Three-dimensional warning area object recognition early warning control method and system based on RF radar and medium
Through the three-dimensional warning area object recognition and early warning control method based on RF radar, the existing technology identification and early warning problems in harsh environments are solved, and high-precision engineering vehicle identification and early warning control is realized, reducing false alarm and omission rates, and improving the rate of false alarms and omissions. Construction safety.
Patent Information
- Application Number
- CN202411100982.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-12
- Publication Date
- 2024-11-22
AI Technical Summary
Existing construction safety technology is difficult to accurately identify and warn of construction machinery and vehicles near high-voltage transmission lines in severe weather and complex environments, resulting in high false alarms, missed and unreported rates, and cannot effectively prevent external damage accidents.
Using the three-dimensional warning area object recognition and early warning control method based on RF radar, through multi-level three-dimensional warning area division and point cloud data processing, the precise identification and early warning control of engineering vehicles is achieved, the false alarm and omission rate is reduced, and the missed rate is adapted to rain and snow. , fog, haze and other complex environments.
It realizes high-precision target object identification and early warning control in harsh environments, reduces the false alarm and omission rate, improves construction safety, and ensures the safe operation of high-voltage lines.
Smart Images

Figure CN119001707A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of construction safety technology, and in particular to a three-dimensional warning zone object recognition early warning control method, system and medium based on RF radar. Background Art
[0002] In recent years, with the continuous development of national construction and the continuous expansion of urban scale, various types of construction have been continuously increasing investment, and all walks of life have shown a momentum of vigorous development. While the electricity consumption is growing, the high requirements for the safety and reliability of power supply of the power grid cannot be ignored. However, accidents such as power line tripping and power outages caused by external factors such as various illegal construction have shown a clear upward trend, and external force damage has become one of the main factors affecting the safe and reliable operation of transmission lines. In particular, large construction mechanical arms such as cranes under high-voltage transmission lines approach discharge and accidentally touch live lines in blind spots, causing line short circuit tripping, wire damage, large-scale power outages, and even major accidents such as car crashes and deaths. Especially at night or in rainy, snowy, foggy and haze weather, when the visibility is poor, it is easier for the crane driver to get too close to or even touch the high-voltage line due to lack of effective observation, which may then cause electric shock accidents. The traditional regular manual inspection method has many problems such as low efficiency, time-consuming and labor-intensive, human negligence, and lax supervision, which cannot achieve the role of timely prevention and on-site prevention. The emergence of traditional transmission line video monitoring and external force damage prevention systems has provided new advanced technical means for the work of external force damage prevention of transmission lines, but it has also exposed problems such as insufficient accuracy, insufficient stability, and weak functionality. Even if it can be monitored, accidents may still occur if they are not handled in time. Therefore, the research and invention of a set of solutions that can efficiently, stably, intuitively, and accurately monitor and identify intrusion target objects, and at the same time conduct real-time monitoring and interlocking control of engineering machinery vehicles and cranes operating at high risk levels under power transmission lines, is extremely important and meaningful for effectively curbing external force damage accidents.
[0003] In the current security prevention and identification applications, each sensor detector can only detect a specific point, line, or surface in the environment space, and its detection output is a fixed and irregular single alarm signal. The system space warning surface cannot be preset at multiple levels, the detection distance is short, the warning surface is very narrow, the recognition rate is low, and the missed alarm, false alarm, and non-alarm rates are very high. It does not have the ability to plan and distinguish between the identification area and the identification area in advance. That is, as long as there is a moving target in a certain point, line, or surface area, regardless of size, it will be reported as soon as it is detected. Especially at night or in rainy, snowy, foggy, and haze weather when the visibility is poor, the false alarm rate is very high. When detecting effective targets in macro-distance semi-dynamic and static movements (such as crane booms), it is almost impossible to miss or report, because the extension and swing of the arm are very slow and it is easy to be covered by static signal noise points and difficult to monitor.
[0004] The main disadvantages of the existing anti-intrusion external force destruction solutions are as follows:
[0005] (1) Video image recognition technology device. A single-level overall two-dimensional warning fence can be set up, and an alarm can be issued by comparing and identifying the fixed target object image stored or set in advance with the real-time monitored video image object. This device requires powerful front-end algorithm computing capabilities. The existing independent front-end control processing chip computing power cannot support the algorithm support of continuous large-capacity image data. It can only compare and filter the images of several preset intrusion objects in advance for alarm. Because the intrusion mechanical vehicles or objects are diverse, there may be omissions or non-reports due to the inability to compare and identify. Therefore, it can only rely on a back-end platform with powerful computing power to achieve it; secondly, the detection distance space is integrated and cannot be hierarchically distinguished and identified. Since image recognition is distinguished by pixels, it works in an overall plane space of the line of sight. In the set area, its near and far pixels are integrated, so It is unable to distinguish and identify the distance, height and other directions. Objects that appear in the warning zone screen will generate an alarm regardless of their distance. Thirdly, the detection distance is short. During the day, the effective distance according to the variable focus lens is between 30 and 250 meters. At night, the effective identification distance of night vision is short, about 15 to 120 meters (does not meet the usage requirements). After focusing, there are large areas of defense area omissions and blind spots that cannot be monitored. At the same time, infrared reflection interferes with the analysis of intruding targets, causing false identification alarms, such as moths flying over the lens. Fourthly, heavy fog, rain and snow weather will cause blurred field of view, resulting in no reporting, missed reporting, and false reporting, especially in the case of low visibility at night.
[0006] (2) Laser ranging radar technology device. The laser beam is emitted and reflected back when encountering the target. The corresponding laser beam is processed by the receiving device to calculate the time difference and displacement angle to obtain the target distance and direction for use in alarm. This type of device is divided into single-point and multi-point scanning. The multi-point array scanning method is proportional to the detection accuracy and power consumption, and it is almost difficult to popularize at this stage; the multi-line array has high reception loss and energy consumption in order to achieve long-distance detection and detection, and it does not have cost advantages. First, it has extremely high requirements on the use environment, detection distance and structural materials. Due to the spectral band mode adopted in its own work, its shell must be made of anti-interference, light-filtering, wave-absorbing and light-transmitting materials, which are easy to age and dust-proof and rain-proof. It is completely unsuitable for outdoor all-weather harsh environment. In terms of multi-line laser radar, it is currently mainly used in autonomous driving. There are also warning areas based on protecting vehicles and pedestrians, but there is no application of multi-level spatial warning fences. At present, the basic application in the power system is single-point ranging laser, which uses the pan-tilt left and right control to launch and detect at a fixed time. It has a large number of defense area omissions and delay gaps, and a high false alarm rate. Secondly, it cannot identify and distinguish the status information of intruding objects. During the detection period, an alarm will be triggered as long as it is detected, resulting in serious false alarms. Thirdly, the detection surface has a fixed height and a small angle, which depends on the pan-tilt motor. Fourthly, it is affected by heavy fog, rain and snow, resulting in high false alarm rates and missed alarms.
[0007] (3) Microwave Doppler effect detection and millimeter wave planar radar technology devices. Microwave signals are transmitted into the defense area through the microwave transmitter antenna. When there is no moving target in the defense area, the frequency of the microwave signal received by the receiver is the same as the frequency of the transmitted signal, and no alarm is issued; when there is a moving target, due to the Doppler effect, the frequency of the microwave signal reflected by the target will shift, and the microwave receiving device will obtain the object displacement signal after analysis and use it to issue an alarm. The false alarm rate of microwaves is high, because the Doppler effect will generate a signal and trigger an alarm as long as the moving object is in the detection area and there is displacement; it can only work in a horizontal plane warning area set at a certain angle, the detection surface is narrow, and it cannot detect target height information and static target information, and cannot distinguish and identify the status information of the intruding object. If a moving object exceeds the preset value, it will issue an early warning, which is very easy to cause false alarms.
[0008] (4) Infrared and laser beam fence technology devices. Through the use of independent infrared single-beam or multi-beam transmitters paired with independent infrared single-beam or multi-beam receivers, they are arranged around the required warning area or on a single side. When an object passes through the infrared beam that is blocked, the receiving device immediately processes and sends an alarm signal to alarm. The passive infrared beam device is complex to install and needs to be installed face to face to form a detection line area. The detection distance is short (generally within 150 meters, conventionally 60 meters, and 150 meters with a lens). Due to the working characteristics of the obstruction alarm, there is a large gap in the defense area, which is prone to false alarms and missed alarms. It cannot be filtered and identified inside and outside the warning area, and the alarm is triggered afterwards.
[0009] The above-mentioned existing anti-intrusion external force destruction solutions can, to a certain extent, provide early warning for external force destruction accidents such as short circuits and tripping caused by insufficient operating safety distances of large and medium-sized ultra-high mechanical vehicles that traditional high-voltage transmission lines cannot solve, but they also have their own technical weaknesses. Summary of the invention
[0010] In view of the deficiencies in the prior art, the present invention provides a three-dimensional warning zone object recognition and early warning control method, system and medium based on RF radar. The entire warning zone is divided into multiple levels of three-dimensional warning zones, and the three-dimensional warning zone level and longitude and latitude of the target object are prepared for identification, and then the engineering vehicle is matched to send the corresponding level of early warning control instructions to the engineering vehicle. The recognition accuracy is high, the false alarm and missed alarm are low, and the linkage between early warning and engineering vehicle control can be realized.
[0011] In a first aspect, a method for early warning control of an object recognition in a three-dimensional warning zone based on an RF radar is provided, comprising the following steps:
[0012] S1: Pre-set multi-level three-dimensional warning zones according to the distance from the high-voltage cables;
[0013] S2: Obtain the point cloud data collected by the radio frequency radar detection sensor, and perform coordinate conversion and filtering preprocessing;
[0014] S3: Cluster the preprocessed point cloud data to achieve target detection;
[0015] S4: Match the three-dimensional warning zone level to which the target belongs based on its spatial coordinates;
[0016] S5: Calculate the longitude and latitude of the target object and match it with the longitude and latitude of the engineering vehicle;
[0017] S6: Sending a warning control instruction of a corresponding level to a matching engineering vehicle according to the level of the three-dimensional warning zone of the target object.
[0018] Furthermore, in step S1, a multi-level three-dimensional warning zone is set directly below the high-voltage cable, and the entire warning zone is divided into multiple three-dimensional warning zones by multiple planes with the same horizontal xy plane coordinates and different vertical z-axis coordinates; the closer the three-dimensional warning zone is to the high-voltage line, the higher its warning level.
[0019] Furthermore, in step S2, the coordinate conversion process is expressed as follows:
[0020] X=X'
[0021] Y=Y'
[0022] Z=Z'+(X*tgθ)+Z0
[0023] Where (X', Y', Z') represents the point cloud coordinates in the radar coordinate system, (X, Y, Z) represents the point cloud coordinates converted to the ground coordinate system, Z0 represents the installation height of the RF radar detection sensor, and θ represents the calibration conversion angle.
[0024] Furthermore, in step S2, the point cloud filtering process includes:
[0025] S21: Estimate the two-dimensional plane of the point cloud data through RANSAC, take the points in the local area as ground points, and other points as non-ground points to achieve the segmentation of ground points and non-ground points;
[0026] S22: Filter out ground points and keep non-ground points;
[0027] S23: Perform straight-through filtering within a limited range on the retained non-ground points (x, y, z, RCS), and only select the points within the multi-level three-dimensional warning area and whose RCS values are within the set range for the non-ground points; the non-engineering vehicle point cloud data can be filtered by pre-setting the engineering vehicle RCS value range.
[0028] Furthermore, step S3 specifically includes:
[0029] The preprocessed point cloud data is clustered using a clustering algorithm based on Euclidean distance, and each point cloud cluster obtained is a target point cloud cluster, thereby realizing target detection.
[0030] Furthermore, after step S3, at least one of target object offset identification and filtering and target object dynamic and static identification is also included:
[0031] Target offset identification and filtering: For other targets within the preset offset range of the stable and effective target, determine whether their RCS values are equal to the RCS value of the stable and effective target. If they are equal, they are considered to be the same target and recorded; if the RCS values are not equal, they are considered to be new targets and recorded;
[0032] Identify whether the target is static or dynamic; define a speed threshold. If the speed of the target is less than the speed threshold, the target is considered to be a static target; if the speed of the target is not less than the speed threshold for a continuous preset number of frames, the target is considered to be a dynamic target; separate static targets and dynamic targets and record and store them separately.
[0033] Furthermore, in step S5, the longitude and latitude of the target object are calculated by the following formula:
[0034] long2=d*sinα / [ARC*cos(lat1)*2π / 360]+long1
[0035] lat2=d*cosα / (ARC*2π / 360)+lat1
[0036] Wherein, long1 and lat1 represent the latitude and longitude of the RF radar detection sensor, respectively; long2 and lat2 represent the latitude and longitude of the target, respectively; d represents the distance between the target and the RF radar detection sensor; ARC represents the radius of the earth; and α represents the radar normal deflection angle.
[0037] Furthermore, as the level of the warning zone increases, the corresponding early warning control instructions increase step by step from warning, control deceleration to control locking.
[0038] Furthermore, it also includes: three-dimensional visual display of the identified target objects and multi-level three-dimensional warning areas.
[0039] Furthermore, between step S3 and step S4, the following steps are also included:
[0040] S01: Match the detected target object of the current frame with the target object list; the target object list records the position information and life value of each target object, which is initially empty;
[0041] S02: Calculate the distance between the target detected in the current frame and each target in the target list. If the distance between the two is less than the set distance threshold, it is considered that the two are the same target, and the position information of the target is updated, and its life value is restored to the initial value N; if the distance between the two is not less than the set distance threshold, it is considered to be a new target, and the target and its position information are added to the target list, and its life value is set to the initial value N;
[0042] S03: if there is a target in the target list that has not matched itself in the current frame, the life value of the target in the target list is reduced by 1; when the life value of a target in the target list is 0, it is deleted from the target list;
[0043] S04: After continuously acquiring M frames and performing the above steps S01 to S03, proceed to step S4; wherein M>N.
[0044] In the second aspect, a three-dimensional warning zone object recognition and early warning control system based on RF radar is provided, including a power supply component module, a main control module, a radio frequency radar detection sensor component module, and a positioning module. The power supply component module supplies power to the entire system; the radio frequency radar detection sensor component module and the positioning module are both communicatively connected to the main control module; the main control module is configured to execute the three-dimensional warning zone object recognition and early warning control method based on RF radar as described above.
[0045] In a third aspect, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the three-dimensional warning zone object recognition and early warning control method based on RF radar as described above is implemented.
[0046] The present invention proposes a method, system and medium for object recognition and early warning control in a three-dimensional warning zone based on RF radar. By defining a multi-level three-dimensional warning zone in a custom three-dimensional space, the division of the degree of danger is realized. Three-dimensional warning zones of different levels send early warning control instructions of different intervention degrees to engineering vehicles, so as to realize the detection of target objects under high-voltage cables and early warning linkage control of engineering vehicles. By filtering and clustering the point cloud data, the target object can be accurately identified, and then the noise interference can be further filtered through target tracking. The target object can still be effectively and stably detected under the interference of complex environments such as rain, snow, fog and haze, so as to prevent the occurrence of false alarms, missed alarms and non-alarms. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0048] Figure 1 It is a flow chart of a three-dimensional warning zone object recognition and early warning control method based on RF radar provided by an embodiment of the present invention;
[0049] Figure 2 It is a schematic diagram of splitting a multi-level three-dimensional warning zone provided by an embodiment of the present invention;
[0050] Figure 3 It is a schematic diagram of the installation of the radio frequency radar detection sensor and the side view of the detection area provided by an embodiment of the present invention;
[0051] Figure 4 is a schematic diagram of coordinate system conversion provided by an embodiment of the present invention;
[0052] Figure 5 1 is a schematic diagram of the principle of calculating the longitude and latitude of a target object provided by an embodiment of the present invention; wherein (a) is a position relationship diagram of the target object relative to the radio frequency radar detection sensor, (b) is a schematic diagram of the principle of calculating the radius of the latitude section, and (c) is a schematic diagram of the principle of calculating the longitude and latitude translation angle;
[0053] Figure 6 It is a schematic diagram of the structure of a three-dimensional warning zone object recognition and early warning control system based on RF radar provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0054] To make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be described in detail below. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other implementation methods obtained by ordinary technicians in this field without creative work belong to the scope of protection of the present invention.
[0055] In the description of the present invention, it should be understood that the terms "upper", "lower", "left", "right", "top", "bottom", "center", "vertical", "horizontal" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, which are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. When an element is considered to be "connected" to another element, it may be directly connected to the other element or there may be a central element at the same time. "Multiple" means at least two.
[0056] like Figure 1 As shown, an embodiment of the present invention provides a three-dimensional warning zone object recognition and early warning control method based on RF radar, comprising the following steps:
[0057] S1: Pre-set multi-level three-dimensional warning zones based on the distance from high-voltage cables.
[0058] The number of three-dimensional warning zones is set according to actual needs. In this embodiment, the three-level three-dimensional warning zone is set as an example. The three-level three-dimensional warning zone is divided into planes such as Figure 2 As shown in the figure, after the independent single layer is split, the point space between planes (A, B, C, D) and planes (A1, B1, C1, D1) is defined as a low-level three-dimensional warning area, the point space between planes (A1, B1, C1, D1) and planes (A2, B2, C2, D2) is defined as a medium-level three-dimensional warning area, and the point space between planes (A2, B2, C2, D2) and planes (A3, B3, C3, D3) is defined as a high-level three-dimensional warning area. Among them, planes (A1, B1, C1, D1) and planes (A2, B2, C2, D2) are superimposed common face domain points.
[0059] like Figure 2 As shown in the figure, in this embodiment, a multi-level three-dimensional warning zone is set directly below the high-voltage cable. The entire warning zone is divided into three three-dimensional warning zones by multiple planes with the same horizontal xy plane coordinates and different vertical z-axis coordinates. The closer the three-dimensional warning zone is to the high-voltage line, the higher the warning level. Figure 3 shown.
[0060] S2: Obtain the point cloud data collected by the radio frequency radar detection sensor, and perform coordinate conversion and filtering preprocessing.
[0061] 1. Radar base data and coordinate calibration
[0062] 1.1 Radar Cluster Base Data
[0063] The original point cloud data is generated by the initial signal processing calculation inside the radar. It is refreshed once in each scanning cycle (i.e. one frame). The data IDs between different scanning cycles are not related. Usually, one target will contain multiple points.
[0064] 1.2 Radar Coordinate System Convention
[0065] Convert the radar coordinate system to the ground coordinate system, such as Figure 4 As shown, the radar ground projection is used as the coordinate origin, the radar normal in the ground projection direction is the X-axis, the vertical X direction is the Y-axis, and the Z-axis is the plumb line of the X,Y plane.
[0066] 1.3 Radar output data type definition
[0067]
[0068] 1.4 Radar output Cluster target point data structure
[0069] 2. Coordinate transformation processing
[0070] 2.1 Spatial coordinate calibration method
[0071] After the radar determines the installation height, the point cloud signal output by the radar is the coordinate point data relative to the generalized space of the radar. If the coordinate calibration and data conversion are not performed, it cannot be used (such as the Z-axis pitch height, etc.). Therefore, it is necessary to calibrate the x, y, and z three-dimensional spatial coordinates of the original output point cloud data and calibrate the calculation conversion. To ensure the accuracy of the spatial coordinate data, 1-3 targets in different directions can be calibrated. The purpose of calibration is to lay the foundation for the next step of spatial coordinate data conversion.
[0072] 2.2 Spatial coordinate data calibration and data conversion calculation implementation
[0073] The schematic diagram of calibration with two calibration objects is as follows: Figure 5 As shown, the coordinate transformation process is expressed as follows:
[0074] X=X'
[0075] Y=Y'
[0076] Z=Z'+(X*tgθ)+Z0
[0077] Where (X', Y', Z') represents the point cloud coordinates in the radar coordinate system, (X, Y, Z) represents the point cloud coordinates converted to the ground coordinate system, Z0 represents the installation height of the RF radar detection sensor, and θ represents the calibration conversion angle.
[0078] 3. Filtration pretreatment
[0079] The point cloud filtering process includes:
[0080] S21: Estimate the two-dimensional plane of the point cloud data through RANSAC, take the local points as ground points, and other points as non-ground points to achieve the segmentation of ground points and non-ground points.
[0081] The input of the RANSAC algorithm is a set of observations, a parameterized model that can explain or adapt to the observations, and some credible parameters. The goal is achieved by repeatedly selecting a set of random subsets of the data. The selected subsets are assumed to be inliers, and the specific process can be described as follows:
[0082] Step 1. Randomly select at least three points to fit the plane;
[0083] Step 2. Use the plane obtained in step 1 to test all other point cloud data. If the distance between a point and the plane is less than the threshold, it is considered to be an internal point, otherwise it is an external point.
[0084] Step 3. If enough points are classified as hypothesized internal points, then the estimated model is reasonable enough;
[0085] Step 4. Then, use all the assumed inliers to re-estimate the plane, since it has only been estimated by the initial assumed inliers.
[0086] step5. Repeat the above steps step1-step4 for a preset number of times to select the plane with the most points in the fitting area. The points in the fitting plane are regarded as ground points, and the other points are regarded as non-ground points.
[0087] S22: Filter out ground points and retain non-ground points. By filtering out ground points, the amount of data calculation in the later stage is reduced, and the target detection process is accelerated.
[0088] S23: Perform straight-through filtering within a limited range on the retained non-ground points (x, y, z, RCS), and only select the points within the multi-level three-dimensional warning area and whose RCS values are within the set range for the non-ground points; the non-engineering vehicle point cloud data can be filtered by pre-setting the engineering vehicle RCS value range.
[0089] First, filter by point coordinates to retain point cloud data within the multi-level 3D warning area. Second, considering that different objects have different RCS values (RCS reflection area refers to the ratio of the target's echo power density to the incident radar wave to the incident power density), by setting the RCS value range of engineering vehicles, some non-target point cloud data can be filtered out, thereby filtering out interference from people, trees, etc. The following table lists the RCS values of some objects for reference.
[0090]
[0091] S3: Cluster the preprocessed point cloud data to achieve target detection.
[0092] This embodiment adopts a clustering method based on Euclidean distance, the core idea of which is to classify points that are close in space into one category, and points that are far away into different categories. Each point cloud cluster obtained is a target point cloud cluster, thereby realizing target detection. This algorithm is usually used for hierarchical clustering and can generate a clustering result with a tree structure, in which each node represents a cluster and the child node represents the child cluster of the node.
[0093] The steps of the Euclidean clustering algorithm in this embodiment mainly include:
[0094] S31: Initialization, each sample point is regarded as an independent cluster;
[0095] S32: Calculate the distance between adjacent clusters, usually using single linkage or complete linkage methods;
[0096] S33: merge the two closest clusters to form a new cluster;
[0097] S34: Repeat the above steps until all points are clustered into one category or the preset number of clusters is reached.
[0098] The following is the pseudo code for the Euclidean clustering algorithm:
[0099] (1) Establish a KD-Tree based on the original target data point P;
[0100] (2) The queue Q is used to store the points to be searched and the set S that indicates whether the points have been visited.
[0101] (3) For each point p belonging to P, traverse the point cloud;
[0102] if(p has not been visited);
[0103] Add p to Q;
[0104] While(Q is not empty);
[0105] Take point t from Q;
[0106] Set the status of t to visited;
[0107] Find points near t whose distance is less than l. If they have not been visited, put them in the queue;
[0108] Keep the above points as clusters.
[0109] In the actual application of all RF radars, due to different surrounding environments, especially when there are multiple strong reflective objects in the distance, according to the electromagnetic wave scattering characteristics, the target will be more or less unstable after multiple echo projections. The point cloud data output by the radar sensor will show offset jumps when performing three-dimensional restoration and display, which can easily cause the accurate judgment of the effective target to be filtered out as a noise signal. Therefore, the effective identification and filtering of the offset part of the effective target is as follows: For other targets within the preset offset range of the stable and effective target, determine whether their RCS values are equal to the RCS value of the stable and effective target. If they are equal, they are considered to be the same target and recorded, but only the effective target is retained in subsequent judgments and visualization displays; if the RCS values are not equal, it is considered to be a new target and recorded.
[0110] In practical applications, there are dynamic and static targets. By defining a speed threshold, if the speed of the target is less than the speed threshold, the target is considered to be a static target; if the speed of the target is not less than the speed threshold in a preset number of consecutive frames, the target is considered to be a dynamic target; the static targets and dynamic targets are separated and recorded and stored separately. After separation, dynamic targets or static targets can be filtered out according to needs to obtain the required targets. It can meet complex separation and filtering requirements such as retaining only dynamic targets, retaining only static targets, or requiring some targets to be statically retained and dynamically monitored. This is especially important in the application of early warning control and anti-electric shock in near-electric operation scenarios under complex working conditions in a small range of close distances in the power grid.
[0111] S4: Match the three-dimensional warning zone level to which the target belongs based on its spatial coordinates.
[0112] Specifically, the judgment rules are set as follows:
[0113] (1)X1(A,B,A1,B1,A2,B2,A3,B3) <X<X2(C、D、C1、D1、C2、D2、C3、D3)
[0114] (2)Y1(B,C,B1,C1,B2,C2,B3,C3) <Y<Y2(A、D、A1、D1、A2、D2、A3、D3)
[0115] (3) Z1(A, B, C, D) <Z<Z2(A1、B1、C1、D1)
[0116] (4) Z2(A1, B1, C1, D1) <Z<Z3(A2、B2、C2、D2)
[0117] (5) Z3 (A2, B2, C2, D2) <Z<Z4(A3、B3、C3、D3)
[0118] According to whether the three-dimensional coordinates (X, Y, Z) of the target object meet the above conditions, if (1), (2), and (3) are met, the target object belongs to the low-level three-dimensional warning zone; if (1), (2), and (4) are met, the target object belongs to the intermediate three-dimensional warning zone; if (1), (2), and (5) are met, the target object belongs to the high-level three-dimensional warning zone.
[0119] S5: Calculate the longitude and latitude of the target object and match it with the longitude and latitude of the engineering vehicle.
[0120] Specifically, the azimuth position of the target object relative to the RF radar detection sensor is calculated, and the absolute latitude and longitude of the target object (the highest point of the target object is used instead of the target object for calculation) is calculated by combining the latitude and longitude and azimuth of the location of the RF radar detection sensor. The relative position relationship and angle information of the target object (point B) and the RF radar detection sensor (point A) are as follows: Figure 5 As shown in (a).
[0121] The average radius ARC refers to the radius of the sphere, which can be simply understood as the radius of the equatorial circle, the radius of the section at latitude 0 degrees. In a sphere, the farther away from the center of the sphere, the smaller the radius of the section. Latitude is a line angle with the equatorial plane as the starting point. Figure 5 As shown in (b), knowing the radius of the sphere ARC and the line angle lat1 (current latitude), we can calculate the radius of the section, that is, the radius of the section at the current latitude.
[0122] The principle of calculating the latitude and longitude translation angle is as follows Figure 5 As shown in (c), in a circle, the angle β is the ratio of the chord to the circumference multiplied by 360.
[0123] In fact, the radius r in the longitude direction is different from the chord length in the latitude direction. Figure 5 It can be seen that:
[0124] The latitude translation angle is: d*sinα / [ARC*cos(lat1)*2π / 360];
[0125] The longitude translation angle is: d*cosα / (ARC*2π / 360);
[0126] The translation angle plus the original longitude and latitude will give the new longitude and latitude. Finally, the longitude and latitude calculation formula of the target object is as follows:
[0127] long2=d*sinα / [ARC*cos(lat1)*2π / 360]+long1
[0128] lat2=d*cosα / (ARC*2π / 360)+lat1
[0129] Wherein, long1 and lat1 represent the latitude and longitude of the RF radar detection sensor, respectively; long2 and lat2 represent the latitude and longitude of the target, respectively; d represents the distance between the target and the RF radar detection sensor; ARC represents the radius of the earth; and α represents the radar normal deflection angle.
[0130] The longitude and latitude of the target object are matched with the longitude and latitude obtained by the positioning module on the engineering vehicle (preferably set at the end of the operation, such as the front end of the boom; of course, many engineering vehicles also have their own positioning modules, and a positioning module can also be added at the end of the operation on this basis, with the positioning module at the end of the operation as the main one). If the match is successful, it can be known which engineering vehicle the detected target object is, and it is convenient to send an early warning control instruction to it. If the match is unsuccessful, only an early warning is issued.
[0131] S6: Sending a warning control instruction of a corresponding level to a matching engineering vehicle according to the level of the three-dimensional warning zone of the target object.
[0132] As the level of the warning zone increases, the corresponding early warning control instructions increase from warning, control deceleration to control lock. The following takes the three-level three-dimensional warning zone as an example for explanation.
[0133] For the low-level warning area, the corresponding low-level early warning control output is processed. For example, if an engineering vehicle is detected to intrude into the low-level warning area in the safety area under the high-voltage line, a timely early warning will be reported to the monitoring center, and a low-level laser will be issued for safety warnings and voice calls, and the vehicle control device will be forced to install.
[0134] For the intermediate warning area, the corresponding early warning control output of the intermediate level is processed, such as: prompting the danger of approaching the high-voltage area and forcing the engineering vehicles to slow down.
[0135] For the high-level warning area, the highest point is close to the high-voltage line layer, and the corresponding high-level early warning control output is processed. If it is detected that a construction vehicle intrudes into the high-level warning area in the safety area under the high-voltage line, a warning will be reported to the monitoring center in time, and at the same time, a control lock command will be issued to the control device of the construction vehicle, and a high-level laser will be activated for safety warnings and voice calls. In this way, the safety of power line facilities and construction engineering vehicles, personnel and property can be protected in time.
[0136] In some preferred embodiments, it also includes: three-dimensional visualization of the identified target object and the multi-level three-dimensional warning zone, so as to facilitate intuitive observation of the state of the target object invading the warning zone.
[0137] In some preferred embodiments, the steps between step S3 and step S4 further include:
[0138] S01: Match the detected target object of the current frame with the target object list; the target object list records the position information and life value of each target object, which is initially empty;
[0139] S02: Calculate the distance between the target detected in the current frame and each target in the target list. If the distance between the two is less than the set distance threshold, it is considered that the two are the same target, and the position information of the target is updated, and its life value is restored to the initial value N; if the distance between the two is not less than the set distance threshold, it is considered to be a new target, and the target and its position information are added to the target list, and its life value is set to the initial value N;
[0140] S03: if there is a target in the target list that has not matched itself in the current frame, the life value of the target in the target list is reduced by 1; when the life value of a target in the target list is 0, it is deleted from the target list;
[0141] S04: After continuously acquiring M frames and performing the above steps S01 to S03, proceed to step S4; wherein M>N.
[0142] In rainy, snowy, foggy, and hazy weather conditions, many target point clouds are detected to appear in only one or two frames and then disappear. By setting the life value, if the target is detected to appear in the current frame, its life value will be restored to the initial value N. If the current frame does not detect the target that appeared before, its life value will be reduced by 1. When there is no target in N consecutive frames, the life value is zero. Therefore, the target is considered to be interference noise, and the target is deleted and does not participate in the subsequent calculation and detection process. It can adapt to complex environments such as rain, snow, fog, and haze to prevent false alarms, missed alarms, and non-alarms.
[0143] The above embodiment discloses a method for object recognition and early warning control in a three-dimensional warning zone based on RF radar. By defining a multi-level three-dimensional warning zone in a custom three-dimensional space, the division of the degree of danger is realized. Three-dimensional warning zones of different levels send early warning control instructions of different intervention degrees to engineering vehicles, thereby realizing the detection of targets under high-voltage cables and early warning linkage control of engineering vehicles. By filtering and clustering point cloud data, accurate identification of targets can be achieved, and then by tracking targets, noise interference can be further filtered. Targets can still be effectively and stably detected under the interference of complex environments such as rain, snow, fog, and haze, preventing the occurrence of false alarms, missed alarms, and non-alarms.
[0144] The embodiment of the present invention also provides a three-dimensional warning zone object recognition and early warning control system based on RF radar, including a power supply component module, a main control module, a radio frequency radar detection sensor component module, and a positioning module. The power supply component module supplies power to the entire system; the radio frequency radar detection sensor component module and the positioning module are both connected to the main control module in communication; the main control module is configured to execute the three-dimensional warning zone object recognition and early warning control method based on RF radar as described above. Figure 6 Give a detailed description of the system.
[0145] Power supply component module: It is composed of an external solar UPS component module, a total power supply management module of the internal core main control module onboard part, a DC12V voltage-stabilized power supply module, a DC5V voltage-stabilized power supply module, and a DC3.3V voltage-stabilized power supply module. The DC12.8V-17.5V linear direct current output by the external solar cell component module is sent to the → total power supply management control module, and at the same time, after internal surge and voltage limiting protection, it is output to the input end of the independent DC12V, DC5V, and DC3.3V voltage-stabilized power supply modules of → the back-end parts. Then, the DC12V, DC5V, and DC3.3V voltage-stabilized power supply modules output stable voltage groups after internal voltage stabilization, filtering, etc., and send them to each electronic module component to provide uninterrupted and stable power supply, ensuring that each module component can operate stably. The main power supply management module has the real-time protection and monitoring communication reporting functions for the input and output voltage, current, overvoltage, overcurrent, and surge pulse. It uses the RS485 bus interface to send the protocol data of related functions to the core main control module (AIU). When the core main control module (AIU) detects an abnormality, it will handle and report it.
[0146] Main control module (AIU) and peripheral modules: The main control module is designed with onboard necessary units such as CPU, NPU, GPU, EMMC, DDR4, hardware watchdog and high-power PMIC. At the same time, a large number of peripheral resource interfaces required for conventional / extended applications are designed in the periphery, including: SDIO, SATA, UART, dual Gigabit Ethernet LAN, CAN-bus, USB3.0, USB2.0, PCIE3.0, I2C, SPI, ADC, PWM, audio, I2S, DP display, HDMI display, LVDS display, MIPI-DSI display, MIPI-CSI camera input and other interfaces, which can meet the human-computer interaction, model, graphics processing and high response requirements of various digital Internet of Things industrial control systems. The core main control module is connected to various peripheral electronic function interface modules, including power management module, radio frequency radar detection sensor component module, video camera component module, Beidou / GPS positioning module, 5G / 4G / WiFi / LAN / VPN / APN internal and external network wireless routing communication component module, RF wireless networking communication component module, SATA / USB / SD data storage unit module, MOS / pulse relay power management control output component module and other sensor access GPIO / PWM / ADC / DAC signal control drive and analog-to-digital conversion acquisition control and other access controller modules. The main control module of the present invention is mainly responsible for the data acquisition and communication of all radars and other monitoring sensor component modules accessed in the onboard system, the data transmission and reception control of different communication protocols of the sound and light warning controller component and the engineering machinery vehicle controller component, the analysis of various data protocols, model calculation, logic analysis and judgment, linkage output control, early warning reporting platform, data storage, on-site WEB three-dimensional display and integrated configuration and other functional data, so as to realize all applications in various different environments and different functional scenarios.
[0147] Detection sensor component module: mainly composed of RF radar sensors, cameras and other sensor components, which send various real-time data monitored to the main core controller of the central core control layer through different transmission channel interfaces for data analysis and processing. The camera can be used to retrieve real-time video on the scene when the alarm is triggered to further confirm the target object intrusion into the warning area.
[0148] External access to on-site sound and light warning device control component module: mainly composed of night laser surface light curtain voice warning integrated controller device component. As slave A, it is responsible for receiving and processing various protocol data instructions issued by the on-site main control module, and outputting long-distance laser array light curtain and voice warning effects with different control functions according to different data protocol instructions, so as to realize accurate and effective sound and light warning on site.
[0149] Externally accessed engineering machinery vehicle control drive component module: mainly composed of engineering vehicle bidirectional control actuator device components. As slave B, it is responsible for receiving and processing various protocol data instructions issued by the on-site main control module, parsing and processing them, and performing relevant real-time monitoring, query and function control on various mechanical over-height vehicles and tower crane facilities that are out of range according to different data protocol instructions, so as to achieve early warning and safe locking and shutdown of mechanical engineering vehicle facilities that are out of the safety warning range, thereby achieving the safety protection of relevant personnel and facilities on site.
[0150] On-site integrated control WEB: The present invention designs an on-site core main control module IE WEB terminal and WIFI access, which is responsible for the on-site integration of the overall monitoring and display configuration functions, and realizes the on-site configuration and debugging of various functions of the entire radar early warning monitoring and control (such as the radar early warning range setting for different functional scenarios, and the setting of different early warning monitoring levels for three-dimensional visualization), and on-site integration functions such as three-dimensional visualization real-time monitoring and display of radar sensor data.
[0151] The main control module (AIU) monitors and compares the longitude and latitude of the target object identified and read through the RF wireless protocol to control the alarm. Radar data processing 3D display integrated WEB integrated management control: It is known that in the current security radar sensor detection application, it is mainly based on on-site local lightweight control, basically based on low data volume control and configuration, and big data processing is basically sent to the back-end platform for analysis and processing, which will cause great safety risks. For example, communication problems will cause serious problems due to untimely data reception. In view of this, the present invention solves the problem of directly building and developing a three-dimensional visualization model through the IE browser inside the core main control module (AIU) controller of the front end, and integrates the CPU, NPU, GPU and other acquisition and control parts in data interconnection, realizing the integrated WEB integration of real-time dynamic data accurate monitoring and acquisition, model calculation, early warning, control, configuration and three-dimensional display of the entire system, realizing the local on-site visualization configuration operation of data, data storage and buffering, local processing control output monitoring, local and wireless remote data synchronous transmission, lightweight upload platform for processed data, i.e. dual backup, and local and remote login, viewing and operation through wireless 4G / network cable / wireless WIFI.
[0152] The above embodiment provides a three-dimensional warning zone object recognition and early warning control system based on RF radar. The localized core main control module can visually customize the setting of different status warnings under the three-dimensional space multi-level interval warning zone, and can customize the setting of the coordinate range of the three-dimensional space visual monitoring warning zone of multiple levels. Different warning levels can be set for each level of the warning zone fence. Each level supports external configuration output monitoring, early warning, external joint control output and configurable marking of different colors for distinction. The front-end localized core main control module accurately identifies the specific distance position, angle, height, etc. of the controlled engineering vehicles and uncontrolled target objects in the radar output during multi-target detection, and can distinguish and monitor the spatial positioning and accurately identify different targets in the warning area for early warning. It realizes accurate identification, tracking, early warning and interlocking control of the height and position space area of the multi-target vehicle crane arm of different engineering vehicles and mechanical crane action targets under the high-voltage line in the warning space area. The front-end localized core main control module effectively monitors and processes the radar output signal in the state of anti-rain, snow, fog, and haze weather signal noise interference to prevent the occurrence of false alarms, missed alarms, and non-alarms.
[0153] The embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the three-dimensional warning zone object recognition and early warning control method based on RF radar as described above is implemented.
[0154] It can be understood that the same or similar parts of the above embodiments can be referenced to each other, and the contents not described in detail in some embodiments can refer to the same or similar contents in other embodiments.
[0155] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.
[0156] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0157] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0158] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0159] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.
Claims
1. A three-dimensional warning area object recognition and early warning control method based on RF radar, characterized in that: The steps include: S1: Pre-set multi-level three-dimensional warning zones according to the distance from the high-voltage cables; S2: Obtain the point cloud data collected by the radio frequency radar detection sensor, and perform coordinate conversion and filtering preprocessing; S3: Cluster the preprocessed point cloud data to achieve target detection; S4: Match the three-dimensional warning zone level to which the target belongs based on its spatial coordinates; S5: Calculate the longitude and latitude of the target object and match it with the longitude and latitude of the engineering vehicle; S6: Sending a warning control instruction of a corresponding level to a matching engineering vehicle according to the warning zone level of the target three-dimensional object.
2. The RF radar-based three-dimensional warning zone object recognition and early warning control method according to claim 1 is characterized in that: In step S1, a multi-level three-dimensional warning zone is set directly below the high-voltage cable, and the entire warning zone is divided into multiple three-dimensional warning zones by multiple planes with the same horizontal xy plane coordinates and different vertical z-axis coordinates; the closer the three-dimensional warning zone is to the high-voltage line, the higher its warning level.
3. The RF radar-based three-dimensional warning zone object recognition and early warning control method according to claim 1 is characterized in that: In step S2, the coordinate conversion process is expressed as follows: X=X' Y=Y' Z=Z'+(X*tgθ)+Z0 Where (X', Y', Z') represents the point cloud coordinates in the radar coordinate system, (X, Y, Z) represents the point cloud coordinates converted to the ground coordinate system, Z0 represents the installation height of the RF radar detection sensor, and θ represents the calibration conversion angle.
4. The RF radar-based three-dimensional warning zone object recognition and early warning control method according to claim 1 is characterized in that: In step S2, the point cloud filtering process includes: S21: Estimate the two-dimensional plane of the point cloud data through RANSAC, take the points in the local area as ground points, and other points as non-ground points to achieve the segmentation of ground points and non-ground points; S22: Filter out ground points and keep non-ground points; S23: Perform straight-through filtering within a limited range on the retained non-ground points (x, y, z, RCS), and only select the points within the multi-level three-dimensional warning area and whose RCS values are within the set range for the non-ground points; filter the non-engineering vehicle point cloud data by pre-setting the engineering vehicle RCS value range.
5. The RF radar-based three-dimensional warning zone object recognition and early warning control method according to claim 1 is characterized in that: After step S3, at least one of target object offset identification and filtering and target object dynamic and static identification is also included: Target offset identification and filtering: For other targets within the preset offset range of the stable and effective target, determine whether their RCS values are equal to the RCS value of the stable and effective target. If they are equal, they are considered to be the same target and recorded; if the RCS values are not equal, they are considered to be new targets and recorded; Identify the dynamic and static targets; define the speed threshold. If the speed of the target is less than the speed threshold, the target is considered to be a static target. If the speed of the target object detected in the preset number of consecutive frames is not less than the speed threshold, the target object is considered to be a dynamic target object; the static target object and the dynamic target object are separated and recorded and stored separately.
6. The RF radar-based three-dimensional warning zone object recognition and early warning control method according to claim 1 is characterized in that: In step S5, the latitude and longitude of the target object are calculated by the following formula: long2=d*sinα / [ARC*cos(lat1)*2π / 360]+long1 lat2=d*cosα / (ARC*2π / 360)+lat1 Wherein, long1 and lat1 represent the latitude and longitude of the RF radar detection sensor, respectively; long2 and lat2 represent the latitude and longitude of the target, respectively; d represents the distance between the target and the RF radar detection sensor; ARC represents the radius of the earth; and α represents the radar normal deflection angle.
7. The RF radar-based three-dimensional warning zone object recognition and early warning control method according to claim 1 is characterized in that: Also includes: The identified targets and multi-level three-dimensional warning areas are displayed in three-dimensional visualization.
8. The RF radar-based three-dimensional warning zone object recognition and early warning control method according to any one of claims 1 to 7, characterized in that: The steps between step S3 and step S4 also include: S01: Match the detected target object of the current frame with the target object list; the target object list records the position information and life value of each target object, which is initially empty; S02: Calculate the distance between the target detected in the current frame and each target in the target list. If the distance between the two is less than the set distance threshold, it is considered that the two are the same target, and the position information of the target is updated, and its life value is restored to the initial value N; if the distance between the two is not less than the set distance threshold, it is considered to be a new target, and the target and its position information are added to the target list, and its life value is set to the initial value N; S03: if there is a target in the target list that has not matched itself in the current frame, the life value of the target in the target list is reduced by 1; when the life value of a target in the target list is 0, it is deleted from the target list; S04: After continuously acquiring M frames and performing the above steps S01 to S03, proceed to step S4; wherein M>N.
9. A three-dimensional warning area object recognition and early warning control system based on RF radar, characterized in that: It includes a power supply component module, a main control module, a radio frequency radar detection sensor component module, and a positioning module. The power supply component module supplies power to the entire system; the radio frequency radar detection sensor component module and the positioning module are both communicatively connected to the main control module; the main control module is configured to execute the three-dimensional warning zone object recognition and early warning control method based on RF radar as described in any one of claims 1 to 8.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for object recognition and early warning control in a three-dimensional warning zone based on RF radar is implemented as described in any one of claims 1 to 8.
Citation Information
Cited By
Engineering vehicle intrusion warning method and system under high-voltage line
CN119068614A
Method and system for warning engineering vehicle intrusion under high-voltage lines
CN119068614B
Radar false target identification method, device and equipment and storage medium
CN119716840A
Device and method for monitoring three-dimensional displacement structure by group microwave radar
CN120703749A
