Dynamic stable overturn-preventing device for intelligent inspection robot dog

Abstract

The invention relates to the technical field of intelligent inspection equipment, in particular to a dynamic stable overturn-preventing device for an intelligent inspection robot dog. The technical problems that most of the existing anti-overturning technologies depend on simple sensor detection and reactive control, are lack of predictability and initiative and cannot effectively cope with dynamic changes in a complex environment are solved; according to the technical scheme, the dynamic stable anti-overturning device for the intelligent inspection robot dog comprises a posture detection module, a posture prediction module, a threshold comparison and alarm module, a control instruction sending module and a posture adjustment execution module; compared with the prior art that the anti-overturning technology mostly depends on simple sensor detection and reactive control, is lack of predictability and initiative, and cannot effectively cope with dynamic changes in a complex environment; through real-time detection, prediction and active control, the overturning risk of the intelligent inspection robot dog can be found and dealt with in advance, and the stability and safety of the robot dog in a complex environment are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent inspection equipment, and in particular to a dynamic stabilization anti-overturning device for an intelligent inspection robot dog. Background Art

[0002] With the development of intelligent technology, smart patrol robot dogs have been widely used in industries such as electricity, petrochemicals, and mining. These robot dogs can perform patrol tasks in complex and dangerous environments, thereby improving work efficiency and reducing personnel risks. However, since the working environment is often complex and changeable, smart patrol robot dogs may encounter various unstable factors during walking, leading to the risk of overturning. Existing anti-overturning technologies mostly rely on simple sensor detection and reactive control, lack predictability and initiative, and cannot effectively respond to dynamic changes in complex environments. Summary of the invention

[0003] To overcome this problem, existing anti-rollover technologies mostly rely on simple sensor detection and reactive control, lack predictability and initiative, and cannot effectively cope with dynamic changes in complex environments.

[0004] The technical solution of the present invention is: a dynamic stability anti-overturning device for an intelligent inspection robot dog, comprising:

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

[0006] The posture prediction module is used to predict the posture 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 result of the posture prediction module with the preset threshold to determine whether the intelligent inspection robot dog has the risk of overturning;

[0008] The control instruction issuing module is used to calculate the control instructions required to adjust the posture according to the current posture information and prediction results of the intelligent inspection robot dog after receiving the alarm;

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

[0010] Preferably, the posture detection module is responsible for acquiring the posture information of the intelligent patrol robot dog in real time, including the tilt angle, angular velocity, and acceleration; the posture prediction module adopts a recurrent neural network algorithm to predict the posture state of the intelligent patrol robot dog; the threshold comparison and alarm module compares the output result of the posture prediction module with the preset threshold to determine whether the intelligent patrol robot dog has a risk of overturning; after receiving the alarm, the control instruction issuing module calculates the control instructions required to adjust the posture according to the current posture information and prediction results of the intelligent patrol robot dog; the posture adjustment execution module is responsible for executing the control instructions generated by the control instruction issuing module to adjust the posture of the intelligent patrol robot dog.

[0011] Preferably, the posture 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, an accelerometer and a visual sensor device for real-time acquisition of the posture information of the intelligent inspection robot dog, including tilt angle, angular velocity, acceleration and visual image; the data preprocessing unit is used to filter and denoise the raw data collected by the sensor unit to improve the accuracy and reliability of the data; the data fusion unit is used to fuse the data of multiple sensors using a data fusion algorithm to obtain more accurate and comprehensive posture information; the communication interface unit is used to transmit the processed posture 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 dynamic stability and anti-overturning function of the device.

[0013] Preferably, the posture prediction module includes a data receiving unit, an RNN model unit, a prediction result output unit and a model updating unit; the data receiving unit is used to receive real-time posture information transmitted by the posture detection module, including tilt angle, angular velocity and acceleration; the RNN model unit is used to adopt a recurrent neural network algorithm to construct a trained RNN model, predict the received posture information, and obtain the posture change trend in the future; the prediction result output unit is used to output the prediction result of the RNN model to other modules of the system for subsequent overturning risk assessment; the model updating unit is used to continuously optimize and update the RNN model according to feedback data and new posture information in actual applications, so as to improve prediction accuracy and adaptability.

[0014] Preferably, the posture prediction module comprises the following steps when working:

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

[0016] S102: Arrange the historical data into a time series format as the input of the RNN model;

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

[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 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 square error to calculate the loss between the predicted value and the true value; the loss value is propagated back to the model through the back propagation algorithm to calculate the gradient of each parameter; according to the direction and size of the gradient, the gradient descent optimization algorithm is used to update the weights and biases of the model;

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

[0022] S108: The pre-processed real-time posture information is input into the trained RNN model unit, and the RNN model predicts the posture change trend in the future period of time based on the historical posture information and time dependency;

[0023] 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;

[0024] S110: According to the feedback data and the new posture information, the model updating unit continuously optimizes and updates the RNN model, including adjusting the model parameters and improving the model structure;

[0025] S111: The posture prediction module continuously performs data reception, preprocessing, posture prediction, prediction result output and model update operations, and performs real-time and continuous prediction of the posture changes of the intelligent inspection robot dog.

[0026] Preferably, by utilizing the RNN algorithm and continuous optimization mechanism, high-precision prediction of future posture changes of the intelligent inspection robot dog can be achieved, providing strong support for the dynamic stability and anti-overturning function of the system.

[0027] Preferably, the threshold comparison and alarm module includes a data receiving unit, a threshold storage unit, a comparison and judgment unit and an alarm triggering unit; the data receiving unit is used to receive future posture change trend data output by the posture prediction module; the threshold storage unit is used to store preset thresholds determined according to the design parameters, working environment and historical overturning data factors of the intelligent inspection robot dog; the comparison and judgment unit is used to compare the received posture prediction data with the preset threshold to determine whether there is a overturning risk; the alarm triggering unit is used to trigger an alarm when the comparison and judgment unit determines that there is a overturning risk, and send the alarm information to the control instruction issuing module; wherein the comparison and judgment unit adopts a real-time comparison method to ensure that each time new posture prediction data is received, it can be immediately compared with the threshold; the threshold storage unit can dynamically adjust the preset threshold according to the actual operation of the robot dog and the use environment, so as to improve the accuracy and practicality of the early warning.

[0028] Preferably, the threshold comparison and alarm module comprises the following steps when working:

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

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

[0031] S203: The comparison and judgment unit judges whether the intelligent inspection robot dog has a risk of overturning according to the comparison result; when any one or more of the predicted data exceeds a preset threshold, it is judged that there is a risk of overturning;

[0032] S204: When it is determined that there is a capsizing risk, the alarm triggering unit immediately triggers an alarm; the alarm forms include sound alarm and light signal alarm;

[0033] S205: When the alarm is triggered, the alarm triggering unit sends alarm information to the control instruction issuing module, where the alarm information includes the type and degree of the capsizing risk and recommended countermeasures.

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

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

[0036] Preferably, when the control instruction issuing module is working, the following steps are included:

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

[0038] S302: When receiving the alarm information, the instruction calculation unit combines the current posture information, the prediction result and the dynamic model of the intelligent inspection robot dog, and the instruction calculation unit uses the fuzzy control and sliding mode control algorithms to calculate and obtain the control instructions required for adjusting the posture. The control instructions include adjusting the walking speed, steering angle and posture stabilization device of the robot dog;

[0039] S303: The calculated control instruction is transmitted to the instruction sending unit; the instruction sending unit uses a high-speed communication protocol to send the control instruction to the posture adjustment execution module;

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

[0041] S305: The control instruction issuing module continuously collects the operation data and feedback information of the robot dog during operation.

[0042] Preferably, the control instruction issuing module can quickly and accurately calculate the control instruction and send it to the posture adjustment execution module to ensure that the robot dog maintains a high degree of stability in a complex environment.

[0043] Preferably, the posture adjustment execution module includes an instruction receiving unit, a drive control unit, a feedback collection unit and a protection mechanism unit; the instruction receiving unit is used to receive the control instructions generated by the control instruction issuing module, and parse and process them; the drive control unit is used to control the operation of the motor and the actuator according to the parsed control instructions, so as to adjust the posture of the intelligent patrol robot dog; the feedback collection 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 make necessary adjustments and optimizations to the control instructions; the protection mechanism unit is used to ensure the safety of the intelligent patrol robot dog by starting the protection mechanism when an abnormal situation is detected during the posture adjustment process; wherein the protection mechanism unit includes overload protection, short circuit protection, and fault diagnosis functions to ensure the safety of the intelligent patrol robot dog during the posture adjustment process.

[0044] Preferably, the posture adjustment execution module includes the following steps when working:

[0045] 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;

[0046] S402: The received control instruction is parsed and processed and converted into a signal that can be recognized by the drive control unit;

[0047] S403: The drive control unit controls the operation of the motor and the actuator according to the analyzed control instruction, and adjusts the speed and direction of the motor and the action of the actuator to adjust the posture of the intelligent inspection robot dog;

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

[0049] S405: After being processed, the collected data is sent to the control instruction issuing module;

[0050] S406: During the posture adjustment process, if the protection mechanism detects an abnormal situation, the protection mechanism is immediately activated; the protection mechanism ensures the safety of the intelligent inspection robot dog by cutting off the power supply and stopping the operation mode;

[0051] S407: The posture adjustment execution module continuously monitors the posture information of the intelligent inspection robot dog, and makes adjustments according to the control instructions of the control instruction issuing module.

[0052] Preferably, the posture adjustment execution module can execute control instructions quickly and accurately, realize precise adjustment of the posture of the intelligent inspection robot dog, and ensure the safety and stability of the adjustment process through real-time feedback and protection mechanisms.

[0053] Beneficial effects of the present invention:

[0054] 1. Compared with the existing anti-rollover technology, which mostly relies on simple sensor detection and reactive control, lacks predictability and initiative, and cannot effectively respond to dynamic changes in complex environments; the present invention can detect and respond to the rollover risk of the intelligent inspection robot dog in advance through real-time detection, prediction and active control, and significantly improve its stability and safety in complex environments;

[0055] 2. By using the RNN algorithm for posture 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, the present invention can provide the posture information of the intelligent inspection robot dog in real time and accurately, providing strong data support for the dynamic stability and anti-overturning function of the system;

[0056] 3. Through autonomous detection and response to capsizing risks, reliance on manual intervention is reduced, and maintenance costs and workload are reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 Shown is a schematic diagram of the working process of the dynamic stability anti-overturning device for the intelligent inspection robot dog of the present invention;

[0058] Figure 2 Shown is a schematic diagram of the working process of the posture prediction module of the dynamic stability anti-overturning device for the intelligent inspection robot dog of the present invention. DETAILED DESCRIPTION

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

[0060] See also Figure 1-2 The present invention provides an embodiment: a dynamic stability anti-overturning device for an intelligent inspection robot dog, comprising:

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

[0062] The posture prediction module is used to predict the posture 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 result of the posture prediction module with the preset threshold to determine whether the intelligent inspection robot dog has the risk of overturning;

[0064] The control instruction issuing module is used to calculate the control instructions required to adjust the posture according to the current posture information and prediction results of the intelligent inspection robot dog after receiving the alarm;

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

[0066] Preferably, the posture detection module is responsible for acquiring the posture information of the intelligent patrol robot dog in real time, including the tilt angle, angular velocity, and acceleration; the posture prediction module adopts a recurrent neural network algorithm to predict the posture state of the intelligent patrol robot dog; the threshold comparison and alarm module compares the output result of the posture prediction module with the preset threshold to determine whether the intelligent patrol robot dog has a risk of overturning; after receiving the alarm, the control instruction issuing module calculates the control instructions required to adjust the posture according to the current posture information and prediction results of the intelligent patrol robot dog; the posture adjustment execution module is responsible for executing the control instructions generated by the control instruction issuing module to adjust the posture of the intelligent patrol robot dog.

[0067] Preferably, the posture 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, an accelerometer and a visual sensor device for real-time acquisition of the posture information of the intelligent inspection robot dog, including tilt angle, angular velocity, acceleration and visual image; the data preprocessing unit is used to filter and denoise the raw data collected by the sensor unit to improve the accuracy and reliability of the data; the data fusion unit is used to fuse the data of multiple sensors using a data fusion algorithm to obtain more accurate and comprehensive posture information; the communication interface unit is used to transmit the processed posture 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 dynamic stability and anti-overturning function of the device.

[0069] Preferably, the posture prediction module includes a data receiving unit, an RNN model unit, a prediction result output unit and a model updating unit; the data receiving unit is used to receive real-time posture information transmitted by the posture detection module, including tilt angle, angular velocity and acceleration; the RNN model unit is used to adopt a recurrent neural network algorithm to construct a trained RNN model, predict the received posture information, and obtain the posture change trend in the future; the prediction result output unit is used to output the prediction result of the RNN model to other modules of the system for subsequent overturning risk assessment; the model updating unit is used to continuously optimize and update the RNN model according to feedback data and new posture information in actual applications, so as to improve prediction accuracy and adaptability.

[0070] Preferably, the posture prediction module comprises the following steps when working:

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

[0072] S102: Arrange the historical data into a time series format as the input of the RNN model;

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

[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 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 square error to calculate the loss between the predicted value and the true value; the loss value is propagated back to the model through the back propagation algorithm to calculate the gradient of each parameter; according to the direction and size of the gradient, the gradient descent optimization algorithm is used to update the weights and biases of the model;

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

[0078] S108: The pre-processed real-time posture information is input into the trained RNN model unit, and the RNN model predicts the posture change trend in the future period of time based on the historical posture information and time dependency;

[0079] 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;

[0080] S110: According to the feedback data and the new posture information, the model updating unit continuously optimizes and updates the RNN model, including adjusting the model parameters and improving the model structure;

[0081] S111: The posture prediction module continuously performs data reception, preprocessing, posture prediction, prediction result output and model update operations, and performs real-time and continuous prediction of the posture changes of the intelligent inspection robot dog.

[0082] Preferably, by utilizing the RNN algorithm and continuous optimization mechanism, high-precision prediction of future posture changes of the intelligent inspection robot dog can be achieved, providing strong support for the dynamic stability and anti-overturning function of the system.

[0083] Preferably, the threshold comparison and alarm module includes a data receiving unit, a threshold storage unit, a comparison and judgment unit and an alarm triggering unit; the data receiving unit is used to receive future posture change trend data output by the posture prediction module; the threshold storage unit is used to store preset thresholds determined according to the design parameters, working environment and historical overturning data factors of the intelligent inspection robot dog; the comparison and judgment unit is used to compare the received posture prediction data with the preset threshold to determine whether there is a overturning risk; the alarm triggering unit is used to trigger an alarm when the comparison and judgment unit determines that there is a overturning risk, and send the alarm information to the control instruction issuing module; wherein the comparison and judgment unit adopts a real-time comparison method to ensure that each time new posture prediction data is received, it can be immediately compared with the threshold; the threshold storage unit can dynamically adjust the preset threshold according to the actual operation of the robot dog and the use environment, so as to improve the accuracy and practicality of the early warning.

[0084] Preferably, the threshold comparison and alarm module comprises the following steps when working:

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

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

[0087] S203: The comparison and judgment unit judges whether the intelligent inspection robot dog has a risk of overturning according to the comparison result; when any one or more of the predicted data exceeds a preset threshold, it is judged that there is a risk of overturning;

[0088] S204: When it is determined that there is a capsizing risk, the alarm triggering unit immediately triggers an alarm; the alarm forms include sound alarm and light signal alarm;

[0089] S205: When the alarm is triggered, the alarm triggering unit sends alarm information to the control instruction issuing module, where the alarm information includes the type and degree of the capsizing risk and recommended countermeasures.

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

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

[0092] Preferably, when the control instruction issuing module is working, the following steps are included:

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

[0094] S302: When receiving the alarm information, the instruction calculation unit combines the current posture information, the prediction result and the dynamic model of the intelligent inspection robot dog, and the instruction calculation unit uses the fuzzy control and sliding mode control algorithms to calculate and obtain the control instructions required for adjusting the posture. The control instructions include adjusting the walking speed, steering angle and posture stabilization device of the robot dog;

[0095] S303: The calculated control instruction is transmitted to the instruction sending unit; the instruction sending unit uses a high-speed communication protocol to send the control instruction to the posture adjustment execution module;

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

[0097] S305: The control instruction issuing module continuously collects the operation data and feedback information of the robot dog during operation.

[0098] Preferably, the control instruction issuing module can quickly and accurately calculate the control instruction and send it to the posture adjustment execution module to ensure that the robot dog maintains a high degree of stability in a complex environment.

[0099] Preferably, the posture adjustment execution module includes an instruction receiving unit, a drive control unit, a feedback collection unit and a protection mechanism unit; the instruction receiving unit is used to receive the control instructions generated by the control instruction issuing module, and parse and process them; the drive control unit is used to control the operation of the motor and the actuator according to the parsed control instructions, so as to adjust the posture of the intelligent patrol robot dog; the feedback collection 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 make necessary adjustments and optimizations to the control instructions; the protection mechanism unit is used to ensure the safety of the intelligent patrol robot dog by starting the protection mechanism when an abnormal situation is detected during the posture adjustment process; wherein the protection mechanism unit includes overload protection, short circuit protection, and fault diagnosis functions to ensure the safety of the intelligent patrol robot dog during the posture adjustment process.

[0100] Preferably, the posture adjustment execution module includes the following steps when working:

[0101] 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;

[0102] S402: The received control instruction is parsed and processed and converted into a signal that can be recognized by the drive control unit;

[0103] S403: The drive control unit controls the operation of the motor and the actuator according to the analyzed control instruction, and adjusts the speed and direction of the motor and the action of the actuator to adjust the posture of the intelligent inspection robot dog;

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

[0105] S405: After being processed, the collected data is sent to the control instruction issuing module;

[0106] S406: During the posture adjustment process, if the protection mechanism detects an abnormal situation, the protection mechanism is immediately activated; the protection mechanism ensures the safety of the intelligent inspection robot dog by cutting off the power supply and stopping the operation mode;

[0107] S407: The posture adjustment execution module continuously monitors the posture information of the intelligent inspection robot dog, and makes adjustments according to the control instructions of the control instruction issuing module.

[0108] Preferably, the posture adjustment execution module can execute control instructions quickly and accurately, realize precise adjustment of the posture of the intelligent inspection robot dog, and ensure the safety and stability of the adjustment process through real-time feedback and protection mechanisms.

[0109] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge scope of those skilled in the art 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: Included are: The posture detection module is responsible for obtaining the posture information of the intelligent inspection robot dog in real time, including tilt angle, angular velocity, and acceleration; The posture prediction module is used to predict the posture state of the intelligent inspection robot dog using a recurrent neural network algorithm; The threshold comparison and alarm module is used to compare the output result of the posture prediction module with the preset threshold to determine whether the intelligent inspection robot dog has the risk of overturning; The control instruction issuing module is used to calculate the control instructions required to adjust the posture according to the current posture information and prediction results of the intelligent inspection robot dog after receiving the alarm; The posture adjustment execution module is responsible for executing the control instructions generated by the control instruction issuing module to adjust the posture of the intelligent inspection robot dog.

2. The dynamic stability anti-overturning device for an intelligent inspection robot dog according to claim 1, characterized in that: The posture 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, 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 tilt angle, angular velocity, acceleration and visual image; The data pre-processing unit is used to filter and denoise the raw data collected by the sensor unit to improve the accuracy and reliability of the data; the data fusion unit is used to fuse the data of multiple sensors using a data fusion algorithm to obtain more accurate and comprehensive posture information; the communication interface unit is used to transmit the processed posture information to other modules of the device in real time.

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

4. The dynamic stability anti-overturning device for an intelligent inspection robot dog according to claim 3, characterized in that: When the posture prediction module is working, it includes the following steps: S101: The data receiving unit receives the posture information transmitted by the posture detection module in real time, including the tilt angle, angular velocity, and acceleration; S102: Arrange the historical data into a time series format as the input of the RNN model; S103: Initializing the RNN model, including setting the number of layers of the model, the number of neurons in the hidden layer, and the activation function; initialization also includes setting initial values ​​for the weights and biases of the model; S104: During the forward propagation process, the RNN model processes the input data one by one in the order of the time series; 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 uses mean square error to calculate the loss between the predicted value and the true value; the loss value is propagated back to the model through the back propagation algorithm to calculate the gradient of each parameter; according to the direction and size of the gradient, the gradient descent optimization algorithm is used to update the weights and biases of the model; S107: The RNN model is trained by repeatedly performing forward propagation and back propagation until the loss value converges to a smaller range; during the training process, the early stopping method is used to prevent overfitting and accelerate convergence; S108: The pre-processed real-time posture information is input into the trained RNN model unit, and the RNN model predicts the posture change trend in the future period of time based on the historical posture information and time dependency; 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 the feedback data and the new posture information, the model updating unit continuously optimizes and updates the RNN model, including adjusting the model parameters and improving the model structure; S111: The posture prediction module continuously performs data reception, preprocessing, posture prediction, prediction result output and model update operations, and performs real-time and continuous prediction of the posture changes of the intelligent inspection robot dog.

5. The dynamic stability anti-overturning device for an intelligent inspection robot dog according to claim 4, characterized in that: The threshold comparison and alarm module includes a data receiving unit, a threshold storage unit, a comparison and judgment unit and an alarm triggering unit; the data receiving unit is used to receive the future posture change trend data output by the posture prediction module; the threshold storage unit is used to store the preset threshold determined according to the design parameters, working environment and historical overturning data factors of the intelligent inspection robot dog; the comparison and judgment unit is used to compare the received posture prediction data with the preset threshold to determine whether there is a risk of overturning; the alarm triggering unit is used to trigger an alarm when the comparison and judgment unit determines that there is a risk of overturning, and send the alarm information to the control instruction issuing module; wherein, the comparison and judgment unit adopts a real-time comparison method to ensure that each time new posture prediction data is received, it can be immediately compared with the threshold; the threshold storage unit can dynamically adjust the preset threshold according to the actual operation of the robot dog and the use environment, so as to improve the accuracy and practicality of the early warning.

6. A dynamic stability anti-overturning device for an intelligent inspection robot dog according to claim 5, characterized in that: The threshold comparison and alarm module works by following the steps below: S201: The data receiving unit of the threshold comparison and alarm module receives the future posture change trend data output by the posture prediction module in real time; S202: The comparison and judgment unit reads a preset threshold value from the threshold value storage unit, and compares the received posture prediction data with the preset threshold value in real time; S203: The comparison and judgment unit judges whether the intelligent inspection robot dog has a risk of overturning according to the comparison result; when any one or more of the predicted data exceeds a preset threshold, it is judged that there is a risk of overturning; S204: When it is determined that there is a capsizing risk, the alarm triggering unit immediately triggers an alarm; the alarm forms include sound alarm and light signal alarm; S205: When the alarm is triggered, the alarm triggering unit sends alarm information to the control instruction issuing module, where the alarm information includes the type and degree of the capsizing risk and recommended countermeasures.

7. A dynamic stability anti-overturning device for an intelligent inspection robot dog according to claim 6, characterized in that: The control instruction issuing module includes a data receiving unit, an instruction calculating unit and an instruction sending unit; the data receiving unit is used to receive the alarm information sent by the threshold comparison and alarm module, as well as the current posture information and prediction results provided by the posture detection module and the posture prediction module; the instruction calculating unit is used to calculate the control instructions required for adjusting the posture based on the received data, combined with the dynamic model and control algorithm of the intelligent inspection robot dog; the instruction sending unit is used to send the calculated control instructions to the posture adjustment execution module to drive the robot dog to adjust the posture.

8. The dynamic stability anti-overturning device for an intelligent inspection robot dog according to claim 7, characterized in that: When the control instruction issuing module is working, the following steps are included: S301: The data receiving unit of the control instruction issuing module receives in real time the threshold comparison and alarm information sent by the alarm module, as well as the current posture information and prediction results provided by the posture detection module and the posture prediction module; S302: When receiving the alarm information, the instruction calculation unit combines the current posture information, the prediction result and the dynamic model of the intelligent inspection robot dog, and the instruction calculation unit uses the fuzzy control and sliding mode control algorithms to calculate and obtain the control instructions required for adjusting the posture. The control instructions include adjusting the walking speed, steering angle and posture stabilization device of the robot dog; S303: The calculated control instruction is transmitted to the instruction sending unit; the instruction sending unit uses a high-speed communication protocol to send the control instruction to the posture adjustment execution module; S304: After the control instruction is issued, the control instruction issuing module continuously monitors the data of the posture detection module and the posture prediction module to evaluate the control effect; the instruction calculation unit adjusts the control instruction according to the feedback data; S305: The control instruction issuing module continuously collects the operation data and feedback information of the robot dog during operation.

9. The dynamic stability anti-overturning device for an intelligent inspection robot dog according to claim 8, characterized in that: The posture adjustment execution module includes an instruction receiving unit, a drive control unit, a feedback collection unit and a protection mechanism unit; the instruction receiving unit is used to receive the control instructions generated by the control instruction issuing module, and parse and process them; the drive control unit is used to control the operation of the motor and the actuator according to the parsed control instructions to achieve the adjustment of the posture of the intelligent patrol robot dog; the feedback collection 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 make necessary adjustments and optimizations to the control instructions; the protection mechanism unit is used to ensure the safety of the intelligent patrol robot dog by starting the protection mechanism when an abnormal situation is detected during the posture adjustment process; wherein the protection mechanism unit includes overload protection, short circuit protection, and fault diagnosis functions to ensure the safety of the intelligent patrol robot dog during the posture adjustment process.

10. The dynamic stability anti-overturning device for an intelligent inspection robot dog according to claim 9, characterized in that: When the posture adjustment execution module is working, the following steps are included: 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 parsed and processed and converted into a signal that can be recognized by the drive control unit; S403: The drive control unit controls the operation of the motor and the actuator according to the analyzed control instruction, and adjusts the speed and direction of the motor and the action of the actuator to adjust the posture of the intelligent inspection robot dog; S404: During the posture adjustment process, the feedback collection unit monitors the adjustment effect in real time and collects relevant data through sensors; S405: After being processed, the collected data is sent to the control instruction issuing module; S406: During the posture adjustment process, if the protection mechanism detects an abnormal situation, the protection mechanism is immediately activated; the protection mechanism ensures the safety of the intelligent inspection robot dog by cutting off the power supply and stopping the operation mode; S407: The posture adjustment execution module continuously monitors the posture information of the intelligent inspection robot dog, and makes adjustments according to the control instructions of the control instruction issuing module.

Images