A millimeter wave radar power line clustering system and method

Abstract

The application discloses a millimeter wave radar power transmission line clustering system and method, relates to the technical field of radar signal processing and power transmission line monitoring, and aims to solve the problems of the prior art in monitoring accuracy, efficiency and all-weather monitoring, and improve the reliability and intelligent level of the power transmission line monitoring system. The system comprises a millimeter wave radar module, a signal processing module, a clustering and classification module and a risk warning module; foreign matter identification and icing monitoring of the power transmission line are carried out based on the millimeter wave radar, the system has high resolution and strong penetration ability, and can work stably under complex weather conditions; in combination with the K-means++ algorithm and the Mahalanobis distance method, foreign matter and icing on the power transmission line can be accurately classified and identified; the system improves the identification accuracy of different targets in complex scenes, reduces the false detection and missed detection problems in the traditional method, can realize automatic real-time monitoring, reduces the demand for manual intervention, and greatly improves the monitoring efficiency and the safety of power transmission line maintenance.

Description

Technical Field

[0001] This invention relates to the fields of radar signal processing and power transmission line monitoring technology, and in particular to a millimeter-wave radar power transmission line clustering system and method. Background Technology

[0002] With the continued growth in electricity demand, the safe operation and efficient monitoring of transmission lines have become increasingly important. Currently, traditional transmission line monitoring methods mainly rely on optical sensors, video surveillance systems, and manual inspections. While these methods can meet basic monitoring needs to a certain extent, they have many limitations in practical applications.

[0003] Optical sensors and video surveillance systems perform poorly in adverse weather conditions (such as fog, rain, and snow), leading to a significant decrease in monitoring effectiveness. This environmental impact makes it difficult for monitoring systems to achieve real-time, all-weather monitoring, thereby affecting the safety of transmission lines. This limitation is particularly pronounced in complex or variable natural environments, making it impossible to detect potential safety hazards in a timely manner.

[0004] On the other hand, while manual inspection, as a traditional monitoring method, can provide direct detection and feedback, it is inefficient and time-consuming, especially in rugged terrain or adverse weather conditions, where inspection personnel face significant safety hazards. This not only increases operating and maintenance costs but also exposes inspection personnel to potential dangers.

[0005] While existing low-frequency radar monitoring systems possess a certain degree of penetration capability and can provide valuable monitoring information in some situations, their relatively low resolution makes it difficult to effectively identify small foreign objects or ice layers on power transmission lines. This is especially true in complex environments, where they are prone to missed detections or false alarms. This insufficient resolution directly leads to the monitoring system's inadequate ability to handle small targets, failing to meet the safety supervision requirements of modern power transmission.

[0006] Furthermore, existing monitoring systems lack intelligent clustering and classification algorithms, making it difficult to effectively distinguish between various types of risks. For example, different types of foreign objects differ significantly from snow and ice, but due to limitations in conventional monitoring methods, the system struggles to quickly and accurately identify and classify them. This results in insufficient processing capacity for monitoring systems in complex scenarios, preventing them from automatically and in real-time classifying and responding to potential threats.

[0007] In summary, current transmission line monitoring technology faces the following key technical challenges:

[0008] Poor environmental adaptability: Traditional optical sensors and video surveillance perform poorly in adverse weather conditions, impacting safety. Low efficiency: Manual inspections are time-consuming, inefficient, and pose safety hazards. Insufficient resolution: Low-frequency radar monitoring systems have low resolution, making it difficult to identify small foreign objects or ice layers in complex environments. Insufficient intelligence: The lack of effective clustering and classification algorithms hinders the rapid and accurate identification and classification of different types of risks.

[0009] The existence of these technical problems urgently needs to be addressed through innovative monitoring technologies and intelligent algorithms in order to improve the safety monitoring level of transmission lines. Summary of the Invention

[0010] To address the aforementioned issues, this invention proposes a millimeter-wave radar-based transmission line clustering system and method. Based on millimeter-wave radar, it identifies foreign objects and monitors icing on transmission lines, possessing high resolution and strong penetration capabilities. It can operate stably under complex weather conditions such as clouds, fog, rain, and snow, providing all-weather, all-time monitoring capabilities. By combining millimeter-wave radar with intelligent clustering algorithms, employing methods such as K-means++ and Mahalanobis distance, it can accurately classify and identify foreign objects and icing on transmission lines. This system can adaptively adjust distance metrics and clustering methods, improving the accuracy of identifying different targets (such as bird nests, plastic bags, and icing) in complex scenarios, reducing false positives and false negatives in traditional methods. Simultaneously, the system enables automated real-time monitoring, reducing the need for manual intervention and significantly improving monitoring efficiency and the safety of transmission line maintenance. It overcomes the shortcomings of existing technologies in monitoring accuracy, efficiency, and all-weather monitoring, enhancing the reliability and intelligence level of transmission line monitoring systems.

[0011] To achieve the above objectives, the present invention adopts the following technical solution:

[0012] A millimeter-wave radar power transmission line clustering system includes a millimeter-wave radar module, a signal processing module, a clustering and classification module, and a risk warning module.

[0013] The millimeter-wave radar module is electrically connected to an external antenna and includes a millimeter-wave radar antenna, a receiving module, and a transmitter. The signal processing module is electrically connected to the millimeter-wave radar module and includes an amplifier, a filter, and a signal preprocessing circuit. The clustering and classification module is electrically connected to the signal processing module and includes a K-means++ algorithm module, a DBSCAN algorithm module, and an analog-to-digital converter (ADC). The risk warning module is electrically connected to the clustering and classification module and internally includes a risk assessment calculation circuit, an alarm mechanism module, a storage unit, and a wireless communication module. The risk warning module is connected to a remote monitoring center via a signal interface.

[0014] A millimeter-wave radar power transmission line clustering method, based on the aforementioned millimeter-wave radar power transmission line clustering system, includes the following steps:

[0015] Step 1: Data Acquisition. The millimeter-wave radar module is based on the high-frequency band and continuously monitors foreign objects or ice on the power transmission line under adverse weather conditions. The millimeter-wave radar module emits millimeter-wave signals and receives the echo signals reflected back from the target.

[0016] Step 2: Signal preprocessing. The received radar echo signal is sent to the signal processing module for denoising, enhancement, and feature extraction. The signal processing module uses amplifiers and filters to process the signal to remove environmental noise and enhance the target echo signal.

[0017] Step 3: Analog signal to digital signal conversion. The clustering and classification module converts the processed analog signal into a digital signal through an analog-to-digital converter (ADC).

[0018] Step 4: Feature extraction. The signal processing module extracts important features from the digital signal, including signal strength, time delay, and spectral features. Feature extraction reduces the dimensionality of the original data and retains the features that have the greatest impact on the clustering results.

[0019] Step 5: Cluster analysis. The clustering and classification module uses the K-means++ algorithm for preliminary clustering, dividing the data into different clusters based on features such as intensity and time delay. Then, the DBSCAN algorithm is used to further process the clustering results, identify low-density regions, and filter out noise.

[0020] Step 6: Precise Classification. The clustering and classification module uses Mahalanobis distance to accurately classify the clustering results, further distinguishing the differences between foreign objects and ice cover features, and comparing the statistical characteristics of each cluster with known categories to determine the target type. By evaluating the similarity and differences between different clusters through Mahalanobis distance, the system can effectively distinguish and classify data in high-dimensional space, improving the accuracy of classification.

[0021] Step 7, Risk Assessment and Early Warning, transmits the classification results to the risk early warning module, assesses the risk level of each target, and conducts a comprehensive analysis in conjunction with historical data; once a high-risk target is detected, the system immediately triggers the alarm mechanism and records detailed information; through real-time risk assessment, potential dangers are promptly identified, ensuring the safety of transmission lines and reducing the possibility of accidents.

[0022] Furthermore, the echo signals collected in step 1 contain features such as the target's location and intensity, providing basic information for subsequent data processing and analysis.

[0023] Furthermore, in step 2 of the signal preprocessing, the filter is a Butterworth, Chebyshev, or elliptic filter, and the transfer function of the low-pass filter is:

[0024]

[0025] In the formula, H(f) is the frequency response of the filter; f is the frequency; f c The cutoff frequency is n; n is the filter order.

[0026] Apply a filter to perform convolution filtering on the signal:

[0027] x'(t)=∫x(τ)·h(t-τ)dτ

[0028] In the formula, x'(t) is the output signal after filtering; x(τ) is the input signal; h(t-τ) is the impulse response of the filter; τ is the integral variable, representing a delay in time; dτ is the infinitesimal element of the integral, representing a small change in the time variable τ.

[0029] The low-frequency portion of the signal is retained after filtering.

[0030] Furthermore, in step 2 of the signal preprocessing, wavelet denoising is used to remove environmental noise, including the following steps:

[0031] Step 2.1 Signal Decomposition: Perform discrete wavelet transform on the input signal x(t) to decompose it into wavelet coefficients at different scales, including approximate component A and detail component D; the formula is:

[0032]

[0033] In the formula, c j,k The wavelet coefficients are denoted as x(t); x(t) is the input signal. Let be the mother wavelet function of the j-th layer, and k be the translation factor;

[0034] Step 2.2 Thresholding: Apply soft or hard thresholding rules to the detail component D to remove noise.

[0035] Soft threshold:

[0036]

[0037] Hard threshold:

[0038]

[0039] In the formula, D' is the detail component after thresholding; sign(D) represents the sign function of D; |D| is the amplitude of the detail component; λ is the threshold value, which is set according to the noise level or signal-to-noise ratio; D represents the detail component of the input signal or image.

[0040] Step 2.3 Signal Reconstruction: Perform inverse wavelet transform using the denoised wavelet coefficients to reconstruct the denoised signal.

[0041]

[0042] In the formula, x'(t) is the signal generated based on the denoised wavelet coefficients; c j,k These are wavelet coefficients; Let be the mother wavelet function of the j-th layer, k be the translation factor, and t be time. This indicates summation over all scales j and positions k.

[0043] Furthermore, in step 2, during signal preprocessing, the target echo signal is enhanced, including the following steps:

[0044] Step 201 Adaptive Gain Control (AGC): AGC enhances weak echo signals by dynamically adjusting the signal gain; the gain function is:

[0045]

[0046] In the formula, G(t) is the gain; α is the gain adjustment coefficient; x(t) is the input signal; |x(t)| is the amplitude of the signal x(t);

[0047] Enhanced signal:

[0048] x′(t)=G(t)·x(t)

[0049] In the formula, x'(t) is the enhanced signal; G(t) is the gain; and x(t) is the input signal.

[0050] Optimize the dynamic range to ensure that the enhanced signal does not exceed the system's dynamic range;

[0051] Step 202 Fourier transform enhancement: Enhance specific frequency components of the echo signal through frequency domain analysis;

[0052] The signal is transformed from the time domain to the frequency domain through Fourier transform;

[0053] X(f)=∫x(t)e -j2πft dt

[0054] In the formula, X(f) is the frequency domain signal; x(t) is the input signal; e -j2πft is a complex exponential function; e is the base of the natural logarithm; j is the imaginary unit; f is the frequency; t is time;

[0055] Enhance specific frequency components:

[0056] X'(f) = W(f)·X(f)

[0057] In the formula, X'(f) is the enhanced frequency domain signal; W(f) is the gain function; and X(f) is the frequency domain signal.

[0058] The formula for calculating the gain function W(f) is as follows:

[0059]

[0060] In the formula, f is the frequency; exp() is the exponential function; f0 is the target frequency; σ is the bandwidth;

[0061] The enhanced signal is converted back from the frequency domain to the time domain using the inverse Fourier transform:

[0062] x'(t)=∫X'(f)e j2πft df

[0063] In the formula, x'(t) is the enhanced signal; X'(f) is the enhanced frequency domain signal e. j2πft is a complex exponential function; f is the frequency.

[0064] Furthermore, in step 5, during cluster analysis, the K-means++ algorithm is applied for preliminary clustering, dividing the data into different clusters based on features such as intensity and time delay. This includes the following steps:

[0065] Step 5.1: Randomly select a data point as the first cluster center;

[0066] Step 5.2: Calculate the squared distance between each data point and the nearest selected cluster center to generate a probability distribution;

[0067] Step 5.3: Select the next cluster center based on the probability distribution of distance, with points that are farther away being assigned higher weights;

[0068] Step 5.4: Repeat steps 5.2 and 5.3 until k initial cluster centers are selected;

[0069] Step 5.5: Use the K-means algorithm to optimize the clustering results according to the preset iterative formula; the calculation formula of the K-means algorithm is as follows:

[0070]

[0071] In the formula, μ j It is the centroid of the j-th cluster; |C j | represents the number of points in the cluster; This indicates that it belongs to the j-th cluster C. j All sample points x i Summation; x i For the j-th cluster C j The sample points in the data.

[0072] Furthermore, in step 5, during the cluster analysis, the DBSCAN algorithm can identify dense areas in the data and identify discrete points as noise, effectively handling outliers in millimeter-wave data and improving monitoring performance under adverse weather conditions.

[0073] The DBSCAN algorithm is used to process clustering results, identify low-density regions, and filter noise, including the following steps:

[0074] Step 501: Randomly select an unvisited data point from the dataset;

[0075] Step 502: Calculate the number of all data points in the neighborhood of the data point with radius ε;

[0076] Step 503: MinPts is the minimum number of neighboring points to form a core point; if the number of neighboring points is greater than or equal to MinPts, then the point is marked as a core point and a new cluster is created.

[0077] Step 504: If the number of neighboring points is less than MinPts, then the point is marked as a noise point;

[0078] Step 505: Check all neighboring points of the core point. If a point in the neighborhood is also a core point, add the neighboring points of the core point to the same cluster. Expand the cluster in this way until no more points can be added.

[0079] Step 506: Repeat steps 501 to 503 to continue selecting unvisited data points until all points have been visited and marked as belonging to a cluster or noise.

[0080] Furthermore, in the precise classification process of step 6, Mahalanobis distance is introduced as a new distance metric. Mahalanobis distance takes into account the correlation between various features and can better handle non-spherical data distributions. By calculating the Mahalanobis distance between each data point and the cluster center, the classification accuracy of foreign objects and ice accretion is effectively improved.

[0081] The formula for calculating Mahalanobis distance is as follows:

[0082]

[0083] In the formula, d M (x,μ) represents the Mahalanobis distance between the sample point vector x and the mean vector μ; x is the sample point vector to be classified; μ is the mean vector of the data, which is the cluster center or the mean of the class; Σ is the covariance matrix of the data; -1 It is the inverse of the covariance matrix; (x-μ) T This represents the transpose of the difference vector between x and the mean μ.

[0084] Furthermore, in step 7, risk assessment and early warning, the specific implementation method of the risk early warning module is as follows:

[0085] Step 7.1 Data Reception and Processing: Data reception and processing is the starting point of the risk warning module; the risk warning module continuously receives analysis data from the clustering and classification modules through an interface. The analysis data includes:

[0086] The risk warning module verifies and preprocesses data, including target type, size, location, and reflection characteristics, to remove potential noise and errors.

[0087] Step 7.2 Risk Assessment: The early warning module analyzes the current situation using a risk assessment algorithm based on the received data; the risk assessment algorithm is as follows:

[0088] R = ω1·S + ω2·P + ω3·E

[0089] In the formula, R represents the generated risk score; S represents the target size; P represents the relative distance between the target location and the transmission line; E represents the environmental conditions; and ω1, ω2, and ω3 represent the weighting coefficients.

[0090] Risk scores are derived based on risk assessment algorithms, and the assessment results are represented numerically.

[0091] Step 7.3 Risk Level Classification: After generating the risk score, the early warning module classifies the risk level, including:

[0092] High risk: Large bird nests may be covered in ice, requiring immediate action;

[0093] Medium risk: Generate a warning message to notify maintenance personnel to conduct a thorough inspection;

[0094] Low risk: Recorded, no proactive notification will be given at this time;

[0095] Step 7.4 Report Generation and Historical Data Management: The early warning module can automatically generate detailed detection reports based on the detected targets and risk assessment results. The report will include: the specific location of the target, type information, time of occurrence, risk level, and recommended handling measures. The report will be output in PDF or other electronic formats for management personnel to view.

[0096] In addition, the early warning module stores all early warning records in the database, forming a historical data management system. The historical data management system supports storage and retrieval. Based on the historical data management system, maintenance personnel can analyze the line conditions at different time periods and optimize inspection and maintenance strategies.

[0097] Compared with existing technologies, this millimeter-wave radar power transmission line clustering system and method has the following advantages:

[0098] This invention significantly improves the detection efficiency and accuracy of foreign objects and icing on power transmission lines by introducing millimeter-wave radar technology combined with intelligent clustering algorithms. Millimeter-wave radar features high resolution and strong penetration capabilities, enabling stable operation in adverse weather conditions such as fog, rain, and snow, unaffected by lighting or environmental factors, ensuring all-weather, all-time monitoring and safeguarding the safety of power transmission lines.

[0099] Compared with traditional optical sensors or low-frequency radar, this invention is more accurate and can accurately distinguish between foreign objects such as bird nests and plastic bags and targets such as ice through backscattered signals collected by millimeter-wave radar, effectively reducing missed detections and false judgments and improving the reliability of monitoring.

[0100] Regarding intelligent algorithms, this invention employs a combination of the K-means++ algorithm and Mahalanobis distance to adaptively adjust classification criteria, handle non-spherical data distributions, and improve the accuracy of classifying foreign objects and ice accumulation. The introduction of the DBSCAN algorithm further enhances the system's ability to handle noise and abnormal data, ensuring the stability of monitoring results in complex scenarios.

[0101] This invention not only reduces the need for manual inspections and lowers maintenance costs, but also improves the safety and operational efficiency of power transmission lines. In particular, it can provide efficient and accurate monitoring even under extreme weather conditions, resulting in significant social and economic benefits. Attached Figure Description

[0102] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0103] Figure 1 This is a schematic diagram of the overall structure of the millimeter-wave radar power transmission line clustering system of the present invention;

[0104] Figure 2 This is a flowchart of the millimeter-wave radar power transmission line clustering method of the present invention;

[0105] Figure 3 This is a flowchart illustrating the workflow of the risk warning module in the millimeter-wave radar power transmission line clustering system of the present invention. Detailed Implementation

[0106] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0107] Example 1

[0108] To address several key issues in existing power transmission line monitoring technologies, this embodiment provides a millimeter-wave radar power transmission line clustering system, such as... Figure 1 As shown, the millimeter-wave radar power transmission line clustering system includes a millimeter-wave radar module, a signal processing module, a clustering and classification module, and a risk warning module.

[0109] The system includes a millimeter-wave radar module electrically connected to an external antenna, comprising a millimeter-wave radar antenna, a receiver module, and a transmitter; a signal processing module electrically connected to the millimeter-wave radar module, comprising an amplifier, a filter, and a signal preprocessing circuit; a clustering and classification module electrically connected to the signal processing module, comprising a K-means++ algorithm module, a DBSCAN algorithm module, and an analog-to-digital converter (ADC); and a risk warning module electrically connected to the clustering and classification module, internally comprising a risk assessment calculation circuit, an alarm mechanism module, a storage unit, and a wireless communication module. The risk warning module is connected to a remote monitoring center via a signal interface.

[0110] Millimeter-wave radar module: The millimeter-wave radar module is responsible for continuously acquiring echo signals from transmission lines. This module is electrically connected to an external antenna, which transmits the echo signals to the receiving module of the millimeter-wave radar module. Operating at high frequencies, the millimeter-wave radar module possesses high resolution and strong penetration capabilities, enabling it to monitor for foreign objects or icing on transmission lines even in adverse weather conditions (such as rain, snow, and fog). The signal transmission path is: radar antenna → millimeter-wave radar module → signal processing module.

[0111] Signal Processing Module: The received radar echo signal is first transmitted to the signal processing module, which performs signal preprocessing, including noise reduction, echo enhancement, and feature extraction. The core circuitry of the signal processing module includes amplifiers, filters, and signal preprocessing circuitry to enhance the target echo signal and filter out interference signals. The processed data is electrically connected to the clustering and classification module via an interface, transmitting the cleaned signal to the clustering and classification module.

[0112] The clustering and classification module includes the K-means++ algorithm module, the DBSCAN algorithm module, and an analog-to-digital converter (ADC). The K-means++ algorithm is an improved K-means clustering algorithm, primarily used for initial clustering. The DBSCAN algorithm (Density-Based Spatial Clustering of Applications with Noise) is a density-based spatial clustering algorithm that effectively handles noisy data, used for initial clustering and identification of density anomalies, respectively. The clustering and classification module converts analog signals to digital signals via the ADC for subsequent digital processing. The analog signals output from the signal processing module are converted to digital signals and input into the clustering algorithms. The K-means++ algorithm performs clustering analysis based on the echo characteristics of the targets, such as intensity and time delay. The DBSCAN algorithm filters out anomalous noise. Finally, Mahalanobis distance is used to further distinguish the characteristic differences between foreign objects and ice cover, thus achieving accurate classification. Mahalanobis distance is a method of measuring the distance between different sample points, particularly suitable for processing data with different dimensions of correlation. The output data of this clustering and classification module is transmitted to the risk warning module through the communication interface.

[0113] Risk Warning Module: This module contains a risk assessment calculation circuit, an alarm mechanism module, a storage unit, and a wireless communication module. It receives the output from the clustering and classification modules and assesses the risk level of the target based on the classification results. The module uses its internal calculation circuitry to comprehensively analyze the hazard of the target in conjunction with historical data. The risk warning module connects to the remote monitoring center via a signal interface. When a high-risk target is detected, the module immediately triggers the alarm mechanism and sends a warning signal to the remote monitoring center. Simultaneously, the system generates a detailed detection report, recording the target type, location, and risk level, and transmits the report to the maintenance team via the wireless communication module. Furthermore, the module's internal storage unit stores all historical data for future analysis.

[0114] The workflow of this millimeter-wave radar power transmission line clustering system is as follows: The millimeter-wave radar module first collects reflected signals from the power transmission line, and then preprocesses them through the signal processing module. The processed data is input into the clustering and classification module, where feature extraction and classification analysis are performed using algorithms. Finally, the data is transmitted to the risk warning module for assessment and early warning. Each module is electrically connected to form a complete closed-loop system, enabling automatic and real-time monitoring and early warning of foreign objects and icing on power transmission lines in real-world environments.

[0115] The high resolution of millimeter-wave radar ensures effective identification of small targets on power transmission lines, while the noise reduction and enhancement functions of the signal processing module significantly reduce the interference of environmental noise on the system. Intelligent algorithms in the clustering and classification module further guarantee accurate classification of foreign objects and icing in complex scenarios, with Mahalanobis distance playing a crucial role in distinguishing between foreign objects and icing. The risk warning module comprehensively analyzes system output data, generates real-time warning signals, and records and analyzes historical data, greatly improving the safety and monitoring efficiency of power transmission lines.

[0116] Example 2

[0117] Based on the millimeter-wave radar transmission line clustering system provided in Embodiment 1, this embodiment provides a millimeter-wave radar transmission line clustering method, such as... Figure 2 As shown, the millimeter-wave radar transmission line clustering method includes the following steps:

[0118] Step 1: Data Acquisition. The millimeter-wave radar module, operating at high frequencies, continuously monitors for foreign objects or icing on power transmission lines under adverse weather conditions. The module transmits millimeter-wave signals and receives echo signals reflected from the targets. The acquired echo signals contain information such as the target's location and intensity, providing fundamental data for subsequent processing and analysis.

[0119] Step 2: Signal preprocessing. The received radar echo signal is sent to the signal processing module for denoising, enhancement, and feature extraction. The signal processing module uses amplifiers and filters to process the signal to remove environmental noise and enhance the target echo signal.

[0120] During signal preprocessing, the filter is a Butterworth, Chebyshev, or elliptic filter, and the transfer function of the low-pass filter is:

[0121]

[0122] In the formula, H(f) is the frequency response of the filter; f is the frequency; f c The cutoff frequency is n; n is the filter order.

[0123] Apply a filter to perform convolution filtering on the signal:

[0124] x'(t)=∫x(τ)·h(t-τ)dτ

[0125] In the formula, x'(t) is the output signal after filtering; x(τ) is the input signal; h(t-τ) is the impulse response of the filter; τ is the integral variable, representing a delay in time; dτ is the infinitesimal element of the integral, representing a small change in the time variable τ.

[0126] The low-frequency portion of the signal is retained after filtering.

[0127] Furthermore, during signal preprocessing, wavelet denoising is used to remove environmental noise, including the following steps:

[0128] Step 2.1 Signal Decomposition: Perform discrete wavelet transform on the input signal x(t) to decompose it into wavelet coefficients at different scales, including approximate component A and detail component D; the formula is:

[0129]

[0130] In the formula, c j,k The wavelet coefficients are denoted as x(t); x(t) is the input signal. Let be the mother wavelet function of the j-th layer, and k be the translation factor;

[0131] Step 2.2 Thresholding: Apply soft or hard thresholding rules to the detail component D to remove noise.

[0132] Soft threshold:

[0133]

[0134] Hard threshold:

[0135]

[0136] In the formula, D' is the detail component after thresholding; sign(D) represents the sign function of D; |D| is the amplitude of the detail component; λ is the threshold value, which is set according to the noise level or signal-to-noise ratio; D represents the detail component of the input signal or image.

[0137] Step 2.3 Signal Reconstruction: Perform inverse wavelet transform using the denoised wavelet coefficients to reconstruct the denoised signal.

[0138]

[0139] In the formula, x'(t) is the signal generated based on the denoised wavelet coefficients; c j,k These are wavelet coefficients; Let be the mother wavelet function of the j-th layer, k be the translation factor, and t be time. This indicates summation over all scales j and positions k.

[0140] Furthermore, during signal preprocessing, the target echo signal is enhanced, including the following steps:

[0141] Step 201 Adaptive Gain Control (AGC): AGC enhances weak echo signals by dynamically adjusting the signal gain; the gain function is:

[0142]

[0143] In the formula, G(t) is the gain; α is the gain adjustment coefficient; x(t) is the input signal; |x(t)| is the amplitude of the signal x(t);

[0144] Enhanced signal:

[0145] x′(t)=G(t)·x(t)

[0146] In the formula, x'(t) is the enhanced signal; G(t) is the gain; and x(t) is the input signal.

[0147] Optimize the dynamic range to ensure that the enhanced signal does not exceed the system's dynamic range;

[0148] Step 202 Fourier transform enhancement: Enhance specific frequency components of the echo signal through frequency domain analysis;

[0149] The signal is transformed from the time domain to the frequency domain through Fourier transform;

[0150] X(f)=∫x(t)e -j2πft dt

[0151] In the formula, X(f) is the frequency domain signal; x(t) is the input signal; e -j2πft is a complex exponential function; e is the base of the natural logarithm; j is the imaginary unit; f is the frequency; t is time;

[0152] Enhance specific frequency components:

[0153] X'(f) = W(f)·X(f)

[0154] In the formula, X'(f) is the enhanced frequency domain signal; W(f) is the gain function; and X(f) is the frequency domain signal.

[0155] The formula for calculating the gain function W(f) is as follows:

[0156]

[0157] In the formula, f is the frequency; exp() is the exponential function; f0 is the target frequency; σ is the bandwidth;

[0158] The enhanced signal is converted back from the frequency domain to the time domain using the inverse Fourier transform:

[0159] x'(t)=∫X'(f)e j2πft df

[0160] In the formula, x'(t) is the enhanced signal; X'(f) is the enhanced frequency domain signal e. j2πft is a complex exponential function; f is the frequency.

[0161] Step 3: Analog signal to digital signal conversion. The clustering and classification module converts the processed analog signal into a digital signal through an analog-to-digital converter (ADC).

[0162] Step 4: Feature extraction. The signal processing module extracts important features from the digital signal, including signal strength, time delay, and spectral features. Feature extraction reduces the dimensionality of the original data and retains the features that have the greatest impact on the clustering results.

[0163] Step 5: Cluster analysis. The clustering and classification module uses the K-means++ algorithm for preliminary clustering, dividing the data into different clusters based on features such as intensity and time delay. Then, the DBSCAN algorithm is used to further process the clustering results, identify low-density regions, and filter out noise.

[0164] In the clustering analysis process, the K-means++ algorithm is applied for preliminary clustering, dividing the data into different clusters based on features such as intensity and time delay. This includes the following steps:

[0165] Step 5.1: Randomly select a data point as the first cluster center;

[0166] Step 5.2: Calculate the squared distance between each data point and the nearest selected cluster center to generate a probability distribution;

[0167] Step 5.3: Select the next cluster center based on the probability distribution of distance, with points that are farther away being assigned higher weights;

[0168] Step 5.4: Repeat steps 5.2 and 5.3 until k initial cluster centers are selected;

[0169] Step 5.5: Use the K-means algorithm to optimize the clustering results according to the preset iterative formula; the calculation formula of the K-means algorithm is as follows:

[0170]

[0171] In the formula, μ j It is the centroid of the j-th cluster; |C j | represents the number of points in the cluster; This indicates that it belongs to the j-th cluster C. j All sample points x i Summation; x i For the j-th cluster C j The sample points in the data.

[0172] During cluster analysis, the DBSCAN algorithm can identify dense areas in the data and identify discrete points as noise, effectively handling outliers in millimeter-wave data and improving monitoring performance under adverse weather conditions.

[0173] The DBSCAN algorithm is used to process clustering results, identify low-density regions, and filter noise, including the following steps:

[0174] Step 501: Randomly select an unvisited data point from the dataset;

[0175] Step 502: Calculate the number of all data points in the neighborhood of the data point with radius .

[0176] Step 503: MinPts is the minimum number of neighboring points to form a core point; if the number of neighboring points is greater than or equal to MinPts, then the point is marked as a core point and a new cluster is created.

[0177] Step 504: If the number of neighboring points is less than 1, then the point is marked as a noise point;

[0178] Step 505: Check all neighboring points of the core point. If a point in the neighborhood is also a core point, add the neighboring points of the core point to the same cluster. Expand the cluster in this way until no more points can be added.

[0179] Step 506: Repeat steps 501 to 503 to continue selecting unvisited data points until all points have been visited and marked as belonging to a cluster or noise.

[0180] Step 6: Precise Classification. The clustering and classification module uses Mahalanobis distance to accurately classify the clustering results, further distinguishing the differences between foreign objects and ice cover features. It also compares the statistical characteristics of each cluster with known categories to determine the target type. By evaluating the similarity and differences between different clusters through Mahalanobis distance, the system can effectively distinguish and classify data in high-dimensional space, improving the accuracy of classification.

[0181] In the process of accurate classification, Mahalanobis distance is introduced as a new distance metric. Mahalanobis distance takes into account the correlation between various features and can better handle non-spherical data distribution. By calculating the Mahalanobis distance between each data point and the cluster center, the classification accuracy of foreign objects and ice accretion is effectively improved.

[0182] The formula for calculating Mahalanobis distance is as follows:

[0183]

[0184] In the formula, d M (x,μ) represents the Mahalanobis distance between the sample point vector x and the mean vector μ; x is the sample point vector to be classified; μ is the mean vector of the data, which is the cluster center or the mean of the class; Σ is the covariance matrix of the data; -1 It is the inverse of the covariance matrix; (x-μ) TThis represents the transpose of the difference vector between x and the mean μ.

[0185] Step 7, Risk Assessment and Early Warning, transmits the classification results to the risk early warning module, assesses the risk level of each target, and conducts a comprehensive analysis in conjunction with historical data; once a high-risk target is detected, the system immediately triggers the alarm mechanism and records detailed information; through real-time risk assessment, potential dangers are promptly identified, ensuring the safety of transmission lines and reducing the possibility of accidents.

[0186] In risk assessment and early warning, the specific implementation method of the risk early warning module is as follows:

[0187] Step 7.1 Data Reception and Processing: Data reception and processing is the starting point of the risk warning module; the risk warning module continuously receives analysis data from the clustering and classification modules through an interface. The analysis data includes:

[0188] The risk warning module verifies and preprocesses data, including target type, size, location, and reflection characteristics, to remove potential noise and errors.

[0189] Step 7.2 Risk Assessment: (e.g.) Figure 3 As shown, the early warning module analyzes the current situation using a risk assessment algorithm based on the received data; the risk assessment algorithm is as follows:

[0190] R = ω1·S + ω2·P + ω3·E

[0191] In the formula, R represents the generated risk score; S represents the target size; P represents the relative distance between the target location and the transmission line; E represents the environmental conditions; and ω1, ω2, and ω3 represent the weighting coefficients.

[0192] Risk scores are derived based on risk assessment algorithms, and the assessment results are represented numerically.

[0193] Step 7.3 Risk Level Classification: After generating the risk score, the early warning module classifies the risk level, including:

[0194] High risk: Large bird nests may be covered in ice, requiring immediate action;

[0195] Medium risk: Generate a warning message to notify maintenance personnel to conduct a thorough inspection;

[0196] Low risk: Recorded, no proactive notification will be given at this time;

[0197] Step 7.4 Report Generation and Historical Data Management: The early warning module can automatically generate detailed detection reports based on the detected targets and risk assessment results. The report will include: the specific location of the target, type information, time of occurrence, risk level, and recommended handling measures. The report will be output in PDF or other electronic formats for management personnel to view.

[0198] In addition, the early warning module stores all early warning records in the database, forming a historical data management system. The historical data management system supports storage and retrieval. Based on the historical data management system, maintenance personnel can analyze the line conditions at different time periods and optimize inspection and maintenance strategies.

[0199] In summary, the method and system for foreign object identification and icing monitoring of transmission lines based on millimeter-wave radar can solve several key problems in existing transmission line monitoring technologies. Millimeter-wave radar has high resolution and strong penetration capabilities, and can operate stably under complex weather conditions such as clouds, fog, rain, and snow, providing all-weather, all-time monitoring capabilities. By combining millimeter-wave radar with intelligent clustering algorithms, and employing methods such as K-means++ and Mahalanobis distance, foreign objects and icing on transmission lines can be accurately classified and identified.

[0200] This system can adaptively adjust distance metrics and clustering methods to improve the accuracy of identifying different targets (such as bird nests, plastic bags, ice, etc.) in complex scenarios, reducing false detections and missed detections in traditional methods. At the same time, the system can achieve automated real-time monitoring, reducing the need for manual intervention and significantly improving monitoring efficiency and the safety of power transmission line maintenance. It also addresses shortcomings in monitoring accuracy, efficiency, and all-weather monitoring, enhancing the reliability and intelligence level of the power transmission line monitoring system.

[0201] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A millimeter-wave radar transmission line clustering method, characterized in that: The millimeter-wave radar transmission line clustering method is based on a millimeter-wave radar transmission line clustering system, which includes a millimeter-wave radar module, a signal processing module, a clustering and classification module, and a risk warning module. The millimeter-wave radar module is electrically connected to an external antenna and includes a millimeter-wave radar antenna, a receiving module, and a transmitter. The signal processing module is electrically connected to the millimeter-wave radar module and includes an amplifier, a filter, and a signal preprocessing circuit. The clustering and classification module is electrically connected to the signal processing module and includes a K-means++ algorithm module, a DBSCAN algorithm module, and an analog-to-digital converter (ADC). The risk warning module is electrically connected to the clustering and classification module and internally includes a risk assessment calculation circuit, an alarm mechanism module, a storage unit, and a wireless communication module. The risk warning module is connected to a remote monitoring center via a signal interface. The millimeter-wave radar power transmission line clustering method includes the following steps: Step 1: Data Acquisition. The millimeter-wave radar module is based on the high-frequency band and continuously monitors foreign objects or ice on the power transmission line under adverse weather conditions. The millimeter-wave radar module emits millimeter-wave signals and receives the echo signals reflected back from the target. Step 2: Signal preprocessing. The received radar echo signal is sent to the signal processing module for denoising, enhancement, and feature extraction. The signal processing module uses amplifiers and filters to process the signal to remove environmental noise and enhance the target echo signal. Step 3: Analog signal to digital signal conversion. The clustering and classification module converts the processed analog signal into a digital signal through an analog-to-digital converter (ADC). Step 4: Feature extraction. The signal processing module extracts important features from the digital signal, including signal strength, time delay, and spectral features. Feature extraction reduces the dimensionality of the original data and retains the features that have the greatest impact on the clustering results. Step 5: Cluster analysis. The clustering and classification module uses the K-means++ algorithm for preliminary clustering, dividing the data into different clusters based on features such as intensity and time delay. Then, the DBSCAN algorithm is used to further process the clustering results, identify low-density regions, and filter out noise. Step 6: Precise Classification. The clustering and classification module uses Mahalanobis distance to accurately classify the clustering results, further distinguishing the differences between foreign objects and ice cover features, and comparing the statistical characteristics of each cluster with known categories to determine the target type. By evaluating the similarity and differences between different clusters through Mahalanobis distance, the system can effectively distinguish and classify data in high-dimensional space, improving the accuracy of classification. Step 7, Risk Assessment and Early Warning, transmits the classification results to the risk early warning module, assesses the risk level of each target, and conducts a comprehensive analysis in conjunction with historical data; once a high-risk target is detected, the system immediately triggers the alarm mechanism and records detailed information; through real-time risk assessment, potential dangers are promptly identified, ensuring the safety of transmission lines and reducing the possibility of accidents.

2. The millimeter-wave radar transmission line clustering method according to claim 1, characterized in that: The echo signals collected in step 1 contain the target's location and intensity information features, providing basic information for subsequent data processing and analysis.

3. The millimeter-wave radar transmission line clustering method according to claim 1, characterized in that: In step 2, during signal preprocessing, the filter is a Butterworth, Chebyshev, or elliptic filter, and the transfer function of the low-pass filter is: ; In the formula, This refers to the frequency response of the filter. For frequency; The cutoff frequency; The filter order; Apply a filter to perform convolution filtering on the signal: ; In the formula, This is the output signal after filtering; For input signals; This represents the impulse response of the filter; Let be the integral variable, representing a delay in time; The infinitesimal element for integration represents the time variable. The minute changes; The low-frequency portion of the signal is retained after filtering.

4. The millimeter-wave radar transmission line clustering method according to claim 1, characterized in that: In step 2, during signal preprocessing, wavelet denoising is used to remove environmental noise, including the following steps: Step 2.1 Signal decomposition: Decompose the input signal Perform discrete wavelet transform to decompose it into wavelet coefficients of different scales, including approximate components. and details The formula is: ; In the formula, These are wavelet coefficients; For input signals; For the first The mother wavelet function of the layer, The translation factor; Step 2.2 Thresholding: For detail components Apply soft or hard thresholding rules to remove noise: Soft threshold: ; Hard threshold: ; In the formula, These are the detail components after thresholding. The sign function representing D; For the magnitude of the detail component; The threshold value is set based on the noise level or signal-to-noise ratio. Represents the detail components of the input signal or image; Step 2.3 Signal Reconstruction: Perform inverse wavelet transform using the denoised wavelet coefficients to reconstruct the denoised signal. ; In the formula, The signal is generated based on the denoised wavelet coefficients; These are wavelet coefficients; For the first The mother wavelet function of the layer, The translation factor; For time; Indicates all scales and location Perform summation.

5. The millimeter-wave radar transmission line clustering method according to claim 1, characterized in that: Step 2, signal preprocessing, enhances the target echo signal, including the following steps: Step 201 Adaptive Gain Control (AGC): AGC enhances weak echo signals by dynamically adjusting the signal gain; the gain function is: ; In the formula, For gain; This is the gain adjustment coefficient; For input signals; For signal The range; Enhanced signal: ; In the formula, The enhanced signal; For gain; For input signals; Optimize the dynamic range to ensure that the enhanced signal does not exceed the system's dynamic range; Step 202 Fourier transform enhancement: Enhance specific frequency components of the echo signal through frequency domain analysis; The signal is transformed from the time domain to the frequency domain through Fourier transform; ; In the formula, It is a frequency domain signal; For input signals; It is a complex exponential function; is the base of the natural logarithm; The imaginary unit; For frequency; For time; Enhance specific frequency components: ; In the formula, The enhanced frequency domain signal; It is the gain function; It is a frequency domain signal; Gain function The calculation formula is as follows: ; In the formula, For frequency; It is an exponential function; For the target frequency; For bandwidth; The enhanced signal is converted back from the frequency domain to the time domain using the inverse Fourier transform: ; In the formula, The enhanced signal; For the enhanced frequency domain signal It is a complex exponential function; For frequency.

6. The millimeter-wave radar transmission line clustering method according to claim 1, characterized in that: Step 5 involves cluster analysis, where the K-means++ algorithm is applied for initial clustering. Based on features such as intensity and time delay, the data is divided into different clusters, including the following steps: Step 5.1: Randomly select a data point as the first cluster center; Step 5.2: Calculate the squared distance between each data point and the nearest selected cluster center to generate a probability distribution; Step 5.3: Select the next cluster center based on the probability distribution of distance, with points that are farther away being assigned higher weights; Step 5.4: Repeat steps 5.2 and 5.3 until k initial cluster centers are selected; Step 5.5: Use the K-means algorithm to optimize the clustering results according to the preset iterative formula; the calculation formula of the K-means algorithm is as follows: ; In the formula, It is the first The centroid of the cluster; It is the number of points in the cluster; Indicates belonging to the first Cluster All sample points Perform summation; For the first Cluster The sample points in the data.

7. The millimeter-wave radar transmission line clustering method according to claim 1, characterized in that: In step 5, during the cluster analysis, the DBSCAN algorithm can identify dense areas in the data and identify discrete points as noise, effectively handling outliers in millimeter-wave data and improving monitoring performance under adverse weather conditions. The DBSCAN algorithm is used to process clustering results, identify low-density regions, and filter noise, including the following steps: Step 501: Randomly select an unvisited data point from the dataset; Step 502, calculate the radius of the data point. The number of all data points in the neighborhood; Step 503, The minimum number of neighboring points required to form the core point; if the number of neighboring points is greater than or equal to... If so, the point is marked as a core point, and a new cluster is created. Step 504, if the number of neighboring points is less than If so, the point is marked as a noise point; Step 505: Check all neighboring points of the core point. If a point in the neighborhood is also a core point, add the neighboring points of the core point to the same cluster. Expand the cluster in this way until no more points can be added. Step 506: Repeat steps 501 to 503 to continue selecting unvisited data points until all points have been visited and marked as belonging to a cluster or noise.

8. The millimeter-wave radar transmission line clustering method according to claim 1, characterized in that: In step 6, during the precise classification process, Mahalanobis distance was introduced as a new distance metric. Mahalanobis distance takes into account the correlation between various features and can better handle non-spherical data distributions. By calculating the Mahalanobis distance between each data point and the cluster center, the classification accuracy of foreign objects and ice accretion is effectively improved. The formula for calculating Mahalanobis distance is as follows: ; In the formula, Represents the sample point vector With mean vector Mahalanobis distance between them; It is the vector of sample points to be classified; It is the mean vector of the data, which represents the center of the cluster or the mean of the class; It is the covariance matrix of the data; It is the inverse of the covariance matrix; express with the mean The transpose of the difference vector.

9. The millimeter-wave radar transmission line clustering method according to claim 2, characterized in that: In step 7, risk assessment and early warning, the specific implementation method of the risk early warning module is as follows: Step 7.1 Data Reception and Processing: Data reception and processing is the starting point of the risk warning module; The risk warning module continuously receives analysis data from the clustering and classification modules through an interface. The analysis data includes: target type, target size, target location, and reflection characteristics. The risk warning module performs data verification and preprocessing to remove possible noise and erroneous data. Step 7.2 Risk Assessment: The early warning module analyzes the current situation using a risk assessment algorithm based on the received data; the risk assessment algorithm is as follows: ; In the formula, To generate a risk score; For target size; The relative distance between the target location and the transmission line; For environmental conditions; , , These are the weighting coefficients; Risk scores are derived based on risk assessment algorithms, and the assessment results are represented numerically. Step 7.3 Risk Level Classification: After generating the risk score, the early warning module classifies the risk level, including: High risk: Large bird nests may be covered in ice, requiring immediate action; Medium risk: Generate a warning message to notify maintenance personnel to conduct a thorough inspection; Low risk: Recorded, no proactive notification will be given at this time; Step 7.4 Report Generation and Historical Data Management: The early warning module can automatically generate detailed detection reports based on the detected targets and risk assessment results. The report will include: the specific location of the target, type information, time of occurrence, risk level, and recommended handling measures. The report will be output in PDF format for management personnel to view. In addition, the early warning module stores all early warning records in the database, forming a historical data management system. The historical data management system supports storage and retrieval. Based on the historical data management system, maintenance personnel can analyze the line conditions at different time periods and optimize inspection and maintenance strategies.

Images