Power cable temperature monitoring method and system based on surface acoustic wave sensor

Abstract

This application discloses a method and system for monitoring the temperature of power cables based on surface acoustic wave (SAW) sensors, belonging to the field of power detection. This application is applicable to power cables with dynamic loads, by simultaneously acquiring frequency drift data and real-time current waveform data from a SAW sensor array along the power cable. The frequency drift data is converted into multiple temperature points and subjected to spatial interpolation and differentiation to obtain measured temperature gradient information. Harmonic component information is extracted from the current waveform and input into a deep neural network to predict theoretical temperature gradient information. Kalman filtering is used to iteratively fuse the predicted temperature gradient information with the measured temperature gradient information to generate an optimized target temperature gradient. Finally, based on this target temperature gradient and starting from a known temperature point, the complete temperature distribution data of the power cable is reconstructed through integral calculation, enabling comprehensive and accurate temperature monitoring of the power cable.

Description

Technical Field

[0001] This application belongs to the field of power detection, and in particular relates to a method and system for monitoring the temperature of power cables based on a surface acoustic wave sensor. Background Technology

[0002] As a critical carrier of electrical energy transmission, the safe and stable operation of power cables directly affects the reliability of the entire power system. Cable temperature is one of the core indicators for measuring its operating status and health. Real-time and accurate temperature monitoring of power cables can effectively prevent serious accidents such as accelerated insulation aging, reduced current carrying capacity, and even fires caused by overheating. This has crucial application prospects for ensuring power grid safety and improving power supply reliability.

[0003] In existing technologies, surface acoustic wave (SAW) sensor technology has the advantages of being wireless, passive, and resistant to electromagnetic interference. The method of measuring temperature using SAW sensors has been applied to the measurement of the surface temperature of power cables. This method involves deploying SAW sensors at key locations on the cable surface and utilizing the characteristic that their resonant frequency changes with temperature to obtain temperature values.

[0004] However, existing methods for temperature monitoring of power cables connected to numerous nonlinear loads do not adequately consider the impact of dynamic loads on the rapid and non-uniform changes in the overall temperature field of the cable. This leads to decreased measurement accuracy and an inability to accurately determine the temperature distribution of the power cable in real time, thus limiting the ability to monitor and warn of potential overheating risks. Therefore, existing methods suffer from poor temperature measurement accuracy and an inability to comprehensively and effectively monitor the temperature of power cables. Summary of the Invention

[0005] This application provides a method, system, device, and computer storage medium for monitoring the temperature of power cables based on a surface acoustic wave sensor, which can perform comprehensive and accurate temperature monitoring of power cables.

[0006] In a first aspect, this application provides a method for monitoring the temperature of power cables based on a surface acoustic wave sensor, applicable to power cables with dynamic loads, the method comprising: The frequency drift data and current waveform timing data of the target power cable at multiple preset locations are acquired. The frequency drift data is collected by a surface acoustic wave sensor array deployed along the target power cable. Based on the preset frequency-temperature mapping relationship, the frequency drift value of the frequency drift data is converted into multiple temperature points, and spatial interpolation and differentiation operations are performed based on the multiple temperature points to obtain temperature gradient information. The amplitude and phase angle of the preset order harmonics are extracted from the spectrum of the current waveform time sequence data to obtain the harmonic component information; Harmonic component information is input into a trained deep neural network to obtain predicted temperature gradient information. The trained deep neural network is trained using training samples that include historical harmonic component information and corresponding historical temperature gradient information. Based on the predicted temperature gradient information and the temperature gradient information, Kalman filtering is used for iterative calculation to obtain the target temperature gradient information. Based on the target temperature gradient information, starting from any one of multiple temperature points, the temperature distribution data of the target power cable is determined through integral calculation, so as to realize the temperature monitoring of the target power cable.

[0007] In one feasible implementation, the method further includes: Calculate the residual vector between the predicted temperature gradient information and the temperature gradient feature vector of the temperature gradient information; The norm of the residual vector is compared with a preset threshold. If the norm of the residual vector is greater than the preset threshold, it is determined that the target power cable has non-electrical damage.

[0008] In one feasible implementation, the frequency drift data includes high-frequency frequency drift data and low-frequency frequency drift data; the method further includes: Based on high-frequency drift data and low-frequency drift data, multiple differential frequency response values ​​are calculated using a preset conversion function, and spatial interpolation is performed based on these multiple differential frequency response values ​​to generate differential frequency response distribution data for the target power cable. The target temperature information that is greater than the preset temperature threshold is determined from the temperature distribution data, and the heat source attribute information of the target power cable at the location corresponding to the target temperature information is determined based on the target differential frequency response value corresponding to the target temperature information in the differential frequency response distribution data.

[0009] In one feasible implementation, the heat source attribute information includes internal electrical heat sources and external environmental heat sources; Based on the target differential frequency response value corresponding to the target temperature information in the differential frequency response distribution data, determine the heat source attribute information of the target power cable at the location corresponding to the target temperature information, including: Determine the target difference frequency response value corresponding to the location in the difference frequency response distribution data based on the target temperature information; When the target differential frequency response value is positive, the heat source attribute of the target power cable at the location corresponding to the target temperature information is an internal electrical heat source; When the target differential frequency response value is non-positive, the heat source attribute of the target power cable at the location corresponding to the target temperature information is the external environment heat source.

[0010] In one feasible implementation, the amplitude and phase angle of a preset order harmonic are extracted from the spectrum of the current waveform time-series data to obtain harmonic component information, including: The current waveform time series data is frequency domain transformed to generate a spectrum, and the fundamental peak value is determined in the spectrum; Based on the fundamental peak value, the harmonic peak values ​​of the preset order harmonics are extracted from the spectrum, as well as the phase difference of each harmonic peak value relative to the fundamental peak value. Harmonic feature vectors are constructed using the amplitude of the peak values ​​of the preset order harmonics and their corresponding phase differences to obtain harmonic component information.

[0011] In one feasible implementation, based on the predicted temperature gradient information and the temperature gradient information, Kalman filtering is used for iterative calculation to obtain the target temperature gradient information, including: The predicted temperature gradient information is used as the state prediction value, and the temperature gradient information is used as the observation value. The correction factor is calculated based on the deviation between the predicted and observed values, as well as the preset process noise covariance and observation noise covariance. The predicted state values ​​are corrected using a correction factor, and the target temperature gradient information is generated through iterative calculation.

[0012] In one feasible implementation, based on target temperature gradient information, starting from any one of multiple temperature points, the temperature distribution data of the target power cable is determined through integration calculation, including: Select one temperature point from multiple temperature points as the starting temperature point; Based on the temperature gradient value of the target temperature gradient information, the temperature change of different locations of the target power cable relative to the starting temperature point is calculated by integral operation. The temperature value at the starting temperature point is superimposed with the temperature change to determine the temperature value at various locations on the target power cable, generating temperature distribution data.

[0013] Secondly, this application provides a power cable temperature monitoring system based on a surface acoustic wave sensor, suitable for power cables with dynamic loads. The system includes: The acquisition module is used to acquire frequency drift data and current waveform timing data of the target power cable at multiple preset locations. The frequency drift data is collected by a surface acoustic wave sensor array deployed along the target power cable. The conversion module is used to convert the frequency drift value of the frequency drift data into multiple temperature points according to the preset frequency-temperature mapping relationship, and to perform spatial interpolation and differentiation operations based on the multiple temperature points to obtain temperature gradient information. The extraction module is used to extract the amplitude and phase angle of a preset order harmonic from the spectrum of current waveform time-series data to obtain harmonic component information. The calculation module is used to input harmonic component information into the trained deep neural network to obtain predicted temperature gradient information. The trained deep neural network is trained using training samples that include historical harmonic component information and corresponding historical temperature gradient information. The calculation module is also used to perform iterative calculations using Kalman filtering based on the predicted temperature gradient information and the temperature gradient information to obtain the target temperature gradient information; The calculation module is also used to determine the temperature distribution data of the target power cable by integral calculation based on the target temperature gradient information, starting from any one of multiple temperature points, so as to realize the temperature monitoring of the target power cable.

[0014] Thirdly, this application provides an electronic device, the device comprising: a processor, and a memory storing computer program instructions; the processor reads and executes the computer program instructions to implement the power cable temperature monitoring method based on a surface acoustic wave sensor as described in any embodiment of the first aspect.

[0015] Fourthly, this application provides a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the power cable temperature monitoring method based on a surface acoustic wave sensor as described in any embodiment of the first aspect.

[0016] This application discloses a method, system, device, and computer storage medium for monitoring the temperature of power cables based on surface acoustic wave (SAW) sensors. By utilizing a SAW sensor array to acquire discrete temperature points along the cable and calculate preliminary temperature gradient information, it innovatively introduces real-time analysis of the power cable current waveform. A deep neural network is used to predict temperature gradient changes caused by dynamic loads from harmonic components. Subsequently, Kalman filtering is used to optimally fuse the physically measured temperature gradient with the harmonic-predicted temperature gradient, effectively correcting the deviation of a single measurement source under dynamic conditions and obtaining a more accurate temperature gradient. Finally, based on this temperature gradient information and known temperature points, integration and reconstruction are performed to achieve continuous real-time temperature distribution monitoring of the entire power cable. This improves the accuracy of power cable temperature measurement and the comprehensiveness of temperature monitoring under dynamic loads.

[0017] Furthermore, by employing a dual-frequency surface acoustic wave sensor array, both high-frequency and low-frequency drift data reflecting the thermal effects at different depths of the cable are simultaneously acquired. The difference between these two data sets is then used to calculate the differential frequency response distribution data, characterizing the temperature difference between the cable's interior and surface. Finally, when a potential hot spot is detected, the differential frequency response value corresponding to the hot spot location is used to identify whether the heat source originates from internal electrical factors within the cable or from external environmental interference. Therefore, this enhances the diagnostic and assessment capabilities for the actual thermal risks of power cables, achieving effective and comprehensive temperature monitoring of power cables. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic flowchart of a power cable temperature monitoring method based on a surface acoustic wave sensor provided in one embodiment of this application; Figure 2 This is a flowchart illustrating a method for calculating target temperature gradient information using Kalman filtering, provided in one embodiment of this application. Figure 3 This is a flowchart illustrating a method for determining temperature distribution data of a target power cable according to an embodiment of this application; Figure 4 This is a schematic diagram of the structure of a power cable temperature monitoring system based on a surface acoustic wave sensor provided in one embodiment of this application; Figure 5 This is a schematic diagram of the hardware structure of an electronic device provided in one embodiment of this application. Detailed Implementation

[0020] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0021] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0022] In existing technologies, surface acoustic wave (SAW) sensor technology has the advantages of being wireless, passive, and resistant to electromagnetic interference. The method of measuring temperature using SAW sensors has been applied to the measurement of the surface temperature of power cables. This method involves deploying SAW sensors at key locations on the cable surface and utilizing the characteristic that their resonant frequency changes with temperature to obtain temperature values.

[0023] However, existing methods for temperature monitoring of power cables connected to numerous nonlinear loads do not adequately consider the impact of dynamic loads on the rapid and non-uniform changes in the overall temperature field of the cable. This leads to decreased measurement accuracy and an inability to accurately determine the temperature distribution of the power cable in real time, thus limiting the ability to monitor and warn of potential overheating risks. Therefore, existing methods suffer from poor temperature measurement accuracy and an inability to comprehensively and effectively monitor the temperature of power cables.

[0024] To address the problems of the prior art, this application provides a method, system, device, and computer storage medium for monitoring the temperature of power cables based on a surface acoustic wave (SAW) sensor. The method for monitoring the temperature of power cables based on a SAW sensor provided in this application will be described first.

[0025] Figure 1 A schematic flowchart of a power cable temperature monitoring method based on a surface acoustic wave sensor according to an embodiment of this application is shown. This method is applicable to power cables with dynamic loads, such as… Figure 1 As shown, the method includes steps S110 to S160.

[0026] Dynamic load power cables refer to power transmission lines whose current and power do not change steadily or slowly in a periodic manner, but rather exhibit rapid, irregular, and non-sinusoidal waveform characteristics. This situation is usually caused by a large number of nonlinear or impulsive electrical devices connected to the end of the line. For example, power cables supplying power to large data centers, communication base stations, rail transit traction systems, or centralized electric vehicle charging stations are typical examples of dynamic load power cables. In these scenarios, switching power supplies, frequency converters, rectifiers, and other equipment inject a large amount of high-order harmonic current into the power grid during operation, causing severe distortion of the waveform of the current actually flowing through the cable, which is no longer a smooth sine wave.

[0027] S110: Acquire frequency drift data and current waveform timing data of the target power cable at multiple preset locations. The frequency drift data is collected by a surface acoustic wave sensor array deployed along the target power cable.

[0028] A surface acoustic wave (SAW) sensor array refers to a collection of multiple independent SAW sensors distributed along the length of a power cable at predetermined spatial locations. This array is used to acquire temperature information at multiple key points along the cable, and can include sensors deployed at cable joints, bends, or long straight sections. For example, a SAW sensor array could consist of three sensors deployed at 0 meters, 5 meters, and 10 meters along the cable.

[0029] Frequency drift data refers to the deviation of the center resonant frequency of a surface acoustic wave (SAW) sensor relative to a reference frequency due to temperature changes. This data directly reflects the temperature change at the sensor's location and can be positive, negative, or zero. For example, a frequency drift of -15.8 kHz indicates that the current frequency has dropped by 15.8 kHz compared to the reference frequency. Current waveform time-series data refers to a sequence of instantaneous values ​​acquired at a fixed sampling frequency over a continuous time period, which comprehensively depicts the change of current over time. This can include the fundamental current and harmonic currents generated by nonlinear loads. For example, a current waveform time-series data can be a digital sequence containing multiple sampling points.

[0030] For example, to monitor a power cable supplying power to a large data center, five surface acoustic wave (SAW) sensors are non-uniformly deployed along the cable surface at its outgoing end, intermediate location, and junction box to form a SAW sensor array. A high-frequency response through-hole current transformer is installed at the cable outgoing end. Data acquisition is performed synchronously after data acquisition begins. Firstly, excitation signals are sent to the five sensors in the SAW sensor array at a frequency of once per second, and echoes are received. By comparing the resonant frequency of the echo signal with a preset 25-degree Celsius reference frequency, a set of five frequency drift data points is obtained, for example, -10.5kHz, -11.2kHz, -15.8kHz, -12.1kHz, and -11.5kHz. At the same time, the output signal of the current transformer is continuously acquired at a high sampling rate of 20,000 times per second to generate a digital sequence containing 2,000 sampling points, such as [0.5A, 1.2A, 2.8A, ..., -1.3A, -0.7A], which serves as the current waveform timing data for that second.

[0031] S120: Based on the preset frequency-temperature mapping relationship, the frequency drift value of the frequency drift data is converted into multiple temperature points, and spatial interpolation and differentiation operations are performed based on the multiple temperature points to obtain temperature gradient information.

[0032] The preset frequency-temperature mapping relationship refers to the predetermined deterministic correspondence between the frequency drift of the surface acoustic wave sensor and its ambient temperature. This can be obtained through calibration experiments on the sensor in a controlled temperature chamber and can be a polynomial function. Temperature gradient information refers to data characterizing the rate of temperature change along the length of the power cable. A larger gradient value indicates a more drastic temperature change at that location, potentially indicating heat concentration or poor heat dissipation. This can be a vector containing multiple gradient values, such as [0.5℃ / m, 1.2℃ / m, -0.3℃ / m].

[0033] First, the preset frequency-temperature mapping relationship is invoked to transform each frequency drift value in the acquired frequency drift data set one by one. For example, a pre-calibrated second-order polynomial function, as shown in formula (1), is used to calculate the corresponding temperature value for each frequency drift value, thereby obtaining a set of discrete temperature points that correspond one-to-one with the positions of the surface acoustic wave sensor array.

[0034] (1) Where T represents temperature, Δf is the frequency drift value, and a, b, and c are calibration coefficients.

[0035] To obtain a continuous temperature gradient, spatial interpolation and differentiation operations are then employed. For example, a cubic spline interpolation algorithm can be used to fit this set of discrete temperature points into a smooth, continuous temperature distribution curve along the cable axis. Finally, performing a first-order derivative operation on this fitted continuous temperature distribution curve yields a new curve, namely the temperature gradient distribution curve. By sampling along this gradient curve at preset spatial nodes, temperature gradient information characterizing the non-uniform heat distribution throughout the monitoring section can be obtained.

[0036] For example, after obtaining a set of 5 frequency drift data, namely -10.5kHz, -11.2kHz, -15.8kHz, -12.1kHz, and -11.5kHz, which correspond to cable positions of 0 meters, 5 meters, 10 meters, 15 meters, and 20 meters respectively, a preset frequency-temperature mapping relationship is first invoked. Specifically, it can be a second-order polynomial function as shown in formula (1), assuming that the calibration coefficients a, b, and c are 0.01, -2.0, and 30 respectively. Each frequency drift value is substituted into the function for calculation. For example, for the frequency drift value of -15.8kHz, the calculated temperature is approximately 64.1℃. By performing the same transformation on all 5 frequency drift values, a set of 5 discrete temperature points is finally obtained, namely 52.0℃, 53.4℃, 64.1℃, 55.5℃, and 54.2℃. Subsequently, in order to obtain temperature gradient information that can characterize the uneven heat distribution throughout the monitoring section, spatial interpolation and differentiation operations were performed on these five discrete temperature points. Specifically, a cubic spline interpolation algorithm was used to fit these five points into a smooth, continuous temperature distribution function along the cable axis. Then, the first derivative of this function was performed to obtain the gradient curve. Finally, by sampling along the cable every 1 meter on this gradient curve, a vector containing 21 gradient values ​​can be obtained, such as [0.32℃ / m, 0.35℃ / m, ..., 2.37℃ / m, ..., -0.28℃ / m]. This vector is the final generated temperature gradient information.

[0037] S130: Extract the amplitude and phase angle of the preset order harmonic from the spectrum of the current waveform time sequence data to obtain the harmonic component information.

[0038] Preset harmonic orders refer to a set of harmonic orders predefined based on the typical electrical characteristics of dynamic loads. Since harmonics generated by different types of nonlinear loads are mainly concentrated on specific odd harmonics, the harmonics with the greatest impact on cable thermal effects are usually selected as the analysis objects. For example, a set of preset harmonic orders can be the 3rd, 5th, 7th, and 11th harmonics. Harmonic component information refers to a structured data set containing the amplitude and phase angle information of all preset harmonic orders. For example, a harmonic component information can be a multidimensional vector, such as [A3, P3, A5, P5, A7, P7], where A represents the amplitude and P represents the phase angle.

[0039] First, a Fast Fourier Transform (FFT) is performed on the current waveform time-series data to transform it from the time domain to the frequency domain, generating a spectrum that reflects the energy distribution of the signal at different frequencies. Next, to accurately locate harmonics, the peak value corresponding to the fundamental frequency of the power grid is searched and determined in the spectrum; this fundamental peak value is typically around 50Hz or 60Hz. Then, based on a preset harmonic order list and using the determined fundamental peak value as a reference, the peak value of each target harmonic is extracted one by one at the corresponding integer multiples of the frequency in the spectrum. For each extracted harmonic peak value, its signal strength is recorded as the harmonic amplitude, and its phase difference relative to the fundamental peak value is calculated as the harmonic phase angle. Finally, all extracted harmonic amplitudes and phase angles are concatenated and combined according to the preset order to form a fixed-dimensional vector; this vector is the final generated harmonic component information.

[0040] S140: Input the harmonic component information into the trained deep neural network to obtain the predicted temperature gradient information. The trained deep neural network is trained using training samples that include historical harmonic component information and corresponding historical temperature gradient information.

[0041] Predicting temperature gradient information refers to the distribution of temperature gradients that should theoretically be caused by harmonics, calculated using a deep neural network model based on the input harmonic component information. A trained deep neural network is a multi-layered artificial neural network that has completed model parameter learning and optimization through a large amount of historical data.

[0042] A trained deep neural network can include an input layer matching the dimension of the harmonic component information, several hidden layers for nonlinear transformation, and an output layer matching the dimension of the temperature gradient information. When harmonic component information is received, such as a six-dimensional vector containing the amplitudes and phases of the 3rd, 5th, and 7th harmonics, this vector is first fed into the network's input layer. The data then propagates forward through the network, passing through each hidden layer sequentially. In each hidden layer, the data from the previous layer is multiplied by a fixed weight matrix and a bias vector is added. It is then processed by a nonlinear activation function, such as the Rectified Linear Unit (ReLU), to extract deeper features. After being transformed layer by layer through all hidden layers, the data is finally passed to the output layer. The output layer performs a linear transformation on the output of the last hidden layer to generate a vector with the same dimension as the temperature gradient information. This vector, ultimately output from the output layer, represents the predicted temperature gradient information, indicating the model's calculation of the theoretical temperature gradient distribution of the cable based on the input harmonics.

[0043] For example, the architecture of a pre-trained deep neural network can be a 6-dimensional input layer for matching harmonic component information containing the amplitude and phase of the 3rd, 5th, and 7th harmonics; two hidden layers, each containing 128 neurons, for nonlinear feature extraction; and a 21-dimensional output layer for outputting temperature gradient information corresponding to the cable length from 0 meters to 20 meters, with a gradient value per meter. When the generated harmonic component information vector, i.e., [10.2, -30, 5.8, 150, 2.1, -120], is received, this vector is fed into the network's input layer. Subsequently, the data propagates forward through the network, passing through the two hidden layers sequentially. In each layer, the data is multiplied by the layer's fixed weight matrix and a bias vector is added, then nonlinearly transformed using the ReLU activation function. After being processed layer by layer through the two hidden layers, the data is finally passed to the output layer. The output layer performs a linear transformation on the output of the last hidden layer, ultimately generating a 21-dimensional vector, for example, [0.30℃ / m, 0.33℃ / m, ..., 2.45℃ / m, ..., -0.31℃ / m].

[0044] As a specific implementation of this application, before step S140: inputting the harmonic component information into the trained deep neural network to obtain the predicted temperature gradient information, the method further includes: a process of training the deep neural network model. First, a training sample set containing multiple training samples is obtained, where each training sample includes historical harmonic component information collected at a historical time, and a corresponding historical temperature gradient information label obtained through physical measurement at the same time. At the start of training, the model parameters of the deep neural network, such as the connection weights and biases between layers, are initialized to a set of random values.

[0045] Subsequently, the model training iterative process begins. In each iteration, one or more training samples are selected from the training sample set. The historical harmonic component information from each training sample is input into the current deep neural network, and a predicted temperature gradient information label is calculated through the network's forward propagation. Next, based on this predicted temperature gradient information label and the corresponding real historical temperature gradient information label in the training sample, a preset loss function, such as the mean squared error function, is used to calculate the prediction error value between the two. If the prediction error value does not meet the training stopping condition, such as the error being less than a preset threshold or the number of iterations reaching the upper limit, an optimization algorithm, such as the adaptive moment estimation (Adam optimizer), is used to calculate the gradient based on the prediction error value through backpropagation. This gradient is then used to update all model parameters of the deep neural network to obtain an updated deep neural network model. Afterward, the process returns to selecting the next batch of training samples and repeating the above forward propagation, error calculation, and parameter update process until the training stopping condition is met. The model obtained at this point is the final trained deep neural network that accurately reflects the electrothermal conversion relationship.

[0046] S150: Based on the predicted temperature gradient information and the temperature gradient information, the target temperature gradient information is obtained by iterative calculation using Kalman filtering.

[0047] Kalman filtering is a highly efficient recursive state estimation algorithm that can provide an optimal estimate of the system state by fusing multiple information sources with uncertainties. Target temperature gradient information refers to the optimized temperature gradient data obtained after Kalman filtering. Compared to the original predicted or observed values, this information combines the advantages of both sources and has higher accuracy.

[0048] Based on predicted and actual temperature gradient information, the target temperature gradient information is obtained through iterative calculation using Kalman filtering, mainly comprising two stages: prediction and update. First, in the prediction stage, the predicted temperature gradient information output by the deep neural network is used as the current state prediction value. Second, in the update stage, the temperature gradient information measured and processed by the surface acoustic wave sensor array is used as the observed value. Subsequently, a correction factor, namely the Kalman gain, is calculated based on the deviation between the state prediction value and the observed value, combined with two preset key parameters: process noise covariance and observation noise covariance. The process noise covariance characterizes the uncertainty of the neural network model's prediction result, while the observation noise covariance characterizes the uncertainty of the physical sensor's measurement result. Finally, the calculated correction factor is used to weight and correct the state prediction value; the corrected result is the target temperature gradient information output in the current iteration cycle. This output target temperature gradient information serves as the final result of this calculation and is also used as historical state information in the next time step iteration, thus achieving continuous, real-time state tracking and optimal estimation.

[0049] S160: Based on the target temperature gradient information, starting from any one of multiple temperature points, the temperature distribution data of the target power cable is determined through integral calculation to achieve temperature monitoring of the target power cable.

[0050] Temperature distribution data refers to a structured dataset that shows how temperature is continuously distributed along the length of a power cable. It can include the temperature value at each specific location, as well as the location and value of the hottest point, i.e., the hot spot. For example, a temperature distribution dataset can be a vector containing multiple temperature values, such as [52.0℃, 52.4℃, 52.9℃, ..., 64.1℃, ..., 54.2℃], where each value corresponds to a specific spatial location on the cable.

[0051] First, from multiple discrete temperature points derived from frequency drift data, a starting temperature point is selected as the basis for integration. The temperature value at this point serves as the benchmark for reconstructing the entire temperature distribution. Next, based on the fused target temperature gradient information, numerical integration is used to calculate the temperature changes at all other locations on the cable relative to this starting temperature point. For example, the Euler method or Runge-Kutta method can be used to divide the cable along its length into multiple small segments. The gradient value at each segment is then multiplied by the segment length to accumulate and calculate the temperature increment at each location. Finally, the temperature value at the starting temperature point is superimposed with the calculated temperature changes at each location to determine the precise temperature values ​​at those locations. Combining all these calculated temperature values ​​in spatial order generates the final temperature distribution data that comprehensively reflects the thermal state of the cable.

[0052] Suppose a temperature distribution data vector containing 21 temperature values ​​is generated, such as [52.0℃, 52.32℃, 52.70℃, ..., 64.1℃, ..., 54.2℃]. To monitor the temperature of the target power cable, this temperature distribution data is analyzed. For example, by traversing all temperature values ​​in the vector, the maximum value of 64.1℃ can be identified, corresponding to a location 10 meters along the cable, thus determining the existence of a localized overheating hotspot on the power cable. Simultaneously, all temperature values ​​can be compared with a preset alarm threshold, such as 60℃. Since 64.1℃ is greater than 60℃, an overheating alarm can be triggered, and the hotspot location and temperature information are reported to the monitoring center, thereby completing real-time comprehensive temperature monitoring and risk warning for the power cable.

[0053] This embodiment utilizes a surface acoustic wave (SAW) sensor array to acquire discrete temperature points along the cable and calculate preliminary temperature gradient information. More innovatively, it introduces real-time analysis of the power cable current waveform, using a deep neural network to predict temperature gradient changes caused by dynamic loads from harmonic components. Subsequently, Kalman filtering is used to optimally fuse the physically measured temperature gradient with the harmonic-predicted temperature gradient, effectively correcting the deviation of a single measurement source under dynamic conditions and obtaining a more accurate temperature gradient. Finally, based on this temperature gradient information and known temperature points, integration and reconstruction are performed to achieve continuous real-time temperature distribution monitoring of the entire power cable. This improves the accuracy of power cable temperature measurement and the comprehensiveness of temperature monitoring under dynamic loads.

[0054] In one feasible implementation, the method further includes: Calculate the residual vector between the predicted temperature gradient information and the temperature gradient feature vector of the temperature gradient information.

[0055] A temperature gradient feature vector is a vector representation of temperature gradient information, where each element corresponds to the gradient value at a specific location on the cable. A residual vector is the difference vector obtained by subtracting the predicted temperature gradient feature vector from the measured temperature gradient feature vector element-wise. Each element of this residual vector represents the deviation between the model-predicted temperature gradient and the physically measured temperature gradient at the corresponding location.

[0056] The predicted temperature gradient information generated in S140 and the temperature gradient information generated in S120 were constructed as temperature gradient feature vectors with the same dimension and spatial correspondence during generation. The residual vector between these two vectors can be calculated by performing vector subtraction. Specifically, each element of the predicted temperature gradient feature vector is subtracted from the corresponding element of the temperature gradient feature vector at the same spatial location, and the result forms a new vector. This newly generated vector is the residual vector, and its magnitude and direction comprehensively reflect the degree of deviation between the model prediction and physical reality. An ideal residual vector should have all elements close to zero.

[0057] The norm of the residual vector is compared with a preset threshold. If the norm of the residual vector is greater than the preset threshold, it is determined that the target power cable has non-electrical damage.

[0058] First, the norm of the residual vector calculated in the previous step is calculated. For example, the L2 norm, also known as the Euclidean norm, can be used. The squares of all elements in the residual vector are summed, and the square root of the sum is taken to obtain a non-negative real value, which is the norm of the residual vector. Then, this calculated norm value is compared with a preset threshold. This preset threshold is determined through statistical analysis of the residual norms generated by the cable under healthy operating conditions over a long period. For example, the average of historical normal norm values ​​can be taken plus three standard deviations. If the comparison finds that the norm of the currently calculated residual vector is greater than this preset threshold, it is determined that the cable has non-electrical damage. This is because a significantly large residual indicates an abnormal thermal gradient on the cable that cannot be explained by the model, and this anomaly is most likely caused by physical damage not covered by the model.

[0059] For example, after obtaining the temperature distribution data, an additional step is included for diagnosing damage caused by non-electrical factors. Suppose that at a certain moment, the predicted temperature gradient information generated in S140 and the temperature gradient information generated in S120 are obtained. Both of these information were constructed as 21-dimensional temperature gradient feature vectors with the same dimensions and spatial correspondence during generation. For example, the predicted gradient vector V_pred can be [0.32℃ / m, ..., 2.45℃ / m, ..., -0.28℃ / m], while the measured gradient vector V_meas can be [0.32℃ / m, ..., 1.85℃ / m, ..., -0.29℃ / m].

[0060] First, the residual vector V_res between the two vectors is calculated by performing vector subtraction. This operation subtracts the corresponding element in the measured gradient vector at the same spatial location from each element in the predicted gradient vector. At the index corresponding to the 10-meter position on the cable, the element value of V_pred is 2.45℃ / m, while the corresponding element value of V_meas is 1.85℃ / m. Therefore, the element value of V_res at that position is 2.45 - 1.85 = 0.6℃ / m. By performing subtraction on the elements at all 21 positions, a complete residual vector V_res can be obtained.

[0061] Subsequently, the norm of the residual vector is compared with a preset threshold to determine whether non-electrical damage exists. Specifically, the norm of the residual vector V_res calculated in the previous step is calculated, for example, using the L2 norm (Euclidean norm), by summing the squares of all 21 elements in the residual vector and then taking the square root of the sum. Assume the calculated norm value is 3.5. This value is compared with a preset threshold, such as 2.0. Since the calculated norm value of 3.5 is greater than the preset threshold of 2.0, this indicates a significant systematic deviation between the model prediction and physical measurements. Because the deep neural network model has learned the normal thermal effects caused by harmonics, this large deviation is likely not caused by conventional electrical factors. Therefore, it can be determined that the target power cable has non-electrical damage, such as a poor connection at the cable joint at the 10-meter mark, causing additional heating that the model has not learned, and this can trigger corresponding warnings or maintenance work orders.

[0062] In one feasible implementation, the frequency drift data includes high-frequency drift data and low-frequency drift data. The high-frequency drift data and low-frequency drift data refer to two frequency drift signals synchronously output by the dual-frequency surface acoustic wave sensor. The high-frequency acoustic wave has a shallower penetration depth, and its frequency drift primarily reflects temperature changes on the cable surface; while the low-frequency acoustic wave has a deeper penetration depth, and its frequency drift reflects more the heat effects from within the cable.

[0063] The method further includes: calculating multiple differential frequency response values ​​based on high-frequency drift data and low-frequency drift data using a preset conversion function, and generating differential frequency response distribution data of the target power cable by spatial interpolation based on the multiple differential frequency response values.

[0064] The differential frequency response value is a scalar value characterizing the temperature difference between the inside and surface of a cable; its sign or magnitude can indicate the direction of heat flow. Differential frequency response distribution data refers to a set of data showing how the differential frequency response values ​​are continuously distributed along the length of the cable.

[0065] The surface wave sensor of this application can be a dual-frequency surface acoustic wave sensor, that is, a sensor that integrates two independent working channels on a single surface wave sensor to simultaneously generate and detect high-frequency and low-frequency surface acoustic waves, thereby sensing the physical characteristics of objects at different depths. During data acquisition, high-frequency drift data and low-frequency drift data for each sensor location are obtained simultaneously. Then, a differential frequency response value is obtained by calculating each pair of high- and low-frequency drift data using a preset conversion function. This conversion function can be a weighted difference function, as shown in formula (2).

[0066] (2) Where Vd represents the difference frequency response value, ΔfL and ΔfH represent the low-frequency drift data and high-frequency drift data, respectively, and kL and kH represent the temperature sensitivity coefficients of the low-frequency mode and high-frequency mode obtained through experimental calibration, respectively.

[0067] By performing the same calculation on all sensors in the sensor array, a set of discrete difference frequency response values ​​corresponding one-to-one with the sensor positions can be obtained. For example, assuming that at a certain position, ΔfL is -16.2kHz, ΔfH is -15.8kHz, and the corresponding sensitivity coefficients kL and kH are -2.6℃ / kHz and -2.5℃ / kHz, the difference frequency response value calculated according to formula (2) is approximately 2.62. Finally, in order to obtain the difference frequency response value at any position, a spatial interpolation method similar to that in S120 is used, such as cubic spline interpolation, to fit this set of discrete difference frequency response values ​​into a continuous difference frequency response distribution curve. The data of this curve is the difference frequency response distribution data, which can be represented as a vector including multiple interpolation points, such as [1.50, 1.82, ..., 2.62, ..., 1.95].

[0068] The target temperature information that is greater than the preset temperature threshold is determined from the temperature distribution data, and the heat source attribute information of the target power cable at the location corresponding to the target temperature information is determined based on the target differential frequency response value corresponding to the target temperature information in the differential frequency response distribution data.

[0069] The target differential frequency response value refers to the differential frequency response value corresponding to a specific target location in continuous differential frequency response distribution data. Heat source attribute information refers to the category of the root cause leading to temperature rise at a specific location on the cable, which can include internal electrical heat sources and external environmental heat sources.

[0070] First, the temperature value of each location area in the temperature distribution data is compared with a preset temperature threshold to determine if there are any locations where the temperature exceeds the limit. The temperature threshold can be 60℃. The temperatures exceeding the threshold and their corresponding locations in the temperature distribution data constitute the target temperature information. Based on the location of the target temperature information, the data point corresponding to the location of the target temperature information is determined in the difference frequency response distribution data, and the difference frequency response value at that location is extracted to obtain the target difference frequency response value. For example, if an overheated spot is located 10 meters from the cable, then the target difference frequency response value is the value of the difference frequency response distribution data at the 10-meter location.

[0071] In one feasible implementation, the heat source attribute information includes internal electrical heat sources and external environmental heat sources. Based on the target differential frequency response value corresponding to the target temperature information in the differential frequency response distribution data, the heat source attribute information of the target power cable at the location corresponding to the target temperature information is determined, including: Based on the target temperature information, determine the target difference frequency response value corresponding to the location in the difference frequency response distribution data.

[0072] After obtaining the location of the target temperature information, the location information is used to perform a query or index operation on the generated continuous difference frequency response distribution data. By inputting the location coordinates of the target temperature information, the difference frequency response value that perfectly matches the location can be extracted from the distribution data, which is the final determined target difference frequency response value.

[0073] Internal electrical heat sources and external environmental heat sources are two specific classifications used to describe heat source attribute information. Internal electrical heat sources refer to heat generated by the cable's own electrical components, such as conductor resistance heating, additional losses caused by harmonics, or poor contact at joints. External environmental heat sources refer to heat generated by the environment in which the cable is located, such as strong solar radiation or proximity to other high-temperature equipment.

[0074] When the target differential frequency response value is positive, the heat source attribute of the target power cable at the location corresponding to the target temperature information is an internal electrical heat source. When the target differential frequency response value is non-positive, the heat source attribute of the target power cable at the location corresponding to the target temperature information is an external environmental heat source.

[0075] When heat is generated inside the cable and conducted outwards, the internal temperature will be higher than the surface temperature, resulting in a positive target differential frequency response value. Conversely, when heat is transferred from the external environment into the cable, the surface temperature will be higher than or equal to the internal temperature, resulting in a non-positive target differential frequency response value. Therefore, by considering the sign of the target differential frequency response value, if it is positive, the heat source attribute information at that location is determined to be an internal electrical heat source. If the target differential frequency response value is non-positive, i.e., zero or negative, the heat source attribute information is determined to be an external environmental heat source.

[0076] For example, assuming the temperature distribution data is a vector containing 21 temperature values, each temperature value in the vector is first compared with a preset temperature threshold, such as 60°C. The comparison reveals that the temperature value at the 10-meter mark on the cable is 64.1°C, which is greater than the 60°C threshold. Therefore, this temperature value and its corresponding 10-meter location are jointly identified as the target temperature information. Subsequently, based on this 10-meter location information, a query is performed on the difference frequency response distribution data, which also covers the range from 0 to 20 meters and was generated in the previous step, for example, [1.50, 1.82, ..., 2.62, ..., 1.95]. By indexing to the data point corresponding to the 10-meter location, the difference frequency response value at that location is extracted, resulting in the target difference frequency response value of 2.62.

[0077] Next, the target differential frequency response value of 2.62 is analyzed to determine the heat source attribute information. Since the current target differential frequency response value of 2.62 is positive, the heat source attribute information of this overheated area is determined to be an internal electrical heat source. This indicates that the overheating at 10 meters is caused by an electrical problem with the cable itself, such as increased contact resistance due to aging of the connector, and is not affected by external environmental interference or has minimal interference. A high-priority maintenance work order can be generated to guide personnel to accurately inspect and repair the connector at the 10-meter location, effectively avoiding unnecessary power outages due to misjudgment and significantly improving maintenance efficiency and power supply reliability.

[0078] This embodiment employs a dual-frequency surface acoustic wave (SAW) sensor array to simultaneously acquire high-frequency and low-frequency drift data that reflect the thermal effects at different depths of the cable. The difference between these two data sets is used to calculate the differential frequency response distribution data, characterizing the temperature difference between the cable's interior and surface. Finally, when a potential hot spot is detected, the differential frequency response value corresponding to the hot spot location is used to identify whether the heat source originates from internal electrical factors within the cable or from external environmental interference. Therefore, this improves the ability to diagnose and assess the actual thermal risks of power cables, achieving effective and comprehensive temperature monitoring of power cables.

[0079] In one feasible implementation, S130: The amplitude and phase angle of a preset order harmonic are extracted from the spectrum of the current waveform time-series data to obtain harmonic component information, including: The current waveform timing data is frequency domain transformed to generate a spectrum, and the fundamental peak value is determined in the spectrum.

[0080] First, the acquired current waveform time-series data undergoes frequency domain transformation. A Fast Fourier Transform (FFT) algorithm can be used to generate the spectrum of this current signal. The generated spectrum contains peaks at multiple different frequencies, with the most concentrated energy being the fundamental frequency component. By searching and locating the frequency point with the strongest signal intensity within the expected fundamental frequency range, such as 45Hz to 55Hz, the fundamental peak value can be determined. The fundamental peak value refers to the peak with the strongest signal intensity in the spectrum, and its corresponding frequency is generally the fundamental frequency of the power grid.

[0081] Using the fundamental peak value as a reference, the harmonic peak values ​​of preset order harmonics are extracted from the spectrum, as well as the phase difference of each harmonic peak value relative to the fundamental peak value.

[0082] Harmonic peaks are signal peaks located in the frequency spectrum at positions that are integer multiples of the fundamental frequency. Each harmonic peak represents the intensity of a specific harmonic component. Phase difference is the difference between the phase angle of a harmonic peak and the phase angle of the fundamental peak.

[0083] Based on a pre-defined list of harmonic orders and using the frequency of the fundamental peak determined in the previous step as a reference, precise harmonic localization and extraction are performed in the frequency spectrum. For example, if the pre-defined order is 3 and the fundamental frequency is 50Hz, harmonic peaks are searched and extracted around 150Hz. For each harmonic peak located according to the pre-defined order, the signal strength of the peak is recorded, and the difference between the phase angle of the harmonic peak and the phase angle of the fundamental peak is calculated. This difference is the phase difference of that harmonic. By performing the same extraction operation on all pre-defined order harmonics, a set of harmonic peaks and their corresponding phase differences are obtained.

[0084] Harmonic feature vectors are constructed using the amplitude of the peak values ​​of the preset order harmonics and their corresponding phase differences to obtain harmonic component information.

[0085] The peak intensities (i.e., harmonic amplitudes) of all extracted preset harmonic orders, along with their corresponding phase differences, are integrated. Following the preset harmonic order order, for example from low to high, the harmonic amplitudes and phase differences of each order are concatenated to form a fixed-dimensional one-dimensional vector. This final vector is the harmonic characteristic vector, which represents the harmonic component information.

[0086] For example, assume the acquired current waveform timing data is a digital sequence containing 2000 sampling points. First, the sequence is transformed into the frequency domain by performing a Fast Fourier Transform to generate a spectrum. A search is performed in the range of 45Hz to 55Hz of this spectrum to determine the peak with the strongest signal intensity, i.e., the fundamental peak is located at 50Hz.

[0087] Subsequently, using the 50Hz fundamental peak as a reference, and based on a set of preset harmonic orders, such as the 3rd, 5th, and 7th, corresponding harmonic information is extracted from the spectrum. Specifically, the corresponding harmonic peaks are extracted near 150Hz, 250Hz, and 350Hz, and their signal strengths are recorded as amplitudes, for example, 10.2A, 5.8A, and 2.1A, respectively. Simultaneously, the phase differences between these three harmonic peaks and the 50Hz fundamental peak are calculated, for example, -30 degrees, 150 degrees, and -120 degrees, respectively. Finally, according to the preset order of the 3rd, 5th, and 7th harmonics, these amplitudes and corresponding phase differences are constructed into a harmonic feature vector. By concatenating the data in the format [amplitude 3, phase difference 3, amplitude 5, phase difference 5, amplitude 7, phase difference 7], a six-dimensional vector is finally obtained: [10.2, -30, 5.8, 150, 2.1, -120].

[0088] Figure 2 A flowchart illustrating a method for calculating target temperature gradient information using Kalman filtering according to an embodiment of this application is shown. This scheme is applicable to power cables with dynamic loads, such as... Figure 2 As shown, the method includes steps S210 to S230.

[0089] In one feasible implementation, S150: Based on the predicted temperature gradient information and the temperature gradient information, iterative calculation is performed using Kalman filtering to obtain the target temperature gradient information, including: S210: Use the predicted temperature gradient information as the state prediction value and the temperature gradient information as the observation value.

[0090] The predicted temperature gradient information calculated by the deep neural network based on real-time harmonic component information is used as the state prediction value in the Kalman filter algorithm. Simultaneously, the measured temperature gradient information obtained by processing the frequency drift data returned by the surface acoustic wave sensor array is used as the observation value in the Kalman filter algorithm.

[0091] S220: Calculate the correction factor based on the deviation between the predicted and observed values, as well as the preset process noise covariance and observation noise covariance.

[0092] Process noise covariance is a preset parameter used to quantify the uncertainty of state predictions, characterizing the degree of confidence in the prediction accuracy of deep neural network models. Observation noise covariance is a preset parameter used to quantify the uncertainty of observations, characterizing the degree of confidence in the accuracy of physical sensor measurement systems. The correction factor, also known as Kalman gain, is a weighted coefficient dynamically calculated based on the relative magnitudes of process noise and observation noise.

[0093] First, the deviation between the predicted and observed values ​​is calculated, i.e., a deviation vector is obtained through vector subtraction. Then, two pre-defined parameter matrices are invoked: the process noise covariance matrix and the observation noise covariance matrix. The values ​​of these two matrices are pre-set based on statistical analysis of the prediction errors of the neural network model and the errors of the sensor measurement system. Using these two noise covariance matrices and the current deviation information, matrix operations are performed using the Kalman gain equation to finally calculate a correction factor. The correction factor is essentially a dynamic weight; when the observation noise is much smaller than the process noise, the correction factor will be larger, meaning the system trusts the measured data more.

[0094] S230: The state prediction value is corrected using a correction factor, and the target temperature gradient information is generated through iterative calculation.

[0095] Using the correction factor calculated in the previous step, the state prediction value is weighted and corrected. Specifically, the deviation vector is multiplied by the correction factor to obtain a correction amount, which is then added to the original state prediction value. The new vector obtained after this correction is the final output target temperature gradient information for the current iteration cycle. This output target temperature gradient information serves as the final result of this calculation, and its related states are also used as historical state information in the next time step iteration, thus achieving continuous real-time state tracking and optimal estimation.

[0096] First, in S210, the generated 21-dimensional predicted temperature gradient information vector, such as [0.30℃ / m, ..., 2.45℃ / m, ...], is used as the state prediction value; simultaneously, the generated 21-dimensional measured temperature gradient information vector, such as [0.32℃ / m, ..., 2.37℃ / m, ...], is used as the observed value. The deviation vector between these two vectors is calculated first. Then, the preset process noise covariance matrix and observation noise covariance matrix are called, and a 21x21-dimensional correction factor matrix is ​​calculated using the Kalman gain equation in conjunction with the deviation information.

[0097] The state prediction vector is corrected using this correction factor matrix to generate target temperature gradient information. For example, at the gradient peak position, the predicted value is 2.45℃ / m, the measured value is 2.37℃ / m, and the deviation is 0.08℃ / m. Assuming the correction amount calculated based on the correction factor is -0.05℃ / m, then the corrected target gradient value is 2.40℃ / m. By performing the same correction operation on all 21 elements of the state prediction vector, a new 21-dimensional vector is finally generated, for example, [0.31℃ / m, ..., 2.40℃ / m, ...]. This vector is the target temperature gradient information finally generated in one iteration cycle in this embodiment. The target temperature gradient information obtained through this fusion method utilizes the deep understanding of harmonic thermal effects by the deep neural network model and is calibrated and constrained by real-time observation data from physical sensors. Therefore, compared with single prediction or observation, it can more smoothly filter out random noise and more accurately reflect the true temperature gradient caused by dynamic load.

[0098] Figure 3 A flowchart illustrating a method for determining temperature distribution data of a target power cable according to an embodiment of this application is shown. This method is applicable to power cables with dynamic loads, such as... Figure 3 As shown, the method includes steps S310 to S330.

[0099] In one feasible implementation, S160: Based on the target temperature gradient information, taking any one of multiple temperature points as a starting point, the temperature distribution data of the target power cable is determined through integration calculation, including: S310: Select one temperature point from multiple temperature points as the starting temperature point.

[0100] From the set of multiple discrete temperature points obtained from S120 that correspond one-to-one with the positions of the surface acoustic wave sensor array, one of them is selected as the starting temperature point. For the convenience and consistency of the calculation, the temperature point corresponding to the sensor located at the physical starting point, such as the 0-meter position, can be selected.

[0101] S320: Based on the temperature gradient value of the target temperature gradient information, calculate the temperature change of different locations of the target power cable relative to the starting temperature point through integral calculation.

[0102] Numerical integration is used to calculate the temperature change at all other discrete locations on the cable relative to the starting temperature point. For example, the cable length can be divided into multiple small spatial segments between the starting point and other points. Then, the gradient value on each segment is multiplied by the segment length, and the total temperature increment from the starting point to each target location is calculated by summing the results. This total temperature increment is the temperature change at that target location.

[0103] S330: The temperature value at the starting temperature point is superimposed with the temperature change to determine the temperature value at each location on the target power cable and generate temperature distribution data.

[0104] The temperature change at each location point calculated in step S320 is superimposed with the absolute temperature value of the selected starting temperature point in S310; that is, the final absolute temperature value of each location point is determined through addition. By performing the same superposition operation on all target locations, a precise temperature value sequence covering the entire monitoring section is obtained. Combining all these calculated temperature values, including the starting temperature point itself, in spatial order generates the final temperature distribution data that comprehensively reflects the thermal state of the cable.

[0105] For example, from the generated set of five discrete temperature points, namely [52.0℃, 53.4℃, 64.1℃, 55.5℃, 54.2℃], the temperature point at 0 meters is selected as the starting temperature point. Then, based on the generated target temperature gradient information vector, such as [0.31℃ / m, 0.38℃ / m, ..., 2.40℃ / m, ...], the temperature change at all other locations on the cable relative to this starting temperature point is calculated through numerical integration. For example, to calculate the temperature change at 10 meters, the path from 0 meters to 10 meters is divided into 10 small spatial segments of 1 meter in length, and the gradient value on each segment is multiplied by the segment length and then summed. Assume the calculated cumulative temperature increment from 0 meters to 10 meters is 12.1℃. This temperature change is then superimposed on the temperature value at the starting temperature point to determine the final absolute temperature at 10 meters as 64.1℃. By performing the same integration and superposition operations on all other locations, a precise temperature value sequence covering the entire monitoring section is obtained. Combining these temperature values ​​in spatial order generates a 21-dimensional temperature distribution data vector, such as [52.0℃, 52.31℃, ..., 64.1℃, ..., 54.2℃], thus achieving a comprehensive reconstruction of the temperature along the cable.

[0106] Based on the same concept, this application provides a power cable temperature monitoring system based on a surface acoustic wave sensor, which is described below in conjunction with... Figure 4The power cable temperature monitoring system based on a surface acoustic wave sensor provided in this application embodiment will be described in detail.

[0107] Figure 4 This is a structural block diagram of a power cable temperature monitoring system based on a surface acoustic wave sensor, as shown in an embodiment of this application.

[0108] like Figure 4 As shown, this system is suitable for power cables under dynamic loads. The power cable temperature monitoring system based on a surface acoustic wave sensor may include: The acquisition module 410 is used to acquire frequency drift data and current waveform timing data of the target power cable at multiple preset locations. The frequency drift data is collected by a surface acoustic wave sensor array deployed along the target power cable. The conversion module 420 is used to convert the frequency drift value of the frequency drift data into multiple temperature points according to the preset frequency-temperature mapping relationship, and to perform spatial interpolation and differentiation operations based on the multiple temperature points to obtain temperature gradient information. The extraction module 430 is used to extract the amplitude and phase angle of a preset order harmonic from the spectrum of the current waveform time series data to obtain harmonic component information. The calculation module 440 is used to input the harmonic component information into the trained deep neural network to obtain the predicted temperature gradient information. The trained deep neural network is trained by training samples including historical harmonic component information and corresponding historical temperature gradient information. The calculation module 440 is also used to perform iterative calculations using Kalman filtering based on the predicted temperature gradient information and the temperature gradient information to obtain the target temperature gradient information. The calculation module 440 is also used to determine the temperature distribution data of the target power cable by integral calculation based on the target temperature gradient information, starting from any one of multiple temperature points, so as to realize the temperature monitoring of the target power cable.

[0109] In one embodiment, the calculation module 440 is further configured to calculate the residual vector between the predicted temperature gradient information and the temperature gradient feature vector of the temperature gradient information; compare the norm of the residual vector with a preset threshold, and determine that the target power cable has non-electrical damage if the norm of the residual vector is greater than the preset threshold.

[0110] In one embodiment, the frequency drift data includes high-frequency drift data and low-frequency drift data; the calculation module 440 is further configured to calculate multiple differential frequency response values ​​based on the high-frequency drift data and low-frequency drift data using a preset conversion function, and generate differential frequency response distribution data of the target power cable by spatial interpolation based on the multiple differential frequency response values; determine the target temperature information that is greater than a preset temperature threshold in the temperature distribution data, and determine the heat source attribute information of the target power cable at the location corresponding to the target temperature information based on the target differential frequency response value corresponding to the target temperature information in the differential frequency response distribution data.

[0111] In one embodiment, the heat source attribute information includes internal electrical heat sources and external environmental heat sources; the calculation module 440 is specifically used to determine the target differential frequency response value corresponding to the location in the differential frequency response distribution data based on the target temperature information; when the target differential frequency response value is positive, the heat source attribute of the target power cable at the location corresponding to the target temperature information is an internal electrical heat source; when the target differential frequency response value is non-positive, the heat source attribute of the target power cable at the location corresponding to the target temperature information is an external environmental heat source.

[0112] In one embodiment, the extraction module 430 is used to perform frequency domain conversion on the current waveform timing data to generate a spectrum, and determine the fundamental peak value in the spectrum; based on the fundamental peak value, extract the harmonic peak values ​​of the preset order harmonics in the spectrum, as well as the phase difference of each harmonic peak value relative to the fundamental peak value; construct a harmonic feature vector with the amplitude of the harmonic peak values ​​of the preset order harmonics and the corresponding phase difference to obtain harmonic component information.

[0113] In one embodiment, the calculation module 440 is specifically used to use the predicted temperature gradient information as the state prediction value and the temperature gradient information as the observation value; calculate the correction factor based on the deviation between the state prediction value and the observation value, as well as the preset process noise covariance and observation noise covariance; correct the state prediction value using the correction factor, and generate the target temperature gradient information through iterative calculation.

[0114] In one embodiment, the calculation module 440 is specifically used to select a temperature point from multiple temperature points as the starting temperature point; calculate the temperature change of different positions of the target power cable relative to the starting temperature point through integral operation based on the temperature gradient value of the target temperature gradient information; and superimpose the temperature value of the starting temperature point with the temperature change to determine the temperature value of each position on the target power cable and generate temperature distribution data.

[0115] Figure 4 Each module in the system shown has an implementation Figures 1 to 3 The functions of each step in the process and their corresponding technical effects are described briefly and will not be elaborated here.

[0116] Figure 5 A schematic diagram of the hardware structure of an electronic device provided in one embodiment of this application is shown.

[0117] The electronic device may include a processor 510 and a memory 520 storing computer program instructions.

[0118] Specifically, the processor 510 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0119] Memory 520 may include mass storage for data or instructions. For example, and not limitingly, memory 520 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 520 may include removable or non-removable (or fixed) media. Where appropriate, memory 520 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 520 is non-volatile solid-state memory.

[0120] Memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to the first aspect of this disclosure.

[0121] The processor 510 reads and executes computer program instructions stored in the memory 520 to implement any of the power cable temperature monitoring methods based on surface acoustic wave sensors in the above embodiments.

[0122] In one example, the electronic device may also include a communication interface 530 and a bus 540. Wherein, such as Figure 5 As shown, the processor 510, memory 520, and communication interface 530 are connected through bus 540 and complete communication with each other.

[0123] The communication interface 530 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0124] Bus 540 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 540 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.

[0125] This electronic device can execute the power cable temperature monitoring method based on a surface acoustic wave sensor as described in this application embodiment, thereby achieving a combination of... Figures 1 to 3 A method for monitoring the temperature of power cables based on a surface acoustic wave sensor is described.

[0126] Furthermore, in conjunction with the power cable temperature monitoring method based on a surface acoustic wave sensor in the above embodiments, this application embodiment can provide a computer-readable storage medium for implementation. This computer-readable storage medium stores computer program instructions; when executed by a processor, these computer program instructions implement any of the power cable temperature monitoring methods based on a surface acoustic wave sensor in the above embodiments.

[0127] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0128] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0129] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0130] The aspects of this application have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by dedicated hardware performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0131] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A method for monitoring the temperature of power cables based on a surface acoustic wave sensor, applicable to power cables with dynamic loads, characterized in that, The method includes: The frequency drift data and current waveform timing data of the target power cable at multiple preset locations are acquired. The frequency drift data is collected by a surface acoustic wave sensor array deployed along the target power cable. According to the preset frequency-temperature mapping relationship, the frequency drift value of the frequency drift data is converted into multiple temperature points, and spatial interpolation and differentiation operations are performed based on the multiple temperature points to obtain temperature gradient information. The amplitude and phase angle of the preset order harmonics are extracted from the spectrum of the current waveform time series data to obtain harmonic component information; The harmonic component information is input into a trained deep neural network to obtain predicted temperature gradient information. The trained deep neural network is trained using training samples that include historical harmonic component information and corresponding historical temperature gradient information. Based on the predicted temperature gradient information and the temperature gradient information, Kalman filtering is used for iterative calculation to obtain the target temperature gradient information. Based on the target temperature gradient information, starting from any one of the multiple temperature points, the temperature distribution data of the target power cable is determined through integral calculation to achieve temperature monitoring of the target power cable.

2. The method according to claim 1, characterized in that, The method further includes: Calculate the residual vector between the predicted temperature gradient information and the temperature gradient feature vector of the temperature gradient information; The norm of the residual vector is compared with a preset threshold. If the norm of the residual vector is greater than the preset threshold, it is determined that the target power cable has non-electrical damage.

3. The method according to claim 1, characterized in that, The frequency drift data includes high-frequency drift data and low-frequency drift data; the method further includes: Based on the high-frequency drift data and the low-frequency drift data, multiple differential frequency response values ​​are calculated using a preset conversion function, and spatial interpolation is performed based on the multiple differential frequency response values ​​to generate differential frequency response distribution data of the target power cable; In the temperature distribution data, a target temperature information greater than a preset temperature threshold is determined, and based on the target temperature information and the target differential frequency response value corresponding to the target temperature information in the differential frequency response distribution data, the heat source attribute information of the target power cable at the location corresponding to the target temperature information is determined.

4. The method according to claim 3, characterized in that, The heat source attribute information includes internal electrical heat sources and external environmental heat sources; The step of determining the heat source attribute information of the target power cable at the location corresponding to the target temperature information based on the target difference frequency response value in the difference frequency response distribution data according to the target temperature information includes: Based on the target temperature information, determine the target difference frequency response value corresponding to the location in the difference frequency response distribution data; When the target differential frequency response value is positive, the heat source attribute of the target power cable at the location corresponding to the target temperature information is the internal electrical heat source; When the target differential frequency response value is not positive, the heat source attribute of the target power cable at the location corresponding to the target temperature information is the external environmental heat source.

5. The method according to claim 1, characterized in that, The step involves extracting the amplitude and phase angle of a preset order harmonic from the spectrum of the current waveform time-series data to obtain harmonic component information, including: The current waveform timing data is frequency domain transformed to generate a spectrum, and the fundamental peak value is determined in the spectrum; Using the fundamental peak value as a reference, the harmonic peak values ​​of the preset order harmonics are extracted from the spectrum, as well as the phase difference of each harmonic peak value relative to the fundamental peak value; Harmonic feature vectors are constructed using the amplitude of the peak value of the preset order harmonic and the corresponding phase difference to obtain the harmonic component information.

6. The method according to claim 1, characterized in that, The step of obtaining the target temperature gradient information by iterative calculation using Kalman filtering based on the predicted temperature gradient information and the temperature gradient information includes: The predicted temperature gradient information is used as the state prediction value, and the temperature gradient information is used as the observation value. Based on the deviation between the predicted state value and the observed value, as well as the preset process noise covariance and observation noise covariance, a correction factor is calculated. The predicted state value is corrected using the correction factor, and the target temperature gradient information is generated through iterative calculation.

7. The method according to claim 1, characterized in that, The step of determining the temperature distribution data of the target power cable based on the target temperature gradient information, starting from any one of the plurality of temperature points, through integration calculation includes: Select one temperature point from the plurality of temperature points as the starting temperature point; Based on the temperature gradient value of the target temperature gradient information, the temperature change of the target power cable at different locations relative to the starting temperature point is calculated by integral operation. The temperature value at the starting temperature point is superimposed with the temperature change to determine the temperature value at each location on the target power cable, thus generating the temperature distribution data.

8. A power cable temperature monitoring system based on a surface acoustic wave sensor, suitable for power cables with dynamic loads, characterized in that, The system includes: The acquisition module is used to acquire frequency drift data and current waveform timing data of the target power cable at multiple preset locations. The frequency drift data is collected by a surface acoustic wave sensor array deployed along the target power cable. The conversion module is used to convert the frequency drift value of the frequency drift data into multiple temperature points according to a preset frequency-temperature mapping relationship, and to perform spatial interpolation and differentiation operations based on the multiple temperature points to obtain temperature gradient information. The extraction module is used to extract the amplitude and phase angle of a preset order harmonic from the spectrum of the current waveform time series data to obtain harmonic component information. The calculation module is used to input the harmonic component information into the trained deep neural network to obtain the predicted temperature gradient information. The trained deep neural network is trained by training samples including historical harmonic component information and corresponding historical temperature gradient information. The calculation module is also used to perform iterative calculations using Kalman filtering based on the predicted temperature gradient information and the temperature gradient information to obtain the target temperature gradient information; The calculation module is also used to determine the temperature distribution data of the target power cable by integral calculation based on the target temperature gradient information, taking any one of the multiple temperature points as the starting point, so as to realize the temperature monitoring of the target power cable.

9. An electronic device, characterized in that, The device includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the power cable temperature monitoring method based on a surface acoustic wave sensor as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions, which, when executed by a processor, implement the power cable temperature monitoring method based on a surface acoustic wave sensor as described in any one of claims 1-7.

Images