CN-121977860-A - Intelligent bionic dummy system for automobile crash test and cooperative control method

CN121977860ACN 121977860 ACN121977860 ACN 121977860ACN-121977860-A

Abstract

The application relates to the technical field of collision tests, in particular to an intelligent bionic dummy system for an automobile collision test and a cooperative control method. The system comprises a bionic dummy body, a joint driving module arranged at a plurality of main joints of the bionic dummy body, a central control module in communication connection with the joint driving module, and a local execution controller, wherein the joint driving module comprises a magnetorheological damping channel, an active traction channel, a multi-source sensor and a local execution controller, the local execution controller constructs a target moment according to equivalent stiffness parameters, equivalent damping parameters, a target balance angle and a target moment amplitude which is allowed to be applied in a collision stage, and obtains a first sub-target moment distributed to the magnetorheological damping channel and a second sub-target moment distributed to the active traction channel according to a distribution proportion. The application solves the problems of limited adjustment capability of the dynamic characteristics of main joints, insufficient multi-joint coordination and the like of the traditional dummy system.

Inventors

LIU ZHIXIN
LIU WEIDONG
LIU BOSONG
ZHANG HANXIAO
HE YONGLONG
LIN YONGFENG
CUI JINGKAI
FAN ZHENGQI

Assignees

中国汽车技术研究中心有限公司

Dates

Publication Date: 20260505
Application Date: 20260408

Claims (8)

1. An intelligent bionic dummy system for an automobile crash test, comprising: a bionic dummy body; the joint driving module is arranged at a plurality of main joints of the bionic dummy body and comprises a magneto-rheological damping channel, an active traction channel, a multi-source sensor and a local execution controller; The central control module receives data of the multisource sensor, combines the reference attitude information and the collision working condition information to generate control action instructions of all main joints, and transmits the control action instructions to the corresponding joint driving modules, wherein the control action instructions at least comprise equivalent stiffness parameters, equivalent damping parameters, target balance angles and target moment amplitude which are used for constructing joint target moment and are allowed to be applied in a collision stage, and the distribution proportion of the target moment between a magnetorheological damping channel and an active traction channel; The local execution controller constructs a target moment according to the equivalent stiffness parameter, the equivalent damping parameter, the target balance angle and the target moment amplitude which is allowed to be applied in a collision stage, and obtains a first sub-target moment distributed to the magnetorheological damping channel according to the distribution proportion, and a second sub-target moment distributed to the active traction channel; The active traction channel is used for executing a second sub-target moment to provide active joint posture adjustment and recovery control.
2. The system of claim 1, wherein the multi-source sensor is configured to collect joint angle, joint angular velocity, joint stress information, limb segment acceleration, contact state quantity, and execution channel temperature in real time; The joint driving modules are distributed at the positions of the upper arm joint, the forearm joint and the thigh joint; the multisource sensor and the magneto-rheological damping channel are also distributed at key positions of the dummy body except the joint driving module.
3. An intelligent bionic dummy cooperative control method for an automobile crash test is characterized by being applied to the intelligent bionic dummy system for the automobile crash test provided in any one of claims 1-2; The method comprises the following steps: The central control module receives data of the multisource sensor, combines the reference attitude information and the collision working condition information to generate control action instructions of all main joints, and transmits the control action instructions to the corresponding joint driving modules, wherein the control action instructions at least comprise equivalent stiffness parameters, equivalent damping parameters, target balance angles and target moment amplitudes which are used for constructing joint target moment and are allowed to be applied in a collision stage, and the distribution proportion of the target moment between the magnetorheological damping channel and the active traction channel; The local execution controller constructs a target moment according to the equivalent stiffness parameter, the equivalent damping parameter, the target balance angle and the target moment amplitude which is allowed to be applied in a collision stage, obtains a first sub-target moment distributed to the magnetorheological damping channel according to the distribution proportion and a second sub-target moment distributed to the active traction channel, executes the first sub-target moment through the magnetorheological damping channel to provide adjustable damping and passive energy consumption, and executes the second sub-target moment through the active traction channel to provide active joint posture adjustment and recovery control.
4. The method of claim 3, wherein the locally executing controller constructs a target torque based on the equivalent stiffness parameter, the equivalent damping parameter, the target equilibrium angle, and a target torque magnitude allowed to be applied during the crash phase, and derives a first sub-target torque to be distributed to the magnetorheological damping channel and a second sub-target torque to be distributed to the active traction channel based on the distribution ratio, comprising: ; ; Wherein, the Is the first Target moment of each joint; Is the first Angle of individual joints; Is the first Angular acceleration of individual joints; Is the first The target balance angle of each joint, Is the first Actual angular deviations of the individual joints; For characterising the first Equivalent stiffness parameters of the individual joints; Characterization of the first embodiment The equivalent damping parameters of the individual joints are, Is the first Target moment amplitudes of the joints, sat () is a saturation function; Is the first The distribution proportion of the target moment of each joint in the magnetorheological damping channel, Is the first The individual joints are assigned to a first sub-target moment of the magnetorheological damping channel, Is the first The individual joints are assigned to a second sub-target torque of the active traction channel.
5. The method according to claim 4, wherein the method further comprises: the local execution controller monitors the temperature and the thermal load of the magnetorheological damping channel and the active traction channel in real time, and the communication state of the local execution controller and the central control module, and reports the communication state to the central control module; if the central control module determines that the temperature or thermal load exceeds a set threshold, or the communication fails, either of the following schemes is implemented: Narrowing the sub-target moment of the channel exceeding the set threshold; reducing or cutting off a second sub-target torque of the active traction channel; and improving the first sub-target moment of the magnetorheological damping channel.
6. The method of claim 5, wherein the central control module receives data from the multisource sensor and combines the reference pose information with the collision condition information to generate control action commands for each primary joint, comprising: the central control module inputs the data of the multisource sensor, the reference gesture information and the collision working condition information into the state coding unit to obtain a coding vector; Inputting the coding vector into a strategy network to obtain control action instructions of all main joints; the strategy network adopts a feedforward neural network structure and comprises an input layer, three hidden layers and an output layer.
7. The method of claim 6, wherein before inputting the encoded vector into the policy network to obtain the control action command for each primary joint, further comprising: The method comprises the steps of collecting training samples, wherein the training samples comprise posture information, collision working condition information and joint multisource physical data of main joints of a human body, and the training samples are time sequences; inputting the training samples to a state coding unit to obtain coding vectors; inputting the training sample coded at the previous moment into a strategy network to obtain control action instructions of all main joints predicted by the strategy network; Applying the predicted control action instruction to each local execution controller to obtain the posture information of the main joint of the human body at the next moment; And iterating parameters in the strategy network to enable the gesture information of the next moment obtained through the control action instruction to approach the reference gesture information of the next moment.
8. The method of claim 7, wherein iterating parameters in the policy network to approximate the next time pose information derived via control action instructions to the next time reference pose information comprises: iterating parameters in the policy network by optimizing an objective function by: ; ; Wherein, the A long-term cumulative expected return obtained for a strategic network with parameters θ over a training round; A policy network with a parameter theta; Representing the policy network The generated state and action set; T is the upper time step limit of a single training round; Is a discount factor; as a function of the reward, The weight coefficients of the joint angle tracking item, the joint angular velocity tracking item, the key part impact punishment item, the control smoothing item and the energy consumption item are respectively; Is a time sequence of joint angles, t represents time, For reference to the time series of joint angles, Is a time series of the angular velocity of the joint, Is a time series of reference joint angular velocity, Is the peak value of acceleration of the key part of the dummy body, Is a time sequence of control action instructions, Is a time-consuming sequence.

Description

Intelligent bionic dummy system for automobile crash test and cooperative control method Technical Field The application relates to the technical field of collision tests, in particular to an intelligent bionic dummy system for an automobile collision test and a cooperative control method. Background The existing crash test dummy system mainly adopts a preset joint damping and fixing mechanical structure so as to simulate partial gesture response of a human body in the process of collision. Part of the system can carry out limited dynamic adjustment on few key joints and is applied to automobile safety compliance tests and collision protection performance tests. However, the control mode of such a system is mostly independent control of a single joint, and only a simple posture simulation can be realized at a specific stage of the collision process (usually limited to passive response after collision impact), an effective whole-body multi-joint cooperative control mechanism is not formed yet, and full-stage posture control matched with human response is difficult to realize. The existing dummy system usually adopts joint rigidity and damping parameters preset before the test, and is difficult to adjust in real time according to collision working conditions. Meanwhile, the existing control modes are mostly single-joint independent control, so that linkage response of multiple joints such as head and neck, trunk, limbs and the like in the collision process is difficult to coordinate, and differentiated physiological responses of various stages before collision, instant collision and after collision are difficult to cover. Due to the limitations, existing dummy systems still have shortcomings in terms of pose response fidelity, key biomechanical data validity, and repeatability under different conditions. Therefore, it is necessary to provide a new intelligent bionic dummy system and a cooperative control method thereof to improve the shortcomings of the prior art in the aspects of main joint dynamics adjustment, multi-joint cooperative control and collision overall process posture and muscle response simulation. Disclosure of Invention The application aims to provide an intelligent bionic dummy system and a cooperative control method for an automobile collision test, which are used for solving the problems of limited adjustment capacity of main joint dynamics characteristics, insufficient multi-joint cooperation and insufficient simulation of the gesture and muscle response in the whole collision process of the traditional dummy system. In order to achieve the above purpose, the present application adopts the following technical scheme: In a first aspect, the present application provides an intelligent bionic dummy system for an automobile crash test, including: a bionic dummy body; the joint driving module is arranged at a plurality of main joints of the bionic dummy body and comprises a magneto-rheological damping channel, an active traction channel, a multi-source sensor and a local execution controller; The central control module receives data of the multisource sensor, combines the reference attitude information and the collision working condition information to generate control action instructions of all main joints, and transmits the control action instructions to the corresponding joint driving modules, wherein the control action instructions at least comprise equivalent stiffness parameters, equivalent damping parameters, target balance angles and target moment amplitude which are used for constructing joint target moment and are allowed to be applied in a collision stage, and the distribution proportion of the target moment between a magnetorheological damping channel and an active traction channel; The local execution controller constructs a target moment according to the equivalent stiffness parameter, the equivalent damping parameter, the target balance angle and the target moment amplitude which is allowed to be applied in a collision stage, and obtains a first sub-target moment distributed to the magnetorheological damping channel according to the distribution proportion, and a second sub-target moment distributed to the active traction channel; The active traction channel is used for executing a second sub-target moment to provide active joint posture adjustment and recovery control. In a second aspect, the application provides an intelligent bionic dummy cooperative control method for an automobile crash test, which applies an intelligent bionic dummy system for the automobile crash test; The method comprises the following steps: The central control module receives data of the multisource sensor, combines the reference attitude information and the collision working condition information to generate control action instructions of all main joints, and transmits the control action instructions to the corresponding joint driving modules, wherein the control action instructions at least comprise e