A method, device and system for detecting lighting parameters based on a drone

Abstract

The application discloses a kind of based on unmanned aerial vehicle's lighting parameter detection method, device and system, it is related to lighting parameter detection technical field.This scheme considers that handheld device detection exists blind area, and lighting parameter detection of lamp is realized using unmanned aerial vehicle;Meanwhile, to avoid the data deviation caused by atmospheric transmissivity and pose change in unmanned aerial vehicle measurement, multi-spectral imager and multi-axis attitude sensor are added in unmanned aerial vehicle;In the process of lamp lighting parameter detection, according to the optical data, attitude data and the environmental data obtained by the above-mentioned module acquisition executes physical simulation, determines the equivalent illuminance value of working surface after height compensation, and corrects the data deviation caused by atmospheric transmissivity and pose change;Finally, according to the working surface equivalent illuminance value and the relevant data collected, the light attenuation rate is predicted through light attenuation prediction model, and the detection time is greatly shortened compared with traditional method, effectively reduces the failure risk of urban lighting system, and guarantees the stability of urban lighting.

Description

Technical Field

[0001] This application relates to the field of lighting parameter detection technology, and in particular to a method, device and system for lighting parameter detection based on unmanned aerial vehicles (UAVs). Background Technology

[0002] Urban lighting facility inspection is a comprehensive assessment and inspection process for various lighting facilities in a city, aiming to ensure their normal operation, energy efficiency, and compliance with relevant standards. The inspection includes the optical performance of the luminaires, such as luminous flux, luminous efficacy, and color temperature, as well as electrical safety aspects such as electrical wiring, grounding systems, and leakage protection.

[0003] Currently, the testing of lighting parameters for urban lighting facilities faces challenges: traditional handheld devices struggle to perform non-contact measurements on high-altitude lighting fixtures such as those on overpasses and tunnel ceilings, as well as on expressways, creating blind spots. While drone measurements can partially address these issues, they often overlook the influence of environmental factors such as atmospheric transmittance and drone attitude. Under different weather conditions, such as heavy fog or sandstorms, atmospheric transmittance varies significantly, and changes in drone attitude also affect measurement data. The lack of mechanisms to address these complex environmental factors leads to substantial data deviations under different environments, compromising reliability.

[0004] In light of the above, how to address the blind spots in handheld device detection and the data deviations caused by atmospheric transmittance and pose changes in drone measurements during the detection of lighting parameters for urban lighting facilities, and the lack of effective countermeasures that affect data reliability, are urgent problems that technical personnel in this field need to solve. Summary of the Invention

[0005] The purpose of this application is to provide a method, device, and system for detecting lighting parameters based on unmanned aerial vehicles (UAVs), in order to solve the problems of blind spots in handheld device detection and data deviation caused by changes in atmospheric transmittance and pose during UAV measurement in current urban lighting facility lighting parameter detection, which lack effective countermeasures and affect data reliability.

[0006] To address the aforementioned technical problems, this application provides a method for detecting lighting parameters based on a UAV, applied to the UAV's data processing unit; the UAV includes at least a multispectral imager, a multi-axis attitude sensor, a main control module, and a data processing unit; the multispectral imager and the multi-axis attitude sensor are respectively communicatively connected to the main control module, and the main control module is communicatively connected to the data processing unit; the method includes:

[0007] Acquire relevant data of the target luminaire transmitted by the main control module; the relevant data includes at least optical data collected by the multispectral imager, attitude data collected by the multi-axis attitude sensor, and environmental data of the location of the target luminaire;

[0008] Physical simulations are performed based on optical data, attitude data, and environmental data to determine the equivalent illuminance value of the working surface after height compensation.

[0009] Optical data, attitude data, environmental data, and equivalent illuminance values ​​of the working surface are input into a pre-built light decay prediction model to output the light decay rate of the target luminaire. The light decay prediction model is a long short-term memory network model trained based on relevant data and corresponding light decay rates of multiple luminaires.

[0010] On the one hand, before performing physical simulation based on optical data, attitude data, and environmental data, after acquiring the relevant data of the target luminaire transmitted by the main control module, it also includes:

[0011] The Kalman filter algorithm is used to filter noise from the relevant data.

[0012] Unify the timestamps of all relevant data after noise filtering to the same time base;

[0013] The transmission delay of each relevant data after noise filtering is compensated using a linear interpolation method.

[0014] Based on the feature point matching algorithm, the relevant data after noise filtering are mapped to a unified coordinate system.

[0015] On the other hand, physical simulations are performed based on optical data, attitude data, and environmental data to determine the equivalent illuminance value of the working surface after height compensation, including:

[0016] Determine the standard reference altitude and obtain the atmospheric transmittance coefficient;

[0017] Based on optical data, attitude data, and environmental data, determine the original illuminance data, UAV flight altitude, UAV pitch angle, and atmospheric particulate matter concentration.

[0018] Call the physics simulation engine to determine whether the difference between the standard reference altitude and the drone's flight altitude is greater than a preset value;

[0019] If not, the equivalent illuminance value of the working surface will be directly generated based on the original illuminance data and the drone's pitch angle.

[0020] If so, then the atmospheric transmittance compensation factor is determined based on the atmospheric transmittance coefficient, atmospheric particulate matter concentration, and UAV flight altitude.

[0021] Based on the original illuminance data, standard reference altitude, UAV flight altitude, UAV pitch angle, and atmospheric transmittance compensation factor, the equivalent illuminance value of the working surface after altitude compensation is generated.

[0022] On the other hand, the process of constructing the light decay prediction model includes:

[0023] Acquire the location data, light decay label, and corresponding equivalent illuminance value of the working surface of the luminaire;

[0024] Acquire relevant data about the lighting equipment collected by the drone; the relevant data includes at least optical data collected by the multispectral imager, attitude data collected by the multi-axis attitude sensor, and environmental data collected by the environmental parameter acquisition module.

[0025] The location data, light decay labels, corresponding equivalent illuminance values ​​of the working surface, and related data of the lamps were normalized to obtain normalized data.

[0026] Time series samples are generated using sliding window technology and normalized data, and training set data and validation set data are divided according to the time series samples;

[0027] An initial light decay prediction model is constructed based on the training set data; the output of the initial light decay prediction model includes the light decay rate, the importance score of the luminaire, the atmospheric transmittance coefficient, and the prediction time range.

[0028] The model parameters of the initial light decay prediction model are adjusted by an adaptive moment estimation optimizer;

[0029] The prediction error of the initial light decay prediction model is determined using the mean squared error loss function.

[0030] The initial light decay prediction model is validated based on the validation set data to determine whether the prediction error of the initial light decay prediction model meets the preset requirements.

[0031] If not, return to the step of adjusting the model parameters of the initial light decay prediction model through the adaptive moment estimator optimizer to iteratively train the initial light decay prediction model;

[0032] If so, the initial light decay prediction model will be determined as the light decay prediction model.

[0033] On the other hand, after obtaining the light decay rate of the target luminaire, the following is also included:

[0034] Obtain the new atmospheric transmittance coefficient and prediction time range output by the light decay prediction model;

[0035] The new atmospheric transmittance coefficient is fed back into the physics simulation engine;

[0036] The predicted time range, the light decay rate of the target luminaire, the importance score, and related data are transmitted to the control center so that the control center can generate operation and maintenance decisions for the target luminaire based on the predicted time range, the light decay rate of the target luminaire, and the importance score.

[0037] On the other hand, the control center generates operation and maintenance decisions for the target luminaires based on the predicted time range, the light decay rate of the target luminaires, and their importance score, including:

[0038] Determine the failure risk score of the target luminaire based on relevant data;

[0039] The first weight for the light decay rate of the target luminaire, the second weight for the importance score of the target luminaire, and the third weight for the failure risk score of the target luminaire are obtained.

[0040] The overall score of the target luminaire is determined based on its light decay rate and its corresponding first weight, its importance score and its corresponding second weight, and its failure risk score and its corresponding third weight.

[0041] The overall score of the target luminaire is sorted in descending order with the overall scores of the other luminaires to determine the maintenance priority of the target luminaire;

[0042] Maintenance decisions for the target lighting fixtures are generated based on their maintenance priority and predicted timeframe.

[0043] On the other hand, the drone collects relevant data about the target lighting fixtures, including:

[0044] Acquire basic information about each light fixture and each drone within the detection area; the basic information about the light fixtures includes at least the location, number, and distribution density of the light fixtures; the basic information about the drones includes at least the current location, endurance, and flight speed of the drones.

[0045] The detection region is divided into multiple sub-regions based on the Voronoi diagram algorithm;

[0046] The testing workload for each sub-area is determined based on the basic information of each lamp.

[0047] Based on the detection workload of each sub-region and the basic information of each UAV, detection tasks are assigned to each UAV so that each UAV can perform the corresponding detection tasks.

[0048] All testing tasks include the task of collecting relevant data on the target lighting fixtures.

[0049] On the other hand, after each drone performs its corresponding detection task, the process also includes:

[0050] Synchronize the time of each drone and add a timestamp to the data collected by the corresponding detection task;

[0051] The collected data is sorted according to the timestamp of the data collected by each drone;

[0052] The data collected by each drone in the same time period are aligned using a sliding window interpolation algorithm;

[0053] Determine the location of each drone, calibrate the collected data of each drone to the same coordinate system based on the location of each drone, and verify the consistency of each collected data.

[0054] To address the aforementioned technical problems, this application also provides a lighting parameter detection device based on a UAV, applied to the data processing unit of the UAV; the UAV includes at least a multispectral imager, a multi-axis attitude sensor, a main control module, and a data processing unit; the multispectral imager and the multi-axis attitude sensor are respectively communicatively connected to the main control module, and the main control module is communicatively connected to the data processing unit; the device includes:

[0055] The acquisition module is used to acquire relevant data of the target lamp transmitted by the main control module; the relevant data includes at least optical data collected by the multispectral imager, attitude data collected by the multi-axis attitude sensor, and environmental data of the location of the target lamp.

[0056] The simulation module is used to perform physical simulations based on optical data, attitude data, and environmental data to determine the equivalent illuminance value of the working surface after height compensation.

[0057] The prediction module is used to input optical data, attitude data, environmental data, and equivalent illuminance values ​​of the working surface into a pre-built light decay prediction model to output the light decay rate of the target luminaire. The light decay prediction model is a long short-term memory network model trained based on relevant data and corresponding light decay rates of multiple luminaires.

[0058] To address the aforementioned technical problems, this application also provides a UAV-based lighting parameter detection system, comprising a control center and a UAV; the control center is communicatively connected to the UAV; the UAV includes at least a multispectral imager, a multi-axis attitude sensor main control module, and a data processing unit; the UAV's data processing unit is capable of executing the steps of the aforementioned UAV-based lighting parameter detection method during operation.

[0059] The UAV-based lighting parameter detection method provided in this application has the following advantages: Considering the blind spots in handheld device detection, the use of a UAV eliminates these blind spots. Furthermore, to avoid data deviations caused by atmospheric transmittance and pose changes during UAV measurements, a multispectral imager and a multi-axis attitude sensor are added to the UAV. These are used to collect optical and attitude data related to the target lighting fixture, respectively, while also acquiring environmental data of the target lighting fixture's location. During the lighting parameter detection process, physical simulation is performed based on the optical, attitude, and environmental data to determine the equivalent illuminance value of the working surface after height compensation, correcting data deviations caused by atmospheric transmittance and pose changes. This method can operate normally under various complex weather conditions. Finally, based on the equivalent illuminance value of the working surface and the collected relevant data, a pre-constructed light decay prediction model is used to predict the light decay rate. This significantly shortens the detection time compared to traditional methods, providing valuable time for preventative maintenance of lighting fixtures, effectively reducing the failure risk of urban lighting systems, and ensuring the stability of urban lighting.

[0060] In addition, this application also provides a lighting parameter detection device and a lighting parameter detection system based on a drone, with the same effect as above. Attached Figure Description

[0061] To more clearly illustrate the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0062] Figure 1 A schematic diagram of the drone provided in the embodiments of this application;

[0063] Figure 2 A flowchart illustrating a method for detecting lighting parameters based on a drone, provided as an embodiment of this application;

[0064] Figure 3 This application provides an architecture diagram of a drone-based lighting parameter detection system.

[0065] Figure 4 A flowchart of the height compensation algorithm for the physics simulation engine provided in this application embodiment;

[0066] Figure 5 This is an architectural diagram of the optical attenuation prediction model provided in the embodiments of this application;

[0067] Figure 6 This is a timing diagram of multi-machine collaborative detection provided in an embodiment of this application;

[0068] Figure 7 A flowchart for urban road detection provided in this application embodiment;

[0069] Figure 8 A schematic diagram of a lighting parameter detection device based on a drone provided in an embodiment of this application;

[0070] Figure 9 This is a schematic diagram of a drone-based lighting parameter detection system provided in an embodiment of this application. Detailed Implementation

[0071] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of this application.

[0072] The core of this application is to provide a method, device, and system for detecting lighting parameters based on unmanned aerial vehicles (UAVs), in order to solve the problems of blind spots in the detection of lighting parameters of current urban lighting facilities, data deviations caused by changes in atmospheric transmittance and pose during UAV measurements, the lack of effective countermeasures, and the impact on data reliability.

[0073] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0074] Figure 1 This is a schematic diagram of a drone provided in an embodiment of this application. Figure 1 As shown, this application adds a detection platform with a detection probe to the existing UAV carrier. This detection platform includes at least a multispectral imager, a multi-axis attitude sensor, a main control module, and a data processing unit. Specifically, the UAV-based lighting parameter detection method provided in this application is applied to the UAV's data processing unit. The following is a detailed description of each module / unit added to the UAV carrier in this application:

[0075] The testing platform is equipped with a high-resolution multispectral imager that precisely covers the visible and near-infrared bands. It can capture the luminous characteristics of lamps under different spectra, providing data for parameter analysis. This allows for the analysis of the luminous intensity of lamps under different spectra and the determination of parameters such as color temperature and color rendering index.

[0076] A high-precision multi-axis attitude sensor is employed, achieving a measurement accuracy of ±0.1°, to perceive real-time changes in the UAV's attitude and provide crucial data for the altitude compensation algorithm. It should be noted that this embodiment does not limit the specific type of multi-axis attitude sensor; however, considering measurement accuracy, in some embodiments, it is recommended to use at least a nine-axis attitude sensor.

[0077] The main control module is responsible for coordinating the synchronous operation of devices such as the multispectral imager and multi-axis sensors, receiving, buffering, and preprocessing sensor data in real time (such as attitude calculation and environmental parameter calibration). Since a unified bus can lead to bandwidth waste or timing conflicts, reducing system stability, the main control module communicates with other modules through different bus protocols, such as Serial Peripheral Interface (SPI), Inter-Integrated Circuit (I2C), and Controller Area Network (CAN) buses, to control data flow. Simultaneously, the main control module can receive environmental data from the target lighting location via wireless communication, including but not limited to temperature, humidity, and atmospheric particulate matter concentration, providing environmental information for data correction and analysis. For example, when particulate matter concentration is high, the measurement data is corrected based on this parameter. This embodiment does not restrict the source of environmental data; for example, it can be obtained based on authoritative environmental monitoring data or through an external environmental parameter acquisition module. The environmental parameter acquisition module can be independent of the UAV or be a module within the detection platform, depending on the specific implementation.

[0078] In addition, the main control module is communicatively connected to the data processing unit and is responsible for transmitting the data collected by each sensor. This embodiment does not limit the specific method of communication between the main control module and the data processing unit; for example, they can be connected via Universal Serial Bus 3.0 or Peripheral Component Interconnect Express (PCIe) bus.

[0079] Figure 2 This is a flowchart illustrating a method for detecting lighting parameters based on a drone, as provided in an embodiment of this application. Figure 2 As shown, the method includes:

[0080] S10: Obtain relevant data of the target lighting fixture transmitted by the main control module.

[0081] The relevant data includes at least optical data collected by a multispectral imager, attitude data collected by a multi-axis attitude sensor, and environmental data of the location of the target luminaire.

[0082] To detect the lighting parameters of the target light fixture, the first step is to control a drone to fly near it and collect relevant data using the various sensor modules on its newly added detection platform. It's important to note that this data should include at least the optical data collected by the multispectral imager, such as illuminance, color temperature, and display index; the attitude data collected by the multi-axis attitude sensor, such as the drone's pitch and yaw angles; and the environmental data of the target light fixture's location, such as PM2.5 concentration, temperature, and humidity.

[0083] After each sensor module collects its corresponding data, the data is transmitted to the main control module, which then transmits the relevant data of the target lighting fixture to the data processing unit.

[0084] S11: Perform physical simulation based on optical data, attitude data, and environmental data to determine the equivalent illuminance value of the working surface after height compensation.

[0085] Figure 3 This is a diagram illustrating the architecture of a drone-based lighting parameter detection system provided in an embodiment of this application. Figure 3 As shown, the data processing unit embeds a lightweight physics simulation engine optimized for UAV scenarios. With limited hardware resources, it quickly and accurately simulates light distribution, providing data support for altitude compensation and parameter calculation. This embodiment does not limit the specific type of physics simulation engine; for example, DIALux, Relux, or Lighttools can be used, depending on the specific implementation.

[0086] Specifically, the physics simulation engine takes into account collected optical, attitude, and environmental data, such as luminaire type, power, light distribution curve, standard altitude, ambient temperature and humidity, and atmospheric transmittance. It then uses the engine to construct a light intensity distribution cloud map of the luminaire at the standard altitude and outputs a spatial matrix containing parameters such as illuminance and color temperature. Finally, altitude compensation is performed using UAV attitude data to dynamically correct the detection data, ultimately outputting the equivalent illuminance value of the working surface after altitude compensation.

[0087] It should be noted that this embodiment does not restrict the specific process of determining the equivalent illuminance value of the working surface after height compensation through the physical simulation engine.

[0088] S12: Input optical data, attitude data, environmental data, and equivalent illuminance values ​​of the working surface into the pre-built light decay prediction model to output the light decay rate of the target luminaire.

[0089] Among them, the light decay prediction model is a long short-term memory network model trained based on relevant data of multiple lamps and their corresponding light decay rates.

[0090] Furthermore, such as Figure 3As shown, a federated learning framework is pre-built in the data processing unit, supporting multi-machine parameter sharing and collaborative learning, integrating data from multiple UAVs, and improving system detection performance. Specifically, the federated learning framework incorporates a pre-built light decay prediction model, which is a Long Short-Term Memory (LSTM) network model trained based on relevant data and corresponding light decay rates of multiple luminaires. Optical data, attitude data, environmental data, and equivalent illuminance values ​​of the working surface are input into the pre-built light decay prediction model, and the light decay rate of the target luminaire is finally output. This enables the detection of the lighting parameters of the target luminaire, allowing maintenance personnel to perform maintenance on the target luminaire based on its light decay rate.

[0091] It should be noted that this embodiment does not impose restrictions on the specific construction process of the light decay prediction model, and it depends on the specific implementation. Furthermore, data such as the light decay rate of the target luminaires can be transmitted to a 3D visualization terminal via a WebSocket interface, allowing maintenance personnel to maintain the target luminaires based on the information displayed on the terminal.

[0092] In this embodiment, considering the blind spots in handheld device detection, a drone is specifically used to detect the lighting parameters of the lamps, eliminating the detection blind spots. Simultaneously, to avoid data deviations caused by changes in atmospheric transmittance and pose during drone measurements, a multispectral imager and a multi-axis attitude sensor are added to the drone. These are used to collect optical and attitude data related to the target lamps, respectively, while also acquiring environmental data of the target lamp's location. During the lamp lighting parameter detection process, physical simulation is performed based on the optical, attitude, and environmental data to determine the equivalent illuminance value of the working surface after height compensation, correcting the data deviations caused by changes in atmospheric transmittance and pose. This allows for normal operation under various complex weather conditions. Finally, based on the equivalent illuminance value of the working surface and the collected relevant data, a pre-constructed light decay prediction model is used to predict the light decay rate. The detection time is significantly shortened compared to traditional methods, saving valuable time for preventative maintenance of the lamps, effectively reducing the failure risk of urban lighting systems, and ensuring the stability of urban lighting.

[0093] like Figure 3 As shown, to ensure the accuracy and consistency of the relevant data, a data preprocessing module is also included in the data processing unit. Specifically, in some embodiments, before performing physical simulation based on optical data, attitude data, and environmental data, after acquiring the relevant data of the target luminaire transmitted by the main control module, the following steps are also included:

[0094] S13: Perform noise filtering on the relevant data according to the Kalman filter algorithm.

[0095] S14: Unify the timestamps of all relevant data after noise filtering to the same time base.

[0096] S15: Compensate for the transmission delay of each relevant data after noise filtering using a linear interpolation method.

[0097] S16: Map the noise-filtered relevant data to a unified coordinate system based on the feature point matching algorithm.

[0098] In this embodiment, the data preprocessing module can quickly and accurately perform noise filtering and spatiotemporal alignment on the collected relevant data, employing an adaptive filtering algorithm to denoise and ensure data accuracy and consistency. Specifically, a Kalman filter algorithm is used to estimate the noise covariance matrix in real time, dynamically adjust the filter gain, and suppress random noise and periodic interference, thereby performing noise filtering on each relevant data. Furthermore, the timestamps of each relevant data after noise filtering are unified to the same time base, and the transmission delay of each relevant data after noise filtering is compensated using a linear interpolation method. Finally, the relevant data after noise filtering is mapped to a unified coordinate system using a feature point matching algorithm. This ensures the accuracy and consistency of the relevant data.

[0099] Figure 4 A flowchart of the height compensation algorithm for the physics simulation engine provided in this application embodiment. Based on the above embodiments, as... Figure 4 As shown, in some embodiments, physical simulation is performed based on optical data, attitude data, and environmental data to determine the equivalent illuminance value of the work surface after height compensation, including:

[0100] S111: Determine the standard reference altitude and obtain the atmospheric transmittance coefficient.

[0101] S112: Based on optical data, attitude data, and environmental data, determine the raw illuminance data, UAV flight altitude, UAV pitch angle, and atmospheric particulate matter concentration.

[0102] S113: Call the physics simulation engine to determine whether the difference between the standard reference altitude and the drone's flight altitude is greater than a preset value; if not, proceed to step S114; if yes, proceed to step S115.

[0103] S114: Based on the original illuminance data and the UAV pitch angle, directly generate the equivalent illuminance value of the working surface.

[0104] S115: Determine the atmospheric transmittance compensation factor based on the atmospheric transmittance coefficient, atmospheric particulate matter concentration, and UAV flight altitude.

[0105] S116: Generate the equivalent illuminance value of the working surface after altitude compensation based on the original illuminance data, standard reference altitude, UAV flight altitude, UAV pitch angle, and atmospheric transmittance compensation factor.

[0106] Specifically, a detailed and accurate light distribution model at a standard altitude is pre-generated in the physical simulation engine. This model has undergone multiple verifications and calibrations, serving as the benchmark for all subsequent calculations and ensuring their accuracy and reliability. Furthermore, through mathematical derivation and physical model construction, a precise and stable mapping relationship is established between the UAV's real-time pose and the simulation data.

[0107] ;

[0108] in, This is the equivalent illuminance value for the working surface, used to convert the measured data of the UAV into the theoretical illuminance value at the standard altitude, eliminating the influence of altitude, attitude and environmental interference. The measured values ​​are obtained by equipping the drone with an illumination sensor, which is the raw illumination data collected in real time by a multispectral imager. The standard reference height (e.g., 10m) is used as a preset reference height in the physical simulation model to establish the light distribution reference, and is generated based on the physical simulation engine.

[0109] Based on the above mapping relationship, in order to determine the equivalent illuminance value of the working surface after height compensation, it is first necessary to determine the standard reference height. And obtain the atmospheric transmittance coefficient. It should be noted that the atmospheric transmittance coefficient It is the wavelength-dependent attenuation coefficient (unit: The light intensity attenuation coefficient (LQC) reflects the ability of light of a specific wavelength to be absorbed and scattered by fine particulate matter in the atmosphere. This coefficient needs to be obtained through experimental calibration (measuring light intensity attenuation curves at different PM2.5 concentrations using a standard light source in a dark room). The LQC will vary depending on the wavelength of light and environmental conditions. PM2.5 is the concentration of fine particulate matter in the atmosphere, which can be obtained in real time through air quality monitoring equipment. The higher the PM2.5 concentration, the stronger the atmosphere's attenuation effect on light.

[0110] Furthermore, based on the collected optical data, attitude data, and environmental data of the target luminaire, the raw illuminance data is determined. Drone flight altitude UAV pitch angle And atmospheric particulate matter concentration (PM2.5). It should be noted that the drone's flight altitude... This refers to the drone's real-time flight altitude, measured by a high-precision laser altitude sensor on the drone. The drone's pitch angle... It is the angle between the drone's fuselage and the horizontal plane, measured by a multi-axis attitude sensor, used to correct for detection angle deviations caused by the drone's tilt (application of the cosine theorem).

[0111] Subsequently, the physics simulation engine is invoked to determine whether the difference between the standard reference altitude and the drone's flight altitude is greater than a preset value. In this embodiment, the size of the preset value is not limited; for example, it can be set to 0.5m. If it is confirmed that the difference between the standard reference altitude and the drone's flight altitude is not greater than the preset value, it is considered that the difference between the two is small, and no altitude compensation or atmospheric compensation is required. The equivalent illuminance value for the working surface can be directly generated based on the original illuminance data and the drone's pitch angle, as detailed below:

[0112] ;

[0113] in, This is the equivalent illuminance value for the working surface (without height compensation). Actual measured values ​​from the illuminance sensor mounted on the drone. This refers to the pitch angle of the drone.

[0114] If the difference between the standard reference altitude and the UAV's flight altitude is confirmed to be greater than a preset value, it is considered that the difference between the two is significant, and altitude and atmospheric compensation are required to correct data deviations caused by changes in atmospheric transmittance and attitude. Specifically, an atmospheric transmittance compensation factor is determined based on the atmospheric transmittance coefficient, atmospheric particulate matter concentration, and UAV flight altitude, as follows:

[0115] ;

[0116] in, This is the atmospheric transmittance compensation factor. Here, represents the atmospheric transmittance coefficient, and PM2.5 represents the concentration of atmospheric particulate matter. This represents the drone's flight altitude. It is understandable that the above formula fully considers the impact of atmospheric composition on light transmission under different weather conditions.

[0117] Finally, based on the original illuminance data, standard reference altitude, UAV flight altitude, UAV pitch angle, and atmospheric transmittance compensation factor, the equivalent illuminance value of the working surface after altitude compensation is generated, as follows:

[0118] ;

[0119] in, This is the equivalent illuminance value of the working surface (after height compensation). Actual measured values ​​from the illuminance sensor mounted on the drone. For standard reference height, The altitude at which the drone flies. For the drone's pitch angle, This is the atmospheric transmittance compensation factor.

[0120] In summary, this embodiment specifically uses the mapping relationship between the real-time pose of the UAV and the simulation data, and corrects it through inverse square ratio ( To compensate for the attenuation of light intensity due to changes in height, angle correction is used. Eliminate the effects of drone attitude tilt; through environmental correction ( It compensates for atmospheric absorption of light; finally, it outputs the illuminance value of the working surface. This corrects for data biases caused by changes in atmospheric transmittance and pose.

[0121] Figure 5 This is an architecture diagram of the optical attenuation prediction model provided in an embodiment of this application. Based on the above embodiments, in some embodiments, such as... Figure 5 As shown, the construction process of the light decay prediction model includes:

[0122] S121: Obtain the location data of the luminaire, the light decay label, and the corresponding equivalent illuminance value of the working surface.

[0123] S122: Acquire relevant data about the lighting fixtures collected by the drone.

[0124] The relevant data includes at least optical data acquired by the multispectral imager, attitude data acquired by the multi-axis attitude sensor, and environmental data acquired by the environmental parameter acquisition module.

[0125] S123: Normalize the location data of the luminaires, the light decay label, the corresponding equivalent illuminance value of the working surface, and related data to obtain normalized data.

[0126] S124: Generate time series samples based on sliding window technology and normalized data, and divide the training set data and validation set data according to the time series samples.

[0127] S125: Construct an initial light decay prediction model based on the training set data.

[0128] The output of the initial light decay prediction model includes the light decay rate, the importance score of the luminaire, the atmospheric transmittance coefficient, and the prediction time range.

[0129] S126: Adjust the model parameters of the initial light decay prediction model using an adaptive moment estimation optimizer.

[0130] S127: Use the mean squared error loss function to determine the prediction error of the initial light decay prediction model.

[0131] S128: Validate the initial light decay prediction model based on the validation set data and determine whether the prediction error of the initial light decay prediction model meets the preset requirements; if not, return to step S126 to iteratively train the initial light decay prediction model; if yes, determine the initial light decay prediction model as the light decay prediction model.

[0132] To construct a light decay prediction model, historical data from multiple luminaires needs to be collected first. Specifically, this involves acquiring the luminaire's location data, light decay labels, and corresponding equivalent illuminance values ​​for the working surface. The light decay labels are manually annotated historical maintenance records of the luminaires. Simultaneously, relevant data from luminaires collected by a drone is also acquired. This data will include at least optical data collected by a multispectral imager, such as illuminance, color temperature, and color rendering index; attitude data collected by a multi-axis attitude sensor, such as the drone's pitch and yaw angles; and environmental data collected by an environmental parameter acquisition module, such as PM2.5 concentration, temperature, and humidity.

[0133] The location data, light decay labels, corresponding equivalent illuminance values ​​of the working surface, and related data of the luminaires are further normalized to obtain normalized data. Time series samples are generated based on the sliding window technique and the normalized data, and training set data and validation set data are divided according to the time series samples. In this embodiment, there is no limitation on the specific ratio of training set data and validation set data.

[0134] Subsequently, an initial light decay prediction model was constructed based on the training set data. The input included multiple features such as illuminance, color temperature, PM2.5 concentration, flight altitude, and lamp usage time. The output was the predicted light decay rate, along with the lamp's importance score, atmospheric transmittance coefficient, and prediction time range. It should be noted that the lamp's importance score considers factors such as pedestrian and vehicle traffic at its location, and whether it is a critical area (e.g., transportation hub, hospital, school). The importance score is assessed by combining the lamp's location data with relevant statistical data (e.g., traffic flow statistics, population distribution data). The prediction time range represents the future time interval corresponding to the predicted light decay rate, such as the next month or the next three months, and is used to generate corresponding maintenance strategies.

[0135] Furthermore, the model parameters of the initial light attenuation prediction model are adjusted using the Adaptive Moment Estimation (Adam) optimizer, and the prediction error of the initial light attenuation prediction model is determined using the Mean Squared Error (MSE) loss function. In a specific implementation, five complete training iterations (epochs) can be performed, with each parameter update using 32 samples for computation. Finally, the initial light attenuation prediction model is validated based on the validation set data to determine whether the prediction error of the initial light attenuation prediction model meets the preset requirements. In this embodiment, there are no restrictions on the preset requirements; for example, an accuracy improvement of more than 0.5% is acceptable.

[0136] If it is confirmed that the prediction error of the initial light decay prediction model does not meet the preset requirements, the process returns to the step of adjusting the model parameters of the initial light decay prediction model through the adaptive moment estimator optimizer to iteratively train the initial light decay prediction model. If it is confirmed that the prediction error of the initial light decay prediction model does not meet the preset requirements, the initial light decay prediction model is determined as the light decay prediction model, so as to facilitate subsequent use of the light decay prediction model for light decay rate prediction, as well as the prediction and updating of the luminaire importance score, atmospheric transmittance coefficient, and prediction time range.

[0137] Based on the above embodiments, in some embodiments, after obtaining the light decay rate of the target luminaire, the method further includes:

[0138] S17: Obtain the new atmospheric transmittance coefficient and prediction time range output by the light decay prediction model.

[0139] S18: Feeds the new atmospheric transmittance coefficient back to the physics simulation engine.

[0140] S19: Transmit the predicted time range, the light decay rate of the target luminaire, the importance score, and related data to the control center so that the control center can generate operation and maintenance decisions for the target luminaire based on the predicted time range, the light decay rate of the target luminaire, and the importance score.

[0141] like Figure 3 As shown, to better maintain the target lighting equipment, this application also includes a control center. The control center contains an operation and maintenance decision-making unit, used to collect data from multiple UAVs, perform centralized analysis and decision-making, and facilitate operation and monitoring by maintenance personnel. It is understood that the operation and maintenance decision-making unit of the control center is communicatively connected to the data processing unit of the UAVs; this embodiment does not limit the specific communication protocol used.

[0142] Specifically, after obtaining the light decay rate of the target lamp, the UAV's data processing unit obtains the new atmospheric transmittance coefficient and prediction time range output by the light decay prediction model, and feeds the new atmospheric transmittance coefficient back to the physical simulation engine so that the physical simulation engine can use the new atmospheric transmittance coefficient when performing the next altitude compensation, thereby improving the accuracy of the output working surface equivalent illuminance value.

[0143] Subsequently, the UAV's data processing unit transmits the prediction time range, the light decay rate of the target luminaire, its importance score, and related data to the control center. This allows the control center to generate maintenance decisions for the target luminaire based on the prediction time range, the light decay rate of the target luminaire, and its importance score. The details are explained below:

[0144] Specifically, in some embodiments, the control center generates operation and maintenance decisions for the target luminaire based on the predicted time range, the light decay rate of the target luminaire, and its importance score, including:

[0145] S191: Determine the failure risk score of the target luminaire based on the relevant data of the target luminaire.

[0146] S192: Obtain the first weight of the light decay rate of the target luminaire, the second weight of the importance score of the target luminaire, and the third weight of the failure risk score of the target luminaire.

[0147] S193: Determine the comprehensive score of the target luminaire based on its light decay rate and its corresponding first weight, its importance score and its corresponding second weight, and its failure risk score and its corresponding third weight.

[0148] S194: Sort the overall score of the target luminaire in descending order with the overall scores of the other luminaires to determine the maintenance priority of the target luminaire.

[0149] S195: Generate maintenance decisions for the target lighting fixtures based on their maintenance priority and predicted time range.

[0150] First, the failure risk score of the target luminaire is determined based on relevant data. Specifically, this considers factors such as the luminaire's usage time, light decay, environmental factors, and historical failure records. Luminaires with longer usage time, more severe light decay, and harsher environmental conditions generally have a higher failure risk. Machine learning models can be used, taking these factors as input, to predict the failure risk probability of the luminaire; however, this embodiment does not impose any limitations on this approach.

[0151] Further, the first weight of the light decay rate of the target luminaire, the second weight of the importance score of the target luminaire, and the third weight of the failure risk score of the target luminaire are obtained. Based on the light decay rate of the target luminaire and its corresponding first weight, the importance score of the target luminaire and its corresponding second weight, and the failure risk score of the target luminaire and its corresponding third weight, the comprehensive score of the target luminaire is determined as follows:

[0152] Overall score = light decay rate First weight + importance score Second weighting + failure risk score Third weight.

[0153] It should be noted that in this embodiment, the magnitudes of the first, second, and third weights are not limited and will be determined according to the specific implementation.

[0154] Subsequently, the overall score of the target luminaire is sorted in descending order compared to the overall scores of the remaining luminaires to determine the maintenance priority of the target luminaire. It is understandable that maintenance personnel may need to maintain multiple luminaires, including the target luminaire. To determine the maintenance order, the maintenance priority of the target luminaire needs to be determined, so that maintenance decisions for the target luminaire can be generated based on its maintenance priority and the predicted time range. It should also be noted that the predicted time range indicates when the target luminaire will be maintained.

[0155] Figure 6 This is a timing diagram for multi-drone collaborative detection provided in an embodiment of this application. To ensure that the detection tasks of each drone are balanced and efficient, and to avoid situations where some drones have excessively heavy or light workloads, based on the above embodiments, in some embodiments, such as... Figure 6 As shown, the drone collects relevant data on the target lighting fixtures, including:

[0156] S101: Obtain basic information about each light fixture and each drone within the detection area.

[0157] The basic information for lighting fixtures includes at least their location, quantity, and distribution density; the basic information for drones includes at least their current location, endurance, and flight speed.

[0158] S102: Divide the detection region into multiple sub-regions according to the Voronoi diagram algorithm.

[0159] S103: Determine the testing workload for each sub-area based on the basic information of each lamp.

[0160] S104: Based on the detection workload of each sub-region and the basic information of each UAV, assign detection tasks to each UAV so that each UAV can perform the corresponding detection tasks.

[0161] All testing tasks include the task of collecting relevant data on the target lighting fixtures.

[0162] This application also provides a process for balancing the distribution of multi-machine inspection tasks. Specifically, it obtains basic information about each light fixture and each drone within the inspection area. It is important to note that the basic information for light fixtures includes at least their location, quantity, and distribution density; the basic information for drones includes at least their current location, endurance, and flight speed. By analyzing the distribution of light fixtures, the inspection difficulty and workload for different areas are determined.

[0163] The detection area is further divided into multiple sub-regions using the Voronoi diagram algorithm. The Voronoi diagram algorithm is a distance-based geometric partitioning method designed to divide a plane or space into multiple regions, each corresponding to a specific seed point, where all points within that region are closer to the seed point than to other seed points. Specifically, the algorithm is implemented through the following steps: First, a set of discrete points are selected as seed points. Then, for each seed point, all points closer to that point than to other seed points are found; the region formed by these points is the Voronoi region of that seed point. The boundary between adjacent Voronoi regions is the perpendicular bisector (in two dimensions) or perpendicular bisector plane (in three dimensions) between the two seed points. In this embodiment, the Voronoi diagram algorithm is used to divide the detection area into multiple sub-regions, ensuring that the lights within each sub-region are relatively concentrated. Simultaneously, the flight range and endurance of the drone are considered to ensure that each sub-region can be effectively covered by the drone.

[0164] Subsequently, the detection workload for each sub-area is determined based on the basic information of each lamp. Then, based on the detection workload of each sub-area and the basic information of each drone, detection tasks are assigned to each drone. The goal is to balance the workload of each drone as much as possible, which can be achieved by minimizing the variance of the task time for each drone. It is understood that all detection tasks include the collection of relevant data from the target lamps. Furthermore, the task progress and status of the drones can be monitored in real time during the detection process. If a drone completes its task ahead of schedule or malfunctions, a genetic algorithm can be used to reassign tasks, ensuring the overall efficiency of the detection task.

[0165] To ensure the accuracy and reliability of the collected data, avoid analytical errors caused by data inconsistencies, and improve data quality and usability, based on the above embodiments, in some embodiments, after each UAV performs its corresponding detection task, the following is also included:

[0166] S105: Synchronize the time of each UAV and add a timestamp to the data collected by the corresponding detection task.

[0167] S106: Sort the collected data according to the timestamp of the data collected by each UAV.

[0168] S107: Align the data collected by each UAV in the same time period using a sliding window interpolation algorithm.

[0169] S108: Determine the position of each UAV, calibrate the collected data of each UAV to the same coordinate system according to the position of each UAV, and verify the consistency of each collected data.

[0170] During the data aggregation phase, rigorous and meticulous spatiotemporal consistency verification is performed. A specialized verification algorithm is designed to precisely align and verify data collected by multiple drones in both time and space. Specifically, all drones are first synchronized using GPS or other high-precision timing systems to ensure accurate and consistent timestamps in the collected data. During data acquisition, a precise timestamp is added to each data point. The collected data is sorted by timestamp, and a sliding window interpolation algorithm is used to align data collected by different drones within the same time period. A time deviation threshold is set; if the time deviation of a data point exceeds the threshold, it is marked as abnormal data. Subsequently, the spatial location information of the drones is obtained using their positioning systems. Feature points in the images collected by different drones are extracted using feature matching algorithms (such as SIFT and SURF), the matching relationships between feature points are calculated, and the spatial transformation matrix is ​​solved to calibrate the data collected by different drones to the same coordinate system. Finally, the consistency of the calibrated data is verified using statistical analysis methods, such as calculating the mean, variance, and standard deviation of the data. A data deviation threshold is set; if the deviation of a data point from other data points exceeds the threshold, the data is considered potentially erroneous and further inspection and correction are performed. This ensures that the aggregated data is accurate and reliable, avoids analytical errors caused by data inconsistencies, and improves data quality and usability.

[0171] To enable those skilled in the art to better understand this solution, the following description is provided in conjunction with the appendix. Figure 7 Further explanation of the technical solution of this application:

[0172] Figure 7 A flowchart illustrating urban road detection provided in an embodiment of this application. Figure 7 As shown, detailed and accurate Geographic Information System (GIS) data is first imported, which has undergone on-site measurements and multiple calibrations. Using this data, combined with a self-optimized physical simulation engine, a highly accurate standard altitude simulation model is generated. This model fully considers actual terrain factors such as road slope and curves, providing accurate and reliable references for subsequent detection and data analysis. Further considering various practical factors such as roadside tree distribution, traffic flow, and intersection locations, a self-developed path planning algorithm is used to plan the detection paths for the drone swarm. This ensures that the drones can operate safely and efficiently during the detection process, avoiding collisions with roadside trees while minimizing the impact on normal traffic, thus protecting the safety and normal order of the surrounding environment.

[0173] Furthermore, a formation of three specially modified and debugged drones (i.e., including the detection platform mentioned in the above embodiments) can be used to perform the detection task: the main drone is equipped with high-precision illuminance and color temperature sensors, responsible for accurately collecting key parameters such as illuminance and color temperature. The sensors undergo rigorous calibration and testing to ensure the accuracy of the collected data. The subordinate drones focus on monitoring environmental interference factors, such as atmospheric temperature and humidity, and particulate matter concentration. Through onboard professional environmental monitoring equipment, environmental data is acquired in real time and stably, providing necessary environmental information for data correction. Subsequently, the drone system performs altitude compensation calculations in real time, refreshing the data 20 times per second. Through efficient algorithm and hardware architecture design, it is ensured that the measurement data can be corrected in a timely and accurate manner during drone flight, improving the reliability and accuracy of the detection data.

[0174] Finally, the collected data undergoes in-depth and comprehensive analysis to generate a detailed luminaire health assessment report. This report not only includes an assessment of the luminaires' current status, such as whether brightness meets standards and color temperature is normal, but also uses a light decay prediction model to scientifically predict the remaining lifespan of the luminaires, providing comprehensive information support for operation and maintenance decisions. Based on the data analysis results, a visually intuitive 3D light pollution distribution heatmap is output using professional graphics rendering algorithms. This heatmap clearly displays the light pollution situation in urban road lighting, including the severity and distribution range of light pollution, providing important reference for urban lighting planning and management, and helping relevant departments formulate reasonable lighting optimization plans.

[0175] Furthermore, considering the maintenance of tunnel lighting systems, the structural characteristics of the tunnel can be studied in depth during implementation, including parameters such as tunnel length, width, height, and curve radius, as well as the material properties of different parts of the tunnel, such as the reflectivity of sidewalls, road surfaces, and ceilings. Through on-site measurements and material analysis, a comprehensive and accurate database of tunnel material reflectivity can be constructed. This data, combined with a self-optimized physical simulation engine, can then be used to pre-calculate the illuminance gradient distribution in different sections. Considering the complexities of multiple reflections and absorptions of light within the tunnel, precise physical model simulations provide accurate theoretical basis for subsequent detection and analysis, helping maintenance personnel understand the lighting conditions at different locations within the tunnel in advance. Subsequently, the measured data is compared meticulously and comprehensively with the pre-established simulation model, and a self-developed precise algorithm is used to locate luminaires with installation angle deviations. This algorithm can quickly and accurately determine whether there are deviations in the installation angle of the luminaires based on actual measurement data and theoretical data from the simulation model, ensuring timely detection and correction of installation problems and improving the effectiveness of tunnel lighting. Simultaneously, a light decay prediction model is used to mark the numbers of luminaires with abnormal light decay rates. By monitoring and analyzing the changes in luminous flux of luminaires over time, luminaires with excessively rapid light decay rates can be identified in a timely manner, providing clear targets for the maintenance of tunnel lighting systems, improving the pertinence and efficiency of maintenance work, and ensuring the stable operation of tunnel lighting systems.

[0176] In the above embodiments, the method for detecting lighting parameters based on UAVs has been described in detail. This application also provides embodiments of a lighting parameter detection device based on UAVs.

[0177] Figure 8 This is a schematic diagram of a drone-based lighting parameter detection device provided in an embodiment of this application. Figure 8 As shown, the device includes:

[0178] The acquisition module 10 is used to acquire relevant data of the target lamp transmitted by the main control module; wherein, the relevant data includes at least optical data collected by the multispectral imager, attitude data collected by the multi-axis attitude sensor, and environmental data of the location of the target lamp.

[0179] Simulation module 11 is used to perform physical simulation based on optical data, attitude data and environmental data to determine the equivalent illuminance value of the working surface after height compensation;

[0180] The prediction module 12 is used to input optical data, attitude data, environmental data and equivalent illuminance values ​​of the working surface into a pre-built light decay prediction model to output the light decay rate of the target luminaire; wherein, the light decay prediction model is a long short-term memory network model trained based on the relevant data of multiple luminaires and their corresponding light decay rates.

[0181] Since the embodiments of the apparatus and the embodiments of the method correspond to each other, please refer to the description of the embodiments of the method for the embodiments of the apparatus, which will not be repeated here.

[0182] Figure 9 This is a schematic diagram of a drone-based lighting parameter detection system provided in an embodiment of this application. Figure 9 As shown, the UAV-based lighting parameter detection system includes a control center 20 and a UAV 21; the control center is communicatively connected to the UAV; the UAV includes at least a multispectral imager, a multi-axis attitude sensor, a main control module, and a data processing unit; the UAV's data processing unit is capable of executing the steps of the UAV-based lighting parameter detection method described in the above embodiments during operation.

[0183] The above provides a detailed description of a method, apparatus, and system for detecting lighting parameters based on a drone, as provided in this application. The various embodiments in the specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section. It should be noted that those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

[0184] It should also be noted that, in this specification, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

Claims

1. A method for detecting lighting parameters based on unmanned aerial vehicles (UAVs), characterized in that, A data processing unit for use in a drone; the drone includes at least a multispectral imager, a multi-axis attitude sensor, a main control module, and the data processing unit; the multispectral imager and the multi-axis attitude sensor are respectively communicatively connected to the main control module, and the main control module is communicatively connected to the data processing unit; the method includes: The relevant data of the target lamp transmitted by the main control module is acquired; wherein the relevant data includes at least the optical data collected by the multispectral imager, the attitude data collected by the multi-axis attitude sensor, and the environmental data of the location of the target lamp. A physical simulation is performed based on the optical data, the attitude data, and the environmental data to determine the equivalent illuminance value of the working surface after height compensation. The optical data, the attitude data, the environmental data, and the equivalent illuminance value of the working surface are input into a pre-constructed light decay prediction model to output the light decay rate of the target luminaire; wherein, the light decay prediction model is a long short-term memory network model trained based on the relevant data of multiple luminaires and their corresponding light decay rates; A physical simulation is performed based on the optical data, the attitude data, and the environmental data to determine the equivalent illuminance value of the working surface after height compensation, including: Determine the standard reference altitude and obtain the atmospheric transmittance coefficient; Based on the optical data, the attitude data, and the environmental data, determine the original illuminance data, the UAV flight altitude, the UAV pitch angle, and the atmospheric particulate matter concentration. The physical simulation engine is invoked to determine whether the difference between the standard reference altitude and the flight altitude of the UAV is greater than a preset value. If not, the equivalent illuminance value of the working surface is directly generated based on the original illuminance data and the UAV pitch angle. If so, then the atmospheric transmittance compensation factor is determined based on the atmospheric transmittance coefficient, the atmospheric particulate matter concentration, and the UAV flight altitude; Based on the original illuminance data, the standard reference altitude, the UAV flight altitude, the UAV pitch angle, and the atmospheric transmittance compensation factor, the equivalent illuminance value of the working surface after altitude compensation is generated.

2. The method for detecting lighting parameters based on unmanned aerial vehicles according to claim 1, characterized in that, Before performing physical simulation based on the optical data, the attitude data, and the environmental data, after acquiring the relevant data of the target lighting fixture transmitted by the main control module, the process further includes: The relevant data are subjected to noise filtering processing according to the Kalman filter algorithm; The timestamps of all the relevant data after noise filtering are unified to the same time base; The transmission delay of each of the relevant data after noise filtering is compensated using a linear interpolation method. The noise-filtered data are mapped to a unified coordinate system based on the feature point matching algorithm.

3. The method for detecting lighting parameters based on unmanned aerial vehicles according to claim 1, characterized in that, The process of constructing the light attenuation prediction model includes: Acquire the location data, light decay label, and corresponding equivalent illuminance value of the working surface of the luminaire; Acquire relevant data about the lighting equipment collected by the drone; the relevant data includes at least optical data collected by the multispectral imager, attitude data collected by the multi-axis attitude sensor, and environmental data collected by the environmental parameter acquisition module. The location data, light decay labels, corresponding equivalent illuminance values ​​of the working surface, and related data of the lamps were normalized to obtain normalized data. Time series samples are generated based on the sliding window technique and the normalized data, and training set data and validation set data are divided based on the time series samples; An initial light decay prediction model is constructed based on the training set data; wherein, the output of the initial light decay prediction model includes light decay rate, luminaire importance score, atmospheric transmittance coefficient, and prediction time range; The model parameters of the initial light decay prediction model are adjusted by an adaptive moment estimation optimizer; The prediction error of the initial light decay prediction model is determined using the mean squared error loss function. The initial light decay prediction model is validated based on the validation set data to determine whether the prediction error of the initial light decay prediction model meets the preset requirements. If not, return to the step of adjusting the model parameters of the initial light decay prediction model through the adaptive moment estimator optimizer to iteratively train the initial light decay prediction model; If so, the initial light decay prediction model is determined as the light decay prediction model.

4. The method for detecting lighting parameters based on unmanned aerial vehicles according to claim 3, characterized in that, After obtaining the light decay rate of the target luminaire, the process further includes: Obtain the new atmospheric transmittance coefficient and prediction time range output by the light decay prediction model; The new atmospheric transmittance coefficient is fed back to the physical simulation engine; The predicted time range, the light decay rate of the target luminaire, the importance score, and the relevant data are transmitted to the control center so that the control center can generate operation and maintenance decisions for the target luminaire based on the predicted time range, the light decay rate of the target luminaire, and the importance score.

5. The method for detecting lighting parameters based on unmanned aerial vehicles according to claim 4, characterized in that, The control center generates maintenance decisions for the target luminaire based on the predicted time range, the light decay rate of the target luminaire, and its importance score, including: The failure risk score of the target luminaire is determined based on the relevant data of the target luminaire; Obtain the first weight of the light decay rate of the target luminaire, the second weight of the importance score of the target luminaire, and the third weight of the failure risk score of the target luminaire; The comprehensive score of the target luminaire is determined based on its light decay rate and its corresponding first weight, its importance score and its corresponding second weight, and its failure risk score and its corresponding third weight. The overall score of the target luminaire is sorted in descending order with the overall scores of the other luminaires to determine the maintenance priority of the target luminaire; The operation and maintenance decision for the target lighting fixture is generated based on the maintenance priority of the target lighting fixture and the predicted time range.

6. The method for detecting lighting parameters based on a UAV according to any one of claims 1 to 5, characterized in that, The drone collects the relevant data of the target lighting fixture, including: Acquire basic information about each light fixture and each drone within the detection area; the basic information about the light fixtures includes at least the location, number, and distribution density of the light fixtures; the basic information about the drones includes at least the current location, endurance, and flight speed of the drones. The detection region is divided into multiple sub-regions according to the Voronoi diagram algorithm; The detection workload for each sub-region is determined based on the basic information of each lamp. Based on the detection workload of each sub-region and the basic information of each UAV, detection tasks are assigned to each UAV so that each UAV can perform the corresponding detection tasks. All detection tasks include the task of collecting relevant data on the target lighting fixture.

7. The method for detecting lighting parameters based on a UAV according to claim 6, characterized in that, After each drone performs its corresponding detection task, the process also includes: Synchronize the time of each drone and add a timestamp to the data collected by the corresponding detection task; The collected data is sorted according to the timestamp of the data collected by each drone; The data collected by each drone in the same time period are aligned using a sliding window interpolation algorithm; Determine the location of each drone, calibrate the collected data of each drone to the same coordinate system based on the location of each drone, and verify the consistency of each collected data.

8. A lighting parameter detection device based on a drone, characterized in that, A data processing unit for use in unmanned aerial vehicles (UAVs); the UAV includes at least a multispectral imager, a multi-axis attitude sensor, a main control module, and the data processing unit; the multispectral imager and the multi-axis attitude sensor are respectively communicatively connected to the main control module, and the main control module is communicatively connected to the data processing unit; the device includes: The acquisition module is used to acquire relevant data of the target lamp transmitted by the main control module; wherein, the relevant data includes at least the optical data collected by the multispectral imager, the attitude data collected by the multi-axis attitude sensor, and the environmental data of the location of the target lamp; The simulation module is used to perform physical simulation based on the optical data, the attitude data and the environmental data to determine the equivalent illuminance value of the working surface after height compensation; The prediction module is used to input the optical data, the attitude data, the environmental data, and the equivalent illuminance value of the working surface into a pre-built light decay prediction model to output the light decay rate of the target luminaire; wherein, the light decay prediction model is a long short-term memory network model trained based on the relevant data of multiple luminaires and their corresponding light decay rates; A physical simulation is performed based on the optical data, the attitude data, and the environmental data to determine the equivalent illuminance value of the working surface after height compensation, including: Determine the standard reference altitude and obtain the atmospheric transmittance coefficient; Based on the optical data, the attitude data, and the environmental data, determine the original illuminance data, the UAV flight altitude, the UAV pitch angle, and the atmospheric particulate matter concentration. The physical simulation engine is invoked to determine whether the difference between the standard reference altitude and the flight altitude of the UAV is greater than a preset value. If not, the equivalent illuminance value of the working surface is directly generated based on the original illuminance data and the UAV pitch angle. If so, then the atmospheric transmittance compensation factor is determined based on the atmospheric transmittance coefficient, the atmospheric particulate matter concentration, and the UAV flight altitude; Based on the original illuminance data, the standard reference altitude, the UAV flight altitude, the UAV pitch angle, and the atmospheric transmittance compensation factor, the equivalent illuminance value of the working surface after altitude compensation is generated.

9. A lighting parameter detection system based on an unmanned aerial vehicle (UAV), characterized in that, The system includes a control center and a drone; the control center is communicatively connected to the drone; the drone includes at least a multispectral imager, a multi-axis attitude sensor, a main control module, and a data processing unit; the data processing unit of the drone is capable of executing the steps of the drone-based lighting parameter detection method as described in any one of claims 1 to 7 during operation.

Images