A dynamic stability anti-overturning device for intelligent inspection robot dog

Abstract

The present application relates to the technical field of intelligent inspection equipment, especially to a dynamic stability anti-overturning device for intelligent inspection robot dog; technical problem: the existing anti-overturning technology mostly depends on simple sensor detection and reactive control, lacks predictability and initiative, and cannot effectively respond to dynamic changes in complex environment; technical scheme: a dynamic stability anti-overturning device for intelligent inspection robot dog, comprising a posture detection module, a posture prediction module, a threshold comparison and alarm module, a control instruction issuing module and a posture adjustment execution module; compared with the anti-overturning technology under the prior art, the present application mostly depends on simple sensor detection and reactive control, lacks predictability and initiative, and cannot effectively respond to dynamic changes in complex environment; through real-time detection, prediction and active control, the present application can discover and respond to the overturning risk of intelligent inspection robot dog in advance, and significantly improve its stability and safety in complex environment.

Description

Technical Field

[0001] This invention relates to the field of intelligent inspection equipment technology, and in particular to a dynamic stabilization and anti-tipping device for an intelligent inspection robot dog. Background Technology

[0002] With the development of intelligent technology, intelligent inspection robot dogs have been widely used in industries such as power, petrochemical, and mining. These robot dogs can perform inspection tasks in complex and dangerous environments, thereby improving work efficiency and reducing personnel risks. However, due to the often complex and changeable working environment, intelligent inspection robot dogs may encounter various unstable factors during their movement, leading to the risk of tipping over. Most existing anti-tipping technologies rely on simple sensor detection and reactive control, lacking predictability and initiative, and cannot effectively cope with dynamic changes in complex environments. Summary of the Invention

[0003] To overcome the fact that most existing anti-tipping technologies rely on simple sensor detection and reactive control, lacking predictability and initiative, and unable to effectively cope with dynamic changes in complex environments.

[0004] The technical solution of this invention is: a dynamic stabilization and anti-tipping device for an intelligent inspection robot dog, comprising:

[0005] The attitude detection module is responsible for acquiring the attitude information of the intelligent inspection robot dog in real time, including tilt angle, angular velocity, and acceleration.

[0006] The attitude prediction module is used to predict the attitude state of the intelligent inspection robot dog using a recurrent neural network algorithm.

[0007] The threshold comparison and alarm module is used to compare the output of the attitude prediction module with the preset threshold to determine whether the intelligent inspection robot dog has a risk of tipping over.

[0008] The control command issuing module is used to calculate the control command required to adjust the posture based on the current posture information and prediction results of the intelligent inspection robot dog after receiving an alarm.

[0009] The attitude adjustment execution module is responsible for executing the control commands generated by the control command issuing module to adjust the attitude of the intelligent inspection robot dog.

[0010] Preferably, the attitude detection module is responsible for acquiring the attitude information of the intelligent inspection robot dog in real time, including tilt angle, angular velocity, and acceleration; the attitude prediction module uses a recurrent neural network algorithm to predict the attitude state of the intelligent inspection robot dog; the threshold comparison and alarm module compares the output of the attitude prediction module with a preset threshold to determine whether the intelligent inspection robot dog has a risk of tipping over; after receiving an alarm, the control command issuing module calculates the control command required to adjust the attitude based on the current attitude information and prediction results of the intelligent inspection robot dog; and the attitude adjustment execution module is responsible for executing the control command generated by the control command issuing module to adjust the attitude of the intelligent inspection robot dog.

[0011] Preferably, the attitude detection module includes a sensor unit, a data preprocessing unit, a data fusion algorithm unit, and a communication interface unit. The sensor unit integrates a gyroscope, accelerometer, and vision sensor devices to collect the attitude information of the intelligent inspection robot dog in real time, including tilt angle, angular velocity, acceleration, and visual images. The data preprocessing unit filters and denoises the raw data collected by the sensor unit to improve the accuracy and reliability of the data. The data fusion unit uses a data fusion algorithm to fuse data from multiple sensors to obtain more accurate and comprehensive attitude information. The communication interface unit transmits the processed attitude information to other modules of the device in real time.

[0012] Preferably, through sensor integration and data fusion algorithms, the posture information of the intelligent inspection robot dog can be provided in real time and accurately, providing strong data support for the device's dynamic stability and anti-tipping function.

[0013] Preferably, the attitude prediction module includes a data receiving unit, an RNN model unit, a prediction result output unit, and a model update unit. The data receiving unit receives real-time attitude information transmitted by the attitude detection module, including tilt angle, angular velocity, and acceleration. The RNN model unit uses a recurrent neural network algorithm to construct a trained RNN model, predicts the received attitude information, and obtains the attitude change trend over a future period. The prediction result output unit outputs the prediction results of the RNN model to other modules of the system for subsequent overturning risk assessment. The model update unit continuously optimizes and updates the RNN model based on feedback data and new attitude information from actual applications to improve prediction accuracy and adaptability.

[0014] Preferably, the attitude prediction module includes the following steps during operation:

[0015] S101: The data receiving unit receives attitude information transmitted by the attitude detection module in real time, including tilt angle, angular velocity, and acceleration;

[0016] S102: Organize historical data into time series format and use it as input to the RNN model;

[0017] S103: Initialize the RNN model, including setting the number of layers, the number of neurons in the hidden layers, and the activation function; initialization also includes setting initial values ​​for the model's weights and biases;

[0018] S104: During the forward propagation process, the RNN model processes the input data one by one in the order of the time series;

[0019] S105: For each time step, the RNN model receives the input data of the current time step and combines it with the hidden state of the previous time step to calculate the hidden state and output of the current time step.

[0020] S106: The RNN model uses mean squared error to calculate the loss between the predicted and actual values; the loss value is propagated back to the model through the backpropagation algorithm to calculate the gradient of each parameter; based on the direction and magnitude of the gradient, the gradient descent optimization algorithm is used to update the model's weights and biases.

[0021] S107: The RNN model is trained by repeatedly performing forward and backward propagation until the loss value converges to a small range; during training, early stopping is used to prevent overfitting and accelerate convergence;

[0022] S108: The preprocessed real-time pose information is input into the trained RNN model unit. The RNN model predicts the pose change trend in the future based on historical pose information and time dependence.

[0023] S109: The prediction results of the RNN model are output to the prediction result output unit and transmitted to other modules of the device in real time.

[0024] S110: Based on feedback data and new pose information, the model update unit continuously optimizes and updates the RNN model, including adjusting model parameters and improving model structure;

[0025] S111: The attitude prediction module continuously receives data, preprocesses it, predicts its attitude, outputs the prediction results, and updates the model, enabling real-time and continuous prediction of the attitude changes of the intelligent inspection robot dog.

[0026] Preferably, by utilizing the RNN algorithm and continuous optimization mechanism, it is possible to achieve high-precision prediction of future posture changes of the intelligent inspection robot dog, providing strong support for the system's dynamic stability and anti-tipping function.

[0027] Preferably, the threshold comparison and alarm module includes a data receiving unit, a threshold storage unit, a comparison judgment unit, and an alarm triggering unit. The data receiving unit receives future attitude change trend data output by the attitude prediction module. The threshold storage unit stores preset thresholds determined based on the intelligent inspection robot's design parameters, working environment, and historical overturning data. The comparison judgment unit compares the received attitude prediction data with the preset thresholds to determine if there is a risk of overturning. The alarm triggering unit triggers an alarm when the comparison judgment unit determines that there is a risk of overturning and sends the alarm information to the control command issuing module. The comparison judgment unit uses a real-time comparison method to ensure that each time new attitude prediction data is received, it can be immediately compared with the threshold. The threshold storage unit can dynamically adjust the preset thresholds according to the robot's actual operating conditions and usage environment to improve the accuracy and practicality of the warning.

[0028] Preferably, the threshold comparison and alarm module includes the following steps during operation:

[0029] S201: The data receiving unit of the threshold comparison and alarm module receives the future attitude change trend data output by the attitude prediction module in real time;

[0030] S202: The comparison and judgment unit reads the preset threshold from the threshold storage unit and compares the received attitude prediction data with the preset threshold in real time;

[0031] S203: The comparison and judgment unit determines whether the intelligent inspection robot dog has a risk of tipping over based on the comparison results; when any one or more of the predicted data exceeds the preset threshold, it is judged that there is a risk of tipping over.

[0032] S204: When a risk of overturning is determined to exist, the alarm triggering unit immediately triggers an alarm; the alarm form may include audible alarm and visual alarm.

[0033] S205: When the alarm is triggered, the alarm triggering unit sends the alarm information to the control command issuing module. The alarm information includes the type and degree of overturning risk and the recommended response measures.

[0034] Preferably, by comparing the posture prediction data with the preset threshold in real time, the risk of tipping over can be detected and warned in a timely manner, providing a strong guarantee for the safe operation of the robot dog.

[0035] Preferably, the control command issuing module includes a data receiving unit, a command calculation unit, and a command sending unit. The data receiving unit is used to receive alarm information sent by the threshold comparison and alarm module, as well as current attitude information and prediction results provided by the attitude detection module and attitude prediction module. The command calculation unit is used to calculate the control command required to adjust the attitude based on the received data, combined with the dynamic model and control algorithm of the intelligent inspection robot dog. The command sending unit is used to send the calculated control command to the attitude adjustment execution module to drive the robot dog to perform attitude adjustment.

[0036] Preferably, the control command issuing module includes the following steps during operation:

[0037] S301: The data receiving unit of the control command issuing module receives in real time the alarm information sent by the threshold comparison and alarm module, as well as the current attitude information and prediction results provided by the attitude detection module and attitude prediction module;

[0038] S302: When an alarm message is received, the instruction calculation unit combines the current posture information, prediction results and the dynamic model of the intelligent inspection robot dog. The instruction calculation unit uses fuzzy control and sliding mode control algorithms to calculate and obtain the control instructions required to adjust the posture. The control instructions include adjusting the robot dog's walking speed, turning angle and posture stabilization device.

[0039] S303: The calculated control commands are transmitted to the command sending unit; the command sending unit uses a high-speed communication protocol to send the control commands to the attitude adjustment execution module;

[0040] S304: After the control command is issued, the control command issuing module continuously monitors the data of the attitude detection module and the attitude prediction module to evaluate the control effect; the command calculation unit adjusts the control command based on the feedback data;

[0041] S305: The control command issuing module continuously collects the robot dog's operating data and feedback information during operation.

[0042] Preferably, the control command issuing module can quickly and accurately calculate the control commands and send them to the attitude adjustment execution module to ensure that the robot dog maintains high stability in complex environments.

[0043] Preferably, the attitude adjustment execution module includes an instruction receiving unit, a drive control unit, a feedback acquisition unit, and a protection mechanism unit. The instruction receiving unit receives control instructions generated by the control instruction issuing module and parses and processes them. The drive control unit controls the operation of the motor and actuators according to the parsed control instructions to adjust the attitude of the intelligent inspection robot dog. The feedback acquisition unit monitors the execution of the attitude adjustment in real time and sends feedback data to the control instruction issuing module for necessary adjustments and optimizations to the control instructions. The protection mechanism unit ensures the safety of the intelligent inspection robot dog by activating a protection mechanism when an abnormal situation is detected during attitude adjustment. The protection mechanism unit includes overload protection, short-circuit protection, and fault diagnosis functions to ensure the safety of the intelligent inspection robot dog during attitude adjustment.

[0044] Preferably, the attitude adjustment execution module includes the following steps when it is working:

[0045] S401: The instruction receiving unit of the attitude adjustment execution module receives control instructions in real time and sends out control instructions generated by the module.

[0046] S402: The received control command is parsed and processed, and converted into a signal that the drive control unit can recognize;

[0047] S403: The drive control unit controls the operation of the motor and actuator according to the parsed control instructions. By adjusting the speed and direction of the motor and the action of the actuator, the posture of the intelligent inspection robot dog can be adjusted.

[0048] S404: During the attitude adjustment process, the feedback acquisition unit monitors the adjustment effect in real time and collects relevant data through sensors;

[0049] S405: The collected data is processed and then sent to the control command issuing module;

[0050] S406: If the protection mechanism detects an abnormality during the attitude adjustment process, it will be activated immediately. The protection mechanism ensures the safety of the intelligent inspection robot dog by cutting off the power and stopping operation.

[0051] S407: The attitude adjustment execution module continuously monitors the attitude information of the intelligent inspection robot dog and issues control commands to the module to make adjustments according to the control commands.

[0052] Preferably, the attitude adjustment execution module can quickly and accurately execute control commands to achieve precise adjustment of the intelligent inspection robot dog's attitude, and ensure the safety and stability of the adjustment process through real-time feedback and protection mechanisms.

[0053] The beneficial effects of this invention are:

[0054] 1. Compared with existing anti-tipping technologies, which mostly rely on simple sensor detection and reactive control, lacking predictability and initiative, and unable to effectively cope with dynamic changes in complex environments, this invention can detect and respond to the tipping risk of intelligent inspection robot dogs in advance through real-time detection, prediction and active control, significantly improving their stability and safety in complex environments.

[0055] 2. By using the RNN algorithm for attitude prediction and combining it with real-time feedback data for model optimization, the system can continuously learn and adapt to different working environments and conditions, enhancing its adaptability and flexibility. By integrating multiple sensors and adopting data fusion algorithms, this invention can provide real-time and accurate attitude information of the intelligent inspection robot dog, providing strong data support for the system's dynamic stability and anti-tipping function.

[0056] 3. By autonomously detecting and responding to the risk of overturning, the reliance on manual intervention is reduced, thus lowering maintenance costs and workload. Attached Figure Description

[0057] Figure 1 The diagram shown is a schematic representation of the working process of the dynamic stabilization and anti-tipping device for an intelligent inspection robot dog according to the present invention.

[0058] Figure 2 The diagram shows the workflow of the attitude prediction module of the dynamic stabilization and anti-tipping device for intelligent inspection robot dogs according to the present invention. Detailed Implementation

[0059] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0060] Please see Figure 1-2 The present invention provides an embodiment: a dynamic stabilization and anti-tipping device for an intelligent inspection robot dog, comprising:

[0061] The attitude detection module is responsible for acquiring the attitude information of the intelligent inspection robot dog in real time, including tilt angle, angular velocity, and acceleration.

[0062] The attitude prediction module is used to predict the attitude state of the intelligent inspection robot dog using a recurrent neural network algorithm.

[0063] The threshold comparison and alarm module is used to compare the output of the attitude prediction module with the preset threshold to determine whether the intelligent inspection robot dog has a risk of tipping over.

[0064] The control command issuing module is used to calculate the control command required to adjust the posture based on the current posture information and prediction results of the intelligent inspection robot dog after receiving an alarm.

[0065] The attitude adjustment execution module is responsible for executing the control commands generated by the control command issuing module to adjust the attitude of the intelligent inspection robot dog.

[0066] Preferably, the attitude detection module is responsible for acquiring the attitude information of the intelligent inspection robot dog in real time, including tilt angle, angular velocity, and acceleration; the attitude prediction module uses a recurrent neural network algorithm to predict the attitude state of the intelligent inspection robot dog; the threshold comparison and alarm module compares the output of the attitude prediction module with a preset threshold to determine whether the intelligent inspection robot dog has a risk of tipping over; after receiving an alarm, the control command issuing module calculates the control command required to adjust the attitude based on the current attitude information and prediction results of the intelligent inspection robot dog; and the attitude adjustment execution module is responsible for executing the control command generated by the control command issuing module to adjust the attitude of the intelligent inspection robot dog.

[0067] Preferably, the attitude detection module includes a sensor unit, a data preprocessing unit, a data fusion algorithm unit, and a communication interface unit. The sensor unit integrates a gyroscope, accelerometer, and vision sensor devices to collect the attitude information of the intelligent inspection robot dog in real time, including tilt angle, angular velocity, acceleration, and visual images. The data preprocessing unit filters and denoises the raw data collected by the sensor unit to improve the accuracy and reliability of the data. The data fusion unit uses a data fusion algorithm to fuse data from multiple sensors to obtain more accurate and comprehensive attitude information. The communication interface unit transmits the processed attitude information to other modules of the device in real time.

[0068] Preferably, through sensor integration and data fusion algorithms, the posture information of the intelligent inspection robot dog can be provided in real time and accurately, providing strong data support for the device's dynamic stability and anti-tipping function.

[0069] Preferably, the attitude prediction module includes a data receiving unit, an RNN model unit, a prediction result output unit, and a model update unit. The data receiving unit receives real-time attitude information transmitted by the attitude detection module, including tilt angle, angular velocity, and acceleration. The RNN model unit uses a recurrent neural network algorithm to construct a trained RNN model, predicts the received attitude information, and obtains the attitude change trend over a future period. The prediction result output unit outputs the prediction results of the RNN model to other modules of the system for subsequent overturning risk assessment. The model update unit continuously optimizes and updates the RNN model based on feedback data and new attitude information from actual applications to improve prediction accuracy and adaptability.

[0070] Preferably, the attitude prediction module includes the following steps during operation:

[0071] S101: The data receiving unit receives attitude information transmitted by the attitude detection module in real time, including tilt angle, angular velocity, and acceleration;

[0072] S102: Organize historical data into time series format and use it as input to the RNN model;

[0073] S103: Initialize the RNN model, including setting the number of layers, the number of neurons in the hidden layers, and the activation function; initialization also includes setting initial values ​​for the model's weights and biases;

[0074] S104: During the forward propagation process, the RNN model processes the input data one by one in the order of the time series;

[0075] S105: For each time step, the RNN model receives the input data of the current time step and combines it with the hidden state of the previous time step to calculate the hidden state and output of the current time step.

[0076] S106: The RNN model uses mean squared error to calculate the loss between the predicted and actual values; the loss value is propagated back to the model through the backpropagation algorithm to calculate the gradient of each parameter; based on the direction and magnitude of the gradient, the gradient descent optimization algorithm is used to update the model's weights and biases.

[0077] S107: The RNN model is trained by repeatedly performing forward and backward propagation until the loss value converges to a small range; during training, early stopping is used to prevent overfitting and accelerate convergence;

[0078] S108: The preprocessed real-time pose information is input into the trained RNN model unit. The RNN model predicts the pose change trend in the future based on historical pose information and time dependence.

[0079] S109: The prediction results of the RNN model are output to the prediction result output unit and transmitted to other modules of the device in real time.

[0080] S110: Based on feedback data and new pose information, the model update unit continuously optimizes and updates the RNN model, including adjusting model parameters and improving model structure;

[0081] S111: The attitude prediction module continuously receives data, preprocesses it, predicts its attitude, outputs the prediction results, and updates the model, enabling real-time and continuous prediction of the attitude changes of the intelligent inspection robot dog.

[0082] Preferably, by utilizing the RNN algorithm and continuous optimization mechanism, it is possible to achieve high-precision prediction of future posture changes of the intelligent inspection robot dog, providing strong support for the system's dynamic stability and anti-tipping function.

[0083] Preferably, the threshold comparison and alarm module includes a data receiving unit, a threshold storage unit, a comparison judgment unit, and an alarm triggering unit. The data receiving unit receives future attitude change trend data output by the attitude prediction module. The threshold storage unit stores preset thresholds determined based on the intelligent inspection robot's design parameters, working environment, and historical overturning data. The comparison judgment unit compares the received attitude prediction data with the preset thresholds to determine if there is a risk of overturning. The alarm triggering unit triggers an alarm when the comparison judgment unit determines that there is a risk of overturning and sends the alarm information to the control command issuing module. The comparison judgment unit uses a real-time comparison method to ensure that each time new attitude prediction data is received, it can be immediately compared with the threshold. The threshold storage unit can dynamically adjust the preset thresholds according to the robot's actual operating conditions and usage environment to improve the accuracy and practicality of the warning.

[0084] Preferably, the threshold comparison and alarm module includes the following steps during operation:

[0085] S201: The data receiving unit of the threshold comparison and alarm module receives the future attitude change trend data output by the attitude prediction module in real time;

[0086] S202: The comparison and judgment unit reads the preset threshold from the threshold storage unit and compares the received attitude prediction data with the preset threshold in real time;

[0087] S203: The comparison and judgment unit determines whether the intelligent inspection robot dog has a risk of tipping over based on the comparison results; when any one or more of the predicted data exceeds the preset threshold, it is judged that there is a risk of tipping over.

[0088] S204: When a risk of overturning is determined to exist, the alarm triggering unit immediately triggers an alarm; the alarm form may include audible alarm and visual alarm.

[0089] S205: When the alarm is triggered, the alarm triggering unit sends the alarm information to the control command issuing module. The alarm information includes the type and degree of overturning risk and the recommended response measures.

[0090] Preferably, by comparing the posture prediction data with the preset threshold in real time, the risk of tipping over can be detected and warned in a timely manner, providing a strong guarantee for the safe operation of the robot dog.

[0091] Preferably, the control command issuing module includes a data receiving unit, a command calculation unit, and a command sending unit. The data receiving unit is used to receive alarm information sent by the threshold comparison and alarm module, as well as current attitude information and prediction results provided by the attitude detection module and attitude prediction module. The command calculation unit is used to calculate the control command required to adjust the attitude based on the received data, combined with the dynamic model and control algorithm of the intelligent inspection robot dog. The command sending unit is used to send the calculated control command to the attitude adjustment execution module to drive the robot dog to perform attitude adjustment.

[0092] Preferably, the control command issuing module includes the following steps during operation:

[0093] S301: The data receiving unit of the control command issuing module receives in real time the alarm information sent by the threshold comparison and alarm module, as well as the current attitude information and prediction results provided by the attitude detection module and attitude prediction module;

[0094] S302: When an alarm message is received, the instruction calculation unit combines the current posture information, prediction results and the dynamic model of the intelligent inspection robot dog. The instruction calculation unit uses fuzzy control and sliding mode control algorithms to calculate and obtain the control instructions required to adjust the posture. The control instructions include adjusting the robot dog's walking speed, turning angle and posture stabilization device.

[0095] S303: The calculated control commands are transmitted to the command sending unit; the command sending unit uses a high-speed communication protocol to send the control commands to the attitude adjustment execution module;

[0096] S304: After the control command is issued, the control command issuing module continuously monitors the data of the attitude detection module and the attitude prediction module to evaluate the control effect; the command calculation unit adjusts the control command based on the feedback data;

[0097] S305: The control command issuing module continuously collects the robot dog's operating data and feedback information during operation.

[0098] Preferably, the control command issuing module can quickly and accurately calculate the control commands and send them to the attitude adjustment execution module to ensure that the robot dog maintains high stability in complex environments.

[0099] Preferably, the attitude adjustment execution module includes an instruction receiving unit, a drive control unit, a feedback acquisition unit, and a protection mechanism unit. The instruction receiving unit receives control instructions generated by the control instruction issuing module and parses and processes them. The drive control unit controls the operation of the motor and actuators according to the parsed control instructions to adjust the attitude of the intelligent inspection robot dog. The feedback acquisition unit monitors the execution of the attitude adjustment in real time and sends feedback data to the control instruction issuing module for necessary adjustments and optimizations to the control instructions. The protection mechanism unit ensures the safety of the intelligent inspection robot dog by activating a protection mechanism when an abnormal situation is detected during attitude adjustment. The protection mechanism unit includes overload protection, short-circuit protection, and fault diagnosis functions to ensure the safety of the intelligent inspection robot dog during attitude adjustment.

[0100] Preferably, the attitude adjustment execution module includes the following steps when it is working:

[0101] S401: The instruction receiving unit of the attitude adjustment execution module receives control instructions in real time and sends out control instructions generated by the module.

[0102] S402: The received control command is parsed and processed, and converted into a signal that the drive control unit can recognize;

[0103] S403: The drive control unit controls the operation of the motor and actuator according to the parsed control instructions. By adjusting the speed and direction of the motor and the action of the actuator, the posture of the intelligent inspection robot dog can be adjusted.

[0104] S404: During the attitude adjustment process, the feedback acquisition unit monitors the adjustment effect in real time and collects relevant data through sensors;

[0105] S405: The collected data is processed and then sent to the control command issuing module;

[0106] S406: If the protection mechanism detects an abnormality during the attitude adjustment process, it will be activated immediately. The protection mechanism ensures the safety of the intelligent inspection robot dog by cutting off the power and stopping operation.

[0107] S407: The attitude adjustment execution module continuously monitors the attitude information of the intelligent inspection robot dog and issues control commands to the module to make adjustments according to the control commands.

[0108] Preferably, the attitude adjustment execution module can quickly and accurately execute control commands to achieve precise adjustment of the intelligent inspection robot dog's attitude, and ensure the safety and stability of the adjustment process through real-time feedback and protection mechanisms.

[0109] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. A dynamic stability anti-overturning device for an intelligent inspection robot dog, characterized in that: Comprise: The posture detection module is responsible for real-time acquisition of the posture information of the intelligent inspection robot dog, including the inclination angle, angular velocity and acceleration; The posture prediction module is used to predict the posture state of the intelligent inspection robot dog by using a recurrent neural network algorithm; The threshold comparison and alarm module compares the output result of the posture prediction module with the preset threshold to determine whether the intelligent inspection robot dog has a risk of overturning; The control instruction issuing module calculates the control instruction required for adjusting the posture according to the current posture information and the prediction result of the intelligent inspection robot dog after receiving the alarm; The posture adjustment execution module is responsible for executing the control instruction generated by the control instruction issuing module to adjust the posture of the intelligent inspection robot dog; The posture adjustment execution module comprises an instruction receiving unit, a drive control unit, a feedback acquisition unit and a protection mechanism unit; the instruction receiving unit is used to receive and analyze and process the control instruction generated by the control instruction issuing module; the drive control unit is used to control the motor and the actuator to run according to the analyzed control instruction, so as to adjust the posture of the intelligent inspection robot dog; the feedback acquisition unit is used to monitor the execution of the posture adjustment in real time and send the feedback data to the control instruction issuing module, so as to adjust and optimize the control instruction as needed; the protection mechanism unit is used to ensure the safety of the intelligent inspection robot dog by starting the protection mechanism when an abnormal condition is detected during the posture adjustment; wherein the protection mechanism unit comprises overload protection, short circuit protection and fault diagnosis function to ensure the safety of the intelligent inspection robot dog during the posture adjustment; The posture adjustment execution module comprises the following steps when working: S401: The instruction receiving unit of the posture adjustment execution module receives the control instruction generated by the control instruction issuing module in real time; S402: The received control instruction is analyzed and processed to be converted into a signal that can be recognized by the drive control unit; S403: The drive control unit controls the running of the motor and the actuator according to the analyzed control instruction, adjusts the speed, steering of the motor and the action of the actuator, and adjusts the posture of the intelligent inspection robot dog; S404: During the posture adjustment, the feedback acquisition unit monitors the adjustment effect in real time and collects relevant data through sensors; S405: The collected data is sent to the control instruction issuing module after processing; S406: During the posture adjustment, if the protection mechanism detects an abnormal condition, the protection mechanism is started immediately; the protection mechanism ensures the safety of the intelligent inspection robot dog by cutting off the power supply and stopping the running mode; S407: The posture adjustment execution module continuously monitors the posture information of the intelligent inspection robot dog and adjusts it according to the control instruction of the control instruction issuing module.

2. The dynamic stability anti-overturning device for the intelligent inspection robot dog according to claim 1, characterized in that: The posture detection module comprises a sensor unit, a data preprocessing unit, a data fusion algorithm unit and a communication interface unit; the sensor unit is integrated with a gyroscope, an accelerometer and a visual sensor device, which is used to collect the posture information of the intelligent inspection robot dog in real time, including the inclination angle, angular velocity, acceleration and visual image; The data preprocessing unit is configured to filter and denoise the raw data collected by the sensor unit, so as to improve the accuracy and reliability of the data; the data fusion unit is configured to fuse the data of multiple sensors by using a data fusion algorithm, so as to obtain more accurate and comprehensive attitude information; and the communication interface unit is configured to transmit the processed attitude information to other modules of the device in real time.

3. The dynamic stability anti-overturning device for the intelligent inspection robot dog according to claim 2, characterized in that: The attitude prediction module comprises a data receiving unit, an RNN model unit, a prediction result output unit and a model updating unit; the data receiving unit is configured to receive real-time attitude information transmitted by the attitude detection module, including a tilt angle, an angular velocity and an acceleration; the RNN model unit is configured to use a recurrent neural network algorithm to construct a trained RNN model, to predict the received attitude information, and to obtain a trend of attitude change in a future period of time; and the prediction result output unit is configured to output the prediction result of the RNN model to other modules of the system, for subsequent roll-over risk assessment. The model updating unit is configured to continuously optimize and update the RNN model according to feedback data and new attitude information in actual application, so as to improve the prediction accuracy and adaptability.

4. The dynamic stability anti-overturning device for the intelligent inspection robot dog according to claim 3, characterized in that: The attitude prediction module comprises the following steps when working: S101: The data receiving unit receives real-time attitude information transmitted by the attitude detection module, including a tilt angle, an angular velocity and an acceleration; S102: Historical data is arranged in the form of a time sequence as input of the RNN model; S103: The RNN model is initialized, including setting the number of layers of the model, the number of neurons of the hidden layer and the activation function; initialization also includes setting initial values for the weights and biases of the model; S104: In the process of forward propagation, the RNN model processes the input data one by one in the order of the time sequence; S105: For each time step, the RNN model receives the input data of the current time step, and combines the hidden state of the previous time step to calculate the hidden state and output of the current time step; S106: The RNN model calculates the loss between the predicted value and the true value by using the mean square error; the loss value is propagated back to the model by using the back propagation algorithm to calculate the gradient of each parameter; the weights and biases of the model are updated by using the gradient descent optimization algorithm according to the direction and size of the gradient; S107: The RNN model is trained by repeatedly performing the processes of forward propagation and back propagation until the loss value converges to a small range; during the training process, the early stopping method is used to prevent overfitting and accelerate convergence; S108: The preprocessed real-time attitude information is input into the trained RNN model unit, and the RNN model predicts the trend of attitude change in a future period of time according to historical attitude information and time dependence; S109: The prediction result of the RNN model is output to the prediction result output unit and transmitted to other modules of the device in real time; S110: According to feedback data and new attitude information, the model updating unit continuously optimizes and updates the RNN model, including adjusting the model parameters and improving the model structure. S111: The attitude prediction module continuously performs data receiving, preprocessing, attitude prediction, prediction result outputting and model updating operations to perform real-time and continuous prediction on the attitude change of the intelligent inspection robot dog.

5. The dynamic stability anti-overturning device for the intelligent inspection robot dog according to claim 2, characterized in that: The threshold comparison and alarm module comprises a data receiving unit, a threshold storage unit, a comparison and judgment unit and an alarm triggering unit. The data receiving unit is configured to receive the future attitude change trend data output by the attitude prediction module. The threshold storage unit is configured to store preset thresholds determined according to the design parameters, working environment and historical overturn data factors of the intelligent inspection robot dog. The comparison and judgment unit is configured to compare the received attitude prediction data with the preset thresholds to determine whether there is an overturn risk. The alarm triggering unit is configured to trigger an alarm when the comparison and judgment unit determines that there is an overturn risk, and send the alarm information to the control instruction issuing module. The comparison and judgment unit adopts a real-time comparison mode to ensure that the received new attitude prediction data can be compared with the thresholds immediately. The threshold storage unit can dynamically adjust the preset thresholds according to the actual operation and use environment of the robot dog to improve the accuracy and practicality of the early warning.

6. The dynamic stability anti-overturning device for the intelligent inspection robot dog according to claim 3, characterized in that: The threshold comparison and alarm module comprises the following steps when working: S201: The data receiving unit of the threshold comparison and alarm module receives the future attitude change trend data output by the attitude prediction module in real time; S202: The comparison and judgment unit reads the preset thresholds from the threshold storage unit and compares the received attitude prediction data with the preset thresholds in real time; S203: The comparison and judgment unit determines whether the intelligent inspection robot dog has an overturn risk according to the comparison result. When any one or more of the prediction data exceeds the preset threshold, it is determined that there is an overturn risk; S204: When it is determined that there is an overturn risk, the alarm triggering unit triggers an alarm immediately. The alarm form includes a sound alarm and a light signal alarm; S205: The alarm triggering unit sends the alarm information to the control instruction issuing module when triggering the alarm. The alarm information includes the type and degree of the overturn risk and the recommended countermeasures.

7. The dynamic stability anti-overturning device for the intelligent inspection robot dog according to claim 6, characterized in that: The control instruction issuing module comprises a data receiving unit, an instruction calculation unit and an instruction sending unit. The data receiving unit is configured to receive the alarm information sent by the threshold comparison and alarm module, and the current attitude information and prediction results provided by the attitude detection module and the attitude prediction module. The instruction calculation unit is configured to calculate the control instruction required for adjusting the attitude according to the received data, in combination with the dynamics model and control algorithm of the intelligent inspection robot dog. The instruction sending unit is configured to send the calculated control instruction to the attitude adjustment execution module to drive the robot dog to adjust the attitude.

8. The dynamic stability anti-overturning device for the intelligent inspection robot dog according to claim 5, characterized in that: The control instruction issuing module comprises the following steps when working: S301: The data receiving unit of the control instruction issuing module receives the alarm information sent by the threshold comparison and alarm module, and the current attitude information and prediction results provided by the attitude detection module and the attitude prediction module in real time; S302: When receiving the alarm information, the instruction calculation unit combines the current attitude information, the prediction result and the dynamics model of the intelligent patrol robot dog, and calculates the control instruction required for adjusting the attitude by using the fuzzy control and sliding mode control algorithms. The control instruction includes adjusting the walking speed, turning angle and attitude stabilizing device of the robot dog; S303: The calculated control instruction is transmitted to the instruction sending unit; the instruction sending unit sends the control instruction to the attitude adjustment execution module by using a high-speed communication protocol; S304: After the control instruction is sent out, the control instruction sending module continuously monitors the data of the attitude detection module and the attitude prediction module to evaluate the control effect; the instruction calculation unit adjusts the control instruction according to the feedback data; S305: The control instruction sending module continuously collects the running data and feedback information of the robot dog during the running process.

Images