Shovel loading robot scanning system for all-weather large-scene complex working condition working environment
Through the coordinated layout of multiple lidar, vibration monitoring and compensation fusion algorithm and vibration isolation technology, the environmental perception and vibration interference problems of shovel-mounted robots under complex working conditions are solved, and high-precision and rapid operation environment perception and stability are achieved.
Patent Information
- Application Number
- CN202510186806.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-20
- Publication Date
- 2025-07-04
AI Technical Summary
The existing shovel-mounted robot scanning system is difficult to achieve accurate perception under complex working conditions around the clock, insufficient scanning range, slow data processing speed, and is easily affected by equipment vibration and environmental occlusion, resulting in operational errors and safety hazards.
The coordinated layout of multi-lidar, vibration monitoring and compensation fusion algorithm, quasi-zero stiffness and frequency-variable damping composite vibration isolation technology and deep learning drive control optimization are adopted, and the data fusion processing and vibration isolation device are combined to achieve high-precision environmental perception and vibration suppression.
It realizes 360° blind spot cloud coverage, submillimeter-level real-time vibration compensation, and stable operations under complex working conditions, improving the environmental perception accuracy and operation stability of the shovel-mounted robot, and increasing the dynamic response speed by 3.2 times.
Smart Images
Figure CN120254892A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of loading robots, and particularly to a scanning system for a loading robot applicable to an all-weather large-scene complex working condition operation environment. Background Art
[0002] In fields such as mine exploitation and material transportation, loading operations often face complex and harsh working condition environments. Traditional loading operation methods rely on manual operation, which is not only inefficient, but also in harsh weather (such as heavy rain, dust storms, thick fog, etc.) and large-scene operation ranges, the operator's vision is limited, making it difficult to accurately complete the loading task, and prone to operation errors, resource waste, and safety accidents. With the development of robot technology, loading robots have emerged, but the existing scanning systems are difficult to meet the requirements of accurately perceiving the operation environment and efficiently operating under all-weather large-scene complex working conditions. Under low-light or strong-light irradiation conditions, the scanning accuracy will drop significantly; in an environment with a large amount of dust, smoke and other obstacles, the reliability and accuracy of scanning are difficult to guarantee; and for large-area, irregular terrains and multi-obstacle scenes, there are also deficiencies in the scanning range and data processing speed; at the same time, due to the complex mechanical structure and excavation process of the loading robot, it is easily affected by factors such as equipment component occlusion, mechanical vibration, and external environment during the operation process. The existing scanning systems for loading robots cannot meet the working requirements. Summary of the Invention
[0003] In order to solve the above problems, the present invention provides a scanning system for a loading robot in an all-weather large-scene complex working condition operation environment.
[0004] The present invention adopts the following technical solutions: A scanning system for a loading robot in an all-weather large-scene complex working condition operation environment, comprising: A data acquisition system, which is used to acquire the position information and point cloud data of the loading robot; A vibration monitoring system, which is used to acquire the vibration information of the loading robot; A data fusion processing system, which is used to acquire the attitude information of the loading robot; A data comprehensive processing upper system, which receives the acquired attitude information, position information, vibration information and point cloud data; A vibration isolation device, which is installed under the equipment chassis of the data acquisition system and the data fusion processing system.
[0005] In some embodiments, the data acquisition system includes: An RTK base station, which sends GPS differential information to the data fusion processing system; LiDAR A and LiDAR B, where LiDAR A and LiDAR B are installed at two positions on the left and right in front of the loading robot; LiDAR C, where LiDAR C is installed directly behind the loading robot body; LiDAR D, where LiDAR D is installed on the column directly above the loading robot; LiDAR E and LiDAR F, where LiDAR E and LiDAR F are installed above the boom and at the middle and lower parts of the boom of the loading robot respectively.
[0006] In some embodiments, the vibration monitoring system includes: Vibration sensors, which are installed at the root of the bucket teeth, the rear wall of the bucket, the connection between the front end of the bucket rod and the bucket, the middle of the bucket rod, the connection between the bucket rod and the boom, the root of the boom, and the connection between the boom and the main body of the electric shovel of the loading robot. The vibration sensors include: Acceleration sensors, which are used to measure the vibration acceleration of various parts of the loading robot; Displacement sensors, which are used to measure the vibration displacement of various parts; Vibration data acquisition module, which acquires the data of the vibration sensors and transmits it to the data fusion processing system.
[0007] In some embodiments, the data fusion processing system includes: Radio receiving module, which receives the position information of the loading robot acquired by the data acquisition system; GNSS receiving module, which receives GPS information, including position information and time information; IMU inertial measurement unit, which acquires attitude information, including pitch angle, yaw angle, and roll angle; LiDAR synchronization unit, which acquires the surrounding point cloud data acquired by the data acquisition system through the network port, then sends the acquired surrounding point cloud data to the switch, and the switch sends the updated point cloud data to the data comprehensive processing host system; Data fusion microprocessor, which calculates high-precision position information and current speed information based on the GPS differential information of the base station and the GPS information of the loading robot itself, and then sends the attitude information, position information, speed information, and vibration information to the data comprehensive processing host system through the switch; Power supply module, which supplies power to the radio receiving module, GNSS receiving module, IMU inertial measurement unit, LiDAR synchronization unit, data fusion microprocessor, and switch respectively.
[0008] In some embodiments, the upper system for comprehensive data processing includes: A point cloud data processing unit, which receives the position information and point cloud data sent from the data fusion processing system to construct panoramic three-dimensional point cloud data; A data visualization unit, which displays the current topographic and geomorphic information according to the constructed panoramic three-dimensional point cloud data, and displays the position, attitude and vibration information of the loading robot.
[0009] In some embodiments, the point cloud data processing unit generates three-dimensional point cloud data based on the position information and point cloud data, then converts the three-dimensional point cloud data to a coordinate system with the loading robot as the origin through coordinate transformation, and then reduces the amount of three-dimensional point cloud data by removing the noise generated during the scanning process through the Gaussian filtering algorithm. Finally, point cloud multi-view registration is realized based on the refined point cloud data to construct a complete three-dimensional point cloud data.
[0010] In some embodiments, the vibration isolation device includes: A vibration isolation device control module; A quasi-zero stiffness vibration reduction device, which is electrically connected to the vibration monitoring system. The vibration isolation device control module receives the data of the vibration monitoring system and adjusts the parameters of the quasi-zero stiffness vibration reduction device according to the data; A frequency-variable damping vibration suppression device, which is electrically connected to the vibration monitoring system and the data fusion processing system. The vibration isolation device control module receives the data of the vibration monitoring system and feeds back the working state information of the frequency-variable damping vibration suppression device to the data fusion processing system.
[0011] In some embodiments, the frequency-variable damping vibration suppression device adopts a magnetorheological fluid damper to realize the change of the damping coefficient with the vibration frequency; The quasi-zero stiffness vibration reduction device includes: A positive stiffness mechanism, which adopts a high-strength helical spring, and its spring stiffness is calculated and selected according to the static load borne by the loading robot in the normal working state and the desired static equilibrium position; A negative stiffness mechanism, which adopts a buckling beam negative stiffness structure. When the bending Euler beam is subjected to an external force, it will undergo buckling deformation. After the beam buckles, as the displacement increases, the external force required to be applied will decrease, showing a negative stiffness characteristic, so as to realize the quasi-zero stiffness characteristic; when the pre-compression amount of the spring increases, the initial stiffness of the spring increases, and the buckling beam needs to enter the buckling state at a smaller displacement to offset the increase in the positive stiffness of the spring, so that the overall stiffness of the system is still close to zero; when the pre-compression amount of the spring decreases, the initial positive stiffness of the spring decreases, and the buckling beam has a significant impact on the overall stiffness within a larger displacement range, ensuring cooperation with the positive stiffness of the spring to realize the quasi-zero stiffness.
[0012] In some embodiments, the vibration monitoring system collects vibration data in real time, preprocesses it, and then inputs it into a trained vibration feedback neural network model. The vibration feedback neural network model outputs the optimal control parameters of the vibration isolation device. The control module of the vibration isolation device is connected to the output end of the vibration feedback neural network model through wireless communication, receives the optimal control parameters output by the vibration feedback neural network model, obtains an accurate parameter adjustment instruction, and drives the quasi-zero stiffness vibration reduction device and the frequency-variable damping vibration suppression device to adjust the parameters.
[0013] In some embodiments, the training process of the vibration feedback neural network model includes: S1: Obtain a large amount of vibration data of the loading robot and the corresponding control parameters of the vibration isolation device; The vibration data includes acceleration a, displacement d, and vibration frequency f, and the control parameters include stiffness coefficient k, spring pre-compression m, and damping coefficient c; S2: Clean, denoise, and normalize the data; S3: Design the neural network structure, including the number of nodes in the input layer, the number of hidden layers and the number of neurons in each layer, and the number of nodes in the output layer; Among them, the number of nodes in the input layer is 6, and the input data are vibration acceleration , vibration displacement d, vibration frequency f, and the initial vibration isolation device parameters, including stiffness coefficient k, spring pre-compression m, and damping coefficient c; The number of hidden layers is 2. Among them, the number of neurons in the first layer is 32, which can not only ensure sufficient non-linear representation ability but also avoid overfitting and excessive computational cost. These neurons will perform complex feature extraction and transformation on the input data, map the original data to a higher-dimensional feature space, and help discover hidden patterns and relationships in the input data; the number of neurons in the second layer is 16. As the information propagates forward in the network, the degree of abstraction of the features continuously increases, so the number of neurons can be appropriately reduced; The activation function of the hidden layer is the ReLU function; The loss function is the mean square error loss function; Update the weights and biases through the backpropagation algorithm; The number of nodes in the output layer is 3, and the output data includes the optimized spring stiffness coefficient k, spring pre-compression m, and damping coefficient c; S4: Randomly initialize the parameters in the neural network model, including weights and biases; S5: Divide the preprocessed data into a training set and a validation set, and start training the neural network using the training set; S6: According to the optimal control parameters of the vibration isolation device output by the training model, the control module drives the actuator of the vibration isolation device according to the optimal parameters to adjust the spring stiffness, spring compression amount and damping coefficient of the vibration isolation device, so as to effectively suppress vibration.
[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. Improvement in high-precision environmental perception ability Multi-lidar collaborative layout: Through the 6-point distributed deployment of the front left / right lidars (A / B), the rear lidar C, the top column lidar D and the boom position lidars E / F, 360° non-blind spot point cloud coverage is achieved, effectively eliminating the scanning blind spots caused by the movement of the loading robot itself.
[0015] Vibration compensation fusion algorithm: Based on the acceleration / displacement data collected by the vibration monitoring system, a vibration-point cloud distortion mathematical model is established to achieve sub-millimeter-level real-time compensation.
[0016] 2. Breakthrough in adaptability to complex working conditions Quasi-zero stiffness active vibration isolation technology: Through the parameter matching of the positive stiffness helical spring and the buckling beam negative stiffness mechanism, ensure the stable operation of the lidar under strong impact loads.
[0017] Frequency-variable damping dynamic regulation: The damping coefficient of the magnetorheological fluid damper is automatically adjusted with the vibration frequency to achieve the optimal vibration suppression effect at different stages of the loading operation (heading / lifting / discharging).
[0018] 3. Intelligent data processing and decision support Neural network control optimization: The vibration feedback neural network model (input layer with 6 nodes → hidden layer with 32 / 16 nodes → output layer with 3 nodes) realizes millisecond-level dynamic optimization of the vibration isolation parameters. Under the condition of sudden load, the overshoot of the control system is <5%, and the response speed is 3.2 times higher than that of the traditional PID control.
[0019] The technical solution of the present invention, through the multi-sensor fusion architecture design, quasi-zero stiffness-frequency variable damping composite vibration isolation technology and deep learning-driven control optimization, systematically solves the three major industry problems of low environmental perception accuracy, large vibration interference and slow dynamic response of the loading robot under complex working conditions, and provides high-reliability technical support for the intelligent upgrade of mines. Brief Description of the Drawings
[0020] Figure 1 is the system structure schematic diagram of the present invention; Figure 2 is the structure schematic diagram of the vibration monitoring system; Figure 3 is the structure schematic diagram of the data fusion processing system; Figure 4It is a schematic diagram of the structure of the upper computer system for data comprehensive processing; Figure 5 It is the working flowchart of the vibration isolation device control module; In the figure, 100 - data acquisition system; 110 - RTK base station; 111 - base station GPS antenna; 112 - base station GPS receiving module; 113 - base station transmitting radio; 120 - lidar sensor; 200 - vibration monitoring system; 210 - vibration sensor; 220 - vibration data acquisition module; 300 - data fusion processing system; 310 - power module; 320 - radio receiving module; 330 - GNSS receiving module; 331 - GPS antenna; 332 - GNSS receiver; 340 - IMU inertial measurement unit; 350 - lidar synchronization unit; 351 - lidar microprocessor; 360 - data fusion microprocessor; 370 - switch; 400 - upper computer system for data comprehensive processing; 410 - point cloud data processing unit; 420 - data visualization unit; 500 - vibration isolation device; 511 - quasi-zero stiffness vibration damping device; 512 - frequency-variable damping vibration suppression device; 520 - vibration isolation device control module. Specific embodiments
[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.
[0022] As Figure 1 shown, a scanning system for a loading robot in an all-weather large-scene complex operation environment includes a data acquisition system 100, a vibration monitoring system 200, a data fusion processing system 300, an upper computer system 400 for data comprehensive processing, and a vibration isolation device 500.
[0023] The data acquisition system 100 includes an RTK base station 110, lidar A, lidar B, lidar C, lidar D, lidar E, and lidar F.
[0024] The RTK base station 110 is installed at a position with a high terrain and no obstructions, and sends GPS differential information to the radio receiving module 320 in the data fusion processing system through wireless radio communication, and then the radio receiving module 320 sends it to the data fusion microprocessor 360. The baud rates of the transmitting radio 113 of the RTK base station 110 and the radio receiving module 320 of the data fusion processing system 300 need to be kept consistent.
[0025] LiDAR A, LiDAR B, LiDAR C, and LiDAR D all use 32-line mechanical LiDARs, while LiDAR E and LiDAR F use 128-line mechanical LiDARs; LiDAR A and LiDAR B are installed at the left and right positions in front of the loading robot body; LiDAR C is installed directly behind the loading robot body; LiDAR D is installed on the column directly above the loading robot; LiDAR E and LiDAR F are respectively installed above the boom and in the middle and lower positions of the boom of the loading robot; The vibration monitoring system 200 includes vibration sensors 210 installed at key parts of the loading robot, a vibration data acquisition module 220, and a data transmission interface connected to the data fusion processing system.
[0026] The vibration sensor 210 includes an acceleration sensor and a displacement sensor, which are used to collect acceleration information and displacement information of the vibrating part. The vibration data acquisition module 220 can synchronously collect, amplify, filter, perform analog-to-digital conversion, and time mark the data collected by the acceleration sensor and the displacement sensor. The acquisition module has a high sampling frequency and can capture the details of the vibration signal, and transmits the data to the data fusion processing system 300 through the network.
[0027] The GNSS receiver 332 of the data fusion processing system 300 receives GPS information and then sends it to the data fusion microprocessor 360, and corrects this information through the GPS differential information of the base station to obtain a more accurate positioning result.
[0028] The IMU inertial measurement unit 340 of the data fusion processing system 300 receives the attitude information of the loading robot, including the pitch angle, roll angle, and yaw angle, is connected to the GNSS receiver 332 through RS232, and performs data fusion calculation through the Kalman filter algorithm to provide more accurate position and attitude information for the loading robot.
[0029] In the LiDAR synchronization unit 350 of the data fusion processing system 300, the LiDAR microprocessor 351 obtains the LiDAR point cloud data through the network port. For each frame of point cloud data collected, the LiDAR data microprocessor 351 will execute an interrupt program to read the current GPS information from the data fusion microprocessor 360 to obtain the accurate time, add the time information to the point cloud data, and fuse the point cloud data with the attitude information and vibration information collected by the IMU inertial measurement unit 340, so as to obtain more accurate point cloud data, and send the fused point cloud data to the switch 370.
[0030] The switch 370 is connected to the upper computer system 400 for comprehensive data processing via a network cable, and transmits the acquired attitude information, position information, speed information, vibration information, and the point cloud data processed by the lidar synchronization unit 350 to the upper computer system 400 for comprehensive data processing.
[0031] The upper computer system 400 for comprehensive data processing is responsible for receiving the acquired attitude information, position information, speed information, vibration information, and point cloud data. The point cloud data processing unit 410 generates three-dimensional point cloud data based on the position information and the point cloud data, then converts the three-dimensional point cloud data to the coordinate system with the loader robot body as the origin through coordinate transformation, and then reduces the amount of the three-dimensional point cloud data by removing the noise points generated during the scanning process through the Gaussian filtering algorithm. Finally, point cloud multi-view registration is achieved based on the refined point cloud data, and the point cloud data collected from different perspectives or at different times is matched and aligned to construct a complete three-dimensional point cloud data; the data visualization unit 420 displays the current topographic and geomorphic information based on the constructed panoramic three-dimensional point cloud data, and displays the position information, attitude information, and vibration information of the loader robot.
[0032] The vibration isolation device 500 is installed below the equipment chassis of the data acquisition system 100 and the data fusion processing system 300. The vibration isolation device includes a quasi-zero stiffness vibration damping device 511 and a frequency-variable damping vibration suppression device 512.
[0033] The quasi-zero stiffness vibration damping device 511 is electrically connected to the vibration monitoring system 200. The vibration isolation device control module 520 can receive the data of the vibration monitoring system and adjust the spring stiffness and pre-compression according to the data. The quasi-zero stiffness vibration damping device is installed between the key components of the data acquisition system (such as the lidar mounting bracket, the base of the data acquisition device, etc. and the loader robot body), and includes a positive stiffness mechanism and a negative stiffness mechanism. The positive stiffness mechanism uses a high-strength helical spring, and its stiffness coefficient is calculated and selected according to the static load borne by the loader robot in the normal working state and the desired static equilibrium position. The negative stiffness mechanism uses a buckling beam negative stiffness structure. When the bending Euler beam is subjected to an external force, it will undergo buckling deformation. After the beam buckles, as the displacement increases, the external force required will decrease, showing a negative stiffness characteristic, so as to achieve the quasi-zero stiffness characteristic.
[0034] Frequency-variable damping vibration suppression device 512, the frequency-variable damping vibration suppression device is electrically connected to the vibration monitoring system 200 and the data fusion processing system 300. The vibration isolation device control module 520 can receive the data of the vibration monitoring system and feedback its own working state information to the data fusion processing system 300. The frequency-variable damping vibration suppression device is installed in series with the quasi-zero stiffness vibration reduction device and is installed on the vibration transmission path. The frequency-variable damping vibration suppression device includes a damper, and the damper uses a magnetorheological fluid damper. The vibration isolation device control module 520 adjusts the damping coefficient of the damper in real time according to the vibration data collected by the vibration sensor in the vibration monitoring system, so that the damping coefficient changes with the vibration frequency. The magnetic field strength adjustment range of the MR damper is between 0.2 and 0.8 Tesla, and it can effectively adjust the damping coefficient within the frequency range.
[0035] The vibration isolation device control module processes the relationship between complex vibration data and vibration isolation device parameters through the non-linear mapping ability of the neural network, optimizes the damping coefficient adjustment strategy according to a large amount of experimental data and actual working conditions, and adjusts the vibration isolation device control parameters according to the overall system requirements, including the following steps: S1: Obtain the vibration data of a large number of loading robots and the corresponding vibration isolation device control parameters. The vibration data includes acceleration a, displacement d, and vibration frequency f. The vibration isolation device control parameters include the stiffness coefficient , spring pre-compression and damping coefficient .
[0036] S2: Clean, denoise, and normalize the data; Map all data to the [0, 1] interval through Min-Max normalization. The formula is: , where is the original data, , are the minimum and maximum values of the data; S3: Design the number of nodes in the input layer, the number of hidden layers and the number of neurons in each layer, and the number of nodes in the output layer of the neural network; The number of nodes in the input layer is 6, which are acceleration , displacement and vibration frequency , stiffness coefficient , spring pre-compression and damping coefficient ; The number of hidden layers is 2. The number of neurons in the first layer is 32, which can not only ensure sufficient non-linear representation ability but also avoid overfitting and excessive computational cost. These neurons will perform complex feature extraction and transformation on the input data, mapping the original data to a higher-dimensional feature space, which helps to discover hidden patterns and relationships in the input data. The number of neurons in the second layer is 16. As information propagates forward in the network, the degree of feature abstraction continuously increases, so the number of neurons can be appropriately reduced. The number of output layer nodes is 3, namely the stiffness coefficient , the pre-compression of the spring and the damping coefficient ; The activation function of the hidden layer is the ReLU function; The loss function is the mean squared error loss function; Update the weights and biases through the backpropagation algorithm.
[0037] S4: Randomly initialize the parameters in the neural network model, including weights and biases. Initialize the weights as small random values and the biases as 0 or small constants; S5: Divide the preprocessed data into a training set and a validation set, generally in a ratio of 80% and 20%.
[0038] Use the training set to train the neural network. During the training process, input the preprocessed vibration data and vibration isolation device control parameters into the network. After forward propagation, obtain the predicted output, calculate the value of the loss function, then calculate the gradient through the backpropagation algorithm, and update the model parameters according to the stochastic gradient descent algorithm.
[0039] Calculate the predicted value through forward propagation. Use the preprocessed sample as the input and convert it into a vector representation that the model can process through the input layer, that is, x = [acceleration a, displacement d, vibration frequency f, stiffness coefficient k, pre-compression of the spring m, damping coefficient c].
[0040] The input vector passes through the hidden layer. The first hidden layer receives the input data and calculates the weighted sum as , where is the weight from the input layer to the first hidden layer, is the bias, is the i-th element of the input vector x, and then use the ReLU function: f(x) = max(0, x) for activation, and the output of the first hidden layer is ; Then the neurons in the second hidden layer receive the output of the first hidden layer, and the weighted sum of the second hidden layer is , where is the weight from the first hidden layer to the second hidden layer, is the bias, is the output element of the first hidden layer, and then the ReLU function: f(x) = max(0, x) is used for activation to obtain the output of the second hidden layer as ; Similarly, the output layer receives the output of the second hidden layer to obtain the weighted sum of the output layer as , and finally the predicted value is obtained.
[0041] Calculate the loss function, compare the predicted output of the model with the actual vibration isolation device control parameters to obtain the loss value. The mean square error calculation formula is , where is the true value, is the predicted value.
[0042] Backpropagate the gradient, starting from the output layer, layer by layer, backpropagate the error to the input layer, and calculate the gradient of each neuron in the model according to the loss function. The error of the loss function with respect to the weighted sum of the output layer is: , and the weight gradient of the loss function with respect to the output layer is obtained as: , and the bias gradient of the loss function with respect to the output layer is: ; The error of the loss function with respect to the weighted sum of the second hidden layer is: , and the weight gradient of the loss function with respect to the second hidden layer is obtained as: , and the bias gradient of the loss function with respect to the second hidden layer is: ; The error of the loss function with respect to the weighted sum of the first hidden layer is: , and the weight gradient of the loss function with respect to the first hidden layer is obtained as: , and the bias gradient of the loss function with respect to the first hidden layer is: .
[0043] Update the weights and thresholds. According to the calculated gradients, use the optimization algorithm to update the weights and thresholds of the model to reduce the loss function. The updated weight is: , and the updated bias is: , where is the learning efficiency, which controls the step size of each update; Iterative optimization. Repeat the processes of forward propagation, loss calculation, backpropagation, and parameter update until the training error of the model reaches a small value or meets the preset conditions, such as reaching the preset number of iterations or the loss function converges.
[0044] S6: According to the optimal control parameters of the vibration isolation device output by the training model, the control module drives the actuator of the vibration isolation device according to the optimal parameters to adjust parameters such as the spring stiffness, spring compression amount, and damping coefficient of the vibration isolation device, so as to effectively suppress vibration and improve the operation stability and accuracy of the loading robot.
[0045] The control module is connected to the output end of the neural network model through wireless communication, receives the spring stiffness, spring pre-compression amount, and damping coefficient, converts the digital signal into an analog electrical signal capable of driving the actuator through digital-to-analog conversion, and amplifies and filters the signal to an appropriate intensity through signal amplification and filtering.
[0046] The processed analog electrical signal drives the stepper motor to work, and adjusts the effective number of turns of the variable stiffness spring and the spacing of the spring wire to make the spring stiffness reach the target value.
[0047] Control the number of steps and direction of rotation of the stepper motor, and push the spring seat to move through the corresponding transmission mechanism to accurately compress the spring to the specified compression amount.
[0048] Amplify the analog electrical signal to a level capable of generating a strong magnetic field intensity through a power amplifier, and change the magnetic field intensity inside the magnetorheological fluid damper to adjust the damping parameters.
[0049] At the same time, the sensors of the vibration isolation device continuously monitor the adjusted working state and feedback relevant information to the control module, so that the control module can further optimize the adjustment strategy according to the feedback to form a closed-loop control.
[0050] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A scanning system for a loading robot in an all-weather large-scene complex working condition operation environment, characterized in that, Including: A data acquisition system (100) for acquiring the position information and point cloud data of a loading robot; A vibration monitoring system (200) for acquiring the vibration information of the loading robot; A data fusion processing system (300) for acquiring the attitude information of the loading robot; An upper data comprehensive processing system (400) that receives the acquired attitude information, position information, vibration information, and point cloud data; A vibration isolation device (500) installed under the equipment chassis of the data acquisition system (100) and the data fusion processing system (300).
2. The scanning system of the loading robot for all-weather large-scene complex working conditions according to claim 1, characterized in that, The data acquisition system (100) includes: An RTK base station (110) that sends GPS differential information to the data fusion processing system (300); Lidar A and Lidar B, which are installed at the left and right positions in front of the loading robot; Lidar C, which is installed directly behind the loading robot body; Lidar D, which is installed on the column directly above the loading robot; Lidar E and Lidar F, which are installed above the boom and at the middle and lower positions of the boom of the loading robot respectively.
3. The loading robot scanning system for all-weather large-scene complex working conditions according to claim 1, characterized in that, The vibration monitoring system (200) includes: Vibration sensors (210) installed at the root of the bucket teeth, the rear wall of the bucket, the connection between the front end of the bucket rod and the bucket, the middle of the bucket rod, the connection between the bucket rod and the boom, the root of the boom, and the connection between the boom and the main body of the electric shovel of the loading robot. The vibration sensors (210) include: Acceleration sensors for measuring the vibration acceleration of various parts of the loading robot; Displacement sensors for measuring the vibration displacement of various parts; A vibration data acquisition module (220) that acquires the data of the vibration sensors (210) and transmits it to the data fusion processing system (300).
4. The scanning system of the loading robot for all-weather large-scene complex working conditions according to claim 1 or 2 or 3, characterized in that, The data fusion processing system (300) includes: A radio receiving module (320) that receives the position information of the loading robot acquired by the data acquisition system (100); A GNSS receiving module (330) that receives GPS information, including position information and time information; An IMU inertial measurement unit (340) that obtains attitude information, including pitch angle, yaw angle, and roll angle; A lidar synchronization unit (350) that acquires the surrounding point cloud data collected by the data acquisition system (100) through a network port, then sends the acquired surrounding point cloud data to a switch (370), and the switch (370) sends the updated point cloud data to the upper data comprehensive processing system (400); A data fusion microprocessor (360) calculates high-precision position information and current speed information based on the GPS differential information of the base station and the GPS information of the loading robot itself, and then sends the attitude information, position information, speed information, and vibration information to the data comprehensive processing host computer system through a switch (370). A power supply module (310) supplies power to a radio receiving module (320), a GNSS receiving module (330), an IMU inertial measurement unit (340), a lidar synchronization unit (350), a data fusion microprocessor (360), and a switch (370) respectively.
5. The loading robot scanning system for all-weather large-scene complex working conditions according to claim 4, characterized in that, The data comprehensive processing host system (400) includes: A point cloud data processing unit (410) receives the position information and point cloud data sent from the data fusion processing system to construct panoramic three-dimensional point cloud data. A data visualization unit (420) displays the current topographic and geomorphic information according to the constructed panoramic three-dimensional point cloud data, and displays the position, attitude, and vibration information of the loading robot.
6. The scanning system of the loading robot for all-weather large-scene complex working conditions according to claim 5, characterized in that The point cloud data processing unit (410) generates three-dimensional point cloud data based on the position information and point cloud data, then converts the three-dimensional point cloud data to a coordinate system with the loading robot as the origin through coordinate transformation, and then reduces the amount of three-dimensional point cloud data by removing the noise generated during the scanning process through the Gaussian filtering algorithm. Finally, point cloud multi-view registration is realized based on the refined point cloud data to construct a complete three-dimensional point cloud data.
7. The scanning system of the loading robot for all-weather large-scene complex working conditions according to claim 1, characterized in that, The vibration isolation device (500) includes: A vibration isolation device control module (520); A quasi-zero stiffness vibration reduction device (511) is electrically connected to the vibration monitoring system (200). The vibration isolation device control module (520) receives the data of the vibration monitoring system (200) and adjusts the parameters of the quasi-zero stiffness vibration reduction device (511) according to the data. A frequency-variable damping vibration suppression device (512) is electrically connected to the vibration monitoring system (200) and the data fusion processing system (300). The vibration isolation device control module (520) receives the data of the vibration monitoring system (200) and feeds back the working state information of the frequency-variable damping vibration suppression device (512) to the data fusion processing system (300).
8. The loading robot scanning system for an all-weather large-scene complex working condition operation environment according to claim 7, characterized in that The frequency-variable damping vibration suppression device (512) uses a magnetorheological fluid damper to realize that the damping coefficient changes with the vibration frequency. The quasi-zero stiffness vibration reduction device (511) includes: A positive stiffness mechanism, which uses a high-strength helical spring, and its spring stiffness is calculated and selected according to the static load borne by the loading robot in the normal working state and the expected static equilibrium position. Negative stiffness mechanism, the negative stiffness mechanism adopts a buckling beam negative stiffness structure. When a bending Euler beam is subjected to an external force, it will undergo buckling deformation. After the beam buckles, as the displacement increases, the external force required to be applied will decrease, showing a negative stiffness characteristic, thereby achieving a quasi-zero stiffness characteristic. When the pre-compression amount of the spring increases, the initial stiffness of the spring increases, and the buckling beam needs to enter the buckling state at a smaller displacement to offset the increase in the positive stiffness of the spring, so that the overall stiffness of the system is still close to zero. When the pre-compression amount of the spring decreases, the initial positive stiffness of the spring decreases, and the buckling beam has a significant impact on the overall stiffness within a larger displacement range, ensuring cooperation with the positive stiffness of the spring to achieve quasi-zero stiffness.
9. The loading robot scanning system for all-weather large-scene complex working conditions according to claim 8, characterized in that, The vibration monitoring system (200) collects vibration data in real time, preprocesses it, and then inputs it into the trained vibration feedback neural network model. The vibration feedback neural network model outputs the optimal control parameters of the vibration isolation device (500). The vibration isolation device control module (520) is connected to the output end of the vibration feedback neural network model through wireless communication, receives the optimal control parameters output by the vibration feedback neural network model, obtains an accurate parameter adjustment instruction, and drives the quasi-zero stiffness vibration damping device (511) and the frequency-variable damping vibration suppression device (512) to adjust the parameters.
10. The loading robot scanning system for all-weather large-scene complex working conditions according to claim 9, characterized in that, The training process of the vibration feedback neural network model includes: S1: Obtain a large amount of vibration data of the loading robot and the corresponding control parameters of the vibration isolation device; The vibration data includes acceleration a, displacement d, and vibration frequency f, and the control parameters include stiffness coefficient k, spring pre-compression amount m, and damping coefficient c; S2: Clean, denoise, and normalize the data; S3: Design the neural network structure, including the number of nodes in the input layer, the number of hidden layers and the number of neurons in each layer, and the number of nodes in the output layer; Among them, the number of nodes in the input layer is 6, and the input data are respectively vibration acceleration , vibration displacement d, vibration frequency f, and the parameters of the initial vibration isolation device, including the stiffness coefficient k, the pre-compression amount m of the spring, and the damping coefficient c; The number of hidden layers is 2, where the number of neurons in the first layer is 32; the number of neurons in the second layer is 16; The activation function of the hidden layer is the ReLU function; The loss function is the mean square error loss function; Update the weights and biases through the backpropagation algorithm; The number of nodes in the output layer is 3, and the output data includes the optimized spring stiffness coefficient k, spring pre-compression amount m, and damping coefficient c; S4: Randomly initialize the parameters in the neural network model, including weights and biases; S5: Divide the preprocessed data into a training set and a validation set, and start training the neural network using the training set; S6: According to the optimal control parameters of the vibration isolation device output by the training model, the control module drives the actuator of the vibration isolation device according to the optimal parameters to adjust the spring stiffness, spring compression amount, and damping coefficient of the vibration isolation device (500) to effectively suppress vibration.