A dual-radar cooperative scheduling adaptive perception control method and system

Abstract

The application relates to the technical field of intelligent control of engineering machinery, and provides a self-adaptive sensing control method and system based on double-radar cooperative scheduling. The method comprises the following steps: installing a first rotating radar and a second rotating radar on a rotatable support respectively; acquiring obstacle distribution data of a tower crane hoisting operation area in real time through the first rotating radar and the second rotating radar; acquiring real-time stage information of hoisting operation in real time through an operation state sensing module; calculating a sensing dead angle position according to the real-time stage information; dynamically generating a first task instruction and a second task instruction according to a priority level of a target monitoring area, the sensing dead angle position and the obstacle distribution data; and controlling the first rotating radar and the second rotating radar to adjust a rotating angle and sensing parameters, so that the first rotating radar performs key sensing on the target monitoring area with high priority, and the second rotating radar performs compensatory sensing on an area corresponding to a sensing blind area of the first rotating radar and the sensing dead angle position.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent control of engineering machinery, and in particular to a self-adaptive perception control method and system based on dual-radar cooperative scheduling. BACKGROUND

[0002] In the field of construction, the safety of tower crane hoisting operation depends on the real-time perception of obstacles in the operation area. In the traditional technology, single radar sensor or fixed-angle arrangement of multi-radar system has significant defects: on the one hand, single radar is limited by detection angle and physical obstruction, and it is difficult to cover complex operation area, resulting in large area of perception blind area; on the other hand, the fixed multi-radar system usually adopts a preset scanning strategy, which cannot adjust the monitoring focus according to the dynamic stage of hoisting operation (such as lifting, rotating, landing point), resulting in insufficient monitoring accuracy or resource waste in high-risk areas (such as below the hook, rotating path of the crane boom).

[0003] In the prior art, some solutions try to improve the perception effect by increasing the number of radars or fixedly dividing the detection area, but the following core problems are not solved:

[0004] 1. Lack of dynamic association between operation stage and monitoring area: the priority of the target monitoring area is not dynamically adjusted according to the risk characteristics of different stages such as lifting, rotating and landing point, for example, the high-risk area below the hook is not scanned in the lifting stage, and the front path is not monitored preferentially in the rotating stage;

[0005] 2. Lack of compensation mechanism for dead angle: the traditional multi-radar system simply splices the detection range, and does not calculate the perception dead angle based on the radar installation position and real-time angle, and lacks the strategy of cooperative compensation of main and auxiliary radars, resulting in that the blocked area or radar beam coverage blind area cannot be effectively monitored;

[0006] 3. Lack of intelligence of task allocation strategy: the radar task instruction is not dynamically generated combined with obstacle distribution data and priority rules, making it difficult to balance the detection accuracy and efficiency under different working conditions, for example, the scanning frequency is not improved in the high-risk area, and there is redundant detection in the low-risk area.

[0007] Therefore, a method is needed to solve at least one of the above problems. SUMMARY

[0008] The present application provides a self-adaptive perception control method and system based on dual-radar cooperative scheduling, aiming to solve the problem that the prior art does not propose a dual-radar cooperative scheduling method based on operation stage information, target area priority and perception dead angle compensation, especially lacks a technical solution to realize perception range coverage optimization and redundant monitoring of high-risk areas through dynamic task allocation strategy.

[0009] In a first aspect, the embodiments of the present application provide a dual-radar cooperative scheduling adaptive perception control method; the method comprises:

[0010] The first rotating radar and the second rotating radar are respectively installed on a rotatable support to construct a dual-radar module perception system, and the dual-radar module is in communication connection with a perception task scheduling unit and a working state perception module;

[0011] Obstacle distribution data of a tower crane hoisting operation area is acquired in real time by the first rotating radar and the second rotating radar, and real-time stage information of the hoisting operation is acquired in real time by the working state perception module, wherein the real-time stage information comprises a lifting stage, a rotating stage and a landing point stage;

[0012] According to the real-time stage information, a target monitoring area corresponding to the current operation stage and a corresponding priority level are determined in combination with a preset target area priority rule, a perception dead angle position is calculated according to the installation position and the real-time rotating angle of the first rotating radar and the second rotating radar, and first task instructions and second task instructions are dynamically generated according to the priority level of the target monitoring area, the perception dead angle position and the obstacle distribution data, and according to a preset task allocation strategy;

[0013] According to the first task instructions and the second task instructions, the rotating angle and the perception parameters of the first rotating radar and the second rotating radar are controlled to adjust, so that the first rotating radar performs intensive perception on the target monitoring area with high priority, and the second rotating radar performs compensatory perception on the area corresponding to the perception blind area and the perception dead angle position of the first rotating radar, thereby realizing coverage optimization of the perception range and redundant monitoring of high-risk areas.

[0014] In some embodiments, the real-time acquisition of the obstacle distribution data of the tower crane hoisting operation area by the first rotating radar and the second rotating radar comprises: time synchronization and spatial coordinate calibration are respectively performed on original point cloud data collected by the first rotating radar and the second rotating radar, and noise points are filtered based on a preset adaptive filtering algorithm; the filtered point cloud data is input into a preset target classification neural network model to identify the type of obstacles and mark static obstacles and dynamic obstacles; and structured distribution data containing the position, type and risk level of the obstacles are generated according to the position coordinates, size information and motion state of the obstacles.

[0015] In some embodiments, the real-time stage information of the hoisting operation is acquired in real time by the operation state perception module, including: collecting the data of the lifting distance encoder, the rotation mechanism angle encoder and the hook height sensor of the tower crane, and inputting the multi-dimensional sensor signals into a preset state recognition finite state machine model; the action sequence of the hoisting operation is analyzed in time sequence by the state machine model, and the current operation stage is judged in combination with the motion threshold of each actuator; when the lifting motor speed suddenly changes from zero and the hook height changes, it is determined that the lifting stage is reached; when the rotation mechanism angle changes at a rate exceeding a preset threshold, it is determined that the rotation stage is reached; when the hook height tends to be stable and approaches the target height, it is determined that the landing point stage is reached.

[0016] In some embodiments, the target monitoring area corresponding to the current operation stage and the corresponding priority level are determined according to the real-time stage information and in combination with a preset target area priority rule, including: a multi-area division model including a lifting area, a rotation path area and a landing buffer area is preset, and a dynamic priority mapping table is established for different operation stages; when in the lifting stage, a radius of 5 meters below the hook is set as a first priority monitoring area, and the area directly below the jib is set as a second priority; when in the rotation stage, a 180-degree sector area in front of the current rotation direction of the jib is set as a first priority, and two 60-degree sector areas on the sides are set as a second priority; a priority optimization neural network trained by historical accident data is used to correct the preset priority rule online, and the priority coefficients corresponding to each area are dynamically adjusted according to the real-time obstacle distribution.

[0017] In some embodiments, the dead angle position is calculated according to the installation positions and real-time rotation angles of the first and second rotating radars, including: a three-dimensional coordinate system with the rotation center of the tower crane as the origin is established, and the installation coordinates and orientation angle parameters of the first and second rotating radars are obtained; a mathematical model of a fan-shaped detection area is constructed based on the detection distance threshold and beam angle range of the radar sensor; the detection ranges corresponding to the current rotation angles of the two groups of radars are projected onto the operation area plane by a coordinate transformation algorithm; the union and complement of the detection areas of the two radars are calculated, the blind area region that is not covered is identified, the detection blind area that is blocked is secondarily screened in combination with an obstacle blocking probability model, and the actual dead angle position coordinates are determined.

[0018] In some embodiments, the first task instruction and the second task instruction are dynamically generated according to the priority level of the target monitoring area, the dead angle position and the obstacle distribution data, and a preset task allocation strategy, including: constructing a radar task parameter space containing detection accuracy, scanning frequency and angle adjustment step, and establishing a target function with the priority level as the weight; using an improved particle swarm optimization algorithm, and taking covering the first priority area, compensating for the dead angle and balancing the radar load as the constraint condition, the task parameters of the two groups of radars are optimized and solved; when the priority difference of the target monitoring area exceeds a preset threshold, the first rotating radar is allocated to perform a high-frequency fine scanning task in a high priority area, and the second rotating radar is allocated to perform a reciprocating blind compensation scanning task in a dead angle area; when the priority difference is low, a cooperative scanning strategy is generated through a task scheduling neural network, so that the two radars perform area coverage according to a preset phase difference.

[0019] In some embodiments, the first rotating radar and the second rotating radar are controlled to adjust the rotation angle and the sensing parameter according to the first task instruction and the second task instruction, including: establishing a kinematics model of a radar servo system, and converting the target angle in the task instruction into a motor control pulse sequence; using a PID control algorithm with feedforward compensation to perform closed-loop control on the radar rotating shaft, and correcting the angle deviation in real time; according to the sensing parameter in the task instruction, the transmission and reception module parameters are dynamically configured through a radar hardware drive interface; during the adjustment process, the angular velocity and the acceleration of the rotating mechanism are monitored in real time through a radar self-checking sensor, and a safety speed limiting mechanism is triggered when abnormal vibration is detected; wherein the sensing parameter includes the detection distance, the resolution and the pulse repetition frequency.

[0020] In some embodiments, the method further includes: establishing a historical operation data training set containing operation stage, obstacle distribution, radar scheduling strategy and collision risk event label; training a radar scheduling strategy network through a deep reinforcement learning algorithm, taking the current operation state, the target area priority and the dead angle information as the state input, taking the main second rotating radar task parameter combination as the action output, and taking the maximum risk warning accuracy and the minimum scheduling delay as the reward function; after completing one operation cycle, the deviation data of the actual scheduling effect and the preset strategy are input into the strategy network for online update, so as to realize adaptive optimization of the task allocation strategy.

[0021] In some embodiments, the method further comprises: constructing a multi-dimensional risk assessment model comprising obstacle motion trajectory prediction, priority area risk level, radar perception blind area; using a long short-term memory network to model the position sequence of dynamic obstacles and predict the motion trajectory within the next 5 seconds; calculating the spatial overlap between the predicted trajectory and the target monitoring area, and triggering an emergency scheduling mechanism when the predicted trajectory enters a first priority area and the current radar perception frequency is lower than a safety threshold: forcing the first rotating radar to switch to a full-time tracking scanning mode in the target area, and the second rotating radar to adjust to a pre-warning scanning mode in the predicted trajectory direction.

[0022] In a second aspect, the application provides a dual-radar cooperative scheduling adaptive perception control system, applied to a controller, the system comprising:

[0023] A system construction unit is configured to install a first rotating radar and a second rotating radar on a rotatable support respectively, and construct a dual-radar module perception system, which is in communication connection with a perception task scheduling unit and a working state perception module respectively;

[0024] A data acquisition unit is configured to acquire obstacle distribution data of a tower crane hoisting operation area in real time through the first rotating radar and the second rotating radar, and acquire real-time stage information of the hoisting operation in real time through the working state perception module, wherein the real-time stage information comprises a lifting stage, a rotating stage and a landing point stage;

[0025] An instruction generation unit is configured to determine a target monitoring area corresponding to a current operation stage and a corresponding priority level according to the real-time stage information and in combination with a preset target area priority rule, calculate a perception dead angle position according to the installation positions and real-time rotating angles of the first rotating radar and the second rotating radar, and dynamically generate a first task instruction and a second task instruction according to the priority level of the target monitoring area, the perception dead angle position and the obstacle distribution data, and in accordance with a preset task allocation strategy;

[0026] A perception monitoring unit is configured to control the first rotating radar and the second rotating radar to adjust rotating angles and perception parameters according to the first task instruction and the second task instruction, so that the first rotating radar performs intensive perception on the target monitoring area with high priority, and the second rotating radar performs compensatory perception on the area corresponding to the perception blind area and the perception dead angle position of the first rotating radar, thereby achieving coverage optimization of the perception range and redundant monitoring of high-risk areas.

[0027] The adaptive sensing control method and system provided by the embodiment of the application can solve the monitoring lag problem of the traditional fixed strategy by real-time acquisition of hoisting operation stage information (hoisting, rotating, landing point), dynamic determination of a target monitoring area in combination with preset priority rules, focusing of radar resources on a current high-risk area (for example, the area below the lifting hook is monitored in priority in the hoisting stage, and the front path is covered in priority in the rotating stage), and calculation of a sensing dead angle based on a radar installation position and a real-time angle. Through the main second rotating radar cooperative strategy (the first rotating radar focuses on sensing a high-priority area, and the second rotating radar compensates for a blind area and a dead angle), no-dead-angle coverage of the operation area and double monitoring of the high-risk area are realized, and the collision risk is significantly reduced. Through a preset task allocation strategy, main second rotating radar instructions are dynamically generated, the detection accuracy (for example, the scanning frequency and the angle step) is matched with the monitoring demand, the safety performance is improved, resource waste is avoided, and higher efficiency and flexibility are achieved compared with the traditional fixed parameter setting.

[0028] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory and are not restrictive of the application. BRIEF DESCRIPTION OF DRAWINGS

[0029] In order to more clearly illustrate the technical solutions of the embodiments of the application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0030] Figure 1 is a step schematic flow chart of an adaptive sensing control method of a dual-radar cooperative scheduling provided by an embodiment of the application;

[0031] Figure 2 is a structural schematic diagram of a dual-radar module sensing system provided by an embodiment of the application;

[0032] Figure 3 is a structural schematic block diagram of an adaptive sensing control system of a dual-radar cooperative scheduling provided by an embodiment of the application;

[0033] Figure 4 is a structural schematic block diagram of a controller provided by an embodiment of the application.

[0034] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory and are not restrictive of the application. DETAILED DESCRIPTION

[0035] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described, obviously, the described embodiments are a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all the other embodiments obtained by a person of ordinary skill in the art without creative work are within the scope of protection of the present application.

[0036] The flowcharts shown in the drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor are they necessarily executed in the order described. For example, some operations / steps can be further decomposed, combined or partially merged, so the actual execution order can be changed according to the actual situation.

[0037] It should be understood that, in order to clearly describe the technical solutions of the embodiments of the present application, in the embodiments of the present application, the terms "first", "second", etc. are used to distinguish the same items or similar items with basically the same functions and effects. A person of ordinary skill in the art can understand that the terms "first", "second", etc. do not limit the quantity and execution order, and the terms "first", "second", etc. also do not necessarily mean that they are different.

[0038] It should be understood that the terms used in the present application are only for the purpose of describing specific embodiments and do not intend to limit the present application. As used in the present application specification and the appended claims, unless otherwise clear from the context, the singular forms "a", "an" and "the" are intended to include the plural forms.

[0039] It should also be understood that the term "and / or" used in the present application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0040] In the field of building construction, the safety of tower crane hoisting operations depends on real-time perception of obstacles in the operating area. In traditional technology, single radar sensors or fixed-angle multi-radar systems have significant defects: on the one hand, single radar is limited by detection angle and physical obstruction, making it difficult to cover complex operating areas and resulting in large areas of perception blind spots; on the other hand, fixed multi-radar systems usually use preset scanning strategies and cannot adjust monitoring priorities according to the dynamic stages of hoisting operations (such as lifting, rotating, and landing), resulting in insufficient monitoring accuracy or resource waste in high-risk areas (such as below the hook and the rotating path of the jib).

[0041] In the prior art, some solutions attempt to improve perception by increasing the number of radars or fixedly dividing the detection area, but do not solve the following core problems:

[0042] 1. The dynamic association between the operation stage and the monitoring area is missing: the priority of the target monitoring area is not dynamically adjusted according to the risk characteristics of different stages such as lifting, rotation, landing point, etc. For example, the high-risk area below the lifting hook is not scanned in the lifting stage, and the front path is not monitored in the rotation stage.

[0043] 2. The compensation mechanism for the sensing dead angle is insufficient: the traditional multi-radar system simply splices the detection range, does not calculate the sensing dead angle based on the radar installation position and real-time angle, and lacks a strategy for main and auxiliary radar collaborative compensation, resulting in that the blocked area or radar beam coverage blind area cannot be effectively monitored.

[0044] 3. The intelligence of the task allocation strategy is missing: the radar task instructions are not dynamically generated in combination with obstacle distribution data and priority rules, making it difficult to balance detection accuracy and efficiency under different working conditions, such as not increasing the scanning frequency in high-risk areas and having redundant detection in low-risk areas.

[0045] The prior art does not propose a dual-radar collaborative scheduling method based on operation stage information, target area priority, and sensing dead angle compensation, especially lacking a technical solution for achieving sensing range coverage optimization and redundant monitoring of high-risk areas through a dynamic task allocation strategy.

[0046] To solve the above problems, please refer to Figure 1 , the embodiment of the present application provides a dual-radar collaborative scheduling adaptive sensing control method applied to a controller. It should be noted that the information involved in the method provided by the present application is extracted under the authorization of the relevant users and in accordance with the relevant regulations, and does not infringe on the privacy of the users.

[0047] The dual-radar collaborative scheduling adaptive sensing control method provided includes steps S101 to S104. The details are as follows:

[0048] Step S101. Install the first rotating radar and the second rotating radar on the rotatable support respectively to build a dual-radar module sensing system, and the dual-radar module is in communication connection with the sensing task scheduling unit and the operation state sensing module.

[0049] Specifically, two groups of rotatable radar sensors are installed at key positions of the tower (such as below the trolley), and the radar angle is dynamically adjusted through a rotatable support (such as an electric pan-tilt), to build a dual-radar collaborative sensing hardware system with angle adjustment capability (as shown in Figure 2 Two groups of radars are defined as the first rotating radar and the second rotating radar (non-fixed main and auxiliary, dynamically switchable), and are in bidirectional communication with the sensing task scheduling unit (such as an industrial-grade PLC or an edge computing controller) and the operation state sensing module (integrating multiple types of sensors) to realize data interaction and control instruction transmission.

[0050] The radar selection adopts a 120° field of view laser radar (such as TF-ALS-LIDAR-01 radar, single radar detection distance ≥ 150 meters, precision ≤ 3 cm, support angle resolution ≤ 0.05°), which meets the obstacle detection requirements in complex environments, and the two radars are synchronously scanned at a frequency of 10 Hz. The first rotating radar is installed on the top of the tower crane cab, and the second rotating radar is installed below the front end of the lifting arm, forming a three-dimensional monitoring layout with high and low staggered, reducing the self-structure shielding (such as the blocking of the radar beam by the lifting arm).

[0051] The radar and the dispatching unit are connected through Ethernet (TCP / IP) to transmit point cloud data (including distance, angle, and speed information) in real time; the operation state sensing module transmits operation stage signals (such as hoisting / rotation / landing point state on-off signals) through CAN bus or RS485 interface.

[0052] Step S102. Real-time acquisition of obstacle distribution data of the tower crane hoisting operation area through the first rotating radar and the second rotating radar, real-time acquisition of real-time stage information of the hoisting operation through the operation state sensing module, the real-time stage information including hoisting stage, rotation stage and landing point stage.

[0053] Specifically, through the real-time scanning of the double radars in the operation area (radius 50-100 meters), point cloud data containing obstacle position, size and motion state are generated; at the same time, through the operation state sensing module, the tower crane operation parameters are collected, and the current hoisting operation stage (hoisting, rotation, landing point) is identified, providing input for dynamic task scheduling.

[0054] Obstacle data acquisition includes double radars synchronously scanning at a frequency of 10 Hz, the first rotating radar covering the front 180° area of the tower crane (lifting arm rotation range) by default, and the second rotating radar covering the rear and both sides blind area by default. The original point cloud data is filtered by Kalman filter to remove noise, and a dynamic obstacle list (including coordinates X, Y, Z, speed V, and category label) is generated.

[0055] In the operation stage identification, the operation state sensing module integrates multiple sensors: in the hoisting stage, the hook height sensor (laser range finder) is used to detect whether the hook is off the ground (height change rate > 5 cm / s and load sensor value > 10% of rated load); in the rotation stage, the rotary mechanism encoder (accuracy 0.5°) is used to detect the rotation angle change rate of the lifting arm > 1° / s; in the landing stage, the hook height sensor is used to detect that the hook approaches the target height (error ± 50 cm) and the load sensor value decreases (release heavy objects). Multi-sensor data fusion: finite state machine (FSM) algorithm is used, combined with time window (such as 5 seconds) to confirm stage conversion, avoiding misjudgment.

[0056] Step S103. According to the real-time stage information, in combination with a preset target area priority rule, a target monitoring area corresponding to the current operation stage and a corresponding priority level are determined, a sensing dead angle position is calculated according to the installation positions and real-time rotation angles of the first rotating radar and the second rotating radar, and first task instructions and second task instructions are dynamically generated according to the priority level of the target monitoring area, the sensing dead angle position and obstacle distribution data, and in accordance with a preset task allocation strategy.

[0057] Specifically, in combination with a real-time operation stage, a target monitoring area and its priority (high / medium / low) are determined based on a preset priority rule (such as a 30° sector area below the hook during the hoisting stage); a sensing dead angle (such as a two-radar beam overlap blind area and a tower structure shielding area) at the current radar angle is calculated through geometric modeling; and finally, control instructions such as the angle, scanning frequency and detection accuracy of the first and second rotating radars are generated through a task allocation algorithm according to the priority, dead angle position and obstacle distribution.

[0058] The target area priority division includes: during the hoisting stage, the high-priority area is a 20-meter-diameter circular area directly below the hook (where people are likely to gather), and a 10-meter range below the jib (for collision prevention); the medium-priority area is a 30-meter range around the tower; during the rotating stage, the high-priority area is a 50-meter sector area in front of the jib (in front of the rotating path) and a 15° blind area on both sides; and during the landing stage, the high-priority area is a 10-meter rectangular area around the target landing point (for preventing accidental contact with obstacles). The priority rules are stored in the database of the scheduling unit and can be edited by the user.

[0059] The sensing dead angle calculation establishes a three-dimensional model of the tower crane (including shielding components such as the jib, balance arm and tower body), calculates the area shielded by the tower structure based on the current radar angle (θ1, φ1) and beam range (120° horizontally and 15° vertically) through a ray tracing algorithm, and marks the non-overlapping area of the detection range of the two radars (i.e., the single-radar blind area) as a "compensation area".

[0060] The task allocation strategy includes: first rotating radar task: preferentially scanning the high-priority area using the "key area intensive scanning" strategy (such as increasing the scanning frequency to 100 Hz and the angle resolution to 0.5° in the high-priority area); second rotating radar task: covering the first rotating radar blind area and sensing dead angle using the "dynamic blind compensation scanning" strategy (such as reciprocating scanning of the shielding area at a scanning frequency of 50 Hz); based on the priority queue (Priority Queue) and the ant colony optimization algorithm, radar resources are dynamically allocated to ensure the balance between detection accuracy and efficiency in high-risk areas.

[0061] Step S104. According to the first task instruction and the second task instruction, the first rotating radar and the second rotating radar are controlled to adjust the rotation angle and the sensing parameter, so that the first rotating radar focuses on sensing the target monitoring area with high priority, and the second rotating radar compensates for sensing the area corresponding to the blind area and the dead angle position of the sensing of the first rotating radar, to realize coverage optimization of the sensing range and redundant monitoring of the high-risk area.

[0062] Specifically, according to the task instruction, the first rotating radar adjusts the rotation angle to the target monitoring area (such as aligning below the lifting hook in the lifting stage), and improves the sensing parameter (such as increasing the transmission power and reducing the beam width); the second rotating radar synchronously adjusts the angle, compensates for scanning the blind area (such as the rear of the lifting arm) and the calculated dead angle area not covered by the first rotating radar, to form a cooperative mechanism of “first rotating radar focus monitoring + second rotating radar blind compensation redundancy”, to ensure that the high-risk area is double-covered (the overlap of the main second rotating radar detection range is ≥30%).

[0063] The actuator control includes: the sensing task scheduling unit sends angle instructions (such as the horizontal angle θ of the first rotating radar = 45°, the pitch angle φ = 10°) to the radar holder, and realizes accurate positioning of the angle (error ≤0.5°) through a PID control algorithm; dynamically adjusting the radar sensing parameter: enabling “high-precision mode” in high-priority areas (increasing the pulse repetition frequency by 20%, and improving the signal-to-noise ratio by 10dB), and enabling “energy-saving mode” in low-priority areas (reducing the scanning frequency to 30Hz).

[0064] The cooperative compensation mechanism automatically covers the blind area behind the first rotating radar when the first rotating radar monitors the high-priority area (such as the first rotating radar monitors 180° in front, and the second rotating radar monitors 180° in the rear, with an overlap area of 30°); if a sensing dead angle is detected (such as a 20° blind area caused by tower obstruction), the second rotating radar immediately starts “dead angle compensation scanning”, and performs ±15° reciprocating scanning centered on the area, until the dead angle is eliminated (such as the rotation of the tower crane changes the obstruction angle).

[0065] The redundant monitoring verification includes: the double radars cross-check the detection data of the high-risk area (such as the first rotating radar detects an obstacle A, and the second rotating radar synchronously confirms that the position error is ≤10cm), to ensure data reliability; when any radar fails, the other radar automatically switches to a full-range scanning mode (covering a 180° field of view), and enters an emergency backup state.

[0066] In some embodiments, the real-time acquisition of the obstacle distribution data of the tower crane hoisting operation area by the first and second rotating radars comprises: time synchronization and spatial coordinate calibration of the original point cloud data collected by the first and second rotating radars respectively, filtering of noise points based on a preset adaptive filtering algorithm; inputting the filtered point cloud data into a preset target classification neural network model to identify the obstacle types and mark static obstacles and dynamic obstacles; generating structured distribution data containing obstacle position, type and risk level according to the position coordinates, size information and motion state of the obstacles.

[0067] By performing time and space calibration, noise filtering and target classification on the original point cloud data of the dual radars, structured data containing position, type and risk level is generated to provide accurate input for subsequent task scheduling.

[0068] The time and space synchronization calibration comprises: time synchronization: PTP (Precision Time Protocol) is used to calibrate the sampling clock of the dual radars at the nanosecond level, ensuring that the timestamp error of the point cloud data is less than 1 μs; spatial calibration: a conversion matrix (rotation matrix R and translation vector T) of the radar coordinate system and the global coordinate system of the tower crane is established by the Hand-Eye Calibration method, and the calibration accuracy is ≤2 cm.

[0069] The adaptive filtering uses adaptive Kalman filtering to dynamically adjust the process noise covariance matrix Q and the measurement noise covariance matrix R according to the environmental noise, filter out non-target points (such as raindrops and birds) with a signal-to-noise ratio > 15 dB, and use statistical filtering for static clutter (such as fixed buildings) to determine and remove points that have not moved for 3 consecutive frames as static background.

[0070] Target classification and structuring: neural network model: a lightweight 3D point cloud classification model (such as PointNet-Lite) is used, which inputs local point cloud with ≥100 points per frame and outputs obstacle types (personnel / vehicle / scaffold / hoist, classification accuracy ≥95%); risk level labeling: risk level labels are generated according to the distance between the obstacle and the tower crane (<10 meters for high risk, 10-30 meters for medium risk, >30 meters for low risk) and the motion speed (>5 m / s for dynamic high risk); structured data format: JSON format data is generated, containing {obstacle ID, coordinates (X, Y, Z), type, risk level, size (length x width x height), velocity vector (Vx, Vy, Vz)}.

[0071] In some embodiments, the real-time stage information of the hoisting operation is acquired in real time by the operation state sensing module, including: collecting the lifting distance encoder data of the tower crane, the rotation mechanism angle encoder data and the hook height sensor data, and inputting the multi-dimensional sensor signals into a preset state recognition finite state machine model; the action sequence of the hoisting operation is analyzed in time sequence by the state machine model, and the current operation stage is judged in combination with the motion threshold of each actuator; when the lifting motor speed suddenly changes from zero and the hook height changes, it is judged that the lifting stage is in the lifting stage; when the rotation mechanism angle changes at a rate exceeding a preset threshold, it is judged that the rotation stage is in the rotation stage; when the hook height tends to be stable and approaches the target height, it is judged that the landing point stage is in the landing point stage.

[0072] Through multi-sensor fusion and finite state machine (FSM) model, the lifting, rotation and landing point stages of the hoisting operation are recognized in real time, and the problem of misjudgment of traditional single sensor is solved.

[0073] The sensor configuration includes: lifting motor speed: install magneto type speed sensor (accuracy ±0.5%), real-time acquisition of motor encoder pulse signal (resolution 1024 pulses / revolution); rotation mechanism angle includes: use absolute value encoder (accuracy ±0.1°), record real-time azimuth angle of jib; hook height: use laser range finder (range 200 meters, accuracy ±1cm), measure the vertical distance from the hook to the tower crane bottom reference surface.

[0074] The state machine model construction includes: state transition conditions: lifting stage: speed > 0 rpm and lasts for 2 seconds, and hook height change rate > 2 cm / s (anti-misjudgment idle); rotation stage: angle change rate > 0.5° / s and lasts for 3 seconds (exclude fine adjustment action); landing point stage: height change rate < 1 cm / s and distance error from target height (preset by operation planning system) < 50 cm. Time sequence analysis: use sliding time window (5 seconds) to verify state persistence, avoid transient signal interference (such as sudden speed change caused by short-term current fluctuation).

[0075] In some embodiments, according to the real-time stage information, the target monitoring area corresponding to the current operation stage and the corresponding priority level are determined in combination with the preset target area priority rule, including: a multi-area division model including lifting area, rotation path area and landing buffer area is preset, and a dynamic priority mapping table is established for different operation stages; when in the lifting stage, the area within a radius of 5 meters below the hook is set as a first priority monitoring area, and the area directly below the jib is set as a second priority; when in the rotation stage, the area in front of the jib in the current rotation direction within 180 degrees is set as a first priority, and the areas on both sides within 60 degrees are set as a second priority; the priority optimization neural network trained by historical accident data is used to correct the preset priority rule online, and the priority coefficients of each area are dynamically adjusted according to the real-time obstacle distribution.

[0076] By establishing a multi-region priority model, the priority of the monitoring region is dynamically optimized in combination with historical accident data to solve the problem that the traditional fixed priority cannot adapt to complex working conditions.

[0077] The preset region division and priority mapping includes: a multi-region model: a lifting hook as the center, a 5-meter radius cylinder (first priority, key monitoring personnel intrusion); a rotating path region: a crane boom as the axis, a 180° fan-shaped region in front with a 50-meter radius (first priority, obstacle collision prevention); a drop point buffer region: a target drop point as the center, a 10m x 10m rectangular region (first priority, prevent lifting objects from touching surrounding structures). A dynamic mapping table: stores the priority of each stage region (for example, the 60° fan-shaped region on both sides in the rotating stage is the second priority, focusing on monitoring adjacent tower machines).

[0078] The priority online correction includes: training data: collecting obstacle distribution data in historical collision accidents (such as 80% of accidents occurring within 3 meters below the hook), constructing a priority optimization neural network (input: obstacle type + distance + historical accident weight, output: priority coefficient adjustment value); real-time adjustment: when a high-risk obstacle (such as a moving person) is detected in a certain region for 3 consecutive times, the priority of the region is automatically increased by 1 level (up to 3 levels), and maintained for 10 minutes.

[0079] In some embodiments, the calculation of the dead angle position according to the installation position and real-time rotation angle of the first and second rotating radars includes: establishing a three-dimensional coordinate system with the tower machine rotation center as the origin, obtaining the installation coordinates and orientation angle parameters of the first and second rotating radars; based on the detection distance threshold and beam angle range of the radar sensor, a fan-shaped detection region mathematical model is constructed; through a coordinate transformation algorithm, the detection range corresponding to the current rotation angle of the two radars is projected to the working region plane; the union and complement of the two radar detection regions are calculated, the uncovered blind region is identified, and the obscured detection blind region is further screened in combination with an obstacle shielding probability model to determine the actual dead angle position coordinates.

[0080] By using a three-dimensional coordinate system and geometric modeling, the double-radar detection blind area and the tower machine structure shielding area are accurately calculated to solve the coverage gap problem of traditional fixed splicing radars. The coordinate system and the detection model include:

[0081] Global coordinate system: the origin O is the tower machine rotation center, the X axis points to the initial direction of the crane boom, the Y axis is horizontal and perpendicular to the X axis, and the Z axis is vertically upward; radar detection model: the detection range of a single radar is a fan-shaped region, and the mathematical expression is:

[0082] ;

[0083] Wherein, Rmax=100 meters, horizontal beam angle a=120°, vertical beam angle b=15°, The current horizontal / tilt angle. The dead angle identification algorithm includes: projection transformation: projecting the three-dimensional detection area to the operation plane (Z=0) to generate a two-dimensional fan-shaped coverage area; blind area calculation: calculating the union complement (i.e. the uncovered area) of the double-radar coverage area through polygon Boolean operation, combining the tower crane structure CAD model (such as the shielding area of the jib), and using the ray method to determine whether the area is shielded; secondary screening: marking the blind area with a shielding probability > 70% as an "actual perception dead angle" and outputting a coordinate list (such as the area 20-30 meters behind the tower crane, which is a blind area caused by the balance arm shielding).

[0084] In some embodiments, the first task instruction and the second task instruction are dynamically generated according to the priority level of the target monitoring area, the perception dead angle position, and the obstacle distribution data, and in accordance with a preset task allocation strategy, including: constructing a radar task parameter space containing detection accuracy, scanning frequency, and angle adjustment step, and establishing a target function with priority level as the weight; using an improved particle swarm optimization algorithm, with covering the first priority level area, compensating for the perception dead angle, and balancing the radar load as the constraint condition, the task parameters of the two groups of radars are optimized and solved; when the priority difference of the target monitoring area exceeds the preset threshold, the first rotating radar is allocated to perform a high-frequency fine scanning task for the high priority area, and the second rotating radar is allocated to perform a reciprocating blind compensation scanning task for the dead angle area; when the priority difference is low, a cooperative scanning strategy is generated through a task scheduling neural network, so that the two radars perform area coverage according to a preset phase difference.

[0085] By constructing a radar task parameter optimization model, dynamic task allocation is realized by combining particle swarm algorithm and neural network, and detection accuracy and efficiency are balanced.

[0086] The task parameter space and the target function include: parameter space: detection accuracy (distance resolution Ad, angle resolution Aq), scanning frequency f (30-100 Hz), angle adjustment step As (0.1°-5°); the target function includes:

[0087] ;

[0088] Wherein, wi is the area priority weight, Pi is the area coverage, and l is the load balancing coefficient to avoid high energy consumption caused by high-frequency scanning of the double radars.

[0089] The optimization algorithm implementation includes: improved particle swarm algorithm: introduce inertia weight dynamic adjustment (weight 0.8 in lifting stage, improve convergence speed; weight 0.5 in landing point stage, enhance local search), constraint conditions include: first priority area coverage ≥ 95%, radar load difference ≤ 20%; task scheduling neural network: input current priority difference, obstacle density, dead angle number, output main second rotating radar phase difference (such as phase difference 60° to realize alternate scanning, reduce data redundancy).

[0090] In some embodiments, the control of the first rotating radar and the second rotating radar to adjust the rotation angle and the sensing parameter according to the first task instruction and the second task instruction includes: establishing a kinematic model of a radar servo system, converting the target angle in the task instruction into a motor control pulse sequence; adopting a PID control algorithm with feedforward compensation to perform closed-loop control on the radar rotating shaft, and correcting the angle deviation in real time; according to the sensing parameter in the task instruction, dynamically configuring the transmission and reception module parameters through the radar hardware drive interface; in the adjustment process, the angular velocity and acceleration of the rotating mechanism are monitored in real time through the radar self-checking sensor, and the safety speed limiting mechanism is triggered when abnormal vibration is detected; wherein the sensing parameter includes detection distance, resolution and pulse repetition frequency.

[0091] The kinematic model and the PID control are used to realize accurate adjustment of the radar angle, dynamically configure the sensing parameter, and simultaneously integrate the safety monitoring mechanism to prevent mechanical failure.

[0092] The servo control includes: kinematic model: establish the angle conversion relationship (such as 1 pulse = 0.05°) between the motor pulse number and the angle of the two axes (horizontal H and pitch P) of the holder, and convert the target angle θtar into pulse number N = θtar / 0.05; PID control: adopt position loop PID + speed feedforward compensation, wherein the angle tracking error is ≤0.3°, and the response time is <1 second.

[0093] The sensing parameter configuration includes: hardware drive: configure the radar transmission module parameters through the SPI interface, such as setting the pulse repetition frequency (PRF) = 20kHz in the high priority area (improve the distance resolution to 3cm), and setting the PRF = 10kHz in the low priority area (reduce power consumption); safety mechanism: install a vibration sensor (accuracy ±0.1g), and trigger the speed limiting mechanism (angular velocity ≤5° / s) when the detected vibration acceleration >1.5g to prevent the holder from shaking violently.

[0094] In some embodiments, the method further comprises: establishing a historical operation data training set containing operation stages, obstacle distribution, radar scheduling strategy and collision risk event label; training a radar scheduling strategy network through a deep reinforcement learning algorithm, taking the current operation state, target area priority and perception dead angle information as state input, taking the main second rotating radar task parameter combination as action output, and taking the maximum risk warning accuracy and the minimum scheduling delay as the reward function; and inputting the deviation data of actual scheduling effect and preset strategy into the strategy network for online update every completed operation cycle, to realize adaptive optimization of the task allocation strategy.

[0095] By learning the optimal scheduling strategy from historical data through deep reinforcement learning (DRL), adaptive evolution of the task allocation strategy is realized, and the problem that preset rules cannot cover complex working conditions is solved.

[0096] The training framework includes: state space: S={operation stage, priority area list, dead angle coordinates, obstacle risk distribution} (discretization processing, such as stage encoding 0-2, and region priority normalization [0, 1]); action space: A={first rotating radar parameter combination, second rotating radar parameter combination} (12 parameter combinations in total, such as {angle 1, frequency 1, precision 1}); reward function: R=0.6 risk warning accuracy-0.3 scheduling delay-0.1; radar energy consumption, wherein the risk warning accuracy=correct warning number / total risk event number, and the scheduling delay=the time (seconds) from identification to execution.

[0097] The PPO (Proximal Policy Optimization) algorithm is adopted, the strategy network is updated once every 50 operation cycles (about 2 hours), the experience replay buffer capacity is 100,000, and the preset rule is used to generate actions in the initial stage, and the reinforcement learning strategy is switched to after 200 effective data are accumulated.

[0098] In some embodiments, the method further comprises: constructing a multi-dimensional risk assessment model containing obstacle motion trajectory prediction, priority area risk level and radar perception blind area; using a long short-term memory network to model the position sequence of dynamic obstacles and predict the motion trajectory in the next 5 seconds; calculating the spatial overlap between the predicted trajectory and the target monitoring area, and triggering an emergency scheduling mechanism when the predicted trajectory enters a priority area and the current radar perception frequency is lower than the safety threshold: forcing the first rotating radar to switch to the full-time tracking scanning mode of the target area, and the second rotating radar to adjust to the front warning scanning mode in the predicted trajectory direction.

[0099] The trajectory of the dynamic obstacle is predicted through the LSTM, the emergency scheduling is triggered in combination with the risk assessment model, and the response capability to sudden risks is improved.

[0100] The risk assessment model includes: LSTM trajectory prediction: input the obstacle coordinate sequence of the past 10 time steps (200 ms), output the position prediction of the future 5 seconds (25 steps) (root mean square error ≤ 5 cm), model structure: 2 layers of LSTM (128 units each) + full connection layer; spatial overlap calculation: polygon overlap judgment is performed on the predicted trajectory and the first priority area, when the overlap area > 30% of the area and the current scanning frequency of the radar < 80 Hz, the emergency scheduling is triggered.

[0101] The emergency scheduling mechanism includes: the first rotating radar: forced switching to the target area full-time tracking scanning (frequency 100 Hz, angle resolution 0.5°), suspending other low-priority tasks; the second rotating radar: turning to the 5-meter position in front of the predicted trajectory, performing front warning scanning (scanning range ± 30°, frequency 60 Hz), forming a “tracking + warning” dual protection.

[0102] In some embodiments, by abstracting the two groups of radars as agents, a distributed multi-agent reinforcement learning (MARL) system is constructed, so that the radars can autonomously optimize the detection strategy in dynamic cooperation, solve the communication delay and load imbalance problem of traditional centralized scheduling, and realize real-time collaborative efficiency improvement.

[0103] The agent modeling includes: state space: each agent state includes its own detection parameters (angle, frequency), global obstacle distribution (compressed into a grid heat map), and opponent load state (CPU occupancy, scanning queue length); action space: includes angle adjustment (± 5° step), frequency switching (3 gears: low / medium / high), precision configuration (2 gears: ordinary / fine), a total of 12 combined actions.

[0104] The reward function design includes: joint reward Rjoint=0.5Rcoverage+0.3Rbalance−0.2RCollision; wherein, Rcoverage is the first area coverage degree (reward +1 for each 1% improvement), Rbalance is the load balancing degree (penalty -5 when the frequency difference between the two radars > 20 Hz), and Rcollision is the penalty -10 when the detection area overlap rate > 30%. The QMIX algorithm (Quantile Mixing) is used to process the non-stationary cooperation environment, the central coordination network combines the local Q values of each agent, and the experience replay buffer supports cross-agent data sharing; rule-guided exploration (ε=0.3) is used in the early training, and after 3000 episodes, the stable strategy is converged, realizing a 60% reduction in scanning blind area when the two radars cooperate.

[0105] Please refer to Figure 3 as shown, Figure 3FIG. 1 is a structural schematic diagram of a dual-radar collaborative scheduling adaptive perception control system 200 provided by an embodiment of the present application. The dual-radar collaborative scheduling adaptive perception control system 200 is used to execute the steps of the dual-radar collaborative scheduling adaptive perception control method shown in each of the embodiments described above. The dual-radar collaborative scheduling adaptive perception control system 200 can be a single server or a server cluster, or the dual-radar collaborative scheduling adaptive perception control system 200 can be a terminal, which can be a handheld terminal, a notebook computer, a wearable device, or a robot, etc.

[0106] As shown in FIG. 1, the dual-radar collaborative scheduling adaptive perception control system 200 includes: Figure 3

[0107] a system construction unit 201, configured to install a first rotating radar and a second rotating radar on a rotatable support respectively, to construct a dual-radar module perception system, and to communicate the dual-radar module with a perception task scheduling unit and a working state perception module respectively;

[0108] a data acquisition unit 202, configured to acquire obstacle distribution data of a tower crane hoisting operation area in real time through the first rotating radar and the second rotating radar, and to acquire real-time stage information of the hoisting operation in real time through the working state perception module, the real-time stage information including a lifting stage, a rotating stage, and a landing point stage;

[0109] an instruction generation unit 203, configured to determine a target monitoring area corresponding to a current operation stage and a corresponding priority level according to the real-time stage information and in combination with a preset target area priority rule, to calculate a perception dead angle position according to installation positions and real-time rotating angles of the first rotating radar and the second rotating radar, and to dynamically generate a first task instruction and a second task instruction according to a priority level of the target monitoring area, the perception dead angle position, and the obstacle distribution data, and in accordance with a preset task allocation strategy;

[0110] a perception monitoring unit 204, configured to control the first rotating radar and the second rotating radar to adjust rotating angles and perception parameters according to the first task instruction and the second task instruction, so that the first rotating radar performs intensive perception on a target monitoring area with a high priority, and the second rotating radar performs compensatory perception on an area corresponding to a perception blind area of the first rotating radar and a perception dead angle position, to realize coverage optimization of a perception range and redundant monitoring of a high-risk area.

[0111] It should be noted that, for the convenience and brevity of description, the specific working processes of the dual-radar collaborative scheduling adaptive perception control system and each module described above can be clearly understood by referring to the corresponding content in each embodiment of the dual-radar collaborative scheduling adaptive perception control method described above, which will not be described herein again. ​

[0112] The above-mentioned dual-radar cooperative scheduling adaptive perception control method can be implemented in the form of a computer program, which can run on the device as shown in the figure. Figure 3 The device shown in the figure.

[0113] Please refer to Figure 4 , Figure 4 is a structural schematic block diagram of the controller provided by the embodiment of the application. The controller includes a processor, a memory and a network interface connected through a device bus, wherein the memory can include a storage medium and an internal memory.

[0114] The storage medium can store an operating device and a computer program. The computer program includes program instructions, which, when executed, can cause the processor to execute any kind of dual-radar cooperative scheduling adaptive perception control method.

[0115] The processor is used to provide computing and control capabilities to support the operation of the entire controller.

[0116] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium, which, when executed by the processor, can cause the processor to execute any kind of dual-radar cooperative scheduling adaptive perception control method.

[0117] The network interface is used for network communication, such as sending assigned tasks. Those skilled in the art can understand that Figure 4 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the application, and does not constitute a limitation on the terminal to which the scheme of the application is applied. The specific controller can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0118] It should be understood that the processor can be a central processing unit (CPU), and the processor can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0119] Among them, in one embodiment, the processor is used to run the computer program stored in the memory to implement the following steps:

[0120] The first rotating radar and the second rotating radar are respectively installed on a rotatable support to construct a dual-radar module perception system, and the dual-radar module is in communication connection with a perception task scheduling unit and a working state perception module;

[0121] Obstacle distribution data of a tower crane hoisting operation area is acquired in real time by the first rotating radar and the second rotating radar, and real-time stage information of hoisting operation is acquired in real time by the working state perception module, and the real-time stage information includes a lifting stage, a rotating stage and a landing point stage;

[0122] According to the real-time stage information, a target monitoring area corresponding to a current operation stage and a corresponding priority level are determined in combination with a preset target area priority rule, a perception dead angle position is calculated according to the installation position and the real-time rotating angle of the first rotating radar and the second rotating radar, and first task instructions and second task instructions are dynamically generated according to the priority level of the target monitoring area, the perception dead angle position and the obstacle distribution data, and according to a preset task allocation strategy;

[0123] According to the first task instructions and the second task instructions, the rotating angle and the perception parameters of the first rotating radar and the second rotating radar are controlled to adjust, so that the first rotating radar performs intensive perception on a target monitoring area with high priority, and the second rotating radar performs compensatory perception on a region corresponding to a perception blind area of the first rotating radar and a perception dead angle position, so as to realize coverage optimization of a perception range and redundant monitoring of a high-risk region.

[0124] In some embodiments, the obstacle distribution data of the tower crane hoisting operation area acquired in real time by the first rotating radar and the second rotating radar includes: time synchronization and spatial coordinate calibration are respectively performed on original point cloud data collected by the first rotating radar and the second rotating radar, and noise points are filtered based on a preset adaptive filtering algorithm; the filtered point cloud data is input into a preset target classification neural network model to identify obstacle types and mark static obstacles and dynamic obstacles; and structured distribution data containing obstacle position, type and risk level are generated according to the position coordinates, size information and motion state of the obstacles.

[0125] In some embodiments, the real-time stage information of the hoisting operation is acquired in real time by the operation state perception module, including: collecting the data of the lifting distance encoder, the rotation mechanism angle encoder and the hook height sensor of the tower crane, and inputting the multi-dimensional sensor signals into a preset state recognition finite state machine model; the action sequence of the hoisting operation is analyzed in time sequence by the state machine model, and the current operation stage is judged in combination with the motion threshold of each actuator; when the lifting motor speed suddenly changes from zero and the hook height changes, it is determined that the lifting stage is reached; when the rotation mechanism angle changes at a rate exceeding a preset threshold, it is determined that the rotation stage is reached; when the hook height tends to be stable and approaches the target height, it is determined that the landing point stage is reached.

[0126] In some embodiments, the target monitoring area corresponding to the current operation stage and the corresponding priority level are determined according to the real-time stage information and in combination with a preset target area priority rule, including: a multi-area division model including a lifting area, a rotation path area and a landing buffer area is preset, and a dynamic priority mapping table is established for different operation stages; when in the lifting stage, a radius of 5 meters below the hook is set as a first priority monitoring area, and the area directly below the jib is set as a second priority; when in the rotation stage, a 180-degree sector area in front of the current rotation direction of the jib is set as a first priority, and two 60-degree sector areas on the sides are set as a second priority; a priority optimization neural network trained by historical accident data is used to correct the preset priority rule online, and the priority coefficients corresponding to each area are dynamically adjusted according to the real-time obstacle distribution.

[0127] In some embodiments, the dead angle position is calculated according to the installation positions and real-time rotation angles of the first and second rotating radars, including: a three-dimensional coordinate system with the rotation center of the tower crane as the origin is established, and the installation coordinates and orientation angle parameters of the first and second rotating radars are obtained; a mathematical model of a fan-shaped detection area is constructed based on the detection distance threshold and beam angle range of the radar sensor; the detection ranges corresponding to the current rotation angles of the two groups of radars are projected onto the operation area plane by a coordinate transformation algorithm; the union and complement of the detection areas of the two radars are calculated, the blind area region that is not covered is identified, the detection blind area that is blocked is secondarily screened in combination with an obstacle blocking probability model, and the actual dead angle position coordinates are determined.

[0128] In some embodiments, the first task instruction and the second task instruction are dynamically generated according to the priority level of the target monitoring area, the dead angle position and the obstacle distribution data, and a preset task allocation strategy, including: constructing a radar task parameter space containing detection accuracy, scanning frequency and angle adjustment step, and establishing a target function with the priority level as the weight; using an improved particle swarm optimization algorithm, and taking covering the first priority area, compensating for the dead angle and balancing the radar load as the constraint condition, the task parameters of the two groups of radars are optimized and solved; when the priority difference of the target monitoring area exceeds a preset threshold, the first rotating radar is allocated to perform a high-frequency fine scanning task in a high priority area, and the second rotating radar is allocated to perform a reciprocating blind compensation scanning task in a dead angle area; when the priority difference is low, a cooperative scanning strategy is generated through a task scheduling neural network, so that the two radars perform area coverage according to a preset phase difference.

[0129] In some embodiments, the first task instruction and the second task instruction are dynamically generated according to the priority level of the target monitoring area, the dead angle position and the obstacle distribution data, and a preset task allocation strategy, including: constructing a radar task parameter space containing detection accuracy, scanning frequency and angle adjustment step, and establishing a target function with the priority level as the weight; using an improved particle swarm optimization algorithm, and taking covering the first priority area, compensating for the dead angle and balancing the radar load as the constraint condition, the task parameters of the two groups of radars are optimized and solved; when the priority difference of the target monitoring area exceeds a preset threshold, the first rotating radar is allocated to perform a high-frequency fine scanning task in a high priority area, and the second rotating radar is allocated to perform a reciprocating blind compensation scanning task in a dead angle area; when the priority difference is low, a cooperative scanning strategy is generated through a task scheduling neural network, so that the two radars perform area coverage according to a preset phase difference.

[0130] In some embodiments, the method further comprises: establishing a historical operation data training set containing operation stage, obstacle distribution, radar scheduling strategy and collision risk event label; training a radar scheduling strategy network through a deep reinforcement learning algorithm, taking the current operation state, the target area priority and the dead angle information as the state input, taking the main second rotating radar task parameter combination as the action output, and taking the maximum risk warning accuracy and the minimum scheduling delay as the reward function; every time an operation cycle is completed, the deviation data of the actual scheduling effect and the preset strategy are input into the strategy network for online update, so as to realize adaptive optimization of the task allocation strategy.

[0131] In some embodiments, the method further comprises: constructing a multi-dimensional risk assessment model comprising obstacle motion trajectory prediction, priority area risk level, radar perception blind area; using a long short-term memory network to model the position sequence of dynamic obstacles and predict the motion trajectory within the next 5 seconds; calculating the spatial overlap between the predicted trajectory and the target monitoring area, and triggering an emergency scheduling mechanism when the predicted trajectory enters a first priority area and the current radar perception frequency is lower than a safety threshold: forcing the first rotating radar to switch to a full-time tracking scanning mode in the target area, and synchronously adjusting the second rotating radar to a pre-warning scanning mode in the predicted trajectory direction.

[0132] The above merely provides a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present application, and these modifications or replacements shall be encompassed within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. An adaptive sensing control method for dual-radar cooperative scheduling, characterized in that, include: The first and second rotating radars are respectively mounted on a rotatable bracket to construct a dual-radar module sensing system. The dual radar modules are respectively connected to the sensing task scheduling unit and the operation status sensing module. The first and second rotating radars acquire obstacle distribution data in the tower crane hoisting operation area in real time, and the operation status perception module acquires real-time stage information of the hoisting operation, including the lifting stage, rotation stage and landing stage. Based on the real-time stage information and combined with the preset target area priority rules, the target monitoring area and corresponding priority level corresponding to the current operation stage are determined. Based on the installation position and real-time rotation angle of the first and second rotating radars, the blind spot position is calculated. Based on the priority level of the target monitoring area, the blind spot position, and obstacle distribution data, the first and second task instructions are dynamically generated according to the preset task allocation strategy, including: constructing a radar task parameter space containing detection accuracy, scanning frequency, and angle adjustment step size, and establishing a target function with priority level as the weight. An improved particle swarm optimization algorithm is used to optimize the task parameters of the two sets of radars under the constraints of covering the first-priority area, compensating for blind spots, and balancing radar load. When the priority difference of the target monitoring area exceeds a preset threshold, the first rotating radar is assigned to perform a high-frequency fine scanning task of the first-priority area, and the second rotating radar performs a reciprocating blind spot filling scanning task of the blind spot area. When the priority difference is low, a cooperative scanning strategy is generated through a task scheduling neural network to enable the two radars to cover the area according to a preset phase difference. According to the first and second task instructions, the first and second rotating radars are controlled to adjust their rotation angles and sensing parameters, so that the first rotating radar focuses on sensing high-priority target monitoring areas, and the second rotating radar compensates for sensing the areas corresponding to the sensing blind spots and dead angles of the first rotating radar, thereby achieving coverage optimization of the sensing range and redundant monitoring of high-risk areas.

2. The method according to claim 1, characterized in that, The real-time acquisition of obstacle distribution data in the tower crane hoisting operation area via the first and second rotating radars includes: The raw point cloud data collected by the first and second rotating radars are synchronized in time and calibrated in space coordinates, and noise points are filtered out based on a preset adaptive filtering algorithm. The filtered point cloud data is input into a preset target classification neural network model to identify obstacle types and label static and dynamic obstacles; Based on the location coordinates, size information, and motion status of obstacles, structured distribution data containing the location, type, and risk level of obstacles is generated.

3. The method according to claim 1, characterized in that, The real-time acquisition of hoisting operation stage information through the operation status sensing module includes: Collect data from the tower crane's hoisting sensors, slewing mechanism angle encoders, and hook height sensors, and input the multi-dimensional sensor signals into a preset state recognition finite state machine model; The motion sequence of hoisting operations is analyzed using a finite state machine model for state recognition, and the current operation stage is determined by combining the motion thresholds of each actuator. When a continuous change in hook height is detected, it is determined to be the lifting stage; when the rate of change of the slewing mechanism angle exceeds a preset threshold, it is determined to be the rotation stage; and when the hook height tends to stabilize and approaches the target height, it is determined to be the landing stage.

4. The method according to claim 1, characterized in that, Based on the real-time stage information and combined with preset target area priority rules, the target monitoring area and corresponding priority level corresponding to the current operation stage are determined, including: A multi-area partitioning model is pre-defined, including a lifting area, a rotation path area, and a landing point buffer zone, and a dynamic priority mapping table is established for different operation stages; During the lifting phase, the area within a 5-meter radius below the hook is designated as the first-priority monitoring zone, and the area directly below the boom is designated as the second-priority zone. When in the rotation phase, the 180-degree sector area in front of the current rotation direction of the crane boom is set as the first priority, and the 60-degree sector areas on both sides are set as the second priority. A priority optimization neural network trained using historical accident data is used to correct preset priority rules online and dynamically adjust the priority coefficients of each area based on the real-time obstacle distribution.

5. The method according to claim 1, characterized in that, Based on the installation positions and real-time rotation angles of the first and second rotating radars, the location of the blind spot is calculated, including: Establish a three-dimensional coordinate system with the tower crane's rotation center as the origin, and obtain the installation coordinates and orientation angle parameters of the first and second rotating radars; Based on the detection range threshold and beam angle range of the radar sensor, a mathematical model of the sector detection area is constructed; through a coordinate transformation algorithm, the detection range corresponding to the current rotation angle of the two radars is projected onto the working area plane. The union and complement of the detection areas of the two radars are calculated to identify the uncovered blind areas. Combined with the obstacle occlusion probability model, the obscured detection blind areas are further filtered to determine the actual sensing blind spot coordinates.

6. The method according to claim 1, characterized in that, The step of controlling the first rotating radar and the second rotating radar to adjust their rotation angles and sensing parameters according to the first task instruction and the second task instruction includes: A kinematic model of the radar servo system is established, and the target angle in the mission command is converted into a motor control pulse sequence. A PID control algorithm with feedforward compensation is used to perform closed-loop control of the radar rotation axis and correct the angle deviation in real time. According to the sensing parameters in the mission command, the parameters of the transmit and receive modules are dynamically configured through the radar hardware driver interface. During the adjustment process, the angular velocity and acceleration of the rotating mechanism are monitored in real time by radar self-test sensors. When abnormal vibration is detected, a safety speed limiting mechanism is triggered. The sensing parameters include detection distance, resolution, and pulse repetition frequency.

7. The method according to claim 1, characterized in that, The method further includes: Establish a training set of historical operation data, including operation phases, obstacle distribution, radar scheduling strategies, and collision risk event labels; The radar scheduling strategy network is trained by deep reinforcement learning algorithm, with the current operation status, target area priority, and blind spot information as state inputs, the combination of second rotating radar mission parameters as action outputs, and the reward function being to maximize the risk warning accuracy and minimize the scheduling delay. After each job cycle is completed, the deviation data between the actual scheduling effect and the preset strategy is input into the strategy network for online updates, thereby achieving adaptive optimization of the task allocation strategy.

8. The method according to claim 1, characterized in that, The method further includes: Construct a multi-dimensional risk assessment model that includes obstacle trajectory prediction, priority area risk level, and radar perception blind spot; A long short-term memory network is used to model the position sequence of dynamic obstacles and predict their trajectory within the next 5 seconds. The predicted trajectory is spatially overlapped with the target monitoring area. When the predicted trajectory enters the first-priority area and the radar's current sensing frequency is below the safety threshold, an emergency dispatch mechanism is triggered: the first rotating radar is forced to switch to the full-time tracking and scanning mode of the target area, and the second rotating radar is simultaneously adjusted to the forward warning scanning mode of the trajectory prediction direction.

9. An adaptive sensing control system with dual radar cooperative scheduling, characterized in that, include: The system construction unit is used to install the first rotating radar and the second rotating radar on a rotatable bracket to build a dual radar module sensing system. The dual radar modules are respectively connected to the sensing task scheduling unit and the operation status sensing module. The data acquisition unit is used to acquire obstacle distribution data in the tower crane hoisting operation area in real time through the first rotating radar and the second rotating radar, and to acquire real-time stage information of the hoisting operation through the operation status perception module. The real-time stage information includes the hoisting stage, the rotation stage and the landing stage. The instruction generation unit is used to determine the target monitoring area and its corresponding priority level for the current operation stage based on the real-time stage information and a preset target area priority rule; calculate the blind spot location based on the installation position and real-time rotation angle of the first and second rotating radars; and dynamically generate the first and second task instructions according to the priority level of the target monitoring area, the blind spot location, and obstacle distribution data, in accordance with a preset task allocation strategy. This includes: constructing a radar task parameter space that includes detection accuracy, scanning frequency, and angle adjustment step size; and establishing a target function with priority level as the weight. An improved particle swarm optimization algorithm is used to optimize the task parameters of the two sets of radars under the constraints of covering the first-priority area, compensating for blind spots, and balancing radar load. When the priority difference of the target monitoring area exceeds a preset threshold, the first rotating radar is assigned to perform a high-frequency fine scanning task of the first-priority area, and the second rotating radar performs a reciprocating blind spot filling scanning task of the blind spot area. When the priority difference is low, a cooperative scanning strategy is generated through a task scheduling neural network to enable the two radars to cover the area according to a preset phase difference. The perception and monitoring unit is used to control the first rotating radar and the second rotating radar to adjust their rotation angles and perception parameters according to the first task instruction and the second task instruction. This enables the first rotating radar to focus on sensing high-priority target monitoring areas, and the second rotating radar to compensate for the sensing blind spots and dead angles of the first rotating radar, thereby achieving coverage optimization of the sensing range and redundant monitoring of high-risk areas.

Images