A power distribution network sound-light safety control method

Abstract

The application discloses a kind of power distribution network sound-light safety control method, it is related to power distribution network safety control technical field, comprising the following steps: obtaining the multi-source heterogeneous monitoring data of power distribution network protection target;The monitoring data is carried out data fusion processing, and unified situation matrix is generated;Current environmental safety level and potential threat type are determined based on unified situation matrix, and dynamic protection strategy is generated in combination with preset strategy library;The protection action corresponding to the dynamic protection strategy is executed;The application solves the technical problems of high correlation uncertainty and poor fusion reliability of multi-modal sensing data in complex environment, realizes closed-loop control from accurate perception to accurate protection, and significantly improves the intelligent level and reliability of power distribution network safety protection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power distribution network safety management and control, and more particularly to a power distribution network sound-light safety management and control method. BACKGROUND

[0002] With the deepening of the construction of smart grid and new power system, as the key link connecting the main network and users, the importance of safe, stable and reliable operation of the power distribution network is increasingly prominent. The traditional "civil defense + physical defense" safety management and control mode has been difficult to cope with the increasingly complex non-traditional security threats (such as precise physical intrusion against power facilities, remote spying based on unmanned aerial vehicles, and malicious interference with key equipment, etc.). Under this background, the intelligent safety management and control system integrating advanced perception, artificial intelligence and active protection technology has become an inevitable trend and frontier direction of the development of power distribution network safety. Sound and light technologies have shown broad application prospects in this field due to their unique advantages such as non-contact, holographic perception and directional control.

[0003] At the perception level, existing systems are mostly passive monitoring, relying on the independent response of sensors to environmental changes, lacking active detection and deep perception ability of the environment. Various modal sensors (such as sound and light) usually work independently, and the back-end only performs simple alarm logic linkage, resulting in a high false alarm rate of the system, which cannot effectively distinguish between real threats and environmental disturbances. At the decision and execution level, the response strategy of existing technologies is extensive and low in intelligence. Once the alarm is triggered, a wide-area sound-light alarm (such as long sounding of full-field sirens and diffuse lighting) is usually started. This indiscriminate response method not only has limited defense effectiveness and cannot effectively deter and block specific targets, but also causes serious sound-light pollution, interfering with the normal production and life of surrounding residents or enterprises, making the application of high-power security devices in sensitive areas greatly limited. SUMMARY

[0004] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a power distribution network sound-light safety management and control method, which solves the technical problems of high false alarm rate, extensive response strategy, limited defense effectiveness and large disturbance to the surrounding environment of the existing power distribution network security system through the synergistic mechanism of active perception, multi-modal fusion and intelligent decision-making.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0006] A power distribution network sound-light safety management and control method, comprising the following steps: acquiring multi-source heterogeneous monitoring data of a power distribution network protection target; performing data fusion processing on the monitoring data to generate a unified situation matrix; determining the current environmental safety level and the potential threat type based on the unified situation matrix, and generating a dynamic protection strategy in combination with a pre-set strategy library; and executing the protection actions corresponding to the dynamic protection strategy.

[0007] In a preferred embodiment, the multi-source heterogeneous monitoring data is obtained by obtaining a dynamic risk portrait of the power distribution network protection target, the dynamic risk portrait including real-time safety state evaluation results and predictive threat evolution information; based on the dynamic risk portrait, dynamically adjusting a perception strategy of the sensor cluster, wherein the perception strategy includes at least one of the type, sampling frequency, monitoring range and data accuracy of the sensor; and controlling the sensor cluster to actively collect multi-source heterogeneous monitoring data of the protection target according to the adjusted perception strategy.

[0008] In a preferred embodiment, the dynamic risk portrait is obtained by obtaining a state atlas of the protection space, and performing differential comparison with a pre-stored baseline atlas to obtain a differential atlas; performing pattern recognition on the differential atlas to obtain a risk event set; positioning the risk event to a specific electrical equipment or interval according to the topology structure of the power distribution network and the space mapping relationship; and fusing the positioned risk event and the real-time alarm event to output the dynamic risk portrait.

[0009] In a preferred embodiment, the state atlas of the protection space is obtained by controlling a distributed transmission device to transmit orthogonal coded sound wave and light wave detection signals, and collecting mixed signals fed back by the environment; processing the mixed signals and the original detection signals to extract channel response characteristics of the sound wave and the light wave in space propagation; and generating the state atlas of the protection space based on the channel response characteristics using a diffraction tomography inversion algorithm.

[0010] In a preferred embodiment, the dynamic adjustment of the perception strategy of the sensor cluster is based on the dynamic risk portrait, extracting the type, confidence, three-dimensional space coordinates and space envelope volume of the risk event; querying a pre-stored task-sensor mapping table according to the type of the risk event to activate an optimal multi-sensor collaborative perception pipeline; calculating the optimal observation geometry of the sensor cluster based on the three-dimensional space coordinates and the space envelope volume, and assigning micro perception parameters to each sensor in the pipeline; and dynamically setting the operation mode of the collaborative perception pipeline according to the confidence.

[0011] In a preferred embodiment, the unified situation matrix is generated by extracting feature vectors of acoustic and optical data, and deriving time-series phase trajectories and spatial phase distributions therefrom; generating a spatial three-dimensional coordinate system based on radar point cloud data, and using the spatial three-dimensional coordinate system as a reference frame to align the time-series phase trajectories and the spatial phase distributions in space-time; constructing phase topology loops connecting the time-series phase trajectories and the spatial phase distributions in the aligned space-time region; calculating homotopy invariants of the phase topology loops; using the homotopy invariants to bidirectionally optimize and correct the acoustic feature vectors and the optical feature vectors with environmental parameter data as constraints; and generating the unified situation matrix representing the environmental situation according to the optimized and corrected feature vectors.

[0012] In a preferred embodiment, the bidirectional optimization and correction of the acoustic feature vectors and the optical feature vectors with environmental parameter data as constraints and using the homotopy invariants is implemented by constructing a joint optimization cost function of the acoustic and optical feature vectors, wherein the joint optimization cost function is composed of weighted summation of an acoustic reconstruction error term, an optical reconstruction error term, and a topological consistency constraint term; the topological consistency constraint term is calculated from the homotopy invariants; the environmental parameter data is mapped to a regularization coefficient of the topological consistency constraint term; and the acoustic and optical feature vectors are iteratively updated by solving the minimum value of the joint optimization cost function until a convergence condition is reached, thereby obtaining the optimized and corrected feature vectors.

[0013] In a preferred embodiment, the generation of the dynamic defense strategy is implemented by obtaining an abnormal event type, a confidence level, and a three-dimensional spatial coordinate based on the unified situation matrix; generating a waveform selection instruction by matching a preset acoustic waveform library according to the abnormal event type; calculating an acoustic driving instruction by a beamforming optimization algorithm in combination with a state atlas based on the three-dimensional spatial coordinate; generating an optical driving instruction according to the three-dimensional spatial coordinate; and combining the acoustic driving instruction, the waveform selection instruction, and the optical driving instruction to generate the dynamic defense strategy.

[0014] In a preferred embodiment, the calculation of the acoustic driving instruction by the beamforming optimization algorithm is implemented by identifying and defining a target region and a non-target region according to the unified situation matrix; constructing an optimization problem with the maximization of sound pressure energy in the target region as an objective function and with the constraint that the sound pressure level in the non-target region does not exceed a safety threshold; solving the optimization problem by a second-order cone programming algorithm to obtain optimal complex emission weights of each unit in a distributed loudspeaker array; and generating the acoustic driving instruction in combination with the waveform selection instruction and the optimal complex emission weights.

[0015] In a preferred embodiment, the execution of the corresponding protection action of the dynamic protection strategy specifically comprises: issuing acoustic driving instructions and waveform selection instructions to a loudspeaker array controller to drive it to generate a directional sound beam pointing to the target point; issuing optical driving instructions to an optical device controller to drive the optical searchlight device to focus on the target point; and generating a synchronization signal to control the strobe rhythm of the optical searchlight device to be consistent with the sound pulse rhythm of the directional sound beam.

[0016] The technical effect and advantages of the power distribution network sound-light safety control method of the present application are as follows:

[0017] 1. The existing system adopts a "uniform defense" strategy to monitor and collect data without distinction for all regions and all times. The present application no longer blindly collects data, but dynamically schedules the most suitable sensors (such as optical and acoustic) and sets the optimal parameters (such as focal length and sampling rate) based on the preliminary perceived risk signs, and accurately focuses on the threat target to maximize the value of the collected data.

[0018] 2. The existing security system mainly relies on passive alarm triggered by threshold value, and the alarm information is isolated, the false alarm rate is high, and the response measures are single (such as the long buzzing of a buzzer), which cannot distinguish the threat types and levels, and is easy to cause the "Wolf is coming" effect. The present application can understand and deeply recognize the environment through the technical step of "multi-source data fusion-unified situation matrix", which can not only find abnormalities, but also evaluate the security level and identify the potential threat type, so as to realize the pre-prediction and accurate judgment in the process. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 The present application is a power distribution network sound-light safety control method flowchart.

[0020] Figure 2 The present application is a power distribution network sound-light safety control method flowchart. DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0022] Embodiment 1, Figure 1 The present application is a power distribution network sound-light safety control method, which specifically comprises the following steps:

[0023] S1, acquiring multi-source heterogeneous monitoring data of the power distribution network protection target;

[0024] In the embodiment, the multi-source heterogeneous monitoring data of the power distribution network protection target is acquired, specifically:

[0025] A dynamic risk portrait of the power distribution network protection target is acquired, and the dynamic risk portrait includes real-time safety state evaluation results and predictive threat evolution information;

[0026] Based on the dynamic risk portrait, the perception strategy of the sensor cluster is dynamically adjusted, wherein the perception strategy includes at least one of the type, sampling frequency, monitoring range, and data accuracy of the sensor;

[0027] The sensor cluster is controlled to actively collect multi-source heterogeneous monitoring data of the protection target according to the adjusted perception strategy.

[0028] The control distributed emission device emits orthogonal coded acoustic and optical wave detection signals, specifically:

[0029] A plurality of distributed acoustic and optical emitters are used, and each emitter includes an acoustic emission unit and an optical laser emission unit. The signals of each emitter are precisely adjusted by a central control system to ensure that the frequency range of the acoustic wave signal is in the range of 30 Hz to 5 kHz to ensure spatial resolution, and the frequency range of the optical laser is set in the visible or infrared light wave band to ensure sensitivity to spatial obstacles and device surface profiles. The signals emitted by each emitter use time division multiplexing (TDM) or frequency division multiplexing (FDM) technology to ensure that the signals are orthogonal in time or frequency. The reason for using orthogonal coding is to ensure that the multi-source signals can be completely distinguished at the receiving end, avoiding superposition interference of signals from different emitters during propagation:

[0030] When using time division multiplexing, each emitter emits a signal in a predetermined time window to ensure the time interval of adjacent signals , wherein is the maximum propagation distance from the emitter to the sensor, is the propagation speed of the acoustic or optical wave.

[0031] When using frequency division multiplexing, each emitter works in a different frequency band to ensure a frequency interval , wherein is the minimum frequency interval.

[0032] An acoustic and optical detection signal set with orthogonal coding is output.

[0033] The mixed signal fed back by the environment is collected, specifically:

[0034] The mixed signal is an acoustic wave signal collected by a microphone and the reflection signal collected by the optical sensor constitute.

[0035] arranged in the monitoring area microphone sensors, respectively located at , to ensure coverage of the target area, the sampling frequency of each microphone is This frequency meets the Nyquist sampling theorem, ensuring that the signal will not alias. The sensor arrangement position is determined according to the actual target device or area geometry to ensure coverage of the entire monitoring area.

[0036] Each microphone at time The sound wave signal collected can be expressed as the convolution of the emitted signal and the environmental reflection signal:

[0037]

[0038] wherein, is the impulse response of the th microphone, is the environmental noise, and the noise power is obtained from the measured background noise level.

[0039] arranged optical sensors, respectively located at , using laser radar technology, with a sampling frequency of to ensure that the reflection signal can be accurately captured.

[0040] The reflection signal collected by the optical sensor is the intensity variation of the reflected light, and the formula is:

[0041]

[0042] wherein, is the receiving sensitivity of the sensor, is the attenuation coefficient of the environmental medium, is the path length of the laser beam propagation, is the sensor noise, is the phase, is the angular frequency of the light wave.

[0043] All sensors are calibrated by a time synchronization device to ensure that the collected sound and light signal feedback accurately corresponds to the emission end.

[0044] The mixed signal and the original probe signal are processed to extract the channel response characteristics of the spatial propagation of the sound wave and the light wave, specifically:

[0045] The signal received by each sensor is cross-correlated with the original transmitted signal . The cross-correlation processing is used to extract the weak feedback component that strictly matches the probe signal from the strong noise. The calculation formula is:

[0046]

[0047] where, is the cross-correlation function, is the delay time, is the sampling window, and the delay time between each sensor and the transmitted signal is calculated.

[0048] The minimum variance distortion response (MVDR) method is used to optimize the directivity of the received signal, minimize the noise and maximize the signal strength. The beamforming formula is:

[0049]

[0050] where, is the beam output signal in the direction , represents the spatially weighted received signal strength in the direction , is the phase delay of the th microphone relative to the target direction , is used to compensate for the phase difference of the th microphone signal, so that the signals are coherently superimposed in the target direction, is the weighting coefficient of the th microphone, used to minimize the noise power of the received signal:

[0051]

[0052]

[0053] where, is the target signal frequency, is the sound speed, usually 343 m / s (in normal temperature air), is the projection distance of the th microphone to the reference point along the signal incident direction.

[0054] The channel response characteristics of the spatial propagation of the sound wave and the light wave are extracted , including the propagation delay time and the direction.

[0055] The state atlas of the protection space is generated by using a diffraction tomography inversion algorithm based on the channel response characteristics, and specifically:

[0056] The state atlas includes an acoustic impedance distribution atlas and a three-dimensional contour atlas.

[0057] The diffraction tomography inversion algorithm is a reverse problem solving method, which is used to restore the structure or property inside an object through known channel response data. In the generation process of the protection space acoustic impedance distribution atlas, the channel response characteristics provide actual observation data for inversion.

[0058] Firstly, the space to be protected is divided into a regular three-dimensional grid, and the edge length of the grid is determined by the sampling rate and the sensor arrangement. For example, when the sampling rate of the microphone array is , and the time resolution is microseconds, the corresponding spatial resolution depends on the speed of sound propagation (about ), so the spatial resolution corresponding to a single sampling point is mm, so the three-dimensional space is divided into voxel units with a size of 1 cm to ensure that it matches the sampling accuracy.

[0059] According to the channel response characteristics, the propagation path delay and amplitude between each transmit-receive pair are calculated. The propagation path is regarded as a ray passing through the space grid, and the corresponding propagation attenuation and phase shift are determined by the acoustic impedance distribution of the medium on the path. The observation data of all transmit-receive pairs are combined to obtain a projection matrix, and each row of the matrix corresponds to the weighting of a space unit by an acoustic or optical path.

[0060] A linear equation set is established:

[0061]

[0062] Wherein, is the observed channel response vector, is the projection matrix calculated from the geometric positions of the transmitter and receiver and the path length, is the acoustic impedance distribution vector to be solved, and each element corresponds to a voxel of the space grid.

[0063] The algebraic reconstruction technique (ART) or least squares method is used for iterative inversion, and the initial assumption is that the impedance value of the entire space is the air impedance , impedance values of each grid cell are updated iteratively until the error between the reconstruction result and the observed channel response is less than 5%. After obtaining the acoustic impedance distribution, threshold segmentation is performed on the impedance values, cells below a certain threshold (close to the air impedance) are marked as air regions, and cells above the threshold are marked as solids or device surfaces. The boundary cells are reconstructed into a three-dimensional surface to obtain the three-dimensional profile map of the protection space. The data of the optical sensor is used to correct the boundary details, for example, the spatial distribution of the reflection intensity can be used to compensate for the parts where the acoustic resolution is insufficient.

[0064] The difference between the pre-stored baseline map and the observed data is compared to obtain a difference map, specifically:

[0065] The system establishes a baseline map of the protection area during the deployment phase. The baseline map is obtained by long-term acquisition and statistical averaging under the condition that the device is in a healthy operating state, and records the acoustic impedance distribution and three-dimensional profile under normal conditions. During operation, the new observation results are compared with the baseline point by point. The difference result is represented in the form of a map, where the color or gray scale change reflects the deviation amplitude relative to the baseline. For acoustic impedance, the allowable fluctuation range is set to ±3%, which is derived from experimental data on the influence of common air humidity and temperature changes on impedance. Units exceeding the fluctuation range are marked as acoustic impedance abnormal units. For the contour boundary, the allowable offset range is set to 5mm, which is derived from the influence of electrical equipment installation errors and thermal expansion and contraction. Units exceeding the offset range are marked as contour abnormal units. All abnormal units form an "abnormal region difference map", which directly shows the location and range of abnormalities in space. The significance of this difference step is to avoid relying solely on instantaneous data for judgment, but to use a long-term stable baseline as a reference to improve the reliability of abnormal detection.

[0066] The difference map is subjected to pattern recognition to obtain a risk event set, specifically:

[0067] The difference map is then input into the rule base for pattern recognition. The rule base pre-stores the difference characteristics of multiple typical risk patterns, such as personnel misentry, foreign object falling, device structure deformation, and abnormal reflection on the surface of insulators. The pattern recognition process is realized through feature comparison and classifier calculation: if the difference region is highly consistent with the features in the rule base, the corresponding risk event is output. Finally, a risk event set is formed, each risk event in which has an event category, occurrence region, and confidence score.

[0068] The risk event is located to a specific electrical device or interval according to the topology of the power distribution network and the spatial mapping relationship, specifically:

[0069] The risk event needs to correspond to the power grid equipment topology to determine which device and interval has a problem. The topology is derived from the design drawings and installation records of the power grid equipment, and each device has clear spatial boundary coordinates and topology connection relationships.

[0070] The spatial coordinates of the risk event are compared with the topology model one by one. If the event location falls within the boundary of a device, the device risk is located. If the event location is in the interval area between two devices, the interval risk is located. Each risk event is labeled to a specific electrical device or interval location, for example: "110kV main transformer A phase winding → partial discharge risk". The output is a set of risk events with device or interval location labels.

[0071] The located risk event and real-time alarm event are fused to output a dynamic risk portrait, specifically:

[0072] The located risk event and real-time alarm event are fused using the D-S evidence theory fusion method.

[0073] In addition to acousto-optic detection, the system also accesses other alarm sources such as current surge detection and temperature sensor alarm. These real-time alarm events and located acousto-optic risk events exist in parallel. Through the D-S evidence fusion method, multiple evidence from different sources is weighted and synthesized to obtain a unified risk confidence distribution. This fusion process avoids the limitations of single source false positives and can output a dynamic, time-varying risk portrait. The risk portrait includes not only the risk category, but also the risk development trend, spatial expansion range, and threat level to the specific device operation.

[0074] Based on the dynamic risk portrait, the perception strategy of the sensor cluster is dynamically adjusted, wherein the perception strategy includes at least one of the type, sampling frequency, monitoring range, and data accuracy of the sensor, specifically:

[0075] Based on the dynamic risk portrait, the type, confidence, three-dimensional spatial coordinates, and spatial envelope volume of the risk event are extracted. According to the risk event type, a preconfigured task-sensor mapping table is queried to activate the optimal multi-sensor collaborative perception pipeline. The pipeline can include a joint observation sequence of acoustics, optics, and auxiliary sensors. Through parallel or serial scheduling, high-confidence risk perception is achieved. Based on the three-dimensional spatial coordinates and spatial envelope volume, the optimal observation geometry of the sensor cluster is calculated, and micro-perception parameters such as sampling frequency, monitoring range, and sensitivity are assigned to each sensor in the pipeline. According to the confidence, the operating mode of the collaborative perception pipeline is dynamically set. For high confidence, a high-frequency sampling, all-around coverage, and low-latency transmission operating mode is adopted. For low confidence, an energy-saving mode or extended sampling interval is adopted.

[0076] The control sensor cluster actively collects multi-source heterogeneous monitoring data of the protection target according to the adjusted perception strategy, specifically as follows:

[0077] The sensor cluster re-collects environmental state information according to the new sampling frequency, working mode and observation area. The collected data enters the above processing flow again to form a new channel response and difference map, thereby realizing closed-loop updating. This cyclic mechanism ensures that the protection system can continuously track and dynamically respond to risk events in a complex environment.

[0078] S2, performing data fusion processing on the monitoring data to generate a unified situation matrix;

[0079] The data fusion processing on the monitoring data to generate a unified situation matrix is specifically as follows:

[0080] The acoustic data is pre-processed (such as noise reduction and filtering), then time-frequency features are extracted through short-time Fourier transform, and further analytic signals are calculated using Hilbert transform, so as to obtain the instantaneous phase of the signal. The instantaneous phases of continuous multiple frames are connected to form a time sequence phase trajectory representing the phase evolution of the acoustic event. The optical data (two-dimensional image data) is processed by a Log-Gabor filter in multiple scales and multiple directions, the response of each pixel is calculated, and the phase consistency value of each pixel is calculated using these responses, so as to obtain the spatial phase distribution of the whole image. The radar point cloud data is used to model the equipment area in three dimensions, different targets are separated by point cloud clustering algorithm, the centroid of each target is calculated to obtain the spatial coordinates of the equipment, and then a spatial three-dimensional coordinate system of the equipment area is generated and used as a reference framework to align the time sequence phase trajectory and the spatial phase distribution in space-time. Specifically, the direction of the acoustic signal is determined based on the direction estimation of the acoustic signal (for example, using the MUSIC algorithm), the propagation direction vector of the acoustic signal is converted into a coordinate in the three-dimensional coordinate system, and is aligned with the target centroid in the radar point cloud, so as to realize the spatial alignment of the acoustic data and the radar point cloud. According to the intrinsic and extrinsic parameter matrices of the optical sensor calibrated in advance, the pixel coordinates of the optical signal are converted into three-dimensional space coordinates, and the coordinates are associated with the target centroid in the radar point cloud data to ensure the space-time alignment between the optical data and the radar point cloud. In the space-time aligned area, the acoustic time sequence phase trajectory and the optical spatial phase distribution are connected to form a phase topological loop. For adjacent phase points in the phase topological loop and , the phase difference is calculated:

[0081]

[0082] Each phase difference value mapped to within the interval, calculate the total phase change , calculate the homotopy invariant (i.e. the winding number) of the phase topology loop :

[0083]

[0084] The environmental parameter data is used as a constraint, and the homotopy invariant is used for bidirectional optimization and correction of the acoustic feature vector and the optical feature vector, specifically:

[0085] A joint optimization cost function of the acoustic and optical feature vectors is constructed The joint optimization cost function is composed of an acoustic reconstruction error term, an optical reconstruction error term and a topological consistency constraint term:

[0086]

[0087] wherein, and are the acoustic feature vector and the optical feature vector to be optimized, and are the original acoustic feature vector and the optical feature vector, , are weight coefficients, is a regularization coefficient based on environmental parameter data E (such as signal-to-noise ratio, visibility, temperature, humidity, etc.) dynamic adjustment, is a topological consistency constraint term, denotes the Euclidean distance.

[0088] The L-BFGS optimization algorithm is used to solve the minimum value of the cost function. This algorithm can simultaneously and iteratively update the acoustic feature vector and the optical feature vector (i.e. bidirectional optimization) until the convergence condition is met, i.e. the iteration process will stop when the change in the cost function value is less than the preset tolerance, and the optimized feature vectors and

[0089] A unified situation matrix representing the environmental situation is generated according to the optimized and corrected feature vectors. The unified situation matrix contains optimized acoustic features, optical features, target position coordinates, target speed, event type labels, confidence, etc.

[0090] S3, determine the current environmental safety level and potential threat type based on the unified situation matrix, and generate a dynamic protection strategy combined with a pre-set strategy library;

[0091] The matching dynamic protection strategy is generated, specifically:

[0092] Based on the unified situation matrix, the abnormal event type, confidence and three-dimensional space coordinates are obtained;

[0093] According to the abnormal event type, the preset acoustic waveform library is matched to generate a waveform selection instruction;

[0094] Based on the three-dimensional space coordinates, the state atlas is combined to calculate the acoustic driving instruction through the beamforming optimization algorithm;

[0095] According to the three-dimensional space coordinates, an optical driving instruction is generated;

[0096] The acoustic driving instruction, the waveform selection instruction and the optical driving instruction are combined to generate a dynamic protection strategy.

[0097] Among them, based on the unified situation matrix, the abnormal event type, confidence and three-dimensional space coordinates are obtained, specifically:

[0098] The type of target abnormal event (such as partial discharge, fault vibration, etc.), confidence index (representing event reliability), and three-dimensional space coordinates are analyzed from the unified situation matrix, wherein the three-dimensional space coordinates are derived from the multi-source sensor fusion result, representing the actual position of the abnormal event.

[0099] Among them, according to the abnormal event type, the preset acoustic waveform library is matched to generate a waveform selection instruction, specifically:

[0100] First, through the unified situation matrix and real-time monitoring data, the type of abnormal event currently occurring is identified. These abnormal events may include but are not limited to high-risk areas where electrical equipment fails and power systems are subjected to external interference. According to the type of abnormal event identified, the appropriate acoustic waveform is selected from the preset acoustic waveform library. The acoustic waveform library stores a variety of different waveforms, each of which is suitable for different types of protection needs. For example, when monitoring equipment abnormal vibration, a pulse waveform may be selected, which has fast instantaneous energy output and is suitable for instantaneous acoustic adjustment of equipment or area, and if it is a device aging problem, a more stable sinusoidal waveform is selected for long-term monitoring.

[0101] Among them, based on the three-dimensional space coordinates, the state atlas is combined to calculate the acoustic driving instruction through the beamforming optimization algorithm, specifically:

[0102] According to the unified situation matrix, the target area and the non-target area are identified and defined;

[0103] An optimization problem is constructed with the target area sound pressure energy maximization as the objective function, and the non-target area sound pressure level is constrained not to exceed the safety threshold;

[0104] A second-order cone programming algorithm is used to solve the optimization problem to obtain optimal complex radiation weights of each unit in the distributed loudspeaker array.

[0105] The acoustic driving instruction is generated by combining the waveform selection instruction with the optimal complex radiation weight.

[0106] According to the unified situation matrix, the target area and the non-target area are identified and defined, specifically:

[0107] The three-dimensional space coordinates are taken as the center point, and the event confidence is combined to calculate the protection radius , and the protection radius is no longer a fixed empirical value in the prior art, but is obtained through the following dynamic calculation rule:

[0108]

[0109] wherein, is the minimum protection distance formulated by the power grid operation and maintenance department, is the maximum safety protection distance allowed in the equipment nameplate parameters.

[0110] The obtained through the above formula can be automatically adjusted according to the event identification reliability to cover the range, so that the protection area neither expands excessively due to low confidence nor shrinks blindly due to high confidence, thereby realizing dynamic adaptation.

[0111] Finally, a three-dimensional sphere with as the center and a radius of is defined as the target area , and according to the three-dimensional contour map and the acoustic impedance distribution map, the space outside the target area is divided into a non-target area .

[0112] The optimization problem is constructed with the target area sound pressure energy maximization as the objective function, and the sound pressure level of the non-target area is constrained not to exceed the safety threshold, specifically:

[0113] In the optimization target setting process, the sound pressure energy maximization of the target area is taken as the objective function. Specifically, the sound pressure energy is obtained by weighted sum of the square values of the sound pressure of a plurality of sampling points in the target area, and the number and position of the sampling points are determined by the space coordinate information in the unified situation matrix. The formula can be expressed as:

[0114]

[0115] wherein, is the control vector, which represents the complex radiation weight of each unit in the distributed loudspeaker array, including amplitude and phase parameters, The total sound pressure level energy in the target area represents the energy focusing effect of the loudspeaker array on the target area, and is composed of the sum of the squares of the sound pressure levels at the target points. The number of sampling points within the target area. Indicates in control variables Below, the target area The sound pressure value of each sampling point is derived from the calculation results of the geometric relationship between the emission parameters of the loudspeaker unit and the spatial position of the sampling point (such as distance, azimuth angle, etc.).

[0116] As a constraint, the sound pressure level in non-target areas must not exceed a safety threshold. The sound pressure level is calculated by comparing the sound pressure at the sampling point with a reference sound pressure level.

[0117]

[0118] in, Indicates the sound pressure level in the non-target area. This indicates the safety threshold, which is preset according to national noise standards or equipment operating specifications. Indicates in control variables Below, the sound pressure value at sampling points in non-target areas, The reference sound pressure level is represented by a standard value commonly used in the field of acoustics. .

[0119] Specifically, the optimal complex emission weights of each unit in the distributed loudspeaker array are obtained by solving the optimization problem using a second-order cone programming algorithm:

[0120] Based on the aforementioned optimization problem, the control parameters of each loudspeaker unit in the distributed loudspeaker array are first expressed as complex numbers representing amplitude and phase, and then uniformly represented as control variables in column vector form. The initial value of the vector is set to equal amplitude and zero phase distribution, that is, each unit performs initial transmission with the same power and the same phase.

[0121] The sound pressure energy expression within the target area Convert to matrix form , where the matrix It is constructed from the spatial coordinate information of the sampling points in the target area, reflecting the geometric relationship between the array unit and each sampling point.

[0122] The sound pressure level constraint in the non-target region is converted into a second-order cone inequality constraint, i.e.:

[0123]

[0124] in, Constructed from the coordinates of sampling points in non-target regions, threshold Sourced from safe sound pressure level Compared with reference sound pressure The conversion results are as follows:

[0125] .

[0126] Combining the objective function with the constraints, we form a standard second-order cone programming problem:

[0127]

[0128] The above optimization problem is solved iteratively using a second-order cone programming solver (such as an interior-point solver). During the iteration, the control variables are updated... The amplitude and phase of the sound pressure level are adjusted to gradually increase the sound pressure energy in the target area while ensuring that the sound pressure level in the non-target area remains below the threshold.

[0129] When the iteration converges, the output control variable This is the optimal re-emission weight. Each component corresponds to the re-emission weight of a unit in the distributed loudspeaker array, which explicitly defines the emission amplitude and phase adjustment value of that unit.

[0130] The acoustic driving command is generated by combining the waveform selection command with the optimal re-emission weight, specifically as follows:

[0131] Once a suitable acoustic waveform is determined, the optimal complex emission weights obtained previously using the second-order cone programming algorithm can be applied. To generate acoustic drive instructions. The specific process is as follows:

[0132] The selected acoustic waveform is compared with the optimal reverberation weight. These are combined to generate the drive signal for each speaker unit. Assume the selected waveform is... And the speaker array has N units (corresponding to vectors) For each component in the equation, the output signal of each speaker unit can be expressed as:

[0133]

[0134] in, For the first The output signal of each speaker unit The first in the optimal re-emission weights The nth component represents the nth component. The amplitude and phase of each speaker.

[0135] Through the above operations, each speaker unit will output a signal, and the composite effect of the speaker array will eventually form a directional sound beam pointing to the target area. The output signals of all speaker units are integrated to form a complete control instruction set, and the output signal of each speaker array unit The speaker controller converts it into an electrical signal to drive the speaker to actually emit sound waves.

[0136] Among them, the optical driving instruction is generated according to the three-dimensional space coordinates, specifically:

[0137] Through the three-dimensional space coordinates of the target area, the geometric relationship between the optical equipment and the target area can be determined. Assuming that the target area is located at position and the optical equipment (such as a laser searchlight, a camera, etc.) is located at position The distance between the two is which can be calculated by the Euclidean distance formula. The pointing angle of the optical equipment can also be calculated. The azimuth angle reflects the horizontal direction of the optical equipment relative to the target area, which can be calculated by the dot product and cross product of two vectors. The elevation angle reflects the vertical direction of the optical equipment relative to the target area, which can be calculated by the dot product of two vectors. The optical equipment needs to adjust its emission direction to point to the direction of the target area, which can be determined by the calculated azimuth angle and elevation angle. The rotation angle of the optical equipment is controlled to make its light beam accurately point to the target area.

[0138] The focusing ability of the optical equipment also needs to be adjusted according to the distance of the target area. For example, when the target area is far away, the optical equipment needs to perform more fine focusing operations; when the target area is close, the optical equipment can choose a wider focus range. This adjustment is based on the focal length formula:

[0139]

[0140] wherein, is the focal length, is the focusing angle of the optical equipment, which is usually a fixed design parameter of the optical equipment.

[0141] According to the needs of the target area, the light intensity of the optical equipment needs to be adjusted appropriately. For example, when the target area needs higher intensity lighting, the optical equipment will increase the brightness of the light source according to the light intensity requirement of the target area. This control can be realized by adjusting the output power of the optical equipment.

[0142] The optical driving instruction can be represented as (azimuth angle, elevation angle, focal length, light intensity).

[0143] The dynamic protection strategy is composed of the acoustic driving instruction, the waveform selection instruction and the optical driving instruction.

[0144] S4, performing a protection action corresponding to the dynamic protection strategy according to the dynamic protection strategy.

[0145] In the embodiment, the performing of the protection action specifically includes:

[0146] The acoustic driving instruction and the waveform selection instruction are sent to a loudspeaker array controller to drive the loudspeaker array controller to generate a directional sound beam pointing to the target point, and the optical driving instruction is sent to an optical device controller to drive an optical searchlight device to focus on illuminating the target point, a synchronization signal is generated by adjusting the stroboscopic frequency of the optical device to be consistent with the acoustic pulse frequency, so that the stroboscopic rhythm of the optical searchlight device is consistent with the acoustic pulse rhythm determined by the acoustic driving instruction.

[0147] The above formulas are dimensionless numerical calculations, and the formulas are obtained by software simulation of a large amount of data to obtain a formula closest to the real situation, and the preset parameters in the formula are set by a person skilled in the art according to the actual situation.

[0148] The above embodiments can be realized wholly or partially by software, hardware, firmware or any combination thereof. When realized by software, the above embodiments can be realized wholly or partially in the form of a computer program product.

[0149] Those skilled in the art can realize that the modules and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0150] In addition, each functional module in each embodiment of the present application can be integrated in one processing module, or each module can exist physically alone, or two or more modules can be integrated in one module.

[0151] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0152] Finally: the above only for the preferred embodiments of the present application, and not for limiting the present application, any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application, should be included in the scope of protection of the present application.

Claims

1. A power distribution network sound-light safety management method, characterized in that, The method comprises the following steps: Obtaining multi-source heterogeneous monitoring data of the power distribution network protection target, specifically, obtaining the state atlas of the protection space, and performing differential comparison with the pre-stored baseline atlas to obtain a differential atlas; Performing pattern recognition on the differential atlas to obtain a risk event set; According to the topology structure of the power distribution network and the space mapping relationship, the risk event is located to a specific electrical equipment or interval; Fusing the located risk event and the real-time alarm event to output a dynamic risk portrait; Based on the dynamic risk portrait, the perception strategy of the sensor cluster is dynamically adjusted; The sensor cluster collects multi-source heterogeneous monitoring data of the protection target according to the adjusted perception strategy; The state atlas of the protection space is obtained by: Controlling the distributed transmitting device to transmit orthogonal coded sound waves and light wave detection signals, and collecting mixed signals fed back by the environment; Processing the mixed signals and the original detection signals to extract the channel response characteristics of the sound waves and light waves in space propagation; Based on the channel response characteristics, a diffraction tomography inversion algorithm is used to generate the state atlas of the protection space; Performing data fusion processing on the monitoring data to generate a unified situation matrix; Based on the unified situation matrix, the current environment safety level and the potential threat type are determined, and a dynamic protection strategy is generated in combination with a pre-set strategy library; Performing the protection action corresponding to the dynamic protection strategy.

2. The power distribution network sound-light safety control method according to claim 1, characterized in that, The dynamic risk portrait includes real-time safety state evaluation results and predictive threat evolution information; The perception strategy includes at least one of the type, sampling frequency, monitoring range and data accuracy of the sensor.

3. The power distribution network sound-light safety control method according to claim 2, characterized in that, The dynamic adjustment of the perception strategy of the sensor cluster is specifically: Based on the dynamic risk portrait, the type, confidence, three-dimensional space coordinates and space envelope volume of the risk event are extracted; According to the type of the risk event, a pre-set task-sensor mapping table is queried to activate the optimal multi-sensor collaborative perception pipeline; Based on the three-dimensional space coordinates and the space envelope volume, the optimal observation geometry of the sensor cluster is calculated, and micro perception parameters are assigned to each sensor in the pipeline; According to the confidence, the operating mode of the collaborative perception pipeline is dynamically set.

4. The power distribution network sound-light safety control method according to claim 3, characterized in that, The unified situation matrix is specifically generated by: Extracting the feature vectors of acoustic and optical data, and obtaining the time sequence phase trajectory and spatial phase distribution based on the feature vectors; Based on the radar point cloud data, a three-dimensional coordinate system is generated as a reference framework, and the time sequence phase trajectory and the spatial phase distribution are spatio-temporally aligned; In the aligned spatio-temporal region, a phase topology loop connecting the time sequence phase trajectory and the spatial phase distribution is constructed; The homotopy invariant of the phase topology loop is calculated; Using the homotopy invariant to bidirectionally optimize and correct the acoustic feature vectors and the optical feature vectors with the environmental parameter data as constraints; According to the optimized and corrected feature vectors, a unified situation matrix representing the environmental situation is generated.

5. The power distribution network sound-light safety control method according to claim 4, characterized in that, The bidirectional optimization and correction of the acoustic feature vectors and the optical feature vectors using the homotopy invariant with the environmental parameter data as constraints is specifically: A joint optimization cost function of acoustic and optical feature vectors is constructed, which is composed of an acoustic reconstruction error term, an optical reconstruction error term and a topological consistency constraint term; The topological consistency constraint term is calculated by the homotopy invariant; The environmental parameter data is mapped to the regularization coefficient of the topological consistency constraint term; By solving the minimum value of the joint optimization cost function, the acoustic and optical feature vectors are iteratively updated synchronously until the convergence condition is reached, and the optimized corrected feature vectors are obtained.

6. The power distribution network acousto-optic safety control method according to claim 5, characterized in that, The dynamic protection strategy is generated, specifically: Based on the unified situation matrix, the abnormal event type, confidence and three-dimensional space coordinates are obtained; According to the abnormal event type, the preset acoustic waveform library is matched to generate a waveform selection instruction; Based on the three-dimensional space coordinates, the state atlas is combined to calculate the acoustic driving instruction through the beamforming optimization algorithm; According to the three-dimensional space coordinates, an optical driving instruction is generated; The acoustic driving instruction, the waveform selection instruction and the optical driving instruction are combined to generate a dynamic protection strategy.

7. The power distribution network sound-light safety control method according to claim 6, characterized in that, The acoustic driving instruction is calculated through the beamforming optimization algorithm, specifically: According to the unified situation matrix, the target area and the non-target area are identified and defined; An optimization problem is constructed with the target area sound pressure energy maximization as the objective function, and the non-target area sound pressure level is constrained not to exceed the safety threshold; A second-order cone programming algorithm is used to solve the optimization problem to obtain the optimal complex emission weight of each unit in the distributed loudspeaker array; The acoustic driving instruction is combined with the optimal complex emission weight to generate the acoustic driving instruction.

8. The power distribution network acousto-optic safety control method according to claim 7, characterized in that, The dynamic protection strategy is executed to perform the corresponding protection action, specifically: The acoustic driving instruction and the waveform selection instruction are sent to the loudspeaker array controller to drive it to generate a directional sound beam pointing to the target point; The optical driving instruction is sent to the optical equipment controller to drive the optical searchlight equipment to focus on the target point; A synchronization signal is generated to control the frequency flash rhythm of the optical searchlight equipment to be consistent with the sound pulse rhythm of the directional sound beam.

Images