A charging pile remote monitoring system and a monitoring method

Abstract

The present application relates to the technical field of charging facility monitoring, in particular to a charging pile remote monitoring system and a monitoring method, the system comprising: a high-frequency electrical waveform acquisition module, which acquires high-frequency voltage and current time-domain sequences of a charging loop, and extracts transient harmonic components to construct an electrical fingerprint feature matrix; a thermal and electrical loss mapping analysis module, which calculates theoretical joule heat loss values of each contact node, and generates a dynamic theoretical temperature rise distribution map in combination with environmental thermal resistance parameters; an abnormal deviation traceability determination module, which calculates a thermodynamic divergence index between the dynamic theoretical temperature rise distribution map, and generates a fault blocking control instruction when the divergence index exceeds a dynamic safety zone. In the present application, by establishing a corresponding relationship between high-frequency waveform features and thermal diffusion evolution, in combination with measured temperature deviation and abnormal gradient criteria, heat dissipation disturbance interference can be excluded, abnormal source points of the conductive interface can be locked, and early fault identification, positioning accuracy and grading early warning timeliness can be improved.

Description

Technical Field

[0001] This invention relates to the field of charging facility monitoring technology, and in particular to a remote monitoring system and method for charging piles. Background Technology

[0002] The field of charging facility monitoring technology involves real-time sensing, data acquisition, and safety assessment of the operating status of electric vehicle energy replenishment devices. It is a core component of the interaction between smart grids and new energy vehicles, aiming to ensure electrical safety and energy efficiency management during the charging process through physical quantity detection. Traditional charging pile remote monitoring systems and methods utilize voltage transformers, current transformers, and simple temperature probes installed inside the charging pile to collect power frequency voltage, current amplitude, and single-point temperature data. These data are then compared with preset fixed safety thresholds. When the monitored values ​​exceed the upper limit of the threshold, an alarm or power-off protection is triggered.

[0003] Existing monitoring methods set fixed criteria based on steady-state quantities at power frequency, focusing more on whether the results exceed limits during operation. This makes it difficult to continuously characterize the energy dissipation migration path during load switching. Minor contact degradation is often masked by fluctuations in operating conditions. Temperature observations are biased towards surface results, while electrical observations are biased towards overall load. These two types of information are used in segments. During on-site diagnosis, it is easy to confuse the planar heat diffusion caused by fluctuations in heat exchange conditions with the point-like anomalies caused by the degradation of conductive interfaces, resulting in weak traceability, delayed maintenance triggering, and insufficient reliability of risk interception under fast charging conditions. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a remote monitoring system and method for charging piles.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a remote monitoring system for charging piles includes: The high-frequency electrical waveform acquisition module is used to drive the Rogowski coil and precision voltage divider resistor network embedded in the bus interface of the charging pile to capture the high-frequency voltage and current time domain sequence in real time during the charging process. The module performs fast Fourier transform processing on the high-frequency voltage and current time domain sequence to separate the fundamental wave and higher harmonic components. The module constructs an electrical fingerprint feature matrix based on the amplitude and phase information of each harmonic. The thermoelectric loss mapping analysis module is used to calculate the theoretical Joule heat loss value of each electrical connection node based on the current harmonic components in the electrical fingerprint feature matrix, call the preset material thermal impedance conduction model to simulate the spatial thermal diffusion of the theoretical Joule heat loss value, and generate a dynamic theoretical temperature rise distribution map that reflects the expected thermal field distribution under normal operating conditions. The abnormal deviation tracing and judgment module is used to synchronously receive measured surface temperature field data of key components of the charging pile collected by infrared thermal imaging array sensors. It performs pixel-level differential calculations on the measured surface temperature field data and the dynamic theoretical temperature rise distribution map to extract the coordinates of local hot spots. Based on the temperature difference gradient at the coordinates of the local hot spots, it calculates the thermodynamic divergence index. When the thermodynamic divergence index not only exceeds the statistical safety threshold but also shows a non-linear growth trend, it generates a fault blocking control command.

[0006] As a further aspect of the present invention, the high-frequency electrical waveform acquisition module specifically includes: The signal synchronization conditioning submodule is used to receive the differential current signal from the Rogowski coil and the voltage signal from the precision voltage divider resistor network. It uses a hardware phase-locked loop circuit to lock the zero-crossing moment of the fundamental frequency, and uses the zero-crossing moment as the trigger reference to control the analog-to-digital converter to perform multi-channel synchronous sampling of the voltage and current signals, filter out aliasing noise, and generate a high-frequency voltage and current time domain sequence. The spectrum decomposition and transformation submodule is used to apply the Blackman-Harris window function to the high-frequency voltage and current time domain sequence to suppress spectral leakage, perform radix-2 fast Fourier transform operation, successively extract the real and imaginary parts of the fundamental component and the 2nd to 50th harmonic components from the frequency domain data, and calculate the amplitude and phase angle parameters corresponding to each harmonic component. The fingerprint matrix construction submodule is used to map and fill the calculation results of each harmonic into a preset two-dimensional data structure with the harmonic order as the row index and the voltage amplitude, current amplitude, voltage phase, current phase and active power component as the column vectors. The data is then normalized to generate an electrical fingerprint feature matrix.

[0007] As a further aspect of the present invention, the process by which the fingerprint matrix construction submodule generates the electrical fingerprint feature matrix is ​​as follows: Obtain the amplitude and phase data of each harmonic output by the spectrum decomposition and transformation submodule, construct an augmented feature vector containing the harmonic distortion rate factor and the phase dispersion factor, and use the principal component analysis algorithm to perform dimensionality reduction projection on the augmented feature vector to remove redundant noise features. Based on the preset charging pile operating condition classification labels, the dimensionality-reduced feature vectors are mapped to the corresponding Hilbert space. By calculating the Euclidean distance and cosine similarity between the feature vectors and the standard normal operation reference vectors, an electrical fingerprint feature matrix containing time-domain statistical features and frequency-domain distribution features is established.

[0008] As a further aspect of the present invention, the thermoelectric loss mapping analysis module specifically includes: The skin effect loss calculation submodule is used to analyze the amplitude of each harmonic current in the electrical fingerprint feature matrix, calculate the skin depth by combining the geometric dimension parameters and conductivity parameters of the preset conductor, correct the AC resistance value at each harmonic frequency based on the skin depth, and accumulate the power loss generated by each harmonic to generate the theoretical Joule heat loss value. The thermal resistance mesh loading submodule is used to call the preset material thermal resistance conduction model. This model discretizes the internal space of the charging pile into a three-dimensional finite element thermal resistance mesh, assigns the corresponding material density, specific heat capacity and thermal conductivity to each mesh node, and uses the theoretical Joule heat loss value as a heat source excitation to be loaded onto the corresponding conductor mesh node. The dynamic thermal field simulation submodule is used to iteratively solve the unsteady heat conduction equation based on heat source excitation and environmental boundary conditions using the Crank-Nicholson difference scheme, calculate the temperature rise value of each grid node at the current time step, and map the temperature rise values ​​of all nodes back to the three-dimensional spatial coordinate system to generate a dynamic theoretical temperature rise distribution map.

[0009] As a further aspect of the present invention, the process of calculating the theoretical Joule heat loss value by the skin effect loss calculation submodule is as follows: Obtain the first [character / item] extracted from the electrical fingerprint feature matrix. RMS value of subharmonic current and DC resistance at fundamental frequency The total theoretical Joule heat loss is calculated using the following skin effect correction formula. : ; in, This represents the corrected total theoretical Joule heat loss value. Represents the highest harmonic order. Represents the current harmonic order. Representing the The effective value of the subharmonic current. This represents the resistance of a conductor under direct current conditions. The skin effect coefficient represents the material properties related to the skin effect. The frequency-affected index represents the frequency of influence. Represents the equivalent diameter of the conductor. Represents the skin depth at the fundamental frequency. Represents the equivalent contact resistance. This represents the effective value of the total current.

[0010] As a further aspect of the present invention, the abnormal deviation tracing and determination module specifically includes: The image registration difference module is used to acquire the infrared thermal image output by the infrared thermal imaging array sensor. It uses an affine transformation algorithm to register and align the pixel coordinate system of the infrared thermal image with the spatial coordinate system of the dynamic theoretical temperature rise distribution map. It performs pixel-by-pixel temperature value subtraction to generate a thermal residual image that reflects the deviation between the measured value and the theoretical value. It also extracts the pixel areas in the thermal residual image where the absolute value of the temperature difference exceeds the preset noise threshold to generate local hot spot anomaly coordinates. The divergence index quantification submodule is used to construct a sliding window centered on the coordinates of local hotspot anomalies, calculate the spatial gradient vector magnitude of the temperature difference data within the window, and combine the derivative of the temperature difference with time to calculate the thermodynamic divergence index characterizing the severity of heat accumulation using a weighted fusion algorithm. The safety blocking execution submodule is used to monitor the change trajectory of the thermodynamic divergence index in real time. When the value of the index is continuously higher than the statistical safety threshold and its second time derivative is greater than zero, the system is determined to be in an uncontrolled temperature rise state, and the circuit breaker tripping mechanism is immediately triggered to generate a fault blocking control command.

[0011] As a further aspect of the present invention, the process of calculating the thermodynamic divergence index by the divergence index quantification submodule is as follows: Obtain scalar field data of temperature difference at the coordinates of local hotspot anomalies. The current thermodynamic divergence index is calculated using the following thermodynamic divergence evaluation formula. : ; in, The index representing the calculated thermodynamic divergence. This represents the local integration region centered on the coordinates of the local hotspot anomaly. Represents the spatial gradient weight coefficients. The spatial gradient vector representing the temperature difference field. Represents the differential element of the area. Represents the time trend weighting coefficient. This represents the operation of taking the partial derivative with respect to time. This represents the area of ​​the local integration region. This represents the difference between the measured temperature and the theoretical temperature.

[0012] As a further aspect of the present invention, the specific process by which the security blocking execution submodule determines the nonlinear growth trend is as follows: The thermodynamic divergence index sequence is obtained at multiple consecutive time sampling points, the index change curve is constructed within a sliding time window, and the least squares method is used to perform polynomial fitting on the index change curve. Calculate the second derivative of the fitted polynomial. If the second derivative is always greater than the set nonlinear acceleration threshold for three consecutive sampling periods, and the first derivative continues to increase positively, then it is confirmed that the temperature rise deviation exhibits an exponentially worsening nonlinear growth trend.

[0013] As a further aspect of the present invention, the pixel-level difference operation process in the image registration difference numerator module specifically includes: The radiation temperature values ​​of each pixel in the infrared thermal image are obtained. Based on the background temperature parameters provided by the ambient temperature sensor, the radiation temperature values ​​are corrected for reflectivity and compensated for atmospheric transmittance to obtain the true surface temperature data of the object. The compensated surface temperature data is compared with the theoretical temperature data of the corresponding coordinate points in the dynamic theoretical temperature rise distribution map to generate a thermal residual matrix containing the polarity and magnitude of the positive and negative deviations. The random thermal noise in the thermal residual matrix is ​​then smoothed using a Gaussian filtering algorithm.

[0014] A method for remote monitoring of charging piles, the method being executed based on the aforementioned remote monitoring system for charging piles, includes the following steps: S1. Drive the Rogowski coil and precision voltage divider resistor network embedded in the bus interface of the charging pile to capture the high-frequency voltage and current time domain sequence during the charging process in real time. Perform fast Fourier transform processing on the high-frequency voltage and current time domain sequence to separate the fundamental wave and higher harmonic components. Construct an electrical fingerprint feature matrix based on the amplitude and phase information of each harmonic. S2. Calculate the theoretical Joule heat loss value of each electrical connection node based on the current harmonic components in the electrical fingerprint feature matrix, call the preset material thermal impedance conduction model to simulate the spatial thermal diffusion of the theoretical Joule heat loss value, and generate a dynamic theoretical temperature rise distribution map that reflects the expected thermal field distribution under normal working conditions. S3. Simultaneously receive measured surface temperature field data of key components of the charging pile collected by infrared thermal imaging array sensor, perform pixel-level differential operation on the measured surface temperature field data and dynamic theoretical temperature rise distribution map to extract local hot spot anomaly coordinates, calculate thermodynamic divergence index based on temperature difference gradient at local hot spot anomaly coordinates, and generate fault blocking control command when the thermodynamic divergence index not only exceeds the statistical safety threshold but also shows a nonlinear growth trend.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by constructing a correspondence between high-frequency waveform characteristics and thermal diffusion evolution around the charging process, the change in connection state is transformed into a traceable electrical fingerprint. Based on this fingerprint, the node dissipation path and expected thermal field boundary are deduced, and then fine-grained deviation verification is performed with the measured temperature distribution. The spatial gradient and continuous increase of the abnormal region form a common source criterion, which can isolate the surface thermal expansion characterization caused by external heat dissipation disturbances and lock the abnormal source point corresponding to the instability of the conductive interface, supporting earlier graded treatment and more accurate fault location determination. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the remote monitoring system for charging piles according to the present invention. Figure 2 This is a flowchart illustrating the operation of the high-frequency electrical waveform acquisition module of the present invention. Figure 3 This is a flowchart of the operation of the thermoelectric loss mapping analysis module of the present invention; Figure 4 This is a flowchart of the abnormal deviation tracing and determination module of the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the software-based technical solution is described in detail below with reference to system architecture diagrams and embodiments. It should be understood that the specific embodiments described herein are only for explaining the technical solutions of this invention and do not constitute a limitation on the scope of protection.

[0018] In the description of this invention, the system architecture relationships or data processing flows indicated by terms such as "layer," "module," "interface," "data flow," "client," and "server" are all defined based on the architecture diagram or flowchart corresponding to the embodiments. This way of describing is only used to clearly illustrate the logical relationships between the elements in the technical solution, and not to limit the physical deployment form. The term "multiple" includes two or more technical units, including but not limited to multiple data nodes, processing threads, service instances, or functional components and other scalable elements. The specific number is determined according to the actual business scenario and needs to be specifically specified.

[0019] Please see Figure 1 and Figure 2 The present invention provides a technical solution: a remote monitoring system for charging piles includes: The high-frequency electrical waveform acquisition module is used to drive the Rogowski coil and precision voltage divider resistor network embedded in the bus interface of the charging pile to capture the high-frequency voltage and current time domain sequence in real time during the charging process. The module performs fast Fourier transform processing on the high-frequency voltage and current time domain sequence to separate the fundamental wave and higher harmonic components. Based on the amplitude and phase information of each harmonic, an electrical fingerprint feature matrix is ​​constructed.

[0020] The high-frequency electrical waveform acquisition module specifically includes: The signal synchronization conditioning submodule is used to receive the differential current signal from the Rogowski coil and the voltage signal from the precision voltage divider resistor network. It uses a hardware phase-locked loop circuit to lock the zero-crossing moment of the fundamental frequency, and uses the zero-crossing moment as the trigger reference to control the analog-to-digital converter to perform multi-channel synchronous sampling of the voltage and current signals, filter out aliasing noise, and generate a high-frequency voltage and current time domain sequence. The spectrum decomposition and transformation submodule is used to apply the Blackman-Harris window function to the high-frequency voltage and current time domain sequence to suppress spectral leakage, perform radix-2 fast Fourier transform operation, successively extract the real and imaginary parts of the fundamental component and the 2nd to 50th harmonic components from the frequency domain data, and calculate the amplitude and phase angle parameters corresponding to each harmonic component. The fingerprint matrix construction submodule is used to map and fill the calculation results of each harmonic into a preset two-dimensional data structure with the harmonic order as the row index and the voltage amplitude, current amplitude, voltage phase, current phase and active power component as the column vectors. The data is then normalized to generate an electrical fingerprint feature matrix.

[0021] The process of generating the electrical fingerprint feature matrix by the fingerprint matrix construction submodule is as follows: Obtain the amplitude and phase data of each harmonic output by the spectrum decomposition and transformation submodule, construct an augmented feature vector containing the harmonic distortion rate factor and the phase dispersion factor, and use the principal component analysis algorithm to perform dimensionality reduction projection on the augmented feature vector to remove redundant noise features. Based on the preset charging pile operating condition classification labels, the dimensionality-reduced feature vectors are mapped to the corresponding Hilbert space. By calculating the Euclidean distance and cosine similarity between the feature vectors and the standard normal operation reference vectors, an electrical fingerprint feature matrix containing time-domain statistical features and frequency-domain distribution features is established.

[0022] The high-frequency electrical waveform acquisition module, serving as the system's sensing front-end, employs a heterogeneous collaborative design combining a high-sensitivity analog front-end and a high-performance digital signal processing unit. This module primarily consists of a Rogowski coil sensor array, a precision voltage divider resistor network hardware circuit, a signal synchronization conditioning submodule, a spectrum decomposition and transformation submodule, and a fingerprint matrix construction submodule. The Rogowski coil utilizes a flexible PCB bobbin coil with a sensitivity of 100mV / kA, its internal resistance controlled below 50 ohms, and a bandwidth covering the 10Hz to 10MHz frequency band, used to capture high-frequency current transients on the bus without saturation. The precision voltage divider resistor network employs a low-temperature drift (5ppm per degree Celsius) metal film resistor to achieve a 1000:1 attenuation ratio, coupled with a high-impedance operational amplifier buffer stage, ensuring that the high-voltage side signal does not undergo phase distortion during the step-down process.

[0023] The aforementioned Rogowski coil refers to a hollow toroidal coil made based on the principle of electromagnetic induction. It is mainly used to measure large currents and high-frequency currents, and features a wide measurement range, good frequency response characteristics, no magnetic saturation phenomenon, and small phase shift.

[0024] The signal synchronization conditioning submodule serves as a bridge between the analog physical world and the digital logic world. This submodule integrates a hardware synchronization circuit based on a fully digital phase-locked loop (ADPLL) architecture. Its inputs receive analog differential current signals from the Rogowski coil and analog voltage signals from the voltage divider network. The ADPLL circuit incorporates a phase detector with a center frequency of 50Hz and a capture bandwidth of ±5Hz, enabling real-time tracking of fluctuations in the grid's fundamental frequency. Using the fundamental zero-crossing point as a phase reference, it generates a synchronization trigger pulse with jitter less than 100ns. The analog-to-digital converter (ADC) utilizes a synchronous sampling chip with 16-bit vertical resolution and a single-channel sampling rate of up to 250kSPS. Driven by the trigger pulse, it performs parallel sampling of the voltage and current channels. Before the data enters the digital domain, the submodule's built-in FPGA logic unit implements a 4th-order Butterworth low-pass filter algorithm, setting the cutoff frequency to 50kHz to filter out aliasing noise above the Nyquist frequency, ensuring that the signal-to-noise ratio of the generated original high-frequency voltage and current time-domain sequence is higher than 85dB.

[0025] The aforementioned all-digital phase-locked loop (ADPLL) refers to a phase-locked loop implemented entirely with digital logic circuits. It mainly consists of a digital phase detector, a digital loop filter, and a digitally controlled oscillator, and is used to achieve frequency and phase synchronization of signals.

[0026] The spectrum decomposition and transformation submodule is responsible for frequency domain analysis of massive time-domain data. This submodule first calls the Blackman-Harris window function coefficient table stored in on-chip RAM. This window function has a sidelobe attenuation greater than 92dB, effectively suppressing spectral leakage caused by asynchronous sampling. The submodule's internal DSP core executes a radix-2 decimation-in-time (DIT) Fast Fourier Transform (FFT) logic, setting the FFT points to 4096 to obtain a frequency resolution of approximately 12.2Hz. In the frequency domain data processing stage, the submodule not only extracts the amplitude but also locks the fundamental frequency (50Hz) and the 2nd to 50th harmonics (maximum 2500Hz) separately, extracting the complex spectrum results. Separate the real part from the middle With the imaginary part By using the CORDIC algorithm to transform from Cartesian coordinates to polar coordinates, the precise amplitude of each harmonic and its phase angle relative to the zero-crossing point of the fundamental wave are calculated.

[0027] The aforementioned Blackman-Harris window refers to a generalized cosine window function constructed by summing multiple cosine terms. Its characteristic is that it has extremely low sidelobe levels and is mainly used to minimize energy leakage in spectrum analysis, thereby improving the detection capability of weak signals.

[0028] The fingerprint matrix construction submodule is primarily used for the structured encapsulation and dimensionality reduction of high-dimensional features. This submodule allocates a 50x5 two-dimensional floating-point array in memory as the original fingerprint container. Row indices 1 to 50 correspond to harmonic orders, and column indices correspond to voltage amplitude (normalized to rated voltage), current amplitude (normalized to rated current), voltage phase (radians), current phase (radians), and single-harmonic active power (the product of amplitude and phase difference cosine). In generating the electrical fingerprint feature matrix, the submodule introduces a Principal Component Analysis (PCA) logic unit. This unit first reads the spectral amplitude and phase of the full harmonic data, constructing an augmented feature vector of dimension 1x200, containing four basic parameters of the 50th harmonic. To eliminate differences in data dimensions, Z-Score normalization is used to process the data, making its mean 0 and variance 1. Subsequently, the PCA logic unit calculates the covariance matrix of the augmented feature vector and uses the QR algorithm for eigenvalue decomposition, selecting the top features with a cumulative variance contribution rate exceeding 95%. Principal components, usually The value is between 8 and 15, thus eliminating redundant features such as measurement white noise. The classification and mapping logic projects the dimensionality-reduced feature vectors onto a Hilbert space based on preset operating condition labels, such as constant current charging stage, constant voltage trickle charging stage, and BMS communication handshake stage. Within this space, the submodule calculates the Euclidean distance between the current feature vector and the pre-stored standard normal operating reference vector to measure amplitude deviation, and cosine similarity to measure waveform similarity. When the Euclidean distance is less than 0.1 and the cosine similarity is greater than 0.98, the features are deemed to conform to normal operating conditions, and the final output is an electrical fingerprint feature matrix containing time-domain statistical features such as total distortion (THD) and frequency-domain distribution features.

[0029] Principal Component Analysis (PCA) is a statistical analysis method that uses orthogonal transformation to convert a set of potentially correlated variables into a set of linearly uncorrelated variables. The transformed set of variables is called principal components and is mainly used for dimensionality reduction and feature extraction of data.

[0030] Table 1 lists the performance comparison data of the spectrum decomposition transformation submodule under different window function configurations, verifying the advantages of the Blackman-Harris window in suppressing spectrum leakage.

[0031] Table 1. Window function performance comparison test data; As shown in Table 1, although the main lobe of the Blackman-Harris window is relatively wide, its side lobe attenuation is as high as 92dB, and the amplitude recovery error is only 0.2%, which greatly improves the accuracy of harmonic amplitude extraction and proves the necessity of this configuration in the construction of precision electrical fingerprints.

[0032] Please see Figure 1 and Figure 3 The thermoelectric loss mapping analysis module is used to calculate the theoretical Joule heat loss value of each electrical connection node based on the current harmonic components in the electrical fingerprint feature matrix. It calls the preset material thermal impedance conduction model to simulate the spatial thermal diffusion of the theoretical Joule heat loss value and generates a dynamic theoretical temperature rise distribution map that reflects the expected thermal field distribution under normal operating conditions.

[0033] The thermoelectric loss mapping analysis module specifically includes: The skin effect loss calculation submodule is used to analyze the amplitude of each harmonic current in the electrical fingerprint feature matrix, calculate the skin depth by combining the geometric size parameters and conductivity parameters of the preset conductor, correct the AC resistance value at each harmonic frequency based on the skin depth, and accumulate the power loss generated by each harmonic to generate the theoretical Joule heat loss value.

[0034] The process of calculating the theoretical Joule heat loss value by the skin effect loss calculation submodule is as follows: Obtain the first [character / item] extracted from the electrical fingerprint feature matrix. RMS value of subharmonic current and DC resistance at fundamental frequency The total theoretical Joule heat loss is calculated using the following skin effect correction formula. : ; in, This represents the corrected total theoretical Joule heat loss value. Represents the highest harmonic order. Represents the current harmonic order. Representing the The effective value of the subharmonic current. This represents the resistance of a conductor under direct current conditions. The skin effect coefficient represents the material properties related to the skin effect. The frequency-affected index represents the frequency of influence. Represents the equivalent diameter of the conductor. Represents the skin depth at the fundamental frequency. Represents the equivalent contact resistance. This represents the effective value of the total current.

[0035] The thermal resistance mesh loading submodule is used to call the preset material thermal resistance conduction model. This model discretizes the internal space of the charging pile into a three-dimensional finite element thermal resistance mesh, assigns the corresponding material density, specific heat capacity and thermal conductivity to each mesh node, and uses the theoretical Joule heat loss value as a heat source excitation to be loaded onto the corresponding conductor mesh node. The dynamic thermal field simulation submodule is used to iteratively solve the unsteady heat conduction equation based on heat source excitation and environmental boundary conditions using the Crank-Nicholson difference scheme, calculate the temperature rise value of each grid node at the current time step, and map the temperature rise values ​​of all nodes back to the three-dimensional spatial coordinate system to generate a dynamic theoretical temperature rise distribution map.

[0036] The thermoelectric loss mapping analysis module is the core of the physical field inversion in the system, responsible for transforming abstract electrical data into intuitive physical thermal representations. This module relies on the high parallel computing power of the embedded GPU and consists of a skin effect loss calculation submodule, a thermal resistance mesh loading submodule, and a dynamic thermal field simulation submodule.

[0037] The skin effect loss calculation submodule has a built-in database of material physical properties, storing the conductivity, permeability, and temperature coefficient of conductors. For example, the conductivity of annealed copper at 20 degrees Celsius is... Siemens per meter. This submodule reads the amplitude of each harmonic current in the electrical fingerprint feature matrix and dynamically calculates the skin depth for each harmonic frequency. For example, at a 50Hz fundamental frequency, the skin depth of a copper conductor is approximately 9.3mm, while at 2500Hz (the 50th harmonic), the skin depth decreases to approximately 1.3mm. The submodule adjusts the AC resistance accordingly and calculates the total loss using the superposition principle.

[0038] The skin effect mentioned above refers to the phenomenon that when there is alternating current or alternating electromagnetic field in a conductor, the current distribution inside the conductor is uneven, and the current is concentrated in the "skin" part of the conductor, that is, the surface layer of the conductor, and the current density is smaller the deeper into the conductor.

[0039] In calculating the theoretical Joule heat loss value, this submodule executes formula calculations through a dedicated computing unit. (The remaining text appears to be incomplete and requires further context.) RMS value of subharmonic current DC resistance measured at fundamental frequency The total loss is calculated using the skin effect correction formula: ; in, This represents the corrected total theoretical Joule heat loss value, in watts. Represents the highest harmonic order; Represents the current harmonic order; Representing the The effective value of the current of the second harmonic is expressed in amperes. This represents the resistance of a conductor under direct current conditions, and its unit is ohms. The skin effect coefficient represents the material properties related to the skin effect coefficient. Represents the frequency of influence index; Represents the equivalent diameter of the conductor, in millimeters; The skin depth at the fundamental frequency is expressed in millimeters. This represents the equivalent contact resistance, measured in ohms. This represents the effective value of the total current, in amperes.

[0040] In this formula calculation, the highest harmonic order is set. The value is 50. For a specific operation, the 5th harmonic current is extracted. 12A, DC resistance The skin effect coefficient is 0.005 ohms. The frequency influence index is set to 0.02 based on the properties of copper. Set to 1.5, conductor equivalent diameter 10mm, fundamental wave skin depth The equivalent contact resistance is 9.3 mm. The RMS value of the total current is 0.001 ohms. The value is 100A. Assume the total loss of all harmonics except the 5th harmonic is 50 watts. Substituting the above parameters into the formula, first calculate the correction factor for the harmonic term, i.e. The calculated value is approximately 1.24; the loss component of the 5th harmonic is calculated, i.e. The calculated value is 0.8928 watts; the contact resistance loss term is calculated, i.e. The calculated value is 10 watts. The final corrected total theoretical Joule heat loss value is then obtained. The result is approximately 60.8928 watts. This result indicates that, considering the skin effect and higher harmonics, the theoretical heat generation is higher than that calculated using only DC resistance loss, which is crucial for accurate temperature rise prediction.

[0041] The thermal resistance mesh loading submodule runs in the system's preprocessing layer. This submodule calls the 3D CAD model of the charging pile stored in ROM and uses an octree algorithm to discretize the internal space of the charging pile into a three-dimensional finite element thermal resistance mesh. The mesh element size is set to 2mm x 2mm x 2mm, with a total mesh size reaching 100,000. Based on the material property table, the submodule assigns density, specific heat capacity, and thermal conductivity to each mesh node. For example, copper has a density of 8960 kg / m³, a specific heat capacity of 385 joules / kg Kelvin, and a thermal conductivity of 400 watts / m Kelvin. Subsequently, the submodule calculates the theoretical Joule heat loss value output by the skin effect loss calculation submodule. As the volumetric heat source density, it is weighted and mapped onto the corresponding conductor grid nodes according to the spatial distribution of the current path, forming a heat source excitation matrix.

[0042] The aforementioned three-dimensional finite element thermal resistance mesh refers to a numerical model that discretizes a continuous three-dimensional physical entity into a finite number of tiny units. Each unit is assigned a specific thermal resistance property to approximate the heat conduction process inside complex geometries.

[0043] The dynamic thermal field simulation submodule employs numerical conduction logic to solve the thermodynamic equations. This submodule uses the heat source excitation matrix and real-time ambient temperature as Dirichlet boundary conditions, and the convective heat transfer coefficient under natural convection (5 to 25 watts per square meter Kelvin) as Neumann boundary conditions to construct an unsteady-state heat conduction equation system. The solution process uses the Crank-Nicholson difference scheme, an unconditionally stable implicit method. The submodule sets the time step to 0.5 seconds and solves the large-scale sparse linear equation system using the Gauss-Seidel iterative method. ,in For temperature vectors, and The coefficient matrix, This is the heat source term. After the calculation is completed, the submodule maps the scalar temperature values ​​of each grid node back to the three-dimensional spatial coordinate system through an interpolation algorithm, generating a dynamic theoretical temperature rise distribution map that includes spatial location and temperature values. This distribution map is refreshed at a rate of 2 frames per second, reflecting the theoretical thermal field evolution in real time.

[0044] The aforementioned Crank-Nicholson difference scheme is a finite difference method for numerically solving parabolic partial differential equations. It is second-order accurate in time and numerically stable for all time steps, and is often used to solve the heat conduction equation.

[0045] Please see Figure 1 and Figure 4 The abnormal deviation tracing and judgment module is used to synchronously receive the measured surface temperature field data of the key components of the charging pile collected by the infrared thermal imaging array sensor. It performs pixel-level differential operation on the measured surface temperature field data and the dynamic theoretical temperature rise distribution map to extract the coordinates of local hot spots. Based on the temperature difference gradient at the coordinates of the local hot spots, it calculates the thermodynamic divergence index. When the thermodynamic divergence index not only exceeds the statistical safety threshold but also shows a non-linear growth trend, it generates a fault blocking control command.

[0046] The abnormal deviation tracing and determination module specifically includes: The image registration difference module is used to acquire the infrared thermal image output by the infrared thermal imaging array sensor. It uses an affine transformation algorithm to register and align the pixel coordinate system of the infrared thermal image with the spatial coordinate system of the dynamic theoretical temperature rise distribution map. It performs pixel-by-pixel temperature value subtraction to generate a thermal residual image that reflects the deviation between the measured value and the theoretical value. It also extracts the pixel regions in the thermal residual image where the absolute value of the temperature difference exceeds the preset noise threshold to generate local hot spot anomaly coordinates.

[0047] The pixel-level difference operation process in the image registration difference module specifically includes: The radiation temperature values ​​of each pixel in the infrared thermal image are obtained. Based on the background temperature parameters provided by the ambient temperature sensor, the radiation temperature values ​​are corrected for reflectivity and compensated for atmospheric transmittance to obtain the true surface temperature data of the object. The compensated surface temperature data is compared with the theoretical temperature data of the corresponding coordinate points in the dynamic theoretical temperature rise distribution map to generate a thermal residual matrix containing the polarity and magnitude of the positive and negative deviations. The random thermal noise in the thermal residual matrix is ​​then smoothed using a Gaussian filtering algorithm.

[0048] The divergence index quantification submodule is used to construct a sliding window centered on the coordinates of local hotspot anomalies, calculate the spatial gradient vector magnitude of the temperature difference data within the window, and combine it with the derivative of the temperature difference over time to calculate the thermodynamic divergence index, which characterizes the severity of heat accumulation, using a weighted fusion algorithm.

[0049] The specific process of calculating the thermodynamic divergence index by the divergence index quantification submodule is as follows: Obtain scalar field data of temperature difference at the coordinates of local hotspot anomalies. The current thermodynamic divergence index is calculated using the following thermodynamic divergence evaluation formula. : ; in, The index representing the calculated thermodynamic divergence. This represents the local integration region centered on the coordinates of the local hotspot anomaly. Represents the spatial gradient weight coefficients. The spatial gradient vector representing the temperature difference field. Represents the differential element of the area. Represents the time trend weighting coefficient. This represents the operation of taking the partial derivative with respect to time. This represents the area of ​​the local integration region. This represents the difference between the measured temperature and the theoretical temperature.

[0050] The safety blocking execution submodule is used to monitor the change trajectory of the thermodynamic divergence index in real time. When the value of the index is continuously higher than the statistical safety threshold and its second time derivative is greater than zero, the system is determined to be in an uncontrolled temperature rise state, and the circuit breaker tripping mechanism is immediately triggered to generate a fault blocking control command.

[0051] The specific process by which the security blocking execution submodule determines the non-linear growth trend is as follows: The thermodynamic divergence index sequence is obtained at multiple consecutive time sampling points, the index change curve is constructed within a sliding time window, and the least squares method is used to perform polynomial fitting on the index change curve. Calculate the second derivative of the fitted polynomial. If the second derivative is always greater than the set nonlinear acceleration threshold for three consecutive sampling periods, and the first derivative continues to increase positively, then it is confirmed that the temperature rise deviation exhibits an exponentially worsening nonlinear growth trend.

[0052] The anomaly deviation tracing and judgment module is the decision-making center of the system, responsible for fusing physical simulation data and optical measurement data to identify potential faults. This module integrates a high-resolution infrared thermal imaging processing unit and a safety logic controller, specifically including an image registration difference module, a divergence index quantization submodule, and a safety blocking execution submodule.

[0053] The image registration differential module directly interfaces with an infrared thermal imaging array sensor, which has a resolution of 640 x 480 pixels and a thermal sensitivity (NETD) of less than 40 mK. The module first performs non-uniformity correction (NUC) on the original infrared thermal image, then proceeds to the registration logic. Using an affine transformation algorithm, it identifies the geometric corner points of the charging pile's outer shell as feature anchor points to construct a transformation matrix. This matrix is ​​then used to rotate, scale, and translate the infrared thermal image, ensuring its pixel coordinate system precisely overlaps with the projected view of the dynamic theoretical temperature rise distribution map, with the registration error controlled within one pixel. During the pixel-level differential operation, the submodule first obtains the radiation temperature values ​​of the infrared thermal image pixels, reads ambient temperature sensor data (e.g., 25 degrees Celsius), and performs reflectivity correction based on the target surface emissivity (e.g., PVC insulation layer set to 0.92) to obtain the true surface temperature. Next, this true temperature matrix is ​​subtracted from the corresponding slice matrix of the theoretical temperature rise distribution map to generate a thermal residual image. By setting a noise threshold of 2 degrees Celsius, random measurement errors are filtered out, and only areas with an absolute temperature difference exceeding 2 degrees Celsius are retained. The coordinate centers of local hotspot anomalies are extracted using a connected component labeling algorithm.

[0054] The aforementioned affine transformation algorithm refers to a method in geometry that performs a linear transformation on a vector space followed by a translation to transform it into another vector space. It preserves the "flatness" and "parallelism" of two-dimensional graphics and is often used for geometric correction in image processing.

[0055] The divergence index quantification submodule focuses on the quantitative evaluation of anomalous hotspots. This submodule uses the coordinates of the local hotspot anomalies as the geometric center and extracts a 10x10 pixel sliding window as the local integration region. During the calculation of the thermodynamic divergence index, the submodule acquires the temperature difference scalar field data for that region. And the following thermodynamic divergence evaluation formula is used for calculation: ; in, The thermodynamic divergence index is represented by the calculated value. This represents the local integration region centered on the coordinates of the local hotspot anomaly. This represents the spatial gradient weighting coefficient, used to adjust the contribution of spatial temperature difference variations to the overall index. The spatial gradient vector representing the temperature difference field; Represents the differential element of the area; The time trend weighting coefficient is used to adjust the contribution of the rate of change of temperature difference over time to the overall index. This represents the operation of taking the partial derivative with respect to time; The area represents the local integration region, expressed in square millimeters. This represents the difference between the measured temperature and the theoretical temperature, expressed in degrees Celsius.

[0056] In this computational example, the spatial gradient weight coefficient is set. The time trend weighting coefficient is 0.6. The value is 0.4, representing the area of ​​the local integration region. The area is 100 square millimeters. Calculated by the image processing operator, the integral value of the square of the spatial gradient magnitude of the temperature difference field within the local area, i.e., the integral part in the first term of the formula, is 200, with units of (℃ / mm)²·mm². Simultaneously, the rate of change of the average temperature difference in the region over the two most recent sampling times is calculated, yielding a time partial derivative of 15℃ / s. Substituting the parameters into the formula: Calculated thermodynamic divergence index The value is 126. This value reflects not only the intense spatial distribution of hotspots, i.e., high gradient, but also their rapid deterioration over time, i.e., high temperature rise rate. The advantage of this formula is that it takes into account both the spatial concentration of the fault and the temporal evolution rate, effectively distinguishing between steady-state high temperature differences such as poor heat dissipation and transient fault impacts such as short-circuit arcs, thus improving the accuracy of fault diagnosis.

[0057] The safety blocking execution submodule has a built-in microcontroller with independent power supply. This submodule monitors in real time. The trajectory of indicator changes. The system's preset statistical safety threshold is 80. During the judgment process, the module acquires 20 consecutive sampling points, covering a 10-second time window. The sequence is used to construct an index change curve. The curve is then fitted with a third-order polynomial using the least squares method to obtain the function. The submodule calculates the second derivative of the function. The nonlinear acceleration threshold is set to 5. If the real-time calculation occurs within three consecutive sampling periods (1.5 seconds)... The values ​​are all greater than 5, and the first derivative is... If the temperature rise deviation is consistently greater than 0, it is determined that the temperature rise deviation exhibits an exponentially worsening nonlinear growth trend, indicating that the system is on the verge of thermal runaway. At this point, the submodule outputs a high-level trigger signal to the circuit breaker trip coil within 10 milliseconds, physically cutting off the charging circuit and generating a fault blocking control command which is then uploaded to the cloud server.

[0058] The least squares method mentioned above is a mathematical optimization technique that finds the best function match for data by minimizing the sum of squared errors. It is often used for data fitting and parameter estimation.

[0059] Table 2 shows the test data of thermodynamic divergence index under different fault types, verifying the effectiveness of the index in fault identification.

[0060] Table 2 Comparison of Fault Types and Thermodynamic Dispersion Indices; Table 2 shows the results under normal operation and minor fault conditions. The value is far below the safety threshold of 80, while the index value of serious faults such as short circuit arcing quickly soars to 228. Based on this, the system can achieve precise blocking at the millisecond level to prevent the accident from escalating.

[0061] A method for remote monitoring of charging piles, wherein the method is executed based on the aforementioned remote monitoring system for charging piles, includes the following steps: S1: Drives the Rogowski coil and precision voltage divider resistor network embedded in the bus interface of the charging pile to capture the high-frequency voltage and current time-domain sequence during the charging process in real time. Performs fast Fourier transform processing on the high-frequency voltage and current time-domain sequence to separate the fundamental wave and higher harmonic components. Constructs an electrical fingerprint feature matrix based on the amplitude and phase information of each harmonic. S2: Calculate the theoretical Joule heat loss value of each electrical connection node based on the current harmonic components in the electrical fingerprint feature matrix, call the preset material thermal impedance conduction model to simulate the spatial thermal diffusion of the theoretical Joule heat loss value, and generate a dynamic theoretical temperature rise distribution map that reflects the expected thermal field distribution under normal working conditions. S3: Synchronously receive measured surface temperature field data of key components of the charging pile collected by infrared thermal imaging array sensor, perform pixel-level differential operation on the measured surface temperature field data and dynamic theoretical temperature rise distribution map to extract local hot spot anomaly coordinates, calculate thermodynamic divergence index based on temperature difference gradient at local hot spot anomaly coordinates, and generate fault blocking control command when the thermodynamic divergence index not only exceeds the statistical safety threshold but also shows a non-linear growth trend.

[0062] The above embodiments illustrate preferred embodiments of the present invention. Any equivalent adjustments to the technical solution based on software engineering methods are within the scope of protection, including but not limited to: implementing algorithm logic using different programming languages, refactoring functional modules into services, adjusting data interaction protocols, and optimizing resource scheduling strategies. Any implementation scheme derived from reasonable modifications to the data processing flow, service call chain, or system architecture layer without departing from the core technology of the present invention should be considered within the protection scope defined by the technical solution of the present invention.

Claims

1. A remote monitoring system for charging piles, characterized in that, The system includes: The high-frequency electrical waveform acquisition module is used to drive the Rogowski coil and precision voltage divider resistor network embedded in the bus interface of the charging pile to capture the high-frequency voltage and current time domain sequence in real time during the charging process. The module performs fast Fourier transform processing on the high-frequency voltage and current time domain sequence to separate the fundamental wave and higher harmonic components. The module constructs an electrical fingerprint feature matrix based on the amplitude and phase information of each harmonic. The thermoelectric loss mapping analysis module is used to calculate the theoretical Joule heat loss value of each electrical connection node based on the current harmonic components in the electrical fingerprint feature matrix, call the preset material thermal impedance conduction model to simulate the spatial thermal diffusion of the theoretical Joule heat loss value, and generate a dynamic theoretical temperature rise distribution map that reflects the expected thermal field distribution under normal operating conditions. The abnormal deviation tracing and judgment module is used to synchronously receive measured surface temperature field data of key components of the charging pile collected by infrared thermal imaging array sensors. It performs pixel-level differential calculations on the measured surface temperature field data and the dynamic theoretical temperature rise distribution map to extract the coordinates of local hot spots. It calculates the thermodynamic divergence index based on the temperature difference gradient at the coordinates of the local hot spots. When the thermodynamic divergence index not only exceeds the statistical safety threshold but also shows a non-linear growth trend, it generates a fault blocking control command. The abnormal deviation source tracing and determination module specifically includes: The image registration difference module is used to acquire the infrared thermal image output by the infrared thermal imaging array sensor, use an affine transformation algorithm to register and align the pixel coordinate system of the infrared thermal image with the spatial coordinate system of the dynamic theoretical temperature rise distribution map, perform pixel-by-pixel temperature value subtraction operation, generate a thermal residual image reflecting the deviation between the measured value and the theoretical value, and extract the pixel region in the thermal residual image where the absolute value of the temperature difference exceeds a preset noise threshold to generate the local hot spot anomaly coordinates; The divergence index quantification submodule is used to construct a sliding window centered on the coordinates of the local hotspot anomaly, calculate the spatial gradient vector magnitude of the temperature difference data within the window, and combine the derivative of the temperature difference with time to calculate the thermodynamic divergence index that characterizes the severity of heat accumulation using a weighted fusion algorithm. The safety blocking execution submodule is used to monitor the change trajectory of the thermodynamic divergence index in real time. When the value of the thermodynamic divergence index is continuously higher than the statistical safety threshold and its second time derivative is greater than zero, the system is determined to be in an uncontrolled heating state, and the circuit breaker tripping mechanism is immediately triggered to generate the fault blocking control command. The specific process by which the divergence index quantification submodule calculates the thermodynamic divergence index is as follows: Obtain the temperature difference scalar field data at the coordinates of the local hotspot anomaly. The current thermodynamic divergence index is calculated using the following thermodynamic divergence evaluation formula. : ； in, The thermodynamic divergence index is represented by the calculated value. This represents the local integration region centered on the coordinates of the local hotspot anomaly. Represents the spatial gradient weight coefficients. The spatial gradient vector representing the temperature difference field. Represents the differential element of area. Represents the time trend weighting coefficient. This represents the operation of taking the partial derivative with respect to time. This represents the area of ​​the local integration region. This represents the difference between the measured temperature and the theoretical temperature.

2. The charging pile remote monitoring system according to claim 1, characterized in that, The high-frequency electrical waveform acquisition module specifically includes: The signal synchronization conditioning submodule is used to receive the differential current signal from the Rogowski coil and the voltage signal from the precision voltage divider resistor network. It uses a hardware phase-locked loop circuit to lock the zero-crossing time of the fundamental frequency, and uses the zero-crossing time as a trigger reference to control the analog-to-digital converter to perform multi-channel synchronous sampling of the voltage and current signals, filter out aliasing noise, and generate the high-frequency voltage and current time domain sequence. The spectrum decomposition and transformation submodule is used to apply the Blackman-Harris window function to the high-frequency voltage and current time domain sequence to suppress spectrum leakage, perform radix-2 fast Fourier transform operation, successively extract the real and imaginary parts of the fundamental component and the 2nd to 50th harmonic components from the frequency domain data, and calculate the amplitude parameters and phase angle parameters corresponding to each harmonic component. The fingerprint matrix construction submodule is used to map and fill the calculation results of each harmonic into a preset two-dimensional data structure with the harmonic order as the row index and the voltage amplitude, current amplitude, voltage phase, current phase and active power component as the column vectors, and to normalize the data to generate the electrical fingerprint feature matrix.

3. The charging pile remote monitoring system according to claim 2, characterized in that, The process by which the fingerprint matrix construction submodule generates the electrical fingerprint feature matrix is ​​as follows: The amplitude and phase data of each harmonic output by the spectrum decomposition and transformation submodule are obtained, an augmented feature vector containing the harmonic distortion rate factor and the phase dispersion factor is constructed, and the augmented feature vector is subjected to dimensionality reduction projection using the principal component analysis algorithm to remove redundant noise features. Based on the preset charging pile operating condition classification labels, the dimensionality-reduced feature vectors are mapped to the corresponding Hilbert space. By calculating the Euclidean distance and cosine similarity between the feature vectors and the standard normal operation reference vectors, the electrical fingerprint feature matrix containing time-domain statistical features and frequency-domain distribution features is established.

4. The charging pile remote monitoring system according to claim 1, characterized in that, The thermoelectric loss mapping analysis module specifically includes: The skin effect loss calculation submodule is used to analyze the amplitude of each harmonic current in the electrical fingerprint feature matrix, calculate the skin depth by combining the geometric dimension parameters and conductivity parameters of the preset conductor, correct the AC resistance value at each harmonic frequency based on the skin depth, and accumulate the power loss generated by each harmonic to generate the theoretical Joule heat loss value. The thermal resistance mesh loading submodule is used to call the preset material thermal resistance conduction model. This model discretizes the internal space of the charging pile into a three-dimensional finite element thermal resistance mesh, assigns the corresponding material density, specific heat capacity and thermal conductivity to each mesh node, and uses the theoretical Joule heat loss value as a heat source excitation to be loaded onto the corresponding conductor mesh node. The dynamic thermal field simulation submodule is used to iteratively solve the unsteady heat conduction equation based on heat source excitation and environmental boundary conditions using the Crank-Nicholson difference scheme, calculate the temperature rise value of each grid node at the current time step, and map the temperature rise values ​​of all nodes back to the three-dimensional spatial coordinate system to generate the dynamic theoretical temperature rise distribution map.

5. The remote monitoring system for charging piles according to claim 4, characterized in that, The process by which the skin effect loss calculation submodule calculates the theoretical Joule heat loss value is as follows: Obtain the first extracted from the electrical fingerprint feature matrix RMS value of subharmonic current and DC resistance at fundamental frequency The total theoretical Joule heat loss is calculated using the following skin effect correction formula. : ； in, This represents the corrected total theoretical Joule heat loss value. Represents the highest harmonic order. Represents the current harmonic order. Representing the The effective value of the subharmonic current. This represents the resistance of a conductor under direct current conditions. The skin effect coefficient represents the material properties related to the skin effect. The frequency-affected index represents the frequency of influence. Represents the equivalent diameter of the conductor. Represents the skin depth at the fundamental frequency. Represents the equivalent contact resistance. This represents the effective value of the total current.

6. The remote monitoring system for charging piles according to claim 1, characterized in that, The specific process by which the security blocking execution submodule determines the nonlinear growth trend is as follows: The sequence of the thermodynamic divergence index at multiple consecutive time sampling points is obtained, the index change curve within a sliding time window is constructed, and the index change curve is fitted by polynomial using the least squares method. Calculate the second derivative of the fitted polynomial. If the second derivative is always greater than the set nonlinear acceleration threshold for three consecutive sampling periods, while the first derivative continues to increase positively, then it is confirmed that the temperature rise deviation exhibits the nonlinear growth trend of exponential deterioration.

7. The charging pile remote monitoring system according to claim 1, characterized in that, The pixel-level difference operation process in the image registration difference numerator module specifically includes: The radiation temperature values ​​of each pixel in the infrared thermal image are obtained. Based on the background temperature parameters provided by the ambient temperature sensor, the radiation temperature values ​​are corrected for reflectivity and compensated for atmospheric transmittance to obtain the true surface temperature data of the object. The compensated surface temperature data is compared with the theoretical temperature data of the corresponding coordinate point in the dynamic theoretical temperature rise distribution map to generate a thermal residual matrix containing the polarity and magnitude of the positive and negative deviations. The random thermal noise in the thermal residual matrix is ​​then smoothed using a Gaussian filtering algorithm.

8. A method for remote monitoring of charging piles, characterized in that, The method is used to implement the charging pile remote monitoring system according to any one of claims 1-7, and includes the following steps: S1. Drive the Rogowski coil and precision voltage divider resistor network embedded in the bus interface of the charging pile to capture the high-frequency voltage and current time domain sequence during the charging process in real time. Perform fast Fourier transform processing on the high-frequency voltage and current time domain sequence to separate the fundamental wave and higher harmonic components. Construct an electrical fingerprint feature matrix based on the amplitude and phase information of each harmonic. S2. Calculate the theoretical Joule heat loss value of each electrical connection node based on the current harmonic components in the electrical fingerprint feature matrix, call the preset material thermal impedance conduction model to perform spatial thermal diffusion simulation on the theoretical Joule heat loss value, and generate a dynamic theoretical temperature rise distribution map that reflects the expected thermal field distribution under normal operating conditions. S3. Synchronously receive measured surface temperature field data of key components of the charging pile collected by infrared thermal imaging array sensor, perform pixel-level differential operation on the measured surface temperature field data and the dynamic theoretical temperature rise distribution map to extract local hot spot anomaly coordinates, calculate thermodynamic divergence index based on the temperature difference gradient at the local hot spot anomaly coordinates, and generate fault blocking control command when the thermodynamic divergence index not only exceeds the statistical safety threshold but also shows a nonlinear growth trend.

Images