the device. The execution control module 11 is the final executor of the instructions. Its components are integrated on the PCB board of the controller 10 or on a separate power board: Power relay: used for complete isolation of the main output, using a Hongfa JQC-3FF series high-current relay, whose contact capacity must be greater than the generator's maximum output current. Solid-state relay (SSR): used for independent load switching and load reduction control of multiple output interfaces, Omron G3NA series can be selected to achieve spark-free, fast switching. Fuel control solenoid valve: this is a 12V normally closed solenoid valve, using the CEME5510 series, installed in series in the fuel line from the fuel tank 5 to the engine. Redundant hardware protection circuit: on the main output line, there is a fast-blow fuse, selected according to the rated current, and a 1.5KE series TVS diode for overvoltage protection connected in parallel.Human-Machine Interaction and Communication Unit: In addition to the aforementioned LCD screen, this unit also integrates a multi-functional wireless communication module, which adopts Quectel EC200U series. This module supports 4G LTE network and can be optionally equipped with Wi-Fi and Bluetooth functions to report fault summaries and equipment status data to remote terminals. Specific Implementation Example 4: like Figures 1 to 7 As shown, based on the content of the above specific embodiments, the following content is further disclosed: To further verify the feasibility of the technical solution in this application, the following case study is provided: Case 1: Early Warning of Bearing Wear in Emergency Communication Base Stations in High-Altitude Areas In a mountainous region at an altitude of approximately 3,500 meters in western China, a communications repair team is deploying a temporary emergency communication base station to provide network support for rescue operations in remote areas. They are using the emergency portable power generation equipment of this invention to continuously power the core equipment of the base station. The low air pressure and large temperature differences between day and night in this region pose a severe challenge to the stable operation of the power generation equipment.

[0056] After approximately 72 hours of continuous operation, the internal fault detection module of the power generation equipment began to detect abnormal signals. A MEMS triaxial accelerometer mounted on the gasoline generator body detected subtle changes in the vibration signal. After framing and performing a Fast Fourier Transform on the vibration signal, the data preprocessing unit found that the root mean square (RMS) energy value of the high-frequency band representing the bearing's operating state showed a continuous and slow increase compared to historical baseline data. Simultaneously, the kurtosis value in the time-domain statistical features also slightly increased, which typically indicates initial impact wear. The real-time generated "equipment state feature vector," containing dozens of features including vibration, acoustics, electrical, and temperature, was fed into the deployed variational autoencoder model. The "anomaly score," or reconstruction error (MSE), calculated by the model began to slightly exceed the dynamic threshold calculated from the data of the past five minutes. Fault Fingerprint Diagnosis: The system enters the fault diagnosis phase. By analyzing the "reconstruction error contribution," it is found that the vast majority of the error originates from feature dimensions related to "high-frequency energy RMS" and "kurtosis." Based on the built-in "fault fingerprint" library, the system accurately matches this pattern as "early wear of bearings at generator or engine connection points." Since core electrical parameters such as voltage and current remain stable, the system classifies this fault as "Level 1: Warning." The LCD display on Control Panel 3 immediately displays the message: "Warning - Fault Code: M101 - Suspected abnormal mechanical vibration, inspection recommended." The status indicator light changes from green to flashing yellow. Simultaneously, the human-machine interface and communication unit, through the integrated 4G module, reports a fault evidence package containing the device ID, timestamp, fault code, and key vibration characteristic data summary to the monitoring platform of the rear command center via the MQTT protocol, and pushes a low-priority alarm to the mobile APP of the on-site engineer. During this process, the environmental monitoring module 13 monitored the low air pressure at the current altitude in real time. The environmental compensation subunit has automatically adjusted the acoustic and temperature baseline model for normal engine combustion, thereby avoiding false alarms caused by the influence of the plateau environment on the combustion state and ensuring the accuracy of this vibration anomaly diagnosis.

[0057] Upon receiving the alert, the on-site engineers inspected the generator equipment during a maintenance break. Although the equipment was still running, the engineers focused on the rotating components as instructed and used a handheld stethoscope to confirm a slight abnormal noise in the generator end bearing. The team decided to replace the bearing immediately after completing this emergency power supply mission. This successful early warning prevented the bearing from seizing or disintegrating due to continuous wear, which could have led to catastrophic failures such as generator winding burnout or even complete generator failure. This not only saved the repair team tens of thousands of yuan in equipment losses, but more importantly, it ensured the continuity and reliability of emergency communication, a critical task, demonstrating the core value of predictive maintenance.

[0058] A large outdoor enthusiast group was holding a weekend camping trip in a country park, using an emergency portable generator of this invention to power the campsite's lighting system, sound system, induction cooker, and multiple charging devices. In the evening, due to multiple members simultaneously using high-power appliances, the total load exceeded the generator's rated capacity.

[0059] Fault Detection and Diagnosis Process: The Hall current sensor on the main power output line detected that the output current continuously exceeded 110% of the rated value within 1 minute. Simultaneously, the harmonic acquisition sensor captured that the total harmonic distortion (THD) began to climb to over 8%, indicating that the generator was operating under overload and power quality was deteriorating. The temperature sensor also detected that the generator winding temperature was rising at a rapid rate. The state vector, containing features such as the current RMS, THD, and winding temperature, was fed into the variational autoencoder model, causing the "abnormal score" to spike rapidly, far exceeding the dynamic threshold. Reconstruction error contribution analysis showed that the errors were mainly concentrated in three dimensions: "current RMS," "total harmonic distortion," and "winding temperature." The system immediately diagnosed this as a "severe electrical overload." Based on the triggering conditions in the fault classification logic table, the system classified this fault as "Level 2: Load Reduction and Correction."

[0060] Controller 10 immediately sends a command to execution control module 11. The solid-state relay within the execution module momentarily trips, cutting off power to the two output interface panels 4 labeled "Non-Core / Auxiliary." These interfaces were pre-defined by the user for connecting entertainment and household appliances. The "Core" interface, connecting the campground's main lighting and emergency communication charging, remains powered. The control panel 3 screen displays: "Reduced load operation - Total power exceeded, auxiliary interface disconnected," while a short beep sounds. The remote communication unit also sends a medium-priority alarm to the event organizer's mobile app, explaining the situation and the measures taken.

[0061] With the disconnection of high-power appliances, the generator's total load quickly returned to a safe range. Current, harmonics, and temperature readings returned to normal within seconds, and the "abnormality score" also decreased. The campsite's main lighting remained uninterrupted, preventing chaos. Based on mobile phone prompts, the event organizers coordinated staggered use of high-power appliances among participants. This case fully demonstrates the device's autonomous intervention. Compared to traditional generators that either trip directly under overload causing a complete power outage or continue running until overheating and damage, this device, through intelligent, tiered load management, ensured core power needs while protecting its own safety and providing clear guidance to users, greatly improving user experience and device reliability in complex power usage scenarios.

[0062] Case 3: Identification of Precursor Signs of Fuel Depletion and Emergency Shutdown Protection at Construction Sites At a construction site where work is being carried out at night, a generator of this invention is powering a mobile lighting vehicle and some power tools. Due to negligence in site management, the operators failed to refuel in time.

[0063] Fault Detection and Diagnosis Process: As fuel in fuel tank 5 was about to run out, the air-fuel mixture entering the carburetor became unstable, causing intermittent "surge" and engine speed fluctuations. This phenomenon was captured by multiple sensors. After MFCC feature extraction, the feature vector of the collected audio signal showed a significant deviation from the normal combustion sound pattern, manifested as an increase in the energy of the low-frequency "putt-putt" sound. Due to the unstable engine speed, the voltage and frequency output of the generator also began to fluctuate slightly but sharply. The engine "surge" also produced abnormal energy peaks in the low-frequency band of the vibration spectrum. These three different modal data anomalies together caused the "equipment state feature vector" to deviate significantly from the distribution of normal operating conditions. The "abnormal score" of the variational autoencoder model rose sharply. Fault fingerprint analysis pointed to three dimensions: "MFCC acoustic characteristics," "voltage fluctuation rate," and "low-frequency vibration." The system comprehensively diagnosed it as "combustion abnormality, highly likely due to a fuel supply problem." The system initially classified this as a "Level 2" fault and attempted to observe whether recovery could be achieved by briefly reducing the load. However, due to the fundamental fuel problem, load reduction was ineffective, and the abnormal state persisted. Within seconds, the output voltage deviation exceeded the ±15% threshold. The system immediately escalated the fault level to "Level 3: Critical Fault". Controller 10 immediately implemented dual safety protection measures: it sent a high-level signal to the execution control module 11, driving the power relay to instantly disconnect all power output, protecting the connected lighting equipment and tools from unstable voltage surges. Simultaneously, it sent a shut-off signal to the fuel control solenoid valve installed in the fuel line, completely cutting off the fuel supply and allowing the engine to shut down smoothly. The control panel 3 screen displayed: "Critical Fault - Emergency Stop. Fault Code: E201 - Power System Abnormality, Please Check Fuel".

[0064] The equipment safely and automatically completed the shutdown procedure before causing damage to electrical equipment. The next morning, workers found the fuel tank empty. After refueling, the equipment started successfully on the first try, and all functions were normal. This case highlights the advantages of the equipment's multimodal fusion sensing. It doesn't simply rely on a fuel level sensor, but accurately predicts the root cause of the problem by analyzing the engine's sound, vibration, and electrical output. In traditional generators, the violent surge during fuel depletion can impact electrical equipment and even damage the generator itself due to the back electromotive force generated at the moment of shutdown. The intelligent protection mechanism of this invention transforms this potential damaging process into a controlled and safe shutdown, effectively protecting user assets and the lifespan of the equipment itself.

[0065] It should be noted that, in this document, relational terms such as "first" and "second" are used merely 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 a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising a reference structure" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0066] 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. An emergency portable power generation device with a fault self-detection module, comprising a frame (1), characterized in that: The frame (1) has a frame (2) inside, and a fuel tank (5) is provided on the top of the frame (2). A control panel (3) is provided on one side of the frame (2), and a controller (10) is provided on the other side of the frame (2). An output interface panel (4) is provided on one side of the control panel (3), and the control panel (3) is electrically connected to the controller (10). The control panel (3) and the output interface panel (4) are both installed on the side of the frame (2). A gasoline generator (6) is provided at the bottom of the fuel tank (5), and a battery (7) is provided on one side of the gasoline generator (6). A muffler (9) is provided on the other side of the muffler (9), and an air filter (8) is provided on one side of the muffler (9). The controller (10) is provided with a fault self-detection module, including a sensor module (12) for collecting and analyzing the operating status and an edge detection analysis module for fault identification. The surface of the frame (1), the frame (2) and the gasoline generator (6) is provided with a sensor module (12), an execution control module (11) and an environmental monitoring module (13). The controller (10) is electrically connected to the sensor module (12), the execution control module (11) and the environmental monitoring module (13) respectively through signal lines.

2. The emergency portable power generation device with a fault self-detection module according to claim 1, characterized in that: The frame (2) is L-shaped. The fuel tank (5) is installed on the top surface of the frame (2) and is bolted to the frame (2). The gasoline generator (6) and the battery (7) are both installed inside the frame (1). The frame (2), the gasoline generator (6), and the battery (7) are all bolted to the frame (1). The air filter (8) and the muffler (9) are both connected to the gasoline generator (6) via pipes. The surfaces of the gasoline generator (6) and the frame (2) are provided with a sensor module (12), an execution control module (11), and an environmental monitoring module (13).

3. An emergency portable power generation device with a fault self-detection module according to claim 1, characterized in that: The sensor module (12) collects the operating status parameters of the gasoline generator (6) in real time. The operating status parameters include at least two of the following: voltage, current, vibration, acoustic signal and temperature.

4. An emergency portable power generation device with a fault self-detection module according to claim 3, characterized in that: The sensor module (12) includes a voltage sensor, a current sensor, a harmonic acquisition sensor, a triaxial acceleration sensor, a microphone sensor, and a temperature sensor. The voltage sensor, current sensor, and harmonic acquisition sensor are all installed on the power output line between the control panel (3) and the output interface panel (4) for electrical connection. The triaxial acceleration sensor is installed on the surface of the frame (1) near the engine end. The microphone sensor is installed near the ventilation port of the frame (1) housing. The temperature sensor is installed on the surface of the gasoline generator (6) near the generator windings, the exhaust port, and the controller (10).

5. An emergency portable power generation device with a fault self-detection module according to claim 1, characterized in that: The controller (10) is equipped with an edge detection and analysis module and is electrically connected to the sensor module (12) to perform fusion processing on the collected multimodal data. The edge detection and analysis module consists of a microprocessor, a storage chip and an algorithm program. The embedded microprocessor runs a preset lightweight digital twin algorithm program to realize the fusion analysis of multimodal signals, thereby identifying abnormal operation of the gasoline generator (6).

6. An emergency portable power generation device with a fault self-detection module according to claim 5, characterized in that: The edge detection and analysis module includes a signal acquisition circuit, a data preprocessing unit, and an embedded microprocessor. The embedded microprocessor stores a lightweight model for operational status evaluation to perform fusion analysis on multimodal signals, thereby identifying operational anomalies.

7. An emergency portable power generation device with a fault self-detection module according to claim 1, characterized in that: The controller (10) further includes a fault classification and self-healing control module. The fault classification and self-healing control module has a built-in fault classification logic table. It outputs control signals to the execution control module (11) according to the input parameter range and outputs corresponding control instructions to the execution control module (11). The execution control module (11) is connected to the power output terminal and fuel supply port of the gasoline generator (6). It realizes load switching, output isolation, load reduction operation and emergency shutdown according to the control instructions.

8. An emergency portable power generation device with a fault self-detection module according to claim 7, characterized in that: The execution control module (11) includes a power relay, a solid-state relay and a fuel control solenoid valve, and has redundant hardware protection circuits, including parallel fuses and overvoltage protection units, to prevent equipment damage and safety risks caused by malfunctions. The control panel (3) also includes a human-machine interaction and communication unit, which displays the health status of the equipment, stores fault evidence packages, and reports fault summaries to remote terminals via wired or wireless means.

9. An emergency portable power generation device with a fault self-detection module according to claim 7, characterized in that: The controller (10) is further connected to the environmental monitoring module (13) in communication. The environmental monitoring module (13) includes a temperature sensor, a humidity sensor, an atmospheric pressure sensor and a microphone sensor to detect the external temperature, humidity, air pressure and noise level. The edge detection and analysis module includes an environmental compensation subunit, which corrects the sensor acquisition values ​​based on the output parameters of the environmental monitoring module to improve the accuracy and stability of fault identification. The power generation equipment further includes an independently powered backup battery (7), which automatically switches to power supply mode when the main circuit is powered off to power the edge detection and analysis module and the communication unit.

Images