Millimeter wave radar and microclimate information cross-modal fusion monitoring method for urban power station area

Abstract

The invention discloses a cross-modal fusion monitoring method for millimeter-wave radar and microclimate information of an urban power station area, and the method comprises the steps: S1, collecting the echo amplitude and phase information of the millimeter-wave radar, and carrying out the inversion of the tiny deformation of a structure; s2, synchronously collecting microclimate information and constructing an environment state vector; s3, performing preprocessing such as time synchronization and abnormal value elimination on the multi-source data; s4, extracting structure response features and environment features, and constructing a cross-modal feature set; s5, establishing an incidence relation between the structure and the environment through the fusion mapping function; and S6, constructing a safety state index based on a fusion result, and realizing anomaly identification and risk output. According to the invention, through the cross-modal fusion technology, the non-contact perception advantage of the radar and the environmental characterization effect of microclimate are comprehensively utilized, the problems of single traditional monitoring means, weak anti-interference capability, difficulty in early abnormality identification and the like are solved, and continuous and accurate monitoring of the structure and environment of the electric power transformer area can be realized in a complex urban scene.

Description

Technical Field

[0001] This invention belongs to the field of urban power distribution network operation safety monitoring and intelligent sensing technology, specifically a cross-modal fusion monitoring method for urban power distribution area millimeter-wave radar and micro-meteorological information. Background Technology

[0002] With the continuous advancement of urbanization, the scale of urban power distribution networks is constantly expanding. As the core basic unit directly facing end users in the power distribution network, the number and distribution density of urban power distribution areas are increasing day by day. Urban power distribution areas typically encompass distribution transformers, prefabricated substations, cable terminals, supporting structures, and ancillary building structures. They are widely distributed in residential areas, commercial areas, and along roads, facing the practical problems of complex spatial environments and numerous surrounding interference factors. Their operational safety is directly related to the reliability of urban power supply and public safety.

[0003] During long-term operation, urban power distribution facilities are susceptible to the combined effects of multiple adverse factors. On the one hand, uneven foundation settlement, structural aging, and material degradation can lead to slow deformation and decreased stability of the distribution structure. On the other hand, periodic vibrations caused by vehicle traffic, disturbances from surrounding construction, and micro-meteorological conditions such as heavy rainfall, strong winds, and drastic temperature changes have a continuous impact on the structural condition and operating environment of power distribution facilities. In the early stages, these factors often manifest as minute deformations, abnormal vibration responses, or accumulated tilting at the millimeter level or even smaller. They are highly concealed and difficult to detect. If they are not identified and intervened in a timely manner, they can easily evolve into equipment instability, structural damage, or even partial collapse accidents, posing a serious threat to the safe and stable operation of the urban power supply system.

[0004] Currently, safety monitoring of urban power distribution areas mainly relies on manual inspections, electrical quantity monitoring, and a small number of contact sensors. Manual inspections are limited by personnel experience and inspection cycles, making continuous, real-time monitoring difficult, and their ability to identify early anomalies such as minor deformations is limited. Electrical quantity monitoring mainly reflects the electrical operating status and cannot directly perceive the impact of structural deformation and environmental factors on facilities. Although contact sensors can acquire local physical quantity information, they suffer from high deployment costs and limited coverage, and are susceptible to environmental interference or failure in complex urban environments, making it difficult to meet the actual needs of large-scale, long-term, and refined monitoring.

[0005] Millimeter-wave radar, as a non-contact sensing method, boasts outstanding advantages such as all-weather operation, strong resistance to light interference, and high accuracy in deformation and vibration sensing, and has gradually gained attention in the field of structural health monitoring in recent years. However, in urban power distribution area applications, the monitoring information acquired by millimeter-wave radar mainly reflects the geometric and motion characteristics of the target, making it difficult to characterize the interference of complex environmental factors on the monitoring results. Furthermore, urban environments suffer from severe multipath reflection and diverse target types, and single radar monitoring methods have significant limitations in anomaly identification and risk assessment.

[0006] Furthermore, micrometeorological conditions, as a significant external factor affecting the operating environment of power distribution areas, have a substantial impact on the structural condition of facilities and monitoring signals. In existing technologies, millimeter-wave radar monitoring data and micrometeorological information are typically acquired independently and used in a decentralized manner, lacking effective joint modeling and collaborative analysis methods. This makes it difficult to fully characterize the intrinsic relationship between environmental factors and structural responses, thus limiting the accuracy and reliability of monitoring results.

[0007] Therefore, there is an urgent need for a monitoring method for urban power distribution areas that can effectively integrate millimeter-wave radar monitoring data with micro-meteorological information in complex urban scenarios. Through joint analysis of cross-modal information, it can improve the comprehensive perception of the structural status and environmental effects of power distribution areas, achieve accurate identification and continuous monitoring of early abnormal states, and provide reliable technical support for the safe operation of urban power distribution networks. Summary of the Invention

[0008] To address the technical problems existing in the safety monitoring of urban power distribution areas, such as the single monitoring method, insufficient ability to perceive minor structural deformations and environmental impacts, and difficulty in coordinating the use of multi-source monitoring information, the primary objective of this invention is to provide a cross-modal fusion monitoring method for urban power distribution areas using millimeter-wave radar and micro-meteorological information.

[0009] This invention introduces millimeter-wave radar for non-contact, continuous deformation and vibration monitoring of power distribution facilities. Combined with micro-meteorological information reflecting environmental conditions, it performs unified modeling and joint analysis of multi-source heterogeneous monitoring data, achieving comprehensive perception of the structural state and operating environment of urban power distribution areas. Through cross-modal information fusion processing, the stability and reliability of monitoring results in complex urban scenarios are improved, enhancing the ability to identify early, minor anomalies.

[0010] Another objective of this invention is to reduce the uncertainty caused by environmental changes and interference factors when using a single monitoring method by constructing a standardized monitoring process and data processing method for urban power distribution area application scenarios. This enables continuous monitoring and evaluation of the structural deformation, vibration response, and their relationship with micro-meteorological conditions of power distribution areas, thereby providing technical support for the safe operation and risk prevention of urban power distribution networks.

[0011] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a cross-modal fusion monitoring method for millimeter-wave radar and micro-meteorological information in urban power distribution areas, comprising the following steps:

[0012] S1: Millimeter-wave radar monitoring data acquisition: Millimeter-wave radars are deployed in urban power distribution areas to continuously monitor distribution transformers, prefabricated substations, and their ancillary structures, acquiring radar echo amplitude information. and phase information The phase difference between adjacent time points is obtained based on the phase information. And through the formula Calculate the minute deformation of the target along the radar line of sight. ,in This refers to the operating wavelength of millimeter-wave radar.

[0013] S2: Micro-meteorological information acquisition: Synchronously acquire micro-meteorological information related to the operating environment of the power distribution area to form an environmental state vector that changes over time. The micro-meteorological information is used to characterize the environmental conditions and changes in the power distribution area;

[0014] S3: Multi-source monitoring data preprocessing: The millimeter-wave radar monitoring data and micro-meteorological information are subjected to time synchronization, outlier removal, noise suppression, and data standardization. The data standardization process is performed using the formula... Implementation, in which and These represent the mean and standard deviation of the corresponding features, respectively.

[0015] S4: Cross-modal feature construction: Extracting structural response feature vectors from preprocessed millimeter-wave radar monitoring data ,in Represents the rate of change of deformation over time; extracts environmental feature vectors from preprocessed micrometeorological information. The structural response feature vector and the environmental feature vector are expressed in a unified manner to construct a cross-modal feature set. ;

[0016] S5: Cross-modal fusion analysis: Based on the aforementioned cross-modal feature set, through the cross-modal fusion mapping function... By jointly analyzing millimeter-wave radar characteristics and micro-meteorological characteristics, the correlation between structural state changes and environmental conditions is established, and the fusion results are obtained. ;

[0017] S6: Safety Status Assessment and Monitoring Output: Based on the aforementioned fusion results Through the state evaluation function Table Construction Security Status Indicators When the safety status indicator meets the preset anomaly criteria, the corresponding monitoring results or risk assessment information are output.

[0018] Preferably, the micrometeorological information in step S2 includes ambient temperature. relative humidity Rainfall Wind speed ,wind direction and air pressure At least one of the following, the environmental state vector .

[0019] Preferably, the millimeter-wave radar in step S1 adopts a non-contact observation method to achieve continuous and stable monitoring under different lighting and weather conditions.

[0020] Preferably, the outlier removal in step S3 adopts an outlier identification mechanism based on the 3σ criterion or box plot method, and the noise suppression adopts an adaptive filtering or wavelet denoising algorithm.

[0021] Preferably, the environmental feature vector in step S4 The statistical features are obtained by extracting statistical features from micrometeorological information, including at least one of the following: mean, variance, peak value, valley value, and rate of change.

[0022] Preferably, the cross-modal fusion mapping function in step S5 It is implemented using machine learning models, statistical modeling methods, or deep learning networks. The machine learning models include support vector machines, random forests, or gradient boosting trees, and the deep learning networks include convolutional neural networks, recurrent neural networks, or attention mechanism networks.

[0023] Preferably, the preset anomaly criterion in step S6 is obtained by training with historical normal operation data, and the safety status index Used to quantitatively characterize the structural stability and operating environment safety of power distribution areas, the monitoring results or risk assessment information include anomaly type, anomaly location, risk level, and intervention recommendations.

[0024] Compared with the prior art, the present invention has the following advantages:

[0025] 1. This invention uses millimeter-wave radar as the core sensing device, avoiding the problems of difficult deployment, high maintenance costs, and susceptibility to environmental interference of traditional contact sensors. It can continuously and stably monitor the structural status of power distribution areas under different lighting and weather conditions, such as day and night, and sunny and rainy days, significantly improving the reliability and environmental adaptability of monitoring.

[0026] 2. This invention innovatively integrates millimeter-wave radar monitoring data with micro-meteorological information for cross-modal fusion analysis. This not only accurately acquires the minute deformations and vibration responses of power distribution area structures, but also effectively quantifies the impact of environmental condition changes on structural status. This enables comprehensive perception of the structural status and operating environment of power distribution areas, avoiding the problems of information loss or misjudgment caused by single monitoring methods.

[0027] 3. This invention uses cross-modal fusion processing to jointly model radar features reflecting structural response and micro-meteorological features reflecting environmental changes. This effectively counteracts the impact of interference factors such as environmental changes, multipath reflection, and random noise on monitoring results, significantly improving the stability and consistency of monitoring data in complex urban scenarios and ensuring the accuracy of monitoring results.

[0028] 4. This invention, through deep fusion and comprehensive analysis of cross-modal features, can accurately capture millimeter-level minute anomalies and their evolution trends in the early stages of power distribution area structures. Compared with traditional monitoring methods, it can detect potential risks earlier, providing maintenance personnel with sufficient intervention time and effectively reducing the probability of serious accidents such as equipment instability and structural damage.

[0029] 5. The monitoring method of the present invention does not depend on a specific type of power distribution area structure or a single model of monitoring equipment. It can be flexibly applied to various types of urban power distribution area facilities such as distribution transformers, box-type substations, cable terminals and their ancillary structures, and has wide versatility and good scalability.

[0030] 6. The monitoring results of this invention are output in a standardized data format, which can directly interact and integrate with the existing urban power distribution network operation and maintenance management system without requiring large-scale modification of the existing system, thus reducing the cost of technology implementation and having good engineering application prospects and promotion value. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0032] Figure 1 This is a schematic diagram of the overall architecture of the urban power distribution area monitoring system based on the cross-modal fusion of millimeter-wave radar and micro-meteorological information of the present invention.

[0033] Figure 2 This is a schematic diagram of the deployment of millimeter-wave radar and micro-meteorological equipment in urban power distribution areas according to the present invention;

[0034] Figure 3 This is a flowchart of the cross-modal fusion monitoring method for millimeter-wave radar and micro-meteorological information of the present invention. Detailed Implementation

[0035] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The described embodiments are only used to explain the present invention and are not intended to limit the scope of protection of the present invention.

[0036] This embodiment selects a typical power distribution area in a residential area of ​​a city as the monitoring object. The power distribution area includes a 10kV distribution transformer, a box-type substation and supporting structures. The surrounding area has complex interference factors such as urban road traffic, residents' daily activities and seasonal weather changes, which is consistent with the typical application scenario of urban power distribution areas.

[0037] 1. Monitoring equipment deployment and parameter settings

[0038] (1) Millimeter-wave radar equipment: Select a frequency-modulated continuous wave millimeter-wave radar with a working frequency of 77 GHz and a working wavelength of λ=3.89 mm. Deploy it on fixed structures near the power distribution area. The radar beam covers key monitoring parts such as the distribution transformer box, the box-type substation shell and the supporting columns. The radar sampling frequency is set to 10 Hz to ensure that the dynamic vibration response of the structure can be captured.

[0039] (2) Micro-meteorological monitoring equipment: Micro-meteorological monitoring stations are set up within 50m of the power distribution area, equipped with high-precision temperature and humidity sensors (measurement range: temperature -40℃~85℃, accuracy ±0.1℃; humidity 0~100%RH, accuracy ±2%RH), tipping bucket rain gauges (measurement range 0~4mm / min, accuracy ±0.2mm), ultrasonic anemometers (wind speed measurement range 0~60m / s, accuracy ±0.1m / s; wind direction measurement range 0~360°, accuracy ±1°), and barometric pressure sensors (measurement range 300~1100hPa, accuracy ±0.1hPa). The micro-meteorological data acquisition frequency is set to 10Hz, which is consistent with the radar data acquisition frequency.

[0040] 2. Data Acquisition and Preprocessing

[0041] (1) Data acquisition: The millimeter-wave radar equipment continuously acquires the echo amplitude information of the target structure. and phase information Through formula The minute deformation along the radar line of sight was calculated. Micro-meteorological monitoring equipment synchronously collects ambient temperature. relative humidity Rainfall Wind speed ,wind direction and air pressure This forms an environmental state vector. .

[0042] (2) Data preprocessing:

[0043] Time synchronization: Based on the high-precision timestamps provided by the GPS timing module, radar data and micro-meteorological data are aligned to ensure that the two types of data correspond one-to-one at the same time;

[0044] Outlier removal: The 3σ criterion is used to remove outliers such as abnormal jumps in radar echo amplitude and micro-meteorological data that exceed the reasonable physical range.

[0045] Noise suppression: Wavelet denoising algorithm (using db4 wavelet basis, decomposition level 3) is used to suppress noise in radar phase data, and 5-point moving average method is used to reduce random fluctuations in micro-meteorological data;

[0046] Data standardization: Based on monitoring data from the past 6 months of normal operation of this power distribution area, the mean of each characteristic is calculated. and standard deviation Through formula Standardize all feature data.

[0047] 3. Cross-modal feature construction

[0048] (1) Extraction of structural response feature vectors: Extracting feature vectors from preprocessed radar data (Minor deformation) (The rate of change of deformation is calculated using the first-order difference.) (Standardized echo amplitude) Construct structural response feature vector .

[0049] (2) Environmental feature vector extraction: For the preprocessed micrometeorological data, the 5-minute moving average, variance, peak value and rate of change of each element are extracted to form an environmental feature vector. Its dimensions are 24 (6 micro-meteorological elements × 4 statistical features).

[0050] (3) Construction of cross-modal feature sets: (3D) and (24 dimensions) are concatenated to obtain a cross-modal feature set with 27 dimensions. .

[0051] 4. Cross-modal fusion analysis

[0052] This embodiment uses a Bidirectional Long Short-Term Memory (BiLSTM-Attention) network based on an attention mechanism as the cross-modal fusion mapping function. The specific implementation process is as follows:

[0053] (1) Data set construction: 100,000 sets of cross-modal feature data during the normal operation period of the power distribution area were selected as the training set, and 10,000 sets of data were selected as the validation set; 20,000 sets of data under artificially simulated scenarios such as slight structural tilt and abnormal vibration were selected as the test set.

[0054] (2) Model training: Input the training set data into the BiLSTM-Attention network. The input layer dimension of the network is 27, the hidden layer is set to 2 layers of BiLSTM (64 hidden units per layer), the attention layer adopts the additive attention mechanism, and the output layer dimension is 1 (fusion feature value). The mean squared error loss function is minimized by the Adam optimizer. The number of training iterations is set to 100 rounds, and the batch size is set to 32. Finally, the cross-modal fusion model is trained.

[0055] (3) Calculation of fusion results: The cross-modal feature set acquired and preprocessed in real time is fused together. Input the trained model to obtain the fusion result. .

[0056] 5. Safety Status Assessment and Monitoring Output

[0057] (1) Construction of safety status indicators: based on the fusion results of the training set during normal operation period Calculate the mean Standard deviation =0.15, setting the safety status indicator ,in ≤1.0 is considered normal; 1.0 < ≤2.0 indicates mild abnormality, 2.0 < A score ≤3.0 is considered moderately abnormal. A score of >3.0 indicates a severe abnormality.

[0058] (2) Anomaly detection and output: When an anomaly is detected When the value is 2.3 (moderate anomaly), the system automatically analyzes the factors with the highest contribution in the fusion features (such as the rate of change of deformation accompanying the sudden change in wind speed V(t)). Increase the output of anomaly reports, including: time of anomaly occurrence, location of anomaly (support column of box-type substation), type of anomaly (structural vibration anomaly caused by wind load), risk level (moderate), intervention recommendations (strengthen wind speed monitoring in the area and conduct on-site inspections within 24 hours), and push the report to the distribution network operation and maintenance management terminal in real time.

[0059] Through the application verification of this embodiment, the monitoring method can accurately capture minute abnormal changes in the structure of power distribution areas, effectively combine micro-meteorological conditions for comprehensive evaluation, and the monitoring results are accurate and reliable, providing strong technical support for the safe operation of urban power distribution areas.

[0060] Those skilled in the art should understand that the above embodiments are merely illustrative examples. In practical applications, the radar equipment parameters, micro-meteorological monitoring elements, fusion model types, and anomaly criterion thresholds can be adjusted according to specific monitoring scenarios, all of which fall within the protection scope of this invention.

Claims

1. A cross-modal fusion monitoring method for millimeter-wave radar and micro-meteorological information in urban power distribution areas, characterized in that, Includes the following steps: S1: Millimeter-wave radar monitoring data acquisition: Millimeter-wave radars are deployed in urban power distribution areas to continuously monitor distribution transformers, prefabricated substations, and their ancillary structures, acquiring radar echo amplitude information. and phase information The phase difference between adjacent time points is obtained based on the phase information. And through the formula Calculate the minute deformation of the target along the radar line of sight. ,in This refers to the operating wavelength of millimeter-wave radar. S2: Micro-meteorological information acquisition: Synchronously acquire micro-meteorological information related to the operating environment of the power distribution area to form an environmental state vector that changes over time. The micro-meteorological information is used to characterize the environmental conditions and changes in the power distribution area; S3: Multi-source monitoring data preprocessing: The millimeter-wave radar monitoring data and micro-meteorological information are subjected to time synchronization, outlier removal, noise suppression, and data standardization. The data standardization process is performed using the formula... Implementation, in which and These represent the mean and standard deviation of the corresponding features, respectively. S4: Cross-modal feature construction: Extracting structural response feature vectors from preprocessed millimeter-wave radar monitoring data ,in This represents the rate of change of deformation over time; Environmental feature vectors are extracted from preprocessed micrometeorological information. The structural response feature vector and the environmental feature vector are expressed in a unified manner to construct a cross-modal feature set. ; S5: Cross-modal fusion analysis: Based on the aforementioned cross-modal feature set, through the cross-modal fusion mapping function... By jointly analyzing millimeter-wave radar characteristics and micro-meteorological characteristics, the correlation between structural state changes and environmental conditions is established, and the fusion results are obtained. ; S6: Safety Status Assessment and Monitoring Output: Based on the fusion results Through the state evaluation function Table Construction Security Status Indicators When the safety status indicator meets the preset anomaly criteria, the corresponding monitoring results or risk assessment information are output.

2. The cross-modal fusion monitoring method for urban power distribution area millimeter-wave radar and micro-meteorological information according to claim 1, characterized in that, The micrometeorological information mentioned in step S2 includes ambient temperature. relative humidity Rainfall Wind speed ,wind direction and air pressure At least one of the following, the environmental state vector .

3. The cross-modal fusion monitoring method for urban power distribution area millimeter-wave radar and micro-meteorological information according to claim 1, characterized in that, The millimeter-wave radar described in step S1 adopts a non-contact observation method to achieve continuous and stable monitoring under different lighting and weather conditions.

4. The cross-modal fusion monitoring method for urban power distribution area millimeter-wave radar and micro-meteorological information according to claim 1, characterized in that, The outlier removal in step S3 adopts an outlier identification mechanism based on the 3σ criterion or box plot method, and the noise suppression adopts adaptive filtering or wavelet denoising algorithm.

5. The cross-modal fusion monitoring method for urban power distribution area millimeter-wave radar and micro-meteorological information according to claim 1, characterized in that, The environmental feature vector in step S4 The statistical features are obtained by extracting statistical features from micrometeorological information, including at least one of the following: mean, variance, peak value, valley value, and rate of change.

6. The cross-modal fusion monitoring method for urban power distribution area millimeter-wave radar and micro-meteorological information according to claim 1, characterized in that, The cross-modal fusion mapping function described in step S5 It is implemented using machine learning models, statistical modeling methods, or deep learning networks. The machine learning models include support vector machines, random forests, or gradient boosting trees, and the deep learning networks include convolutional neural networks, recurrent neural networks, or attention mechanism networks.

7. The cross-modal fusion monitoring method for urban power distribution area millimeter-wave radar and micro-meteorological information according to claim 1, characterized in that, The preset anomaly criterion mentioned in step S6 is obtained through training on historical normal operation data, and the safety status index Used to quantitatively characterize the structural stability and operating environment safety of power distribution areas, the monitoring results or risk assessment information include anomaly type, anomaly location, risk level, and intervention recommendations.

Images