An unmanned aerial vehicle fusion identification and positioning method and system for a new energy station

Abstract

The application discloses a kind of unmanned aerial vehicle fusion identification positioning method and system for new energy station, it is related to power grid intelligent security technology field, including obtaining the multi-modal perception data of target unmanned aerial vehicle in the heterogeneous sensor network of new energy station, multi-modal perception data include radio signal, acoustic signal and optical image;Based on radio signal, the first position estimation of target unmanned aerial vehicle is calculated;Based on acoustic signal and optical image, the second position estimation of target unmanned aerial vehicle is calculated;According to the first position estimation and the second position estimation, respectively assess the real-time positioning reliability corresponding thereto;Based on real-time positioning reliability, fusion weight associated with each position estimation is generated;According to fusion weight, the first position estimation and the second position estimation are weighted calculation, generate the three-dimensional fusion positioning result of target unmanned aerial vehicle.The application improves the accuracy of unmanned aerial vehicle identification positioning under complex station environment by fusing radio, acoustic, optical multi-modal data.

Description

Technical Field

[0001] This invention relates to the field of smart grid security technology, and in particular to a method and system for unmanned aerial vehicle (UAV) fusion identification and positioning for new energy power plants. Background Technology

[0002] With the rapid development of the new energy industry, wind power and photovoltaic power stations are mostly located in vast and remote areas with large perimeters and complex environments. Safety hazards such as drones illegally entering inspection areas and touching high-voltage equipment have increased dramatically, seriously threatening the operation of station equipment and the safety of personnel.

[0003] In existing technologies, UAV positioning methods based on single radio signals are widely used, but this technology has shortcomings in the context of new energy power stations: electromagnetic interference generated by the operation of power equipment inside the station will seriously affect the propagation stability of radio signals, resulting in large positioning errors; at the same time, radio signals have limited penetration ability through obstructions, and positioning blind spots are likely to occur in areas obstructed by tall equipment in the station, making it impossible to achieve full coverage positioning of the target airspace and making it difficult to ensure the security of the station perimeter. Summary of the Invention

[0004] To address the shortcomings of existing UAV positioning methods based on single radio signals, which cannot achieve comprehensive coverage of the target airspace and thus fail to ensure perimeter security of power stations, this invention provides a UAV fusion identification and positioning method and system for new energy power stations. By fusing radio, acoustic, and optical multimodal data, it overcomes the deficiencies of single-sensor positioning and improves the accuracy and stability of UAV positioning in complex power station environments. The specific technical solution is as follows: In a first aspect, the present invention provides a method for unmanned aerial vehicle (UAV) fusion identification and positioning for new energy power stations, comprising the following steps: The method acquires multimodal sensing data emitted or generated by a target UAV within the target airspace of a new energy power station in a heterogeneous sensing network. The multimodal sensing data includes at least radio signals, acoustic signals, and optical images. The heterogeneous sensing network is constructed by deploying at least one radio sensing node, at least one acoustic sensing node, and at least one optical sensing node at a predetermined location on the perimeter of the new energy power station. Based on the radio signals, a first position estimate of the target UAV is calculated; A second position estimate of the target UAV is calculated based on acoustic signals and optical images; Based on the first position estimate and the second position estimate, evaluate their corresponding real-time positioning reliability respectively; Based on the real-time location confidence, fusion weights associated with each location estimate are generated; The first position estimate and the second position estimate are weighted according to the fusion weight to generate the three-dimensional fusion positioning result of the target UAV.

[0005] Preferably, calculating a first position estimate of the target UAV based on the radio signal specifically includes: Based on the radio signal, the first position estimate of the target UAV is calculated using the time difference of arrival algorithm.

[0006] Preferably, the second position estimate of the target UAV is calculated based on acoustic signals and optical images, specifically including: Based on the acoustic signal, the initial spatial region of the target UAV is calculated using the time difference of arrival method; Based on the optical image, the three-dimensional coordinates of the target UAV are calculated within the initial spatial region through visual target detection and stereo vision ranging, serving as the second position estimate.

[0007] Preferably, the acquisition of multimodal sensing data launched or generated by a target UAV in the target airspace within the heterogeneous sensor network of the new energy power station further includes: When any sensor node in the heterogeneous sensor network of the new energy power station detects a suspected drone target, a collaborative trigger signal is generated. Based on the aforementioned collaborative trigger signal, the remaining sensor nodes are awakened through the communication network within the site.

[0008] Preferably, before calculating the second position estimate of the target UAV based on acoustic signals and optical images, the method further includes: Call the pre-stored background noise feature library, which contains the noise spectrum features generated by typical equipment in the new energy power station during operation; From the current acoustic signal, components that match the background noise characteristics are suppressed by adaptive filtering or spectral subtraction methods.

[0009] Preferably, a method for unmanned aerial vehicle (UAV) fusion identification and positioning for new energy power stations further includes: The radio signals are analyzed and decoded to obtain remote identification broadcast information that conforms to a preset standard format; If decoding is successful, the drone's unique identifier and autonomously reported real-time location data are extracted from the remote identity broadcast information. The autonomously reported real-time location is used as the third location estimate, and the corresponding real-time location confidence is set based on its communication protocol type. The third location estimate and its location confidence, together with the first location estimate, the second location estimate and their respective location confidence, are input into the weighted calculation process to generate the three-dimensional fusion positioning result.

[0010] Preferably, evaluating the real-time location reliability includes: Based on the signal-to-noise ratio of the radio signal, acoustic signal, or optical image, a position estimation error model for the corresponding sensing mode is established. The current signal-to-noise ratio is matched with the error model to predict the confidence interval of the current estimate; The corresponding real-time location confidence is obtained based on the confidence interval or its reciprocal.

[0011] Preferably, based on the real-time location confidence, generating fusion weights associated with each location estimate includes: The real-time location reliability of each location estimate is normalized. Based on a preset weight allocation strategy function, the normalized confidence scores are mapped to corresponding fusion weights; wherein the strategy function is a linear scaling function, a soft maximum function, or a discriminant function based on expert rules.

[0012] Preferably, the first position estimate and the second position estimate are weighted according to the fusion weights to generate a three-dimensional fusion positioning result for the target UAV, including: The fusion weights are normalized, and based on the normalized fusion weights, the three-dimensional coordinates of the first position estimate and the second position estimate are weighted and summed. The weighted summation result is output as the three-dimensional fusion positioning result of the target UAV, and the result includes at least longitude, latitude and altitude information.

[0013] Secondly, the present invention also provides a drone fusion identification and positioning system for new energy power stations, which applies the aforementioned drone fusion identification and positioning method for new energy power stations, including: The data acquisition unit is used to acquire multimodal sensing data emitted or generated by a target UAV in the target airspace within the heterogeneous sensing network of the new energy power station. The multimodal sensing data includes at least radio signals, acoustic signals, and optical images. The heterogeneous sensing network is composed of at least one radio sensing node, at least one acoustic sensing node, and at least one optical sensing node deployed at a preset location on the perimeter of the new energy power station. The first position calculation unit is used to calculate the first position estimate of the target UAV based on the radio signal; The second position calculation unit is used to calculate the second position estimate of the target UAV based on acoustic signals and optical images; The confidence assessment unit is used to assess the real-time positioning confidence of the first position estimate and the second position estimate, respectively. The weight generation unit is used to generate fusion weights associated with each location estimate based on the real-time location confidence. The fusion positioning unit is used to perform weighted calculations on the first position estimate and the second position estimate according to the fusion weights to generate a three-dimensional fusion positioning result for the target UAV.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention provides a drone fusion identification and positioning method for new energy power stations. By integrating three types of multimodal sensing data—radio, acoustic, and optical—it fully leverages the complementary advantages of different sensing modalities and effectively compensates for the inherent shortcomings of single-sensor positioning technologies. Specifically, radio signal positioning offers long-range detection advantages, acoustic signal positioning has strong resistance to electromagnetic interference, and optical image positioning provides intuitive visual orientation reference. The fusion of these three technologies can collaboratively avoid the adverse effects of complex electromagnetic interference and obstruction by tall equipment at new energy power stations. It also improves the accuracy and stability of drone positioning under variable weather conditions such as strong winds and sandstorms, as well as in scenarios with multiple obstructions, ensuring that positioning errors are controlled within a reasonable range to meet the security requirements of the power station. Through a heterogeneous sensor network deployed around the perimeter of new energy power stations, radio, acoustic, and optical sensor nodes achieve collaborative networking and full-area coverage. Multi-node data interaction enables the identification of blind spots such as gaps between tall wind turbines and photovoltaic arrays, which are difficult to cover using traditional single-sensor positioning technologies. This adapts to the airspace environment of new energy power stations, which have large land areas, complex terrain, and dense equipment distribution. Whether in the core equipment area, inspection channels, or peripheral perimeter areas, it can accurately capture and locate intruding drones, solving the problem of incomplete coverage by traditional positioning technologies in complex station scenarios. Finally, by real-time evaluation of the positioning reliability of the first and second position estimates, corresponding fusion weights are dynamically generated and assigned, ensuring that the positioning results always focus on the more reliable sensor data. This ensures both the real-time nature of the positioning results and improves data reliability. It provides timely data support for the station's security system, reduces the safety risks such as equipment damage and personnel casualties caused by drone intrusion, and strengthens the effectiveness of the new energy power station's security protection system. Attached Figure Description

[0015] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0016] Figure 1This is a flowchart of a drone fusion identification and positioning method for new energy power stations according to the present invention.

[0017] Figure 2 This is a schematic diagram of the heterogeneous sensor network for identifying drones in a local wind farm according to the present invention.

[0018] Figure 3 This is a schematic diagram of the heterogeneous sensor network for global wind farm identification of drones according to the present invention.

[0019] Figure 4 This is a schematic diagram of a drone fusion identification and positioning system for new energy power stations according to the present invention. Detailed Implementation

[0020] 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, not all, of the embodiments of the present invention. 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.

[0021] It should be understood that, when used in this specification, the terms “comprising” and “including” indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0022] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification, unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “the” are intended to include the plural forms.

[0023] It should also be further understood that the term "and / or" as used in this specification refers to any combination of one or more of the associated listed items, as well as all possible combinations, and includes such combinations.

[0024] Please refer to the following examples. Figures 1 to 4 .

[0025] Please see Figure 1 This invention provides a method for unmanned aerial vehicle (UAV) fusion identification and positioning for new energy power stations, comprising the following steps: Step S1: Acquire multimodal sensing data emitted or generated by a target UAV in the target airspace within the heterogeneous sensing network of the new energy power station. The multimodal sensing data includes at least radio signals, acoustic signals, and optical images. The heterogeneous sensing network is constructed by deploying a heterogeneous sensing network consisting of at least one radio sensing node, at least one acoustic sensing node, and at least one optical sensing node at preset locations on the perimeter of the new energy power station, such as at preset locations on wind turbines, photovoltaic arrays, booster stations, and the perimeter of the power station. like Figure 2 and Figure 3 As shown, taking a wind farm as an example, acoustic sensor nodes with microphone arrays are deployed at the bottom of the wind turbine towers to identify rotor noise during hovering or low-speed passage; radio sensor nodes with software-defined radios are deployed on the communication tower of the booster station to obtain a good field of view; and optical sensor nodes with pan-tilt-zoom infrared thermal imagers are deployed on the top of monitoring poles at the edge of the farm area. When the system triggers monitoring, the three nodes start synchronously. The radio nodes collect spectrum signals in the 2.4GHz and 5.8GHz frequency bands; the acoustic nodes collect audio streams at a sampling rate of 48kHz; and the optical nodes collect thermal imaging video streams at a frame rate of 10Hz. All sensor nodes are time-synchronized via a time protocol or GPS timing; and an edge computing gateway receives the data streams from each node.

[0026] Step S2: Calculate the first position estimate of the target UAV based on the radio signal; Geometric positioning is achieved by utilizing the propagation characteristics of radio signals. Possible calculation methods include: Time difference of arrival method: By using distributed radio sensor nodes, the time difference of the same UAV signal arriving at each node is measured, a hyperbolic equation system is established, and the UAV position is calculated.

[0027] Angle of arrival method: Using directional antennas or antenna arrays, the direction of arrival of the signal is estimated at two or more nodes, and the location is determined by the intersection of the directional lines.

[0028] Received signal strength indication method: This method establishes a signal propagation attenuation model and calculates the distance based on the signal strength received by multiple nodes. However, this method is susceptible to the complex electromagnetic environment within the site, resulting in poor accuracy.

[0029] The collected radio signals are preprocessed to extract the target UAV's image transmission or remote control signal. If the Time Difference of Arrival (TDOA) algorithm is used, the timestamps of three or more radio sensor nodes receiving the same signal are used to calculate the intersection area of ​​the two-dimensional hyperboloids where the target is located. Combined with the signal field strength, the altitude is initially estimated, and the first position estimate is output.

[0030] Assume the number of radio sensor nodes deployed at the ZAI new energy power station is . ( ). No. The three-dimensional coordinates of each node are The complex baseband signal transmitted by the drone is , No. The complex signal received by each node is: in, C is a complex attenuation factor that includes path loss and phase; The envelope of the original complex signal; To delay the transmission time, This represents the actual location of the drone. For carrier frequency; This is the initial phase of the receiver; Zero-mean Gaussian white noise, .

[0031] Calculate the first from the received signal Instantaneous signal-to-noise ratio of each node: Expressed in decibels: The average signal-to-noise ratio of all nodes is denoted as . , which serves as the input parameter for subsequent channel evaluation.

[0032] With nodes Using the generalized cross-correlation method as a reference, the signal arrival at the node is estimated. With nodes Time difference : in, for Fourier transform, For the weighting function (preferably PHAT weighting: ).

[0033] The estimated time difference is the delay corresponding to the peak value of the cross-correlation function. Corresponding distance difference: .

[0034] Construct a system of hyperbolic equations: Solve by linearization using Taylor series expansion. In the... In the next iteration, the current position is estimated to be Define the distance function: Directional unit vector: Establish a linear equation: in: for dimensional residual vector, the first One element: for The coefficient matrix of dimension, the first dimension OK: This is the position correction amount; Solve using weighted least squares: Among them, the weight matrix Weighted elements With time difference to estimate variance It is inversely proportional and can be estimated from the peak intensity of the cross-correlation function.

[0035] The updated location is estimated to be: Iterate until convergence. (Preset threshold) to finally obtain the first position estimate .

[0036] Step S3: Calculate the second position estimate of the target UAV based on acoustic signals and optical images; By using coarse acoustic positioning to guide fine optical positioning, complementary advantages are achieved. The specific principle is as follows: Based on the acoustic signal, the initial spatial region of the target UAV is calculated using the time difference of arrival method; The time difference of sound arrival at each element of the microphone array is calculated using the generalized cross-correlation method or the time delay estimation method. Then, the spatial region where the UAV sound source is located is calculated using the time difference of arrival method. In new energy power plants, adaptive filtering is first performed using a pre-stored wind turbine noise library to suppress strong background interference.

[0037] Based on the optical image, the three-dimensional coordinates of the target UAV are calculated within the initial spatial region through visual target detection and stereo vision ranging, serving as the second position estimate.

[0038] Monocular visual ranging: Within an initial area provided by acoustics, deep learning models such as YOLOv5 and YOLOv8 are used to detect drone targets in the image. Based on the target's pixel size in the image and camera intrinsic parameters, the distance between the target and the camera is estimated. Combined with the gimbal's azimuth and pitch angles, the three-dimensional coordinates are calculated.

[0039] Binocular or multi-view stereo vision: If there are two or more optical nodes with known positions within the field of view, the precise three-dimensional coordinates can be directly calculated by matching the pixel positions of the same UAV in different images and using the principle of triangulation. The accuracy is better than that of monocular vision.

[0040] Laser ranging assistance: For extremely high-value areas, a laser ranging module can be integrated into the PTZ camera to directly obtain accurate distance values ​​after optical recognition, thereby improving positioning accuracy.

[0041] In this embodiment, the Generalized Cross-Correlation Method (GCC-PHAT) is used to calculate the time difference of sound arrival at different microphones for coarse acoustic localization, resulting in a possible area. Subsequently, the gimbal thermal imager is guided to be aimed at this area. Target detection algorithms such as YOLO are used to locate the heat source target. Then, using binocular stereo vision or a monocular camera combined with prior knowledge of the target size, the precise three-dimensional coordinates relative to the camera are calculated, converted to geodetic coordinates, and used as a second position estimate. The specific calculation process is as follows: In the initial spatial region calculation for acoustic coarse localization in this embodiment, a deployment is set up. The acoustic sensing node of the microphone array, the first The node positions are The received acoustic signal is: in: For drone sound source signals; For the time delay of sound wave propagation, - For environmental noise, Specific background noise for new energy power plants.

[0042] Using a pre-stored fan noise feature library Perform spectral subtraction: in, The original spectrum, As an over-subtraction factor, take .

[0043] The time difference of arrival (TDOA) estimation uses the generalized cross-correlation method, with node 1 as the reference: TDOA estimation: Corresponding acoustic distance difference: .

[0044] Establish a system of hyperbolic equations: The weighted least squares iterative method is used to solve the problem. In the... During the next iteration: Current estimated location: ; Direction vector: ; Linearized equations: ; in: Weighted matrix Weighted elements .

[0045] Iterative updates until convergence are achieved, yielding coarse acoustic localization. and its geometric precision factor .

[0046] Calculate acoustic positioning reliability: in The average signal-to-noise ratio (dB) These are the parameters for the acoustic modal error model.

[0047] The initial spatial region is defined as a sphere: radius The coverage factor is set to 2.

[0048] In the three-dimensional coordinate calculation of optical precision positioning in this embodiment, there is a set The first optical sensing node, the... The intrinsic and extrinsic parameters of each node are known: Intrinsic parameter matrix Rotation matrix Translation vector .

[0049] acoustic initial region Projected onto each image plane. For the first... There are 1 camera, and the projection of points within the area is: Get the search area on the image .

[0050] In the search area Object detection is performed within the bounding box using a deep learning model (such as YOLO). and confidence level .

[0051] Extract the center pixel coordinates of the target: For at least two cameras (e.g.) Establish the stereo vision equation. Based on the camera model: make The first three columns are The last column is .but eliminate We obtain the following system of linear equations: in, express The OK, express The Each element.

[0052] Solve using the least squares method The second position estimate is obtained. .

[0053] Step S4: Evaluate the real-time positioning reliability of the first position estimate and the second position estimate respectively; Specifically, assessing the real-time location reliability includes: Based on the signal-to-noise ratio of the radio signal, acoustic signal, or optical image, a position estimation error model for the corresponding sensing mode is established. The current signal-to-noise ratio is matched with the error model to predict the confidence interval of the current estimate; The corresponding real-time location confidence is obtained based on the confidence interval or its reciprocal.

[0054] Let the real-time positioning reliability of the first position estimate (based on radio TDOA) be . The real-time positioning reliability of the second position estimation (based on acoustic-optical coordination) is: Both evaluations follow a unified mathematical model, but the input parameters differ.

[0055] Confidence level of the first position estimate Its evaluation relies on the quantization parameters output by the radio positioning process: average signal-to-noise ratio. (dB) and geometric precision factor .

[0056] in, This is the system's reference accuracy constant. The coverage factor is set to 2. The standard deviation of the current radio positioning error predicted based on the error model: Here, These are the parameters of the radio mode-specific error model obtained through calibration using historical data. Dimensionless, determined by the spatial geometric relationship between the sensor node and the UAV during positioning.

[0057] Confidence level of the second position estimate Its evaluation integrates the results of coarse acoustic localization and fine optical localization. Let the confidence level provided by acoustic localization be... The confidence level provided by optical positioning is .

[0058] Acoustic confidence in, GDOP is the average signal-to-noise ratio of the acoustic signal. Geometric precision factor for acoustic positioning. These are the parameters for the acoustic modal error model.

[0059] Optical confidence in, Metrics used to characterize image quality, such as signal-to-noise ratio; The root mean square of the reprojection error for stereo vision; These are the parameters for the optical modal error model.

[0060] Overall confidence level Final confidence level of the second position estimate yes and The function is preferably: Or take the smaller of the two: .

[0061] Step S5: Based on the real-time location confidence, generate fusion weights associated with each location estimate; Specifically, based on the real-time location confidence, fusion weights associated with each location estimate are generated, including: The real-time location confidence scores of each location estimate are normalized; the normalized weights of the fusion are expressed as follows: Among them, satisfying .

[0062] Based on a preset weight allocation strategy function, the normalized confidence scores are mapped to corresponding fusion weights; wherein the strategy function is a linear scaling function, a soft maximum function, or a discriminant function based on expert rules.

[0063] The normalized confidence scores are mapped to the final weights using a predefined policy function F.

[0064] Linear proportional function: Soft maximum function (introducing temperature parameter T>0): The larger T is, the more the weights concentrate on the side with higher confidence.

[0065] Discriminant function based on expert rules: For example, if Below the interference threshold Then, the radio weight will be forcibly reduced: in, As a penalty factor. Then renormalized: Step S6: Weight the first position estimate and the second position estimate according to the fusion weights to generate the 3D fusion positioning result of the target UAV. Specifically, this includes: The fusion weights are normalized, and based on the normalized fusion weights, the three-dimensional coordinates of the first position estimate and the second position estimate are weighted and summed. The weighted summation result is output as the three-dimensional fusion positioning result of the target UAV, and the result includes at least longitude, latitude and altitude information.

[0066] In specific implementation, let the three-dimensional coordinates of the first position estimate be... The estimated three-dimensional coordinates of the second position are: .

[0067] Using the generated fusion weights and (satisfy ), calculate the fusion location: This calculation is performed in a unified local Cartesian coordinate system.

[0068] Will Through coordinate transformation function Convert to standard format in geodetic coordinate system: Output It is a three-dimensional coordinate vector. In practical geographic information systems, this vector is usually converted to include longitude. Latitude and altitude The standard format.

[0069] In other embodiments, a second position estimate of the target UAV can also be calculated based on the acoustic signals or the optical images. At this time, the acoustic ( ) and optics ( Independently generate estimates and share weights Then internal weighting is required first: ,in .

[0070] Specifically, in a preferred embodiment of this application, the acquisition of multimodal sensing data emitted or generated by a target UAV within the target airspace of the heterogeneous sensor network of the new energy power station further includes: When any sensor node in the heterogeneous sensor network of the new energy power station detects a suspected drone target, a collaborative trigger signal is generated. Based on the aforementioned collaborative trigger signal, the remaining sensor nodes are awakened through the communication network within the site.

[0071] In practical implementation, any sensing node in the heterogeneous sensor network can act as a trigger, including radio receiver nodes, acoustic microphone arrays, or optical cameras. When any node determines the presence of a suspected drone target based on its single-modal perception data, a coordinated trigger signal is generated. The judgment criteria can be preset as follows: Radio node: Detects an unknown radio frequency signal that matches the drone's communication frequency band, and the signal strength continuously exceeds the background noise threshold.

[0072] Acoustic node: After suppressing background fan noise, an acoustic signature signal matching the harmonic characteristics of the UAV propeller blades was detected.

[0073] Optical nodes: Using a lightweight moving target detection algorithm, small flying objects that match the size and motion characteristics of a drone are identified in the video stream.

[0074] The generated collaborative trigger signal can be a data packet containing key metadata, such as the trigger node ID and node type.

[0075] Based on a preliminary estimate of the target's location or the area with the strongest signal, the trigger signal is broadcast or multicast through the communication network deployed within the new energy power station. The signal is typically first sent to an edge computing gateway deployed within the station, which is responsible for signal parsing and command distribution. Based on the content of the trigger signal, the edge computing gateway performs intelligent wake-up scheduling: Prioritize waking up and scheduling other types of sensor nodes whose field of view or listening range can cover the location indicated by the trigger signal. For example, if an acoustic node is triggered in area A, optical and radio nodes deployed in area A and its vicinity will be woken up first. Depending on the type of triggering node, its complementary modes will be enhanced. For example, if triggered by a radio node, the telephoto search mode of the optical node will be enhanced simultaneously; if triggered by an acoustic node, the wide-angle tracking and detail recognition modes of the optical node will be emphasized. The woken-up node will be switched from low-power watchdog mode to high-precision tracking mode. Specific actions include increasing the sampling rate and scanning sensitivity of the radio node; adjusting the preset position of the optical pan-tilt camera to point towards the trigger location; and initiating high-resolution beamforming calculations for the acoustic array.

[0076] All awakened and positioned nodes begin synchronous, focused multimodal data acquisition of the target airspace according to a unified time reference. The acquired data is transmitted in real time to the edge computing gateway, entering the positioning and fusion process.

[0077] For most of the time when there is no threat, most nodes in the network can remain in a low-power sleep state. Only when absolutely necessary is the high-performance mode of the entire network or a portion of the network activated briefly. This reduces the enormous energy consumption costs faced when deploying in vast renewable energy power plants. In this embodiment, once a node detects a suspected target, the entire system's sensing resources can be redirected towards the threat, avoiding missed or delayed target detection due to scanning cycles or blind spots. Importantly, the distributed triggering of the wake-up mechanism provides inherent redundancy to the system. Even if one or more nodes in the network fail due to faults, interference, or deliberate sabotage, as long as other nodes are still operational, they can act as triggers to activate the remaining healthy parts of the system, continuing to provide services and avoiding global paralysis caused by a single point of failure.

[0078] Specifically, in a preferred embodiment of this application, before calculating the second position estimate of the target UAV based on acoustic signals and optical images, the method further includes: Call the pre-stored background noise feature library, which contains the noise spectrum features generated by typical equipment in the new energy power station during operation; From the current acoustic signal, components that match the background noise characteristics are suppressed by adaptive filtering or spectral subtraction methods.

[0079] In a preferred embodiment of this application, the acoustic signal preprocessing step is a core anti-interference design for the extremely noisy acoustic environment of new energy power stations. It is implemented by effectively separating and enhancing the weak acoustic signature features of drones from a mixed signal with strong background noise.

[0080] During periods when no drones are in use, the system systematically collects operating noise data from various typical equipment within the renewable energy power plant under different operating conditions. This includes areas near wind turbines, box-type transformers, the main transformer area of ​​the step-up substation, and photovoltaic inverter clusters. Equipment operating status parameters are recorded simultaneously during data collection to correlate noise characteristics.

[0081] The acquired raw noise signal is framed, windowed, and subjected to Fourier transform to extract its long-term spectral features. A feature model is established for each noise source. The feature model, along with its corresponding device ID, location, typical operating conditions, and other metadata, is stored in the database of the edge computing gateway or local server to form a queryable background noise feature library.

[0082] When an acoustic sensor node acquires a real-time signal containing what appears to be drone noise, the system first retrieves the most relevant background noise model from its feature library based on the acoustic node's location information and the site's real-time operating status. The short-time spectrum of the real-time signal is then matched against the retrieved background noise feature model.

[0083] For known, stable, strong line-spectral noise in the feature library, an adaptive notch filter is designed and applied directly in the time or frequency domain to accurately filter out the specific frequency and its harmonic components, while preserving as much UAV acoustic signature information as possible near that frequency. The processed signal is then output for subsequent time difference of arrival (TDOA) calculation and localization. At this point, the components of fixed background noise such as wind turbines in the signal have been significantly suppressed.

[0084] In this preferred embodiment, customized suppression is performed based on prior knowledge of the acoustic fingerprints of specific sites and equipment. It can distinguish noise differences between different wind turbine models or automatically switch processing strategies according to the wind turbine's start-up and shutdown status. In extreme cases where radio signals are subject to strong electromagnetic interference or complete silence, acoustic sensing becomes the primary reliance.

[0085] Specifically, in a preferred embodiment of this application, a method for unmanned aerial vehicle (UAV) fusion identification and positioning for new energy power stations further includes: The radio signals are analyzed and decoded to obtain remote identification broadcast information that conforms to a preset standard format; If decoding is successful, the drone's unique identifier and autonomously reported real-time location data are extracted from the remote identity broadcast information. When scanning frequency bands, radio sensor nodes simultaneously listen for signals in specific frequency bands that conform to the remote identity broadcast standard format.

[0086] The system utilizes a built-in communication protocol parsing engine, which integrates decoding rules for various mainstream protocols, including open standard protocols that define the publicly available information formats that drones must broadcast, as well as proprietary protocols. The engine attempts to decode the captured signals using feature matching and decoding algorithms.

[0087] If decoding is successful, two core fields will be extracted from the data frame: Unique identifiers for drones: such as serial numbers or remote IDs issued by the FAA.

[0088] Autonomously reported real-time location: latitude, longitude, altitude, speed, and heading obtained by the UAV's own GNSS module.

[0089] The autonomously reported real-time location is used as the third location estimate, and the corresponding real-time location confidence is set based on its communication protocol type. The third location estimate and its location confidence, together with the first location estimate, the second location estimate and their respective location confidence, are input into the weighted calculation process to generate the three-dimensional fusion positioning result.

[0090] Estimate the third position and its confidence level , with the first position estimate from radio TDOA (confidence level) ), and second position estimation from acoustic-optical coordination (confidence level) They should be placed on an equal footing.

[0091] According to the aforementioned fusion weight generation algorithm, calculate The integration weights of each of the three Generate the final fusion result and perform a weighted summation: This result represents the optimal localization solution that combines passive detection and active identity reporting.

[0092] In complex electromagnetic environments, when the accuracy of radio TDOA (Time-to-Area Address) degrades due to interference, if the UAV's broadcast signal remains clear, this location information will become the main contributor to the fusion result, effectively maintaining the stability of positioning accuracy and improving the system's survivability in adversarial environments. The GNSS position autonomously reported by the UAV, under conditions of good satellite signal, is itself a high-precision position source. Using it as a third position estimate in the fusion process, especially during system initialization or when a target suddenly appears, can provide a high-quality initial value for the filtering algorithm, accelerating trajectory convergence and reducing initial positioning errors.

[0093] Please see Figure 4 This invention also provides a drone fusion identification and positioning system for new energy power stations, which applies the aforementioned drone fusion identification and positioning method for new energy power stations, including: The data acquisition unit is used to acquire multimodal sensing data emitted or generated by a target UAV in the target airspace within the heterogeneous sensing network of the new energy power station. The multimodal sensing data includes at least radio signals, acoustic signals, and optical images. The heterogeneous sensing network is composed of at least one radio sensing node, at least one acoustic sensing node, and at least one optical sensing node deployed at a preset location on the perimeter of the new energy power station. The first position calculation unit is used to calculate the first position estimate of the target UAV based on the radio signal; The second position calculation unit is used to calculate the second position estimate of the target UAV based on acoustic signals and optical images; The confidence assessment unit is used to assess the real-time positioning confidence of the first position estimate and the second position estimate, respectively. The weight generation unit is used to generate fusion weights associated with each location estimate based on the real-time location confidence. The fusion positioning unit is used to perform weighted calculations on the first position estimate and the second position estimate according to the fusion weights to generate a three-dimensional fusion positioning result for the target UAV.

[0094] The functional explanation of each unit in this embodiment is the same as that of a UAV fusion identification and positioning method for new energy power stations, and the technical effects are the same, so it will not be repeated here.

[0095] Those skilled in the art will recognize that the units of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of the invention.

[0096] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical functional division. In actual implementation, there may be other division methods, such as multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored.

[0097] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0098] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0099] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the specification of the present invention.

Claims

1. A method for unmanned aerial vehicle (UAV) fusion identification and positioning for new energy power stations, characterized in that, Includes the following steps: The method acquires multimodal sensing data emitted or generated by a target UAV within the target airspace of a new energy power station in a heterogeneous sensing network. The multimodal sensing data includes at least radio signals, acoustic signals, and optical images. The heterogeneous sensing network is constructed by deploying at least one radio sensing node, at least one acoustic sensing node, and at least one optical sensing node at a predetermined location on the perimeter of the new energy power station. Based on the radio signals, a first position estimate of the target UAV is calculated; A second position estimate of the target UAV is calculated based on acoustic signals and optical images; Based on the first position estimate and the second position estimate, evaluate their corresponding real-time positioning reliability respectively; Based on the real-time location confidence, fusion weights associated with each location estimate are generated; The first position estimate and the second position estimate are weighted according to the fusion weight to generate the three-dimensional fusion positioning result of the target UAV.

2. The UAV fusion identification and positioning method for new energy power stations according to claim 1, characterized in that, Based on the radio signals, a first position estimate of the target UAV is calculated, specifically including: Based on the radio signal, the first position estimate of the target UAV is calculated using the time difference of arrival algorithm.

3. The UAV fusion identification and positioning method for new energy power stations according to claim 2, characterized in that, The second position estimate of the target UAV is calculated based on acoustic signals and optical images, specifically including: Based on the acoustic signal, the initial spatial region of the target UAV is calculated using the time difference of arrival method; Based on the optical image, the three-dimensional coordinates of the target UAV are calculated within the initial spatial region through visual target detection and stereo vision ranging, serving as the second position estimate.

4. The UAV fusion identification and positioning method for new energy power stations according to claim 1, characterized in that, The multimodal sensing data launched or generated by target UAVs in the target airspace within the heterogeneous sensor network of the new energy power station also includes: When any sensor node in the heterogeneous sensor network of the new energy power station detects a suspected drone target, a collaborative trigger signal is generated. Based on the aforementioned collaborative trigger signal, the remaining sensor nodes are awakened through the communication network within the site.

5. The UAV fusion identification and positioning method for new energy power stations according to claim 1, characterized in that, Before calculating the second position estimate of the target UAV based on acoustic signals and optical images, the method further includes: Call the pre-stored background noise feature library, which contains the noise spectrum features generated by typical equipment in the new energy power station during operation; From the current acoustic signal, components that match the background noise characteristics are suppressed by adaptive filtering or spectral subtraction methods.

6. The UAV fusion identification and positioning method for new energy power stations according to claim 1, characterized in that, Also includes: The radio signals are analyzed and decoded to obtain remote identification broadcast information that conforms to a preset standard format; If decoding is successful, the drone's unique identifier and autonomously reported real-time location data are extracted from the remote identity broadcast information. The autonomously reported real-time location is used as the third location estimate, and the corresponding real-time location confidence is set based on its communication protocol type. The third location estimate and its location confidence, together with the first location estimate, the second location estimate and their respective location confidence, are input into the weighted calculation process to generate the three-dimensional fusion positioning result.

7. The UAV fusion identification and positioning method for new energy power stations according to claim 2, characterized in that, The evaluation of the real-time location reliability includes: Based on the signal-to-noise ratio of the radio signal, acoustic signal, or optical image, a position estimation error model for the corresponding sensing mode is established. The current signal-to-noise ratio is matched with the error model to predict the confidence interval of the current estimate; The corresponding real-time location confidence is obtained based on the confidence interval or its reciprocal.

8. The UAV fusion identification and positioning method for new energy power stations according to claim 7, characterized in that, Based on the real-time location confidence, the fusion weights associated with each location estimate are generated as follows: The real-time location reliability of each location estimate is normalized. Based on a preset weight allocation strategy function, the normalized confidence scores are mapped to corresponding fusion weights; wherein the strategy function is a linear scaling function, a soft maximum function, or a discriminant function based on expert rules.

9. A method for unmanned aerial vehicle (UAV) fusion identification and positioning for new energy power stations according to claim 8, characterized in that, The first position estimate and the second position estimate are weighted according to the fusion weight to generate a 3D fusion localization result for the target UAV, including: The fusion weights are normalized, and based on the normalized fusion weights, the three-dimensional coordinates of the first position estimate and the second position estimate are weighted and summed. The weighted summation result is output as the three-dimensional fusion positioning result of the target UAV, and the result includes at least longitude, latitude and altitude information.

10. A drone fusion identification and positioning system for new energy power stations, characterized in that, The method for unmanned aerial vehicle (UAV) fusion identification and positioning for new energy power stations according to any one of claims 1 to 9 includes: The data acquisition unit is used to acquire multimodal sensing data emitted or generated by a target UAV in the target airspace within the heterogeneous sensing network of the new energy power station. The multimodal sensing data includes at least radio signals, acoustic signals, and optical images. The heterogeneous sensing network is composed of at least one radio sensing node, at least one acoustic sensing node, and at least one optical sensing node deployed at a preset location on the perimeter of the new energy power station. The first position calculation unit is used to calculate the first position estimate of the target UAV based on the radio signal; The second position calculation unit is used to calculate the second position estimate of the target UAV based on acoustic signals and optical images; The confidence assessment unit is used to assess the real-time positioning confidence of the first position estimate and the second position estimate, respectively. The weight generation unit is used to generate fusion weights associated with each location estimate based on the real-time location confidence. The fusion positioning unit is used to perform weighted calculations on the first position estimate and the second position estimate according to the fusion weights to generate a three-dimensional fusion positioning result for the target UAV.

Images