Non-contact household line electric leakage point intelligent diagnostic instrument based on multi-source information fusion

Abstract

The invention belongs to the technical field of electric power safety monitoring, and discloses a non-contact household line electric leakage point intelligent diagnostic instrument based on multi-source information fusion, and the diagnostic instrument obtains a space magnetic field, a temperature field and electric field information through a multi-mode sensing array including a Hall sensor, an infrared thermal imager and an electric field sensor, carries out the filtering and analysis of a dual-core signal processing system FPGA + DSP, and carries out the detection of the electric leakage point of a household line. And reconstructing leakage current distribution in combination with a space magnetic field gradient algorithm based on path integration. A ResNet18 deep learning network is adopted to suppress interference of household appliances, intelligent diagnosis is realized through comparison of an NBIoT module and a cloud platform case library, positioning is guided in cooperation with an augmented reality system, actual measurement shows that the positioning error of leakage current larger than 2 mA is smaller than 10 cm, the detection efficiency is improved by more than 8 times compared with a traditional method, and the safety requirement of the IEC62305 lightning protection standard is met.

Description

Technical Field

[0001] The present invention belongs to the technical field of power safety monitoring, and in particular relates to a contactless household line leakage point intelligent diagnostic instrument based on multi-source information fusion. Background Art

[0002] There are currently significant technical bottlenecks in household leakage detection: the traditional segmented power-off method requires repeated opening and closing of circuit breakers. According to the 2024 data from the "Residential Electrical Safety White Paper", the average inspection time exceeds 2 hours per household, and intermittent leakage faults cannot be captured; contact measuring tools such as multimeters require stripping off the cable insulation layer, and there is a risk of electric shock exceeding the safety voltage of 36V; most non-contact detectors on the market rely on single-point magnetic field strength judgment, and the positioning error is often greater than 1 meter; at the same time, the problem of electromagnetic interference from household appliances is prominent. According to the IEEEStd1159-2019 test report, the false positive alarm rate exceeds 30%, which seriously affects the reliability of detection, so it needs to be improved. Summary of the Invention

[0003] The purpose of the present invention is to provide a contactless household circuit leakage point intelligent diagnostic instrument based on multi-source information fusion to solve the problems raised in the above background technology.

[0004] In order to achieve the above objectives, the present invention provides the following technical solutions: a contactless household line leakage point intelligent diagnostic instrument based on multi-source information fusion, the intelligent diagnostic instrument comprising a multimodal sensor array, a dual-core signal processing system, and an intelligent cloud diagnosis module; The multimodal sensing array includes a Hall sensor, an infrared thermal imaging module and an environmental electric field sensor; The dual-core signal processing system adopts FPGA+DSP architecture, with front-end FPGA architecture and back-end DSP architecture; The intelligent cloud diagnosis module uploads data to the cloud database through the NBIoT module and matches similar cases in real time. It includes a spatial magnetic field gradient calculation unit, a multi-source information fusion model and cloud platform collaborative diagnosis.

[0005] Preferably, in the multimodal sensing array, the Hall sensors are 8 groups of three-axis high-sensitivity Hall sensors with a measuring range of ±50 mT and a resolution of 0.1 μT.

[0006] Preferably, in the multimodal sensing array, the temperature measurement range of the infrared thermal imaging module is 20°C to 150°C, with an accuracy of ±1°C.

[0007] Preferably, in the multimodal sensing array, the frequency band of the ambient electric field sensor is 40-60 Hz, and the dynamic range is 60 dB.

[0008] Preferably, in the dual-core signal processing system, the front-end FPGA realizes 50Hz power band-pass filtering, and the filtering Q value is greater than or equal to 100; the back-end DSP performs wavelet denoising and FFT harmonic analysis, wherein the wavelet denoising adopts a db6 wavelet basis, and the FFT harmonic analysis range is 0-500Hz.

[0009] Preferably, the magnetic field gradient calculation of the spatial magnetic field gradient calculation unit satisfies:

[0010] Wherein, K=3 (confidence coefficient), σ_B` is the standard deviation of the environmental magnetic field; when the gradient value is greater than 5μT / cm, the electric leakage early warning is triggered.

[0011] B: is the gradient operator (Nabla operator) of the magnetic field vector B, and the result is a vector, which represents the change rate and direction of the magnetic field in each direction in space.

[0012] / x: represents the change rate (i.e. partial derivative) of the X-direction component of the magnetic field intensity in the rectangular coordinate system (Bx) in the X-axis direction.

[0013] / y: represents the change rate of the Y-direction component of the magnetic field intensity (By) in the Y-axis direction.

[0014] Bz / z: represents the change rate of the Z-direction component of the magnetic field intensity (Bz) in the Z-axis direction.

[0015] | B|: is the modulus (size) of the magnetic field gradient vector B. It is a scalar, which represents the intensity of the total spatial change rate of the magnetic field, and the calculation formula is the square root of the sum of squares.

[0016] Preferably, the intelligent cloud diagnosis module comprises a pre-trained ResNet18 network, the input layer of the ResNet18 network is [magnetic field fundamental amplitude, 3rd harmonic distortion rate, temperature slope, electric field phase difference], and the interference signals of hair dryer (characteristic frequency 2kHz) and variable frequency air conditioner (15kHz) can be identified, and the false positive rate is less than 5%.

[0017] Preferably, it further comprises an IMU inertial unit (MPU6050 chip) for realizing spatial trajectory recording, and the positioning accuracy error satisfies: ​​ε = 0.02 x L + 0.05 (unit: meter) Wherein, L is the moving distance (meters).

[0018] Preferably, it further comprises an augmented reality guidance system, which projects an AR navigation interface through an OLED display screen to indicate the direction of electric leakage with an arrow and dynamically display a distance value with an accuracy of 0.1 m.

[0019] Preferably, in the multi-source information fusion model of the intelligent cloud diagnosis module, the weight of electromagnetic anomaly is 60%, the weight of temperature mutation is 30%, and the weight of electric field distortion is 10%.

[0020] The beneficial effects of the present application are as follows: 1. The present application solves the problems of blind measurement and contact risk. The diagnostic instrument adopts a non-contact live detection design, with an effective detection distance of 0.1-0.8 m, without the need to strip the cable insulation layer, completely avoiding the risk of electric shock in contact measurement. At the same time, through the multi-modal sensing array, the changes in the electromagnetic field and temperature field can be captured in real time, the intermittent electric leakage can be accurately identified, the detection efficiency is greatly improved, and there is no need to repeatedly open and close the circuit breaker, significantly reducing the operation complexity.

[0021] 2. The present application breaks through the technical limitation of low positioning accuracy. With the help of 8 groups of three-axis high-sensitivity Hall sensors (resolution 0.1 μT), IMU inertial units and spatial magnetic field gradient algorithms based on path integration, combined with a three-dimensional scanning mode (radius 30 cm spherical scanning), the positioning error of the electric leakage point can be ≤10 cm, meeting the accuracy requirements of GB / T16895.31. With the augmented reality guidance system, the direction and distance of electric leakage can be dynamically displayed on the OLED display screen, intuitively guiding the user to quickly lock the fault point, solving the problem of large-scale error caused by single-point judgment of traditional equipment.

[0022] 3. The present application greatly reduces the false alarm rate and improves the intelligence of diagnosis. With the pre-trained ResNet18 network, it can accurately identify household appliance interference signals such as electric hair dryer (2 kHz) and variable frequency air conditioner (15 kHz), with a false alarm rate of less than 5%. At the same time, the multi-source information fusion model compares with the cloud-side over 1000 cases of electric leakage case library, which can intelligently distinguish between line electric leakage (> 5 mA) and equipment electric leakage (> 10 mA), and judge the fault type, such as insulation damage, joint oxidation, etc., effectively solving the problem of high misjudgment caused by household appliance electromagnetic interference, and improving the reliability and intelligence level of the diagnosis result. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 The present application is a household line electric leakage point intelligent diagnostic instrument system composition diagram. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the present application will be clearly and completely described with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort belong to the scope of protection of the present application.

[0025] As shown in Figure 1 The embodiments of the present application provide a non-contact household line leakage point intelligent diagnostic instrument based on multi-source information fusion, which comprises a multi-modal sensor array, a dual-core signal processing system and an intelligent cloud diagnosis module. The multi-modal sensor array comprises a Hall sensor, an infrared thermal imaging module and an environmental electric field sensor. The dual-core signal processing system adopts an FPGA+DSP architecture, a front-end FPGA architecture and a back-end DSP architecture. The intelligent cloud diagnosis module uploads data to a cloud database through an NBIoT module, and matches similar cases in real time, and comprises a spatial magnetic field gradient calculation unit, a multi-source information fusion model and a cloud platform collaborative diagnosis.

[0026] In the multi-modal sensor array, the Hall sensor is an 8-group three-axis high-sensitivity Hall sensor, the range is ±50mT, and the resolution is 0.1μT.

[0027] In the multi-modal sensor array, the temperature measurement range of the infrared thermal imaging module is 20℃~150℃, and the accuracy is ±1℃.

[0028] In the multi-modal sensor array, the frequency band of the environmental electric field sensor is 40~60Hz, and the dynamic range is 60dB.

[0029] In the dual-core signal processing system, the front-end FPGA realizes 50Hz power band pass filtering, and the filtering Q value is ≥100; the back-end DSP performs wavelet denoising and FFT harmonic analysis, wherein the wavelet denoising adopts a db6 wavelet basis, and the FFT harmonic analysis range is 0~500Hz.

[0030] The spatial magnetic field gradient algorithm is:

[0031] Δs: represents a infinitesimal displacement step in space. It is a small distance scalar.

[0032] ΔB: represents the change in the magnetic field vector B after experiencing a displacement of Δs. It is a scalar.

[0033] The magnetic field gradient calculation of the spatial magnetic field gradient calculation unit satisfies:

[0034] Where K = 3 (confidence coefficient), σ_B` is the standard deviation of the ambient magnetic field; when the gradient value is greater than 5μT / cm, a leakage warning is triggered.

[0035] Symbols refer to: B: is the gradient operator (Nabla operator) of the magnetic field vector B, the result of which is a vector representing the rate of change and direction of the magnetic field in all directions in space.

[0036] / x: represents the component of the magnetic field intensity in the X direction in the rectangular coordinate system ( ) in the X-axis direction (i.e., partial derivative).

[0037] / y: represents the Y-direction component of the magnetic field intensity ( ) in the Y-axis direction.

[0038] Bz / z: Indicates the rate of change of the Z-direction component of the magnetic field intensity (Bz) along the Z-axis.

[0039] | B|: is the magnetic field gradient vector The modulus (size) of B. It is a scalar quantity that represents the intensity of the total spatial rate of change of the magnetic field. The calculation formula is the square root of the sum of the squares.

[0040] K is an adjustable amplification factor (or threshold factor), a dimensionless constant. Its specific value is determined based on the on-site electromagnetic environment, sensor accuracy, and experimental data, and is used to adjust the system's sensitivity.

[0041] σ_B: represents the standard deviation of the background noise of the magnetic field strength measured by the system under non-arcing conditions. It represents the inherent measurement fluctuation level of the system.

[0042] Among them, the intelligent cloud diagnosis module includes a pre-trained ResNet18 network. The input layer of the ResNet18 network is [magnetic field fundamental wave amplitude, third harmonic distortion rate, temperature slope, electric field phase difference]. It can identify interference signals from hair dryers (characteristic frequency 2kHz) and variable-frequency air conditioners (15kHz) with a false alarm rate of less than 5%.

[0043] It also includes an IMU inertial unit (MPU6050 chip) for realizing space trajectory recording, and the positioning accuracy error satisfies: `ε=0.02×L+0.05 (unit: meter); Wherein, L is the moving distance (m).

[0044] It also includes an augmented reality guidance system, which projects an AR navigation interface through an OLED display screen to indicate the leakage direction with an arrow and dynamically display the distance value, with an accuracy of 0.1 m.

[0045] In the multi-source information fusion model of the intelligent cloud diagnosis module, the electromagnetic anomaly weight is 60%, the temperature mutation weight is 30%, and the electric field distortion weight is 10%.

[0046] Operation process: ①Start from the distribution box, move the equipment at a uniform speed of 0.5 m / s along the line direction; when moving the equipment along the line direction, record the space trajectory through the IMU inertial unit and calculate the path integral of the magnetic field gradient: L( ×B)·dl= I_leakage; Wherein, : Path integral symbol. Indicates the integral calculation along a closed path or loop L.

[0047] ×: Curl operator (Curl). It is a vector differential operator that acts on a vector field (here it is the magnetic field B), used to measure the rotation degree or vortex source strength of the field at a point. The result of ×B itself is also a vector.

[0048] B: Magnetic field vector. Represents the magnetic field intensity and direction at a point in space. The unit is usually Tesla (T) or Gauss (G). It is a vector with three components: B=( , ,Bz).

[0049] ( ×B): Curl of the magnetic field. According to Maxwell's equations, under the condition of steady current, the curl of the magnetic field is equal to the times of the current density vector at that point.

[0050] dl: Path infinitesimal vector. Represents an infinitely small line segment on the closed path L, with the direction along the tangent direction of the path.

[0051] ( ×B)·dl: represents the component of the magnetic field curl in the direction of path dl.

[0052] : Vacuum permeability. It is a fundamental physical constant, with a value of 4π × 1 N / A² (Newton per Ampere square). It determines the ease of generating a magnetic field in a vacuum.

[0053] I_leakage: Leakage current. This is the target quantity that the formula requires to solve, referring to the net current passing through the measured conductor. For example, in a power system, if the three-phase current is balanced, I_leakage should be zero; if there is an insulation fault or load imbalance, I_leakage is not zero.

[0054] Realize the non-contact quantitative inversion of leakage current value; ② When the magnetic field gradient mutation (ΔB / Δx≥3 μT / cm) is detected, start the three-dimensional scanning mode (make a radius of 30 cm spherical scan with the fault point as the center), record the magnetic field intensity distribution in 8 directions: B azimuth angle θ = arctan( / ), pitch angle = arcsin(Bz / |B|); Where, B: magnetic field vector.

[0055] , Bz: the components of the magnetic field vector B in the X-axis, Y-axis, and Z-axis directions in the Cartesian coordinates, respectively.

[0056] |B|: the modulus (size) of the magnetic field vector B. It is a scalar, representing the total intensity of the magnetic field. The calculation formula is: |B|= .

[0057] θ (Theta): Azimuth angle. In the XY plane, rotate counterclockwise from the positive X-axis to the angle of the magnetic field vector projection in the XY plane (the range is usually 0° to 360°).

[0058] (Phi): Elevation angle or polar angle. Rotate from the XY plane upward (or downward) to the angle of the magnetic field vector itself (the range is -90° to +90°).

[0059] ③ Combine thermal imaging to locate temperature anomaly points (ΔT≥8℃ is the effective criterion); ④ Diagnostic engine output: leakage point coordinates (relative to the starting point ± 10cm), leakage current estimation (error ± 0.5mA), fault type (line insulation damage / connector oxidation / moisture leakage, etc.).

[0060] Example

[0061] In a case of electrical leakage in a 120m² residential kitchen, the traditional method took 135 minutes to locate the fault. The instrument's operation is as follows: Move along the line under the cabinet to a distance of 2.7m, triggering the alarm (magnetic field strength suddenly increases from 0.3μT to 15μT); 3D scanning confirmed that the leakage point was 8 cm behind the wall socket; Thermal imaging showed the temperature at this point was 32.5°C (ambient temperature 24°C); Cloud comparison confirmed that water leakage from the sink caused the socket to oxidize.

[0062] The total detection time was 9 minutes and 38 seconds, and the positioning error was 6 cm.

[0063] Solve the risk of blind detection and contact during power outages, and improve detection efficiency; When moving along wiring beneath cabinets, the system eliminates the need to remove insulation (effective detection range: 0.3m, completely avoiding the electric shock hazard associated with contact-based measurement). Furthermore, the system allows for the entire test process without power outages, allowing appliances like refrigerators and range hoods to operate normally (avoiding the disruption caused by traditional segmented power-off methods). The final test time was a mere 9 minutes and 38 seconds, approximately 14 times faster than the traditional method (135 minutes). This significantly surpasses the technical indicator of "over 8 times the detection efficiency compared to traditional methods," completely resolving the challenges of traditional methods, including power outages, contact hazards, and time-consuming testing.

[0064] Break through the limitations of positioning accuracy and achieve precise guidance; The device moved along the line to a point 2.7 meters away, triggering an alarm (magnetic field strength suddenly increased from 0.3μT to 15μT). During this process, the IMU (MPU6050 chip) recorded the movement trajectory in real time. Combined with a spatial magnetic field gradient algorithm based on path integration, the device performed a three-dimensional scan (spherical scanning with a radius of 30cm centered on the alarm point) and accurately located the leakage point 8cm behind the wall outlet. Simultaneously, an augmented reality guidance system projected an arrow on the OLED display, dynamically indicating "8cm from the fault point" (with an accuracy of 0.1m). This allowed operators to directly locate the target area without extensive wall disassembly. The final positioning error was only 6cm, not only meeting the accuracy requirement of "≤10cm" but also significantly exceeding the positioning error of "greater than 1m" of traditional non-contact detectors, completely resolving the pain points of traditional equipment: single-point judgment and ambiguous range.

[0065] Reduce false positive rate, realize intelligent diagnosis; During detection, the hair dryer (characteristic frequency 2 kHz) is normally used in the kitchen, but the pre-trained ResNet18 network accurately identifies the interference signal of the hair dryer by analyzing the input parameters such as "magnetic field fundamental wave amplitude, 3rd harmonic distortion rate", without false positive alarm (meeting the "false positive rate <5%" index); Subsequently, the intelligent cloud diagnosis module calls the cloud super 1000 leakage case library, combined with the multi-source information fusion model (electromagnetic anomaly weight 60%: magnetic field mutation 14.7μT; Temperature mutation weight 30%: fault point temperature 32.5℃, difference 8.5℃ from ambient temperature 24℃; Electric field distortion weight 10%: environmental electric field frequency band 45Hz, no distortion), not only distinguish "line leakage (leakage current estimated value 6.2mA, error ±0.5mA)" instead of "equipment leakage", but also accurately determine the fault type as "sink water leakage causes socket oxidation", provide clear direction for subsequent maintenance, avoid the problem of "false alarm frequently, fault type difficult to judge" of traditional detector, greatly improve the diagnosis reliability and intelligent level.

[0066] It should be noted that in this text, 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 that there is any such actual relationship or order between these entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or equipment.

[0067] Although the embodiments of the present application have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A non-contact household line leakage point intelligent diagnostic instrument based on multi-source information fusion, characterized in that: The intelligent diagnostic instrument comprises a multi-modal sensor array, a dual-core signal processing system and an intelligent cloud diagnosis module; The multi-modal sensor array comprises a Hall sensor, an infrared thermal imaging module and an environmental electric field sensor; The dual-core signal processing system adopts an FPGA+DSP architecture, with an FPGA architecture at the front end and a DSP architecture at the back end; The intelligent cloud diagnosis module uploads data to a cloud database through an NBIoT module, matches similar cases in real time, and comprises a spatial magnetic field gradient calculation unit, a multi-source information fusion model and a cloud platform collaborative diagnosis.

2. The multi-source information fusion based non-contact household line leakage point intelligent diagnostic instrument according to claim 1, characterized in that: In the multi-modal sensor array, the Hall sensor is an 8-group three-axis high-sensitivity Hall sensor, with a range of ±50 mT and a resolution of 0.1 μT.

3. The multi-source information fusion based non-contact household line leakage point intelligent diagnostic instrument according to claim 1, characterized in that: In the multi-modal sensor array, the temperature measurement range of the infrared thermal imaging module is 20℃~150℃, and the accuracy is ±1℃.

4. The multi-source information fusion based non-contact household line leakage point intelligent diagnostic instrument according to claim 1, characterized in that: In the multi-modal sensor array, the frequency band of the environmental electric field sensor is 40~60 Hz, and the dynamic range is 60 dB.

5. The multi-source information fusion based non-contact household line leakage point intelligent diagnostic instrument according to claim 1, characterized in that: In the dual-core signal processing system, the front-end FPGA realizes 50 Hz power band pass filtering, and the filtering Q value is ≥100; the back-end DSP performs wavelet denoising and FFT harmonic analysis, wherein the wavelet denoising adopts a db6 wavelet basis, and the FFT harmonic analysis range is 0~500 Hz.

6. The multi-source information fusion based non-contact household line leakage point intelligent diagnostic instrument according to claim 1, characterized in that: The magnetic field gradient calculation of the spatial magnetic field gradient calculation unit satisfies: wherein K=3 (confidence coefficient), σ_B` is the standard deviation of the environmental magnetic field; when the gradient value is >5 μT / cm, an electric leakage early warning is triggered; B: is the gradient operator of the magnetic field vector B, which results in a vector indicating the rate and direction of change of the magnetic field in each direction in space; / x: represents the component of the magnetic field intensity in the X direction in the rectangular coordinate system ( ) in the X-axis direction (i.e., partial derivative); / y: represents the Y-direction component of the magnetic field intensity ( ) rate of change in the Y-axis direction; Bz / z: indicates the rate of change of the component of magnetic field strength Z direction (Bz) in the Z-axis direction; | B| is the magnetic field gradient vector The modulus (size) of B; it is a scalar that represents the strength of the total rate of spatial change of the magnetic field, calculated as the square root of the sum of the squares that follows.

7. The multi-source information fusion based non-contact household line leakage point intelligent diagnostic instrument according to claim 1, characterized in that: The intelligent cloud diagnosis module comprises a pre-trained ResNet18 network, the input layer of the ResNet18 network is [magnetic field fundamental wave amplitude, 3rd harmonic distortion rate, temperature slope, electric field phase difference], and can identify interference signals of electric hair dryers (characteristic frequency 2 kHz) and variable frequency air conditioners (15 kHz), with a false alarm rate <5%.

8. The multi-source information fusion based non-contact household line leakage point intelligent diagnostic instrument according to claim 1, characterized in that: It also comprises an IMU inertial unit for realizing spatial trajectory recording, and the positioning accuracy error satisfies: ε=0.02×L+0.05 (unit: meter); wherein L is the moving distance.

9. The multi-source information fusion based non-contact household line leakage point intelligent diagnostic instrument according to claim 1, characterized in that: It also comprises an augmented reality guidance system, which projects an AR navigation interface through an OLED display screen to indicate the electric leakage direction with an arrow and dynamically display the distance value, and the distance value accuracy is 0.1 m.

10. The multi-source information fusion based non-contact household line leakage point intelligent diagnostic instrument according to claim 1, characterized in that: In the multi-source information fusion model of the intelligent cloud diagnosis module, the electromagnetic anomaly weight is 60%, the temperature mutation weight is 30%, and the electric field distortion weight is 10%.

Images