Adaptive control method and system for rescue robot against marine salt spray corrosion

Abstract

The application relates to the technical field of parameter adaptation, and discloses a self-adaptive control method and system of a rescue robot resistant to marine salt mist corrosion, the method comprising the following steps: integrating a current machine motion state of a target process and a machine desired trajectory to form a comprehensive data set, constructing a feedforward feedback composite loop based on a factory power parameter, and generating a basic driving force instruction; setting an auxiliary force signal according to a basic driving force instruction amplitude and a joint servo bandwidth, and obtaining a composite driving force instruction after superposition; calculating a real dynamics parameter by taking an actual driving torque generated by the instruction and a kinematics quantity of the comprehensive data set as the input, reconstructing the feedforward feedback composite loop, and obtaining a final control instruction; and the application can improve the efficiency of self-adaptive control of the rescue robot resistant to marine salt mist corrosion.

Description

Technical Field

[0001] This invention relates to the field of parameter adaptation technology, and in particular to an adaptive control method and system for a rescue robot resistant to marine salt spray corrosion. Background Technology

[0002] In existing control technologies for rescue robots resistant to marine salt spray corrosion, the control system typically relies solely on fixed dynamic parameters calibrated at the factory to generate drive force commands. However, the marine salt spray environment gradually corrodes the robot's joint sealing structure, alters lubrication characteristics, and increases frictional resistance, leading to a significant deviation between the factory parameters and the actual dynamic characteristics. Feedforward feedback controllers using fixed parameters cannot track these time-varying parameter perturbations, causing the mismatch between the actual drive torque and the desired torque to continuously increase. When performing high-precision tasks, rescue robots exhibit trajectory tracking error divergence, joint vibration, and even loss of control, with the control effectiveness declining sharply as the salt spray exposure time increases.

[0003] Another type of existing technology attempts to update model parameters through offline identification or periodic calibration. However, offline identification requires stopping the operation and placing the robot under specific test conditions for excitation, which is not feasible in on-site rescue missions in marine salt spray corrosion environments. Furthermore, the offline identification process cannot capture the nonlinear time-varying characteristics caused by corrosion factors; the identification results are only valid at the calibration time, and subsequent corrosion will lead to new parameter mismatches. In addition, the design of auxiliary excitation signals in existing technologies is often independent of the basic driving force commands. When superimposed, these signals can easily exceed the safe amplitude range of the joint servo actuator, or the excitation frequency may be mismatched with the system bandwidth, preventing the excitation of high-frequency dynamic modes caused by corrosion. Ultimately, this results in insufficient accuracy in estimating the true dynamic parameters, and the adaptive control effect cannot meet the long-term robustness requirements against marine salt spray corrosion. Summary of the Invention

[0004] This invention provides an adaptive control method and system for a rescue robot resistant to marine salt spray corrosion, the main purpose of which is to solve the problem of low efficiency in the adaptive control of the rescue robot resistant to marine salt spray corrosion.

[0005] To achieve the above objectives, the present invention provides an adaptive control method for a rescue robot resistant to marine salt spray corrosion, comprising:

[0006] The current machine motion state and the machine's desired trajectory in the target process are used as the comprehensive dataset of the target process;

[0007] Based on the factory power parameters of the target process, a feedforward feedback composite loop of the target process is constructed, and based on the comprehensive dataset and the feedforward feedback composite loop, the basic driving force command of the target process is generated.

[0008] The auxiliary force signal for the target process is set based on the amplitude of the basic driving force command and the control bandwidth of the joint servo during the target process.

[0009] The auxiliary force signal is superimposed with the basic driving force command to obtain the composite driving force command for the target process;

[0010] The actual driving torque generated by the composite driving force command is used as mechanical response data, and the actual driving torque sequence of the mechanical response data and the kinematics of the comprehensive dataset are used to calculate the true dynamic parameters of the target process.

[0011] The feedforward-feedback composite loop is reconstructed using the actual dynamic parameters as force control gain coefficients to obtain the final control command for the target process.

[0012] In a preferred embodiment, the step of using the current machine motion state and the desired machine trajectory of the target process as a comprehensive dataset of the target process includes:

[0013] The current position, current velocity, and current acceleration of the rescue robot's moving joints are collected during the target process to obtain the current robot motion state during the target process;

[0014] Linear interpolation is performed between the current position of the rescue robot and the target position during the target process to obtain the expected position sequence at the end of the target process. Successive time derivatives are then performed on the expected position sequence at the end to obtain the expected velocity sequence and expected acceleration sequence of the target process.

[0015] The expected position sequence, the expected velocity sequence, and the expected acceleration sequence are used together as the expected machine trajectory of the target process. Sampling times are added to the expected machine trajectory and the current machine motion state to obtain the comprehensive dataset of the target process.

[0016] In a preferred embodiment, the step of constructing a feedforward feedback composite loop for the target process based on the factory power parameters of the target process, and generating a basic driving force command for the target process based on the comprehensive dataset and the feedforward feedback composite loop, includes:

[0017] Separate the factory inertial parameters, Coriolis force parameters, and gravity parameters from the factory dynamic parameters of the target process;

[0018] Based on the factory inertial parameters, the Coriolis force parameters, and the gravity parameters, the torque components of the kinematic quantity sequence in the comprehensive dataset are adjusted to obtain the feedforward feedback composite loop of the target process.

[0019] The desired acceleration of the comprehensive dataset is introduced into the feedforward path of the feedforward-feedback composite loop, and the motion error between the current machine motion state and the desired machine trajectory of the comprehensive dataset is introduced into the feedback path of the feedforward-feedback composite loop to obtain the feedforward driving force component and the feedback driving force component of the target process.

[0020] The feedforward driving force component and the feedback driving force component are combined to obtain the driving torque sequence of the target process, and the driving torque sequence is used as the basic driving force command of the target process.

[0021] In a preferred embodiment, the step of adjusting the torque components of the kinematic quantity sequence in the comprehensive dataset based on the factory inertial term parameters, the Coriolis force term parameters, and the gravity term parameters to obtain the feedforward feedback composite loop of the target process includes:

[0022] Using the factory inertial term parameters as feedforward gain, the desired acceleration sequence of the kinematic quantity sequence is amplified to obtain the feedforward gain channel of the target process;

[0023] The current velocity sequence and the desired velocity sequence, and the current position sequence and the desired position sequence in the kinematic quantity sequence are respectively subjected to difference processing to obtain the velocity motion error sequence and the position motion error sequence of the target process.

[0024] The factory inertial term parameters are used as inertial term feedback gain, the Coriolis force term parameters are used as Coriolis force term feedback gain, and the gravity term parameters are used as gravity term feedback gain. The velocity motion error sequence and the position motion error sequence are amplified to obtain the inertial term feedback gain channel, the Coriolis force term feedback gain channel and the gravity term feedback gain channel of the target process.

[0025] The inertial term feedback gain channel, the Coriolis force term feedback gain channel, and the gravity term feedback gain channel are merged to obtain the feedback gain channel of the target process.

[0026] The feedforward gain channel and the feedback gain channel are connected in parallel to obtain the feedforward-feedback composite loop of the target process.

[0027] In a preferred embodiment, setting the auxiliary force signal for the target process based on the amplitude of the basic driving force command and the control bandwidth of the joint servo during the target process includes:

[0028] The peak amplitude of the basic driving force command is attenuated proportionally to obtain the amplitude of the auxiliary force signal for the target process.

[0029] The cutoff frequency of the joint servo in the target process is shifted to a higher frequency direction to obtain the auxiliary force signal frequency of the target process;

[0030] A sinusoidal excitation signal is generated with the amplitude of the auxiliary force signal as the amplitude and the frequency of the auxiliary force signal as the frequency, and the sinusoidal excitation signal is used as the auxiliary force signal for the target process.

[0031] In a preferred embodiment, the step of superimposing the auxiliary force signal with the basic driving force command to obtain the composite driving force command for the target process includes:

[0032] The values ​​of the basic driving force command and the auxiliary force signal at the sampling time are respectively used as the basic driving force time sequence and the auxiliary force time sequence of the target process;

[0033] The value at the current moment in the basic driving force time series is added to the value at the same moment in the auxiliary force time series to obtain the composite driving force value of the target process.

[0034] The composite driving force values ​​are arranged in chronological order to obtain the composite driving force command for the target process.

[0035] In a preferred embodiment, the step of using the actual driving torque generated by the composite driving force command as mechanical response data, and calculating the true dynamic parameters of the target process based on the actual driving torque sequence of the mechanical response data and the kinematic quantities of the comprehensive dataset, includes:

[0036] Arrange the actual driving torques generated by the composite driving force command according to the sampling time to obtain the mechanical response data of the target process;

[0037] Align the mechanical response data with the current velocity sequence and current acceleration sequence of the comprehensive dataset according to the sampling time to obtain the associated sequence of the target process;

[0038] The actual dynamic parameters of the target process are calculated based on the associated sequence and the number of joint degrees of freedom and link geometry of the rescue robot during the target process.

[0039] In a preferred embodiment, the formula for calculating the actual dynamic parameters includes:

[0040]

[0041] in, These are the actual dynamic parameters. Construct a matrix for the associated sequence. This is the matrix transpose operator. For frequency domain weighted diagonal matrix, For adaptive compromise coefficients, It is a diagonal regularized matrix. This is the actual driving torque vector. This is the vector of nominal dynamic parameters specified by the manufacturer.

[0042] In a preferred embodiment, the step of reconstructing the feedforward-feedback composite loop using the actual dynamic parameters as force control gain coefficients to obtain the final control command for the target process includes:

[0043] The current feedback gain of the target process is obtained by replacing the factory dynamic parameters of the target process with the actual dynamic parameter set.

[0044] Based on the current feedback gain, the expected acceleration sequence of the machine's expected trajectory in the target process is amplified to obtain the current feedforward torque of the target process. Based on the current feedback gain, the position motion error sequence and velocity motion error sequence of the target process are adjusted to obtain the current feedback torque of the target process.

[0045] The current feedforward torque and the current feedback torque are combined to obtain the current feedforward-feedback composite loop of the target process, and the driving torque sequence output by the current feedforward-feedback composite loop is used as the final control command of the target process.

[0046] To address the aforementioned problems, the present invention also provides an adaptive control system for a rescue robot resistant to marine salt spray corrosion, the system comprising:

[0047] The integrated data module uses the current machine motion state and the machine's desired trajectory in the target process as the integrated dataset of the target process;

[0048] The basic driving force command module constructs a feedforward feedback composite loop for the target process based on the factory power parameters of the target process, and generates the basic driving force command for the target process based on the comprehensive dataset and the feedforward feedback composite loop.

[0049] An auxiliary force signal module sets the auxiliary force signal for the target process based on the amplitude of the basic driving force command and the control bandwidth of the joint servo during the target process.

[0050] The composite driving force command module superimposes the auxiliary force signal with the basic driving force command to obtain the composite driving force command for the target process;

[0051] The real dynamics parameter module uses the actual driving torque generated by the composite driving force command as mechanical response data, and calculates the real dynamics parameters of the target process based on the actual driving torque sequence of the mechanical response data and the kinematic quantities of the comprehensive dataset.

[0052] The final control command module reconstructs the feedforward feedback composite loop using the actual dynamic parameters as force control gain coefficients to obtain the final control command for the target process.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] 1. Based on the technical solution described in the document, the adaptive control method for this marine salt spray corrosion-resistant rescue robot superimposes the auxiliary force signal and the basic driving force command to form a composite driving force command. The actual driving torque generated by the composite driving force command is then aligned with the kinematic quantities in the comprehensive dataset to calculate the true dynamic parameters. These true dynamic parameters replace the factory-set dynamic parameters to reconstruct the feedforward feedback composite loop. This method enables the force control gain coefficient upon which the final control command is based to reflect in real time the actual physical effects caused by marine salt spray corrosion, such as increased joint sealing friction, changes in mass distribution, and altered lubrication characteristics. This significantly improves the matching accuracy between the driving torque sequence and the robot's actual dynamic requirements, ensuring that the control bandwidth of the joint servo is effectively utilized when the rescue robot performs tasks in a corrosive environment. The high-frequency excitation of the auxiliary force signal can continuously excite and identify parameter perturbations caused by corrosion, avoiding torque output deviations due to model mismatch.

[0055] 2. The superposition of the basic driving force command and the auxiliary force signal in this method ensures that the amplitude of the composite driving force command always remains within the safe linear operating range of the joint servo actuator. Simultaneously, the frequency of the auxiliary force signal is set to the joint servo cutoff frequency, which is offset towards higher frequencies, allowing the sinusoidal excitation signal to specifically elicit the rapidly changing dynamic characteristics that are prone to change in corrosive environments. Through the construction of associated sequences and frequency-domain weighted identification of real dynamic parameters, this method obtains mechanical response data that is strictly aligned with the current motion state. Furthermore, the feedforward-feedback composite loop reconstructed based on the real dynamic parameters can simultaneously output accurate current feedforward torque and current feedback torque, merging them into the final control command. This method enables the control system to autonomously track and compensate for time-varying parameter perturbations in a marine salt spray corrosion environment, improving trajectory tracking accuracy, joint motion stability, and long-term operational reliability. Attached Figure Description

[0056] Figure 1 This is a flowchart illustrating an adaptive control method for a marine salt spray-resistant rescue robot according to an embodiment of the present invention.

[0057] Figure 2 A functional block diagram of an adaptive control system for a marine salt spray-resistant rescue robot according to an embodiment of the present invention;

[0058] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0059] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0060] This application provides an adaptive control method for a rescue robot resistant to marine salt spray corrosion. The execution entity of the adaptive control method for the rescue robot resistant to marine salt spray corrosion includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the adaptive control method for the rescue robot resistant to marine salt spray corrosion can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0061] Reference Figure 1 The diagram shown is a flowchart illustrating an adaptive control method for a marine salt spray-resistant rescue robot according to an embodiment of the present invention. In this embodiment, the adaptive control method for the marine salt spray-resistant rescue robot includes:

[0062] In this embodiment of the invention, when the current machine motion state and the machine's desired trajectory of the target process are used as the comprehensive dataset of the target process, it is specifically used for:

[0063] The current position, current velocity, and current acceleration of the rescue robot's moving joints are collected during the target process to obtain the current robot motion state during the target process;

[0064] Linear interpolation is performed between the current position of the rescue robot and the target position during the target process to obtain the expected position sequence at the end of the target process. Successive time derivatives are then performed on the expected position sequence at the end to obtain the expected velocity sequence and expected acceleration sequence of the target process.

[0065] The expected position sequence, the expected velocity sequence, and the expected acceleration sequence are used together as the expected machine trajectory of the target process. Sampling times are added to the expected machine trajectory and the current machine motion state to obtain the comprehensive dataset of the target process.

[0066] Specifically, the rotation angle of the joint is directly read by an absolute position encoder installed on each joint of the rescue robot as the current position value. An incremental encoder is used to measure the change in joint angular displacement at a fixed sampling period and divide it by the sampling time to obtain the current speed value.

[0067] Specifically, first, the current coordinates of the rescue robot's end effector in Cartesian space and the coordinates of the target are determined. The straight path between the two is divided into several interpolation steps according to the discrete sampling period of the control system. The end coordinates corresponding to each step are arranged in sequence to form the desired end position sequence. Then, the average speed in the interval is obtained by dividing the coordinate difference between two adjacent positions in the sequence by the sampling time.

[0068] Specifically, the expected position value, expected velocity value, and expected acceleration value of the end point corresponding to the same sampling time are bundled into a data tuple, and the current position, current velocity, and current acceleration values ​​of the current machine motion state collected at the same sampling time are bundled into another data tuple.

[0069] Furthermore, the current velocity value is differentially processed with respect to time to obtain the current acceleration value. These three types of data from all joints are aligned according to the sampling time to jointly constitute the current machine motion state.

[0070] Furthermore, arranging all interval average velocities in chronological order yields the desired velocity sequence. Then, the same processing is applied to adjacent velocity values ​​in the desired velocity sequence to obtain the acceleration change at each sampling time, thus forming the desired acceleration sequence.

[0071] Furthermore, the two types of data tuples are then stored side by side according to the order of sampling time. Each time point contains both the three elements of the expected trajectory and the three elements of the current motion state. The data set of all time points is named the comprehensive dataset.

[0072] In summary, it provides a real and synchronous feedback benchmark for subsequent control, enabling the control system to accurately grasp the real-time movement of each joint of the rescue robot in the corrosive environment of marine salt spray. This avoids state estimation errors caused by sensor drift or delay due to environmental corrosion, thus ensuring that the current machine motion state on which the basic driving force command is based has high fidelity and high timeliness, laying an unshakable data foundation for motion error calculation in the subsequent feedforward feedback composite loop.

[0073] In summary, it generates a completely smooth and continuous motion trajectory from the starting point to the target point. Linear interpolation ensures the monotonicity and overshoot-free characteristics of the position sequence, and successive time differentiation makes the velocity and acceleration sequences naturally satisfy the kinematic consistency constraints, avoiding shocks or oscillations caused by trajectory abrupt changes. It is particularly suitable for scenarios with limited bandwidth of joint servo systems in marine salt spray corrosion environments, effectively reducing the risk of wear and corrosion fatigue of mechanical structures.

[0074] In summary, by strictly aligning the three elements of the desired trajectory with the three elements of the current motion state at a unified sampling time, a data set that is perfectly matched in the time dimension is formed. This allows the feedforward driving force component and the feedback driving force component at each moment in the subsequent feedforward-feedback composite loop to be calculated based on the kinematic quantities at the same moment. This eliminates the phase lag and control error introduced by data asynchrony, thereby significantly improving the parameter identification accuracy and the effectiveness of the final control command in the adaptive control method during the dynamic response process.

[0075] In this embodiment of the invention, when constructing the feedforward feedback composite loop of the target process based on the factory power parameters of the target process, and generating the basic driving force command of the target process according to the comprehensive dataset and the feedforward feedback composite loop, it is specifically used for:

[0076] Separate the factory inertial parameters, Coriolis force parameters, and gravity parameters from the factory dynamic parameters of the target process;

[0077] Based on the factory inertial parameters, the Coriolis force parameters, and the gravity parameters, the torque components of the kinematic quantity sequence in the comprehensive dataset are adjusted to obtain the feedforward feedback composite loop of the target process.

[0078] The desired acceleration of the comprehensive dataset is introduced into the feedforward path of the feedforward-feedback composite loop, and the motion error between the current machine motion state and the desired machine trajectory of the comprehensive dataset is introduced into the feedback path of the feedforward-feedback composite loop to obtain the feedforward driving force component and the feedback driving force component of the target process.

[0079] The feedforward driving force component and the feedback driving force component are combined to obtain the driving torque sequence of the target process, and the driving torque sequence is used as the basic driving force command of the target process.

[0080] Specifically, the dynamic parameter table calibrated when the rescue robot leaves the factory is directly read, and all the coefficients that are proportional to the joint acceleration are extracted and classified as the factory inertial parameters.

[0081] Specifically, the expected acceleration sequence in the comprehensive dataset is first multiplied by the factory inertia term parameter to obtain the feedforward torque component, and then the current velocity sequence in the comprehensive dataset is multiplied by the factory Coriolis force term parameter to obtain the Coriolis torque component.

[0082] Specifically, the branch connected to the desired acceleration is taken out separately from the feedforward feedback composite loop. This branch amplifies the desired acceleration sequence sequentially through the factory inertial term parameter gain module. The amplified output is the feedforward driving force component.

[0083] Specifically, at each sampling moment, the calculated value of the feedforward driving force component is directly added to the value of the feedback driving force component to obtain the total driving torque value at that moment.

[0084] Furthermore, all coefficients related to the product of joint velocities are extracted and classified as the factory-issued Coriolis force parameters, and all coefficients related to the sine and cosine of joint positions are extracted and classified as the factory-issued gravity parameters. These three types of parameters are stored as three independent parameter vectors.

[0085] Furthermore, the current position sequence in the comprehensive dataset is multiplied by the factory gravity parameter to obtain the gravity torque component. At the same time, the position deviation and velocity deviation between the current machine motion state and the desired machine trajectory in the comprehensive dataset are multiplied by the corresponding factory parameters to obtain the feedback torque component. Finally, the calculation paths of all torque components are arranged in parallel according to the feedforward path and the feedback path. The multiplication operation of parameters and kinematic quantities in each path is implemented through the gain module. The parallel structure of these modules constitutes the feedforward-feedback composite loop.

[0086] Furthermore, the current position and current speed in the current machine motion state are extracted from the feedforward feedback composite loop, and the expected position and expected speed in the expected trajectory of the machine are calculated by subtracting them from each time step to obtain a position error sequence and a speed error sequence. The position error sequence is amplified by the factory gravity term parameter gain module to obtain the position feedback force component. The speed error sequence is amplified by the factory inertia term parameter gain module and the factory Coriolis force term parameter gain module to obtain two speed feedback force components. The three feedback force components are added together to obtain the feedback driving force component.

[0087] Furthermore, the total driving torque values ​​at all sampling moments are arranged into a sequence according to time sequence. This sequence is the driving torque sequence. The complete sequence is transmitted to the joint servo drive of the rescue robot as the initial control command, i.e., the basic driving force command.

[0088] In summary, the fixed dynamic characteristics of the rescue robot at the time of manufacture are clearly divided into three independent sets of parameters according to physical meaning. This provides a clear source of gain coefficients for subsequent torque component adjustment, so that even if the sensor signals are interfered with in the marine salt spray corrosion environment, the control system can still maintain the basic control framework based on the factory nominal values, avoiding the confusion in the feedforward feedback structure design caused by mixed parameters.

[0089] In summary, by precisely adjusting the three types of factory-set dynamic parameters with the kinematic quantity sequences such as expected acceleration, current velocity, and current position in the comprehensive dataset according to their physical correspondence, a feedforward feedback composite loop with a clear torque distribution function is formed. This allows the loop to maintain a complete dynamic compensation structure even under parameter perturbations caused by marine salt spray corrosion, providing a structurally stable and physically clear control framework for the generation of subsequent basic driving force commands.

[0090] In summary, the feedforward path directly utilizes the desired acceleration to pre-compensate for the known dynamic torque requirement, while the feedback path utilizes the motion error between the current machine motion state and the machine's desired trajectory to correct unknown disturbances and model mismatches in real time. The separation of the feedforward and feedback components allows the feedback component to independently undertake the task of suppressing errors without retuning the feedforward structure when the nonlinear friction caused by marine salt spray corrosion increases.

[0091] In summary, by directly adding the torque values ​​generated by the feedforward and feedback at the same time, a complete and conflict-free driving torque sequence is formed. This sequence, as the basic driving force command, includes both the nominal torque required for the desired trajectory and the compensation torque required for error correction. This enables the rescue robot to perform its tasks with low tracking error in the corrosive environment of marine salt spray, while also preserving a linear superposition interface structure for subsequent superposition of auxiliary force signals.

[0092] In this embodiment of the invention, when adjusting the torque components of the kinematic quantity sequence in the comprehensive dataset based on the factory inertial term parameters, the Coriolis force term parameters, and the gravity term parameters to obtain the feedforward feedback composite loop of the target process, it is specifically used for:

[0093] Using the factory inertial term parameters as feedforward gain, the desired acceleration sequence of the kinematic quantity sequence is amplified to obtain the feedforward gain channel of the target process;

[0094] The current velocity sequence and the desired velocity sequence, and the current position sequence and the desired position sequence in the kinematic quantity sequence are respectively subjected to difference processing to obtain the velocity motion error sequence and the position motion error sequence of the target process.

[0095] The factory inertial term parameters are used as inertial term feedback gain, the Coriolis force term parameters are used as Coriolis force term feedback gain, and the gravity term parameters are used as gravity term feedback gain. The velocity motion error sequence and the position motion error sequence are amplified to obtain the inertial term feedback gain channel, the Coriolis force term feedback gain channel and the gravity term feedback gain channel of the target process.

[0096] The inertial term feedback gain channel, the Coriolis force term feedback gain channel, and the gravity term feedback gain channel are merged to obtain the feedback gain channel of the target process.

[0097] The feedforward gain channel and the feedback gain channel are connected in parallel to obtain the feedforward-feedback composite loop of the target process.

[0098] Specifically, the factory inertia parameter extracted from the factory dynamic parameters is configured as the gain coefficient input of a multiplier, and the signal input of the multiplier is connected to the expected acceleration value at each sampling moment in the expected acceleration sequence of the comprehensive dataset.

[0099] Specifically, at each sampling moment, the current velocity value and the desired velocity value are taken from the kinematic quantity sequence, and the velocity difference at that moment is obtained by subtracting the desired velocity value from the current velocity value. The velocity differences at all moments are arranged in chronological order to form the velocity motion error sequence.

[0100] Specifically, the velocity motion error sequence is simultaneously fed into two multipliers. The gain coefficient of the first multiplier is connected to the factory inertial term parameter, and the output is the signal sequence of the inertial term feedback gain channel.

[0101] Specifically, the signal values ​​output by the three channels at each sampling time are fed into the same adder. The adder adds the three values ​​to obtain the combined signal value at that time. The combined signal values ​​at all sampling times are arranged in chronological order to form a total signal path, which is the feedback gain channel.

[0102] Specifically, the output signal line of the feedforward gain channel and the output signal line of the feedback gain channel are connected to the two input terminals of the second adder. At each sampling time, the adder adds the amplified signal value output by the feedforward gain channel and the combined signal value output by the feedback gain channel. The signal line led out from the output terminal of the adder is the final output of the feedforward-feedback composite loop. This loop retains the structure of the feedforward path and the feedback path working in parallel.

[0103] Furthermore, the multiplier multiplies the desired acceleration value with the factory-set inertia parameter value and outputs an amplified signal value. The output signals at all sampling times are arranged in chronological order to form a signal path, which is the feedforward gain channel.

[0104] Furthermore, the position difference at that moment is obtained by subtracting the expected position value from the current position value. The position differences at all moments are arranged in chronological order to form the position motion error sequence.

[0105] Furthermore, the gain coefficient of the second multiplier is connected to the Coriolis force parameter, and the output is a signal sequence of the Coriolis force feedback gain channel. The position motion error sequence is then sent to the third multiplier, whose gain coefficient is connected to the gravity parameter, and the output is a signal sequence of the gravity feedback gain channel.

[0106] In summary, by directly using the factory-calibrated inertial parameters to proportionally amplify the desired acceleration, an independent feedforward path is formed. This allows the path to output the torque component required to overcome the inertial load in advance when the joint servo response lags in a marine salt spray corrosive environment, reducing the burden on the feedback loop and thus improving the speed at which the basic driving force command follows the desired trajectory.

[0107] In summary, by performing time-by-time difference calculations, the deviation between the current motion state and the desired trajectory is quantified into two error sequences with clear physical meanings. The velocity error reflects the tracking deviation in terms of damping and Coriolis force, while the position error reflects the position deviation in terms of gravity and elastic force. This provides an independent and uncoupled error input signal for the subsequent feedback gain channel, enabling feedback compensation to apply appropriate gains for different types of errors, thereby improving the anti-disturbance capability of the control system in marine salt spray corrosion environments.

[0108] In summary, by treating the three types of factory-set dynamic parameters as independent feedback gains and applying them to the velocity motion error sequence and position motion error sequence respectively, three physically decoupled feedback gain channels are formed. This allows each channel to independently adjust the feedback compensation amount according to the change characteristics of its own parameters when joint friction and inertia are caused by marine salt spray corrosion. This avoids the problem that a single gain cannot simultaneously match the three different torque characteristics of inertia, Coriolis force and gravity.

[0109] In summary, by adding the outputs of three feedback gain channels with different physical meanings at the same time, a comprehensive feedback gain channel is formed. The total feedback torque output by this channel includes compensation components for inertial error, Coriolis force error, and gravity error. This makes the feedback compensation more consistent with the actual dynamic coupling characteristics of the rescue robot in the marine salt spray corrosion environment, and improves the full-band adaptability and overall stability of the feedback control.

[0110] In summary, by combining the rapid pre-compensation capability of the feedforward gain channel with the error suppression capability of the feedback gain channel in parallel to form a complete control loop, the feedforward channel independently handles the nominal torque demand caused by the desired acceleration, while the feedback channel independently handles the compensation torque demand caused by motion error. The two do not interfere with each other and their outputs are added together. This ensures that even if the high gain of the feedback channel is limited due to sensor noise in the corrosive marine salt spray environment, the feedforward channel can still guarantee basic trajectory tracking accuracy, thus achieving a balance between robustness and speed of the control system.

[0111] In this embodiment of the invention, when setting the auxiliary force signal for the target process based on the amplitude of the basic driving force command and the control bandwidth of the joint servo in the target process, it is specifically used for:

[0112] The peak amplitude of the basic driving force command is attenuated proportionally to obtain the amplitude of the auxiliary force signal for the target process.

[0113] The cutoff frequency of the joint servo in the target process is shifted to a higher frequency direction to obtain the auxiliary force signal frequency of the target process;

[0114] A sinusoidal excitation signal is generated with the amplitude of the auxiliary force signal as the amplitude and the frequency of the auxiliary force signal as the frequency, and the sinusoidal excitation signal is used as the auxiliary force signal for the target process.

[0115] Specifically, the maximum value among all driving torque values ​​in the basic driving force command is taken as the peak amplitude. This peak amplitude is multiplied by a fixed attenuation ratio coefficient, which is a specific value between zero and one. The result of the multiplication operation is the amplitude of the auxiliary force signal.

[0116] Specifically, the original cutoff frequency value specified in the joint servo driver manual is read, and a fixed offset is added to this value. The offset value is positive. The new frequency value obtained after the addition operation is the frequency of the auxiliary force signal.

[0117] Specifically, the amplitude of the auxiliary force signal is taken as the maximum output value of the sine function, and the frequency of the auxiliary force signal is taken as the rate of change of the angle of the sine function.

[0118] Furthermore, at each sampling moment, the corresponding sine function value is calculated. The sine values ​​at all moments are arranged in chronological order to form a continuously changing sine waveform sequence, which is the auxiliary force signal.

[0119] In summary, by controlling the energy of the auxiliary force signal within a fixed proportion of the amplitude of the basic driving force command, the superposition of the signals is prevented from exceeding the output limit of the joint servo driver. This ensures that the amplitude of the composite driving force command remains within the safe linear operating range of the driver in a marine salt spray corrosion environment, preventing torque saturation or actuator damage caused by signal overdrive.

[0120] In summary, setting the frequency of the auxiliary force signal in the high-frequency region above the original cutoff frequency of the joint servo enables the auxiliary force signal to excite the rapid dynamic characteristics of the robot's mechanical system that are prone to change in the marine salt spray corrosion environment, such as the high-frequency component of joint seal friction and the surface micro-motion mode caused by corrosion. This provides excitation data rich in high-frequency information for subsequent identification of real dynamic parameters.

[0121] In summary, a sinusoidal waveform with a single frequency and controllable amplitude is generated as an auxiliary excitation. This sinusoidal signal is linearly superimposed with the basic driving force command to form a composite driving force command that includes both conventional operating torque and continuous periodic excitation. This allows the rescue robot to continuously generate identifiable dynamic responses while performing normal movements, thereby obtaining the mechanical response data required for calculating real dynamic parameters without interrupting operations. This is particularly suitable for application scenarios where offline identification cannot be performed in marine salt spray corrosion environments.

[0122] In this embodiment of the invention, when the auxiliary force signal is superimposed with the basic driving force command to obtain the composite driving force command for the target process, it is specifically used for:

[0123] The values ​​of the basic driving force command and the auxiliary force signal at the sampling time are respectively used as the basic driving force time sequence and the auxiliary force time sequence of the target process;

[0124] The value at the current moment in the basic driving force time series is added to the value at the same moment in the auxiliary force time series to obtain the composite driving force value of the target process.

[0125] The composite driving force values ​​are arranged in chronological order to obtain the composite driving force command for the target process.

[0126] Specifically, the driving torque value corresponding to each moment is extracted from the basic driving force command according to the order of sampling time, and these values ​​are arranged into a sequence, which is named the basic driving force timing sequence.

[0127] Specifically, at the same sampling time, the driving torque value at that time is taken from the basic driving force time series, and the auxiliary force value at that time is taken from the auxiliary force time series. The two values ​​are directly added together, and the result of the addition operation is the composite driving force value at that time.

[0128] Specifically, in ascending order of sampling times, the composite driving force value calculated at each sampling time is sequentially placed into a new sequence. Each element of the sequence corresponds to the composite driving force value at a given time, and the entire sequence is the composite driving force command.

[0129] Furthermore, the sinusoidal excitation signal values ​​corresponding to each moment are extracted from the auxiliary force signal in the same sampling time sequence, and these values ​​are arranged into another sequence, which is named the auxiliary force time sequence.

[0130] In summary, by discretizing the basic driving force command and auxiliary force signal in the continuous time domain according to the same sampling time, two strictly time-aligned numerical sequences are generated. This completely eliminates the timing misalignment that may be caused by electromagnetic interference or communication delay in the marine salt spray corrosion environment, providing a synchronous numerical reference for subsequent time-by-time superposition and ensuring the accuracy and repeatability of the composite process.

[0131] In summary, by linearly superimposing the torque values ​​from two independent sources at the same sampling time, the composite driving force value retains the trajectory tracking capability of the basic driving force command while adding the high-frequency excitation component of the auxiliary force signal. The two are numerically fused into a resultant torque value without conflict, enabling the rescue robot to continuously generate dynamic responses for parameter identification while performing normal operations in the corrosive environment of marine salt spray, and the superposition process does not introduce any nonlinear distortion.

[0132] In summary, the composite driving force values ​​calculated at each sampling moment are organized into a complete instruction sequence in ascending order of time. This sequence is directly used as the input command for the joint servo drive, ensuring the continuity and causality of the driving commands on the time axis. This allows the composite driving force command to smoothly output the resultant torque value at each moment in a smooth sequence even if the basic driving force command changes instantaneously due to external disturbances in a marine salt spray corrosion environment, thus avoiding driving abnormalities caused by command jumps or out-of-order sequences.

[0133] In this embodiment of the invention, when using the actual driving torque generated by the composite driving force command as mechanical response data, and calculating the true dynamic parameters of the target process based on the actual driving torque sequence of the mechanical response data and the kinematic quantities of the comprehensive dataset, it is specifically used for:

[0134] Arrange the actual driving torques generated by the composite driving force command according to the sampling time to obtain the mechanical response data of the target process;

[0135] Align the mechanical response data with the current velocity sequence and current acceleration sequence of the comprehensive dataset according to the sampling time to obtain the associated sequence of the target process;

[0136] The actual dynamic parameters of the target process are calculated based on the associated sequence and the number of joint degrees of freedom and link geometry of the rescue robot during the target process.

[0137] Specifically, during the process of the rescue robot executing the composite driving force command, the actual driving torque value of the joint at each sampling moment is read by the joint torque sensor.

[0138] Specifically, at the same sampling time, the actual driving torque value at that time is extracted from the mechanical response data, and the current velocity value at that time is extracted from the current velocity sequence of the comprehensive dataset.

[0139] Specifically, firstly, a system of linear equations is established based on the number of joint degrees of freedom of the rescue robot and the geometric dimensions of each link. The unknowns of this system of equations are the actual dynamic parameters. Each equation in the system of equations consists of the actual driving torque, current velocity, and current acceleration at a sampling moment in the associated sequence. Then, a frequency-domain weighted diagonal matrix is ​​constructed, and the main diagonal elements of this matrix are assigned different weights according to the frequency domain importance of each sampling moment. Finally, a diagonal regularization matrix is ​​constructed, and the main diagonal elements of this matrix are small positive numbers used for stable solutions.

[0140] Furthermore, these values ​​are recorded and arranged in sequence according to the order of sampling time, and this sequence is the mechanical response data.

[0141] Furthermore, the current acceleration value at that moment is extracted from the current acceleration sequence of the comprehensive dataset, and these three values ​​are used as a combined data tuple. All the combined data tuples at all sampling moments are arranged into a new sequence in chronological order, which is the associated sequence.

[0142] Furthermore, an adaptive compromise coefficient is set to balance the contributions between the original equations and the factory-nominal dynamic parameter vector. The factory-nominal dynamic parameter vector is used as a priori value. Then, the transpose of the associated sequence construction matrix is ​​calculated. This transpose is multiplied by the frequency domain weighted diagonal matrix, and then multiplied by the associated sequence construction matrix to obtain a square matrix. This square matrix is ​​added to the result of multiplying the adaptive compromise coefficient by the diagonal regularization matrix to obtain a new invertible square matrix. At the same time, the result of multiplying the transpose of the associated sequence construction matrix by the frequency domain weighted diagonal matrix and then by the actual driving torque vector is calculated. This result is added to the result of multiplying the adaptive compromise coefficient by the diagonal regularization matrix and then by the aforementioned factory-nominal dynamic parameter vector to obtain the right-hand vector. Finally, this system of linear equations is solved, and the solution vector obtained is the true dynamic parameters.

[0143] In summary, by accurately recording the actual torque values ​​of the joints after the rescue robot actually executes the compound driving force command in chronological order, a complete set of mechanical response data is formed. This data directly reflects the actual dynamic characteristics of the mechanical system under marine salt spray corrosion environment, avoiding the errors caused by using theoretical model estimation, and providing high-fidelity raw input for subsequent calculation of real dynamic parameters.

[0144] In summary, by binding the actual driving torque, current velocity, and current acceleration into a combined tuple at the same sampling moment, and then forming an associated sequence of tuples at all moments in chronological order, the driving torque at each moment can be mapped one-to-one with the corresponding velocity and acceleration. This eliminates model identification distortion caused by data misalignment, and ensures that the column vectors of the associated sequence construction matrix are strictly consistent with the actual motion state in the marine salt spray corrosion environment.

[0145] In summary, by utilizing the velocity, acceleration, and actual driving torque at each moment in the associated sequence, combined with the kinematic constraints determined by the inherent joint degrees of freedom and link geometry of the rescue robot, the current true inertial, Coriolis force, and gravity parameters are obtained by solving a set of linear equations. This ensures that the calculated true dynamic parameters fully reflect the actual physical effects such as changes in mass distribution, increased joint friction, and sealing resistance caused by marine salt spray corrosion, providing accurate force control gain coefficients for subsequent reconstruction of the feedforward feedback composite loop.

[0146] In this embodiment of the invention, the formula for calculating the actual dynamic parameters is specifically used for:

[0147]

[0148] in, These are the actual dynamic parameters. Construct a matrix for the associated sequence. This is the matrix transpose operator. For frequency domain weighted diagonal matrix, For adaptive compromise coefficients, It is a diagonal regularized matrix. This is the actual driving torque vector. This is the vector of nominal dynamic parameters specified by the manufacturer.

[0149] Specifically, the associated sequence construction matrix is ​​a two-dimensional array formed by arranging the current velocity and acceleration values ​​at each sampling moment in the associated sequence according to the number of joint degrees of freedom and link geometry of the rescue robot. The frequency domain weighted diagonal matrix is ​​a diagonal array constructed by the control engineer by assigning different weights to each frequency component based on the frequency domain distribution characteristics of the actual driving torque vector. The adaptive compromise coefficient is a dynamically adjusted value between zero and one, and its value is determined in real time according to the degree of deviation between the current identification process and the factory nominal dynamic parameter vector. The diagonal regularization matrix is ​​a square array with small positive numbers on the main diagonal to ensure the stability of matrix inversion operations. The actual driving torque vector is a one-dimensional array formed by arranging the actual driving torque values ​​at all sampling moments in the mechanical response data in order. The factory nominal dynamic parameter vector is a one-dimensional array formed by arranging all parameter values ​​in the dynamic parameter table calibrated when the rescue robot leaves the factory in a fixed order. The real dynamic parameters are the solution vectors finally output by the calculation formula.

[0150] Furthermore, the significance of this calculation formula lies in generating a transpose matrix from the associated sequence construction matrix using the matrix transpose operator, multiplying it by the frequency domain weighted diagonal matrix, and then multiplying it by the original associated sequence construction matrix to obtain a square matrix. This square matrix is ​​then inverted by adding the result of the adaptive compromise coefficient multiplied by the diagonal regularization matrix, and multiplied by the product of the frequency domain weighted diagonal matrix, the transpose of the associated sequence construction matrix, and the actual driving torque vector. Simultaneously, the result of the adaptive compromise coefficient multiplied by the diagonal regularization matrix and the aforementioned factory nominal dynamic parameter vector is added. Finally, an optimal parameter vector that simultaneously satisfies the actual driving torque measurement value and the factory prior knowledge is obtained, thereby achieving a balance between frequency domain weighted least squares estimation and regularization prior constraints. This ensures that the calculated true dynamic parameters are both consistent with the actual dynamic response characteristics of the current robot and do not deviate too far from the factory nominal parameters.

[0151] In summary, when the adaptive compromise coefficient approaches zero, the calculation formula degenerates into a pure frequency domain weighted least squares estimation. At this time, the true dynamic parameters are completely determined by the associated sequence construction matrix and the actual driving torque vector. When the adaptive compromise coefficient approaches one, the calculation formula strongly depends on the factory nominal dynamic parameter vector. At this time, the true dynamic parameters are pulled towards the factory nominal value. As the value of the main diagonal element in the diagonal regularization matrix increases, the stability of matrix inversion improves, but the deviation of parameter estimation also increases accordingly. As the weight of the low-frequency part in the frequency domain weighted diagonal matrix increases, the calculation formula will pay more attention to the fitting accuracy of the low-frequency component in the actual driving torque vector while sacrificing the fitting accuracy of the high-frequency component.

[0152] In this embodiment of the invention, when reconstructing the feedforward-feedback composite loop using the actual dynamic parameters as force control gain coefficients to obtain the final control command for the target process, it is specifically used for:

[0153] The current feedback gain of the target process is obtained by replacing the factory dynamic parameters of the target process with the actual dynamic parameter set.

[0154] Based on the current feedback gain, the expected acceleration sequence of the machine's expected trajectory in the target process is amplified to obtain the current feedforward torque of the target process. Based on the current feedback gain, the position motion error sequence and velocity motion error sequence of the target process are adjusted to obtain the current feedback torque of the target process.

[0155] The current feedforward torque and the current feedback torque are combined to obtain the current feedforward-feedback composite loop of the target process, and the driving torque sequence output by the current feedforward-feedback composite loop is used as the final control command of the target process.

[0156] Specifically, inertial parameters, Coriolis force parameters, and gravity parameters are extracted from the calculated set of real dynamic parameters. These three parameters are used to cover the original factory inertial parameters, factory Coriolis force parameters, and factory gravity parameters, respectively. The three covered parameters together constitute the current feedback gain.

[0157] Specifically, the inertia term parameter in the current feedback gain is used as the amplification factor and multiplied by each value of the expected acceleration sequence in the machine's expected trajectory. The product result, arranged in chronological order, is the current feedforward torque.

[0158] Specifically, at each sampling moment, the value of the current feedforward torque is directly added to the value of the current feedback torque to obtain the total driving torque value at that moment. The total driving torque values ​​at all moments are arranged into a sequence in chronological order. This sequence is the output of the current feedforward feedback composite loop. This output sequence is transmitted as a command to the joint servo drive of the rescue robot. This sequence is the final control command.

[0159] Furthermore, the inertial term parameter and Coriolis force term parameter in the current feedback gain are multiplied by each value of the velocity motion error sequence, and the gravity term parameter in the current feedback gain is multiplied by each value of the position motion error sequence. The results of these three multiplications are then added together at the same time. The sequence obtained by adding all the results at all times is the current feedback torque.

[0160] In summary, the real dynamic parameter set obtained by actual measurement completely covers the original factory calibration parameters, so that the current feedback gain can accurately reflect the real inertia, Coriolis force and gravity characteristics of the rescue robot in the marine salt spray corrosion environment. This avoids the model mismatch problem caused by environmental erosion and mechanical wear of the factory parameters, thus providing a gain coefficient that is completely matched with the current physical system for subsequent torque calculation.

[0161] In summary, by independently adjusting the desired acceleration sequence, position motion error sequence, and velocity motion error sequence using the updated current feedback gain, the generated current feedforward torque and current feedback torque are both based on real dynamic parameters. This ensures that the feedforward compensation accurately corresponds to the actual inertial requirements, and the feedback compensation accurately corresponds to the actual damping and elastic characteristics, significantly reducing trajectory tracking errors in marine salt spray corrosion environments.

[0162] In summary, by adding the current feedforward torque and the current feedback torque based on the actual dynamic parameters at the same time, a fully updated current feedforward-feedback composite loop is constructed. The driving torque sequence output by this loop simultaneously includes the accurate feedforward nominal torque and the real-time error compensation torque. This enables the rescue robot to adaptively resist parameter perturbations and external disturbances when performing tasks in the corrosive environment of marine salt spray, thus achieving high-precision robust control.

[0163] Compared with the prior art, the present invention has the following beneficial effects:

[0164] 1. Based on the technical solution described in the document, the adaptive control method for this marine salt spray corrosion-resistant rescue robot superimposes the auxiliary force signal and the basic driving force command to form a composite driving force command. The actual driving torque generated by the composite driving force command is then aligned with the kinematic quantities in the comprehensive dataset to calculate the true dynamic parameters. These true dynamic parameters replace the factory-set dynamic parameters to reconstruct the feedforward feedback composite loop. This method enables the force control gain coefficient upon which the final control command is based to reflect in real time the actual physical effects caused by marine salt spray corrosion, such as increased joint sealing friction, changes in mass distribution, and altered lubrication characteristics. This significantly improves the matching accuracy between the driving torque sequence and the robot's actual dynamic requirements, ensuring that the control bandwidth of the joint servo is effectively utilized when the rescue robot performs tasks in a corrosive environment. The high-frequency excitation of the auxiliary force signal can continuously excite and identify parameter perturbations caused by corrosion, avoiding torque output deviations due to model mismatch.

[0165] 2. The superposition of the basic driving force command and the auxiliary force signal in this method ensures that the amplitude of the composite driving force command always remains within the safe linear operating range of the joint servo actuator. Simultaneously, the frequency of the auxiliary force signal is set to the joint servo cutoff frequency, which is offset towards higher frequencies, allowing the sinusoidal excitation signal to specifically elicit the rapidly changing dynamic characteristics that are prone to change in corrosive environments. Through the construction of associated sequences and frequency-domain weighted identification of real dynamic parameters, this method obtains mechanical response data that is strictly aligned with the current motion state. Furthermore, the feedforward-feedback composite loop reconstructed based on the real dynamic parameters can simultaneously output accurate current feedforward torque and current feedback torque, merging them into the final control command. This method enables the control system to autonomously track and compensate for time-varying parameter perturbations in a marine salt spray corrosion environment, improving trajectory tracking accuracy, joint motion stability, and long-term operational reliability.

[0166] like Figure 2 The diagram shown is a functional block diagram of an adaptive control system for a marine salt spray-resistant rescue robot provided in an embodiment of the present invention.

[0167] The adaptive control system 100 of the marine salt spray corrosion-resistant rescue robot described in this invention can be installed in an electronic device. Depending on the functions implemented, the adaptive control system 100 of the marine salt spray corrosion-resistant rescue robot may include a comprehensive data module 101, a basic driving force command module 102, an auxiliary force signal module 103, a composite driving force command module 104, a real dynamic parameter module 105, and a final control command module 106. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and is stored in the memory of the electronic device.

[0168] In this embodiment, the functions of each module / unit are as follows:

[0169] The integrated data module uses the current machine motion state and the machine's desired trajectory in the target process as the integrated dataset of the target process;

[0170] The basic driving force command module constructs a feedforward feedback composite loop for the target process based on the factory power parameters of the target process, and generates the basic driving force command for the target process based on the comprehensive dataset and the feedforward feedback composite loop.

[0171] An auxiliary force signal module sets the auxiliary force signal for the target process based on the amplitude of the basic driving force command and the control bandwidth of the joint servo during the target process.

[0172] The composite driving force command module superimposes the auxiliary force signal with the basic driving force command to obtain the composite driving force command for the target process;

[0173] The real dynamics parameter module uses the actual driving torque generated by the composite driving force command as mechanical response data, and calculates the real dynamics parameters of the target process based on the actual driving torque sequence of the mechanical response data and the kinematic quantities of the comprehensive dataset.

[0174] The final control command module reconstructs the feedforward feedback composite loop using the actual dynamic parameters as force control gain coefficients to obtain the final control command for the target process.

[0175] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0176] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0177] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0178] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0179] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0180] 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 it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An adaptive control method for a rescue robot resistant to marine salt spray corrosion, characterized in that, The method includes: The current machine motion state and the machine's desired trajectory in the target process are used as the comprehensive dataset of the target process; Based on the factory power parameters of the target process, a feedforward feedback composite loop of the target process is constructed, and based on the comprehensive dataset and the feedforward feedback composite loop, the basic driving force command of the target process is generated. The auxiliary force signal for the target process is set based on the amplitude of the basic driving force command and the control bandwidth of the joint servo during the target process. The auxiliary force signal is superimposed with the basic driving force command to obtain the composite driving force command for the target process; The actual driving torque generated by the composite driving force command is used as mechanical response data, and the actual driving torque sequence of the mechanical response data and the kinematics of the comprehensive dataset are used to calculate the true dynamic parameters of the target process. The feedforward-feedback composite loop is reconstructed using the actual dynamic parameters as force control gain coefficients to obtain the final control command for the target process.

2. The adaptive control method for the marine salt spray corrosion-resistant rescue robot as described in claim 1, characterized in that, The step of using the current machine motion state and the machine's desired trajectory as the comprehensive dataset of the target process includes: The current position, current velocity, and current acceleration of the rescue robot's moving joints are collected during the target process to obtain the current robot motion state during the target process; Linear interpolation is performed between the current position of the rescue robot and the target position during the target process to obtain the expected position sequence at the end of the target process. Successive time derivatives are then performed on the expected position sequence at the end to obtain the expected velocity sequence and expected acceleration sequence of the target process. The expected position sequence, the expected velocity sequence, and the expected acceleration sequence are used together as the expected machine trajectory of the target process. Sampling times are added to the expected machine trajectory and the current machine motion state to obtain the comprehensive dataset of the target process.

3. The adaptive control method for the marine salt spray corrosion-resistant rescue robot as described in claim 1, characterized in that, The process involves constructing a feedforward-feedback composite loop for the target process based on the factory-set power parameters, and generating a basic driving force command for the target process based on the comprehensive dataset and the feedforward-feedback composite loop, including: Separate the factory inertial parameters, Coriolis force parameters, and gravity parameters from the factory dynamic parameters of the target process; Based on the factory inertial parameters, the Coriolis force parameters, and the gravity parameters, the torque components of the kinematic quantity sequence in the comprehensive dataset are adjusted to obtain the feedforward feedback composite loop of the target process. The desired acceleration of the comprehensive dataset is introduced into the feedforward path of the feedforward-feedback composite loop, and the motion error between the current machine motion state and the desired machine trajectory of the comprehensive dataset is introduced into the feedback path of the feedforward-feedback composite loop to obtain the feedforward driving force component and the feedback driving force component of the target process. The feedforward driving force component and the feedback driving force component are combined to obtain the driving torque sequence of the target process, and the driving torque sequence is used as the basic driving force command of the target process.

4. The adaptive control method for the marine salt spray corrosion-resistant rescue robot as described in claim 3, characterized in that, The step of adjusting the torque components of the kinematic quantity sequence in the comprehensive dataset based on the factory inertial term parameters, the Coriolis force term parameters, and the gravity term parameters to obtain the feedforward feedback composite loop of the target process includes: Using the factory inertial term parameters as feedforward gain, the desired acceleration sequence of the kinematic quantity sequence is amplified to obtain the feedforward gain channel of the target process. The current velocity sequence and the desired velocity sequence, and the current position sequence and the desired position sequence in the kinematic quantity sequence are respectively subjected to difference processing to obtain the velocity motion error sequence and the position motion error sequence of the target process. The factory inertial term parameters are used as inertial term feedback gain, the Coriolis force term parameters are used as Coriolis force term feedback gain, and the gravity term parameters are used as gravity term feedback gain. The velocity motion error sequence and the position motion error sequence are amplified to obtain the inertial term feedback gain channel, the Coriolis force term feedback gain channel and the gravity term feedback gain channel of the target process. The inertial term feedback gain channel, the Coriolis force term feedback gain channel, and the gravity term feedback gain channel are merged to obtain the feedback gain channel of the target process. The feedforward gain channel and the feedback gain channel are connected in parallel to obtain the feedforward-feedback composite loop of the target process.

5. The adaptive control method for the marine salt spray corrosion-resistant rescue robot as described in claim 1, characterized in that, The auxiliary force signal for the target process, based on the amplitude of the basic driving force command and the control bandwidth of the joint servo during the target process, is set, including: The peak amplitude of the basic driving force command is attenuated proportionally to obtain the amplitude of the auxiliary force signal for the target process. The cutoff frequency of the joint servo in the target process is shifted to a higher frequency direction to obtain the frequency of the auxiliary force signal in the target process; A sinusoidal excitation signal is generated with the amplitude of the auxiliary force signal as the amplitude and the frequency of the auxiliary force signal as the frequency, and the sinusoidal excitation signal is used as the auxiliary force signal for the target process.

6. The adaptive control method for the marine salt spray corrosion-resistant rescue robot as described in claim 1, characterized in that, The step of superimposing the auxiliary force signal with the basic driving force command to obtain the composite driving force command for the target process includes: The values ​​of the basic driving force command and the auxiliary force signal at the sampling time are respectively used as the basic driving force time sequence and the auxiliary force time sequence of the target process; The value at the current moment in the basic driving force time series is added to the value at the same moment in the auxiliary force time series to obtain the composite driving force value of the target process. The composite driving force values ​​are arranged in chronological order to obtain the composite driving force command for the target process.

7. The adaptive control method for the marine salt spray corrosion-resistant rescue robot as described in claim 1, characterized in that, The step of using the actual driving torque generated by the composite driving force command as mechanical response data, and calculating the true dynamic parameters of the target process based on the actual driving torque sequence of the mechanical response data and the kinematic quantities of the comprehensive dataset, includes: The actual driving torques generated by the composite driving force command are arranged according to the sampling time to obtain the mechanical response data of the target process; Align the mechanical response data with the current velocity sequence and current acceleration sequence of the comprehensive dataset according to the sampling time to obtain the associated sequence of the target process; The actual dynamic parameters of the target process are calculated based on the associated sequence and the number of joint degrees of freedom and link geometry of the rescue robot during the target process.

8. The adaptive control method for the marine salt spray corrosion-resistant rescue robot as described in claim 7, characterized in that, The formulas for calculating the actual dynamic parameters include: ； in, For the actual dynamic parameters, Construct a matrix for the associated sequence. This is the matrix transpose operator. For frequency domain weighted diagonal matrices, For adaptive compromise coefficients, It is a diagonal regularized matrix. This is the actual driving torque vector. This is the vector of nominal dynamic parameters as specified by the manufacturer.

9. The adaptive control method for the marine salt spray corrosion-resistant rescue robot as described in claim 1, characterized in that, The process of reconstructing the feedforward-feedback composite loop using the actual dynamic parameters as force control gain coefficients to obtain the final control command for the target process includes: The current feedback gain of the target process is obtained by replacing the factory dynamic parameters of the target process with the actual dynamic parameter set. Based on the current feedback gain, the expected acceleration sequence of the machine's expected trajectory in the target process is amplified to obtain the current feedforward torque of the target process. Based on the current feedback gain, the position motion error sequence and velocity motion error sequence of the target process are adjusted to obtain the current feedback torque of the target process. The current feedforward torque and the current feedback torque are combined to obtain the current feedforward-feedback composite loop of the target process, and the driving torque sequence output by the current feedforward-feedback composite loop is used as the final control command of the target process.

10. An adaptive control system for a marine salt spray corrosion-resistant rescue robot, used to implement the adaptive control method for a marine salt spray corrosion-resistant rescue robot according to any one of claims 1-9, characterized in that, The system includes: The integrated data module uses the current machine motion state and the machine's desired trajectory in the target process as the integrated dataset of the target process; The basic driving force command module constructs a feedforward feedback composite loop for the target process based on the factory power parameters of the target process, and generates the basic driving force command for the target process based on the comprehensive dataset and the feedforward feedback composite loop. An auxiliary force signal module sets the auxiliary force signal for the target process based on the amplitude of the basic driving force command and the control bandwidth of the joint servo during the target process. The composite driving force command module superimposes the auxiliary force signal with the basic driving force command to obtain the composite driving force command for the target process; The real dynamics parameter module uses the actual driving torque generated by the composite driving force command as mechanical response data, and calculates the real dynamics parameters of the target process based on the actual driving torque sequence of the mechanical response data and the kinematic quantities of the comprehensive dataset. The final control command module reconstructs the feedforward feedback composite loop using the actual dynamic parameters as force control gain coefficients to obtain the final control command for the target process.

Images