Upper limb exoskeleton control method, system and equipment for assisting replacement of insulator

Abstract

The invention relates to the technical field of electrical equipment maintenance, and discloses an upper limb exoskeleton control method, system and equipment for assisting replacement of an insulator. The method comprises the following steps: acquiring image data of an insulator and a field environment, and acquiring motion vector information according to the image data based on a preset optical flow method; analyzing the motion vector information based on a deep learning algorithm to obtain motion state information of the insulator; based on a dynamic fusion algorithm, according to the motion state information of the insulator and the position information of the upper limb exoskeleton, obtaining adjustment motion parameters of the upper limb exoskeleton; and issuing a control instruction to the upper limb exoskeleton according to the adjusted motion parameters. Dynamic changes of the insulator and the surrounding environment are monitored through the optical flow method, the insulator in motion is accurately recognized and analyzed in combination with deep learning analysis, the motion trail of the upper limb exoskeleton can be adjusted in time, it is ensured that the upper limb exoskeleton stably and accurately tracks and operates the insulator, the physical burden of operators is relieved, and the working efficiency is improved. And the operation safety and efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment maintenance, and particularly to an upper limb exoskeleton control method, system and device for assisting in replacing insulators. Background Art

[0002] With the rapid development of the power industry, the complexity and maintenance difficulty of the power grid are increasing day by day. During the maintenance of high-voltage transmission lines, replacing insulators is a common and important task. As a key component of the transmission line, the performance of the insulator directly affects the safe and stable operation of the power grid.

[0003] Traditional insulator replacement technologies have many deficiencies. On the one hand, it is difficult for manual operation to accurately control the force and angle during the replacement process, especially in the face of complex and changeable working environments, it is more likely to deviate, affecting the operation quality. On the other hand, traditional technologies lack the ability to monitor and adapt to the working environment in real time, and cannot adjust the operation strategy in a timely manner according to environmental changes, further restricting the operation efficiency and safety.

[0004] In summary, the traditional insulator replacement operation method has the disadvantages of heavy physical burden, low operation efficiency, insufficient accuracy and poor environmental adaptability. These disadvantages not only increase the operation risk, but also limit the efficiency and safety of power equipment maintenance. Therefore, it is particularly important to develop an upper limb exoskeleton control algorithm for assisting in replacing insulators. Summary of the Invention

[0005] The main purpose of the present invention is to provide an upper limb exoskeleton control method, system and device for assisting in replacing insulators, aiming to solve at least one of the above technical problems.

[0006] To achieve the above purpose, the present invention provides an upper limb exoskeleton control method for assisting in replacing insulators, including:

[0007] Collecting image data of the insulator and the on-site environment, and obtaining motion vector information based on the preset optical flow method according to the image data;

[0008] Analyzing the motion vector information based on a deep learning algorithm to obtain the motion state information of the insulator;

[0009] Based on a dynamic fusion algorithm, combining the motion state information of the insulator with the position information of the upper limb exoskeleton to obtain the adjusted motion parameters of the upper limb exoskeleton;

[0010] Issuing a control instruction to the upper limb exoskeleton according to the adjusted motion parameters.

[0011] In some embodiments, collecting image data of the insulator and the on-site environment, and obtaining motion vector information based on a preset optical flow method according to the image data includes:

[0012] Collecting original images of the insulator and the on-site environment in real time, and preprocessing the original images to obtain image data;

[0013] Performing principal component analysis on a number of insulator image samples to extract key feature vectors;

[0014] Constructing a direction weight matrix according to the correlation between the key feature vectors and directions;

[0015] Improving the gradient method in the preset optical flow method based on the direction weight matrix to obtain an improved gradient method;

[0016] Obtaining motion vector information according to the improved gradient method and the image data.

[0017] In some embodiments, obtaining motion vector information according to the improved gradient method and the image data includes:

[0018] Calculating a first direction gradient, a second direction gradient, and a time direction gradient according to the improved gradient method and the image data;

[0019] Calculating a first direction motion vector and a second direction motion vector respectively according to the first direction gradient, the second direction gradient, the time direction gradient, and the direction weight matrix;

[0020] Obtaining motion vector information according to the first direction motion vector and the second direction motion vector.

[0021] In some embodiments, analyzing the motion vector information based on a deep learning algorithm to obtain the motion state information of the insulator includes:

[0022] Constructing a deep learning model based on a multi-level cascaded attention network;

[0023] Inputting the motion vector information into the deep learning model to identify the moving insulator and obtain key motion features related to the insulator; wherein, the key motion features include the motion direction and speed information of the insulator;

[0024] Generating the motion state information of the insulator according to the key motion features.

[0025] In some embodiments, obtaining the adjustment motion parameters of the upper limb exoskeleton based on a dynamic fusion algorithm according to the motion state information of the insulator in combination with the position information of the upper limb exoskeleton includes:

[0026] Set the desired position of the end effector of the upper limb exoskeleton, and obtain the current position of the end effector of the upper limb exoskeleton and the angles of each joint;

[0027] Obtain a dynamic weight vector based on the motion state information of the insulator in combination with the historical operation data of the upper limb exoskeleton;

[0028] Based on the kinetic fusion algorithm, obtain the joint angle adjustment amount according to the desired position, current position, angles of each joint, and the dynamic weight vector;

[0029] Take the joint angle adjustment amount as the adjustment motion parameter of the upper limb exoskeleton.

[0030] In some embodiments, the obtaining a dynamic weight vector based on the motion state information of the insulator in combination with the historical operation data of the upper limb exoskeleton includes:

[0031] Obtain the historical operation data of the upper limb exoskeleton in a preset environment;

[0032] Obtain the current wind direction and wind speed information in real time;

[0033] Evaluate the complexity of the operation environment according to the motion state information of the insulator in combination with the historical operation data and the current wind direction and wind speed information, and obtain a dynamic weight vector.

[0034] In some embodiments, the method further includes:

[0035] Establish a mapping model between environmental parameters and insulator motion interference;

[0036] Monitor the environmental parameters at the operation site in real time;

[0037] Perform real-time analysis according to the environmental parameters to obtain the change trend of the environmental parameters, and judge whether it interferes with the motion of the insulator according to the change trend of the environmental parameters;

[0038] If so, generate a compensation instruction according to the mapping model and the environmental parameters;

[0039] Adjust and correct the motion trajectory of the upper limb exoskeleton based on the compensation instruction.

[0040] In some embodiments, the establishing a mapping model between environmental parameters and insulator motion interference includes:

[0041] Generate original environmental parameters according to wind speed, temperature, humidity, and equipment vibration frequency;

[0042] Generate data samples based on experimental observations of insulator motion under different combinations of original environmental parameters;

[0043] Train a multivariate linear regression model based on particle swarm optimization according to the data samples to obtain a mapping model between environmental parameters and insulator motion interference.

[0044] In addition, to achieve the above object, the present invention also provides an upper limb exoskeleton control system for assisting in replacing insulators, including:

[0045] An optical flow monitoring module, configured to collect image data of the insulator and the on-site environment, and obtain motion vector information based on the preset optical flow method according to the image data;

[0046] An insulator analysis module, configured to analyze the motion vector information based on a deep learning algorithm to obtain motion state information of the insulator;

[0047] A trajectory adjustment module, configured to obtain adjustment motion parameters of the upper limb exoskeleton based on a dynamics fusion algorithm according to the motion state information of the insulator in combination with the position information of the upper limb exoskeleton;

[0048] A motion control module, configured to issue a control instruction to the upper limb exoskeleton according to the adjustment motion parameters.

[0049] In addition, to achieve the above object, the present invention also provides an electronic device, including: a memory, a processor, and an upper limb exoskeleton control program for assisting in replacing insulators stored on the memory and executable on the processor, where the upper limb exoskeleton control program for assisting in replacing insulators is configured to implement the upper limb exoskeleton control method for assisting in replacing insulators as described above.

[0050] The present invention provides an upper limb exoskeleton control method for assisting in replacing insulators, including: collecting image data of the insulator and the on-site environment, and obtaining motion vector information based on the preset optical flow method according to the image data; analyzing the motion vector information based on a deep learning algorithm to obtain motion state information of the insulator; obtaining adjustment motion parameters of the upper limb exoskeleton based on a dynamics fusion algorithm according to the motion state information of the insulator in combination with the position information of the upper limb exoskeleton; and issuing a control instruction to the upper limb exoskeleton according to the adjustment motion parameters. In the present invention, the dynamic changes of the insulator and the surrounding environment are monitored by the optical flow method, and the insulator in motion is accurately identified and analyzed by combining deep learning analysis, so that the motion trajectory of the upper limb exoskeleton can be adjusted in time to ensure that the upper limb exoskeleton stably and accurately tracks and operates the insulator, thereby not only reducing the physical burden of the operator, but also improving the operation safety and efficiency. Description of the Drawings

[0051] Figure 1 It is a schematic structural diagram of an electronic device for the hardware operating environment related to the embodiment solution of the present invention;

[0052] Figure 2 It is a schematic flowchart of an embodiment of the upper limb exoskeleton control method for assisting in replacing insulators according to the present invention;

[0053] Figure 3 It is a schematic flowchart of the control algorithm involved in the embodiment solution of the present invention;

[0054] Figure 4 It is a schematic flowchart of the key control algorithm involved in the embodiment solution of the present invention;

[0055] Figure 5 It is a structural block diagram of an embodiment of the upper limb exoskeleton control system for assisting in replacing insulators according to the present invention;

[0056] Figure 6 It is a schematic diagram of an example module of the control system involved in the embodiment solution of the present invention.

[0057] The realization, functional characteristics and advantages of the object of the present invention will be further described in conjunction with the embodiments with reference to the accompanying drawings. Specific Embodiments

[0058] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0059] It should be noted that all directional indications (such as up, down, left, right, front, back...) in the embodiments of the present invention are only used to explain the relative positional relationship and movement conditions between components in a specific posture (as shown in the accompanying drawings). If this specific posture changes, then the directional indications will also change accordingly.

[0060] In addition, the descriptions involving "first", "second", etc. in the present invention are only for descriptive purposes, and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of such features. In addition, the technical solutions between various embodiments may be combined with each other, but it must be based on the fact that those of ordinary skill in the art can implement them. When the combination of technical solutions results in contradictions or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the protection scope required by the present invention. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0061] Refer to Figure 1 , Figure 1Schematic diagram of the electronic device structure for the hardware operating environment involved in the embodiment of the present invention.

[0062] As shown in Figure 1 , the electronic device may include: a processor 1001, such as a Central Processing Unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display) and an input unit such as a keyboard (Keyboard). Optionally, the user interface 1003 may further include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a Wireless-Fidelity (Wi-Fi) interface). The memory 1005 may be a high-speed Random Access Memory (RAM memory) or a stable Non-Volatile Memory (NVM), such as a disk memory. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0063] Those skilled in the art can understand that Figure 1 the structure shown in

[0064] does not constitute a limitation on the electronic device, and may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements. Figure 1 As shown in

[0065] In Figure 1 the electronic device shown, the network interface 1004 is mainly used for data communication with a network server; the user interface 1003 is mainly used for data interaction with a user; the processor 1001 and the memory 1005 in the electronic device of the present invention may be arranged in the electronic device. The electronic device calls the upper limb exoskeleton control program stored in the memory 1005 through the processor 1001 and executes the upper limb exoskeleton control method for assisting in replacing insulators provided by the embodiment of the present invention.

[0066] The present invention proposes an upper limb exoskeleton control method, system and device for assisting in replacing insulators.

[0067] The embodiment of the present invention provides an upper limb exoskeleton control method for assisting in replacing insulators. Refer to Figure 2, Figure 2 It is a schematic flowchart of an embodiment of the upper limb exoskeleton control method for assisting in replacing insulators according to the present invention.

[0068] As Figure 2 shown, the upper limb exoskeleton control method for assisting in replacing insulators includes:

[0069] Step S100: Collect image data of the insulator and the on-site environment, and obtain motion vector information based on the preset optical flow method according to the image data;

[0070] Step S200: Analyze the motion vector information based on a deep learning algorithm to obtain the motion state information of the insulator;

[0071] Step S300: Based on the dynamic fusion algorithm, combine the motion state information of the insulator with the position information of the upper limb exoskeleton to obtain the adjustment motion parameters of the upper limb exoskeleton;

[0072] Step S400: Send a control command to the upper limb exoskeleton according to the adjustment motion parameters.

[0073] It should be noted that the execution subject in this embodiment can be an electronic device, which can be a computer device with data processing functions, or other devices that can achieve the same or similar functions. This embodiment does not limit this. In this embodiment, a computer device is taken as an example for illustration.

[0074] In one embodiment, collecting image data of the insulator and the on-site environment, and obtaining motion vector information based on the preset optical flow method according to the image data includes: real-time collecting the original images of the insulator and the on-site environment, and preprocessing the original images to obtain image data; performing principal component analysis on a number of insulator image samples to extract key feature vectors; constructing a direction weight matrix according to the correlation between the key feature vectors and directions; improving the gradient method in the preset optical flow method based on the direction weight matrix to obtain an improved gradient method; and obtaining motion vector information according to the improved gradient method and the image data.

[0075] In one embodiment, obtaining motion vector information according to the improved gradient method and the image data includes: calculating a first direction gradient, a second direction gradient, and a time direction gradient according to the improved gradient method and the image data; calculating a first direction motion vector and a second direction motion vector respectively according to the first direction gradient, the second direction gradient, the time direction gradient, and the direction weight matrix; and obtaining motion vector information according to the first direction motion vector and the second direction motion vector.

[0076] Specifically, as Figure 3As shown in the figure, a monitoring device is installed, and optical flow method is used for monitoring: it monitors the dynamic changes of the insulator and the surrounding environment in real time to obtain motion vector information. Optical flow refers to the velocity field of pixel points generated by the movement of objects in an image. When an object moves in an image sequence, the corresponding pixel gray values will change positions between different frames. Optical flow is a physical quantity that describes the direction and speed of the movement of these pixel points between adjacent frames. In this embodiment, the motion information of the insulator, such as the motion direction and speed, can be obtained by analyzing the optical flow. Optical flow algorithms include, but are not limited to, gradient-based methods (such as the Lucas-Kanade algorithm), matching-based methods, energy-based methods, and phase-based methods. In this embodiment, a custom optical flow algorithm (improved gradient method) is adopted, which is an improved algorithm of the gradient-based method.

[0077] Exemplarily, a high-definition camera is installed at a suitable position on the exoskeleton, and an image sequence (original image) of the insulator and the surrounding environment is continuously collected at a fixed frame rate. To reduce the interference of noise on subsequent optical flow calculations, the collected original images are first preprocessed to obtain image data. For example, as Figure 4 shown, Gaussian filtering means are used to perform convolution processing on each frame of the image one by one, effectively filtering out the high-frequency noise in the image and making the image smoother, laying a stable data foundation for optical flow calculation.

[0078] Specifically, in this embodiment, an optical flow calculation method based on the improved gradient method is adopted. First, for a large number of insulator sample images in different scenarios, feature extraction work is carried out to obtain multiple feature vectors (key feature vectors). The correlation degree between each feature vector and the motion direction of the object in the image is deeply analyzed. Based on these correlations, a direction weight matrix is constructed. By integrating this direction weight matrix, the accuracy of capturing the motion vectors of the insulator and the surrounding environment objects is greatly improved. Finally, the motion vectors of pixel points between adjacent frames of the image sequence (or image data) are calculated to obtain motion vector information, providing detailed and reliable motion information for the subsequent deep learning analysis module.

[0079] Exemplarily, during the optical flow method monitoring process, the image sequence captured by a monitoring device such as a camera is processed through a custom optical flow algorithm (improved gradient method). This custom optical flow algorithm is based on the improved gradient method and introduces a direction weight matrix. Let the gradients of the image in the x (first direction) and y directions (second direction) be I x and I y , the gradient in the time direction be I t , and the direction weight matrix be W. Through principal component analysis of a large number of insulator image samples in different scenarios, key feature vectors are extracted, and a weight matrix (direction weight matrix W) is constructed according to the correlation between the key feature vectors and the direction. The motion vector (motion vector information) V = (V x , V y) The calculation formula is as follows:

[0080]

[0081] Among them, V x is the motion vector in the first direction; V y is the motion vector in the second direction; n is the number of pixel points participating in the calculation within the local image area; w i,x and w i,y are the weights at the i-th pixel point in the x and y directions respectively. This algorithm (optical flow calculation based on the improved gradient method) can capture the motion vectors of the insulator and surrounding environmental objects more accurately, providing a more precise data basis for subsequent analysis.

[0082] In one embodiment, the motion vector information is analyzed based on a deep learning algorithm to obtain the motion state information of the insulator, including: constructing a deep learning model based on a multi-level cascaded attention network; inputting the motion vector information into the deep learning model to identify the moving insulator and obtain key motion features related to the insulator; among them, the key motion features include the motion direction and speed information of the insulator; generating the motion state information of the insulator according to the key motion features.

[0083] Specifically, as Figure 3 and Figure 4 shown, a deep learning algorithm is used to analyze the optical flow data (motion vector information) to identify the moving insulator and obtain its motion state information.

[0084] Exemplarily, the deep learning model constructed by deep learning analysis can be a multi-level cascaded attention network. The input of the deep learning model is the motion vector information obtained by the optical flow method. An attention mechanism module is introduced after each convolutional operation. Let the input feature map of the l-th layer be F l , and the attention weight A l is calculated by the following formula:

[0085] A l = sigmoid(W1·(W2·F l + b2) + b1)

[0086] Among them, W1 and W2 are weight matrices; b1 and b2 are bias terms. The weight matrices W1, W2 and the bias terms are determined by backpropagation training on a large number of optical flow data samples containing insulators in different motion states. This model (deep learning model) can automatically focus on the key motion features related to the insulator, suppress irrelevant background interference, more accurately identify the insulator and analyze its motion state. By learning the features of the insulator in different motion directions and speeds in a large number of samples, it can accurately output the motion direction and speed information of the insulator, significantly improving the accuracy of identification and analysis.

[0087] Specifically, for deep learning analysis: a multi-level cascaded attention network for analyzing optical flow data is constructed. The model input is the motion vector information obtained by monitoring with the optical flow method. The structure of the multi-level cascaded attention network includes multiple convolutional layers, pooling layers, and attention mechanism modules. After each convolutional operation, the attention mechanism is introduced to enable the model to automatically focus on capturing key motion features related to the insulator, effectively suppressing interference from irrelevant backgrounds, and enhancing the accuracy of analyzing the motion state of the insulator. For example, optical flow data samples of insulators and their surrounding environments in different scenarios are widely collected, and the position, motion direction, and speed motion state information of the insulators are carefully annotated. These annotated sample data are divided into a training set, a validation set, and a test set. During the training process, the model parameters are updated using the stochastic gradient descent algorithm. Each iteration randomly extracts a batch of samples from the training set, calculates the loss of the model on this batch of samples, and at the same time, regularly evaluates the model performance on the validation set to prevent overfitting of the model. When the performance of the model no longer improves on the validation set, the training is stopped, and a final test is conducted on the test set to obtain the trained multi-level cascaded attention network, which serves as the deep learning model for deep learning analysis, thereby ensuring that the deep learning model can accurately identify the moving insulators and precisely analyze their motion states.

[0088] It can be understood that the motion vector information obtained by monitoring with the optical flow method also undergoes data preprocessing before deep learning analysis. The preprocessing includes noise reduction, normalization, etc. of the motion vector information obtained by the optical flow method. Exemplarily, for noise reduction, an adaptive median filtering algorithm can be used. According to the statistical characteristics of the motion vectors of pixel points in a local area, the size of the filtering window is dynamically adjusted. Let the size of the local area window be S, and by calculating the mean and standard deviation σ V of the motion vectors within the window, when σ V is greater than the threshold T, the window S is increased, and vice versa, the window S is decreased. Exemplarily, for normalization processing, the maximum-minimum normalization algorithm can be used to normalize the motion vector V to the interval [0, 1]. The formula is:

[0089]

[0090] where V min and V max are respectively the minimum and maximum values of the motion vector over a period of time. The preprocessed data (the preprocessed motion vector information) can be better utilized during deep learning analysis, improving the efficiency and accuracy of deep learning analysis.

[0091] In one embodiment, based on the kinematic fusion algorithm, the adjustment motion parameters of the upper limb exoskeleton are obtained according to the motion state information of the insulator in combination with the position information of the upper limb exoskeleton, including: setting the desired position of the end effector of the upper limb exoskeleton, obtaining the current position and each joint angle of the end effector of the upper limb exoskeleton; obtaining the dynamic weight vector according to the motion state information of the insulator in combination with the historical operation data of the upper limb exoskeleton; based on the kinematic fusion algorithm, obtaining the joint angle adjustment amount according to the desired position, current position, each joint angle and the dynamic weight vector; and using the joint angle adjustment amount as the adjustment motion parameters of the upper limb exoskeleton.

[0092] In one embodiment, obtaining the dynamic weight vector according to the motion state information of the insulator in combination with the historical operation data of the upper limb exoskeleton includes: obtaining the historical operation data of the upper limb exoskeleton in a preset environment; obtaining the current wind direction and wind speed information in real time; and evaluating the complexity of the operation environment according to the motion state information of the insulator in combination with the historical operation data and the current wind direction and wind speed information to obtain the dynamic weight vector.

[0093] Specifically, as Figure 3 shown, the motion trajectory adjustment: according to the motion state information of the insulator, in combination with the current position and posture of the exoskeleton, the adjustment motion parameters are obtained to adjust the motion trajectory of the exoskeleton.

[0094] Exemplarily, when calculating the adjustment motion parameters, a kinematic and dynamic fusion algorithm based on dynamic weights can be used. Let the desired position of the end effector of the exoskeleton be P d , the current position be P, and each joint angle be θ = (θ1, θ2,..., θ n ), and the dynamic weight vector W d = (w d1 , w d2 ,..., w dn ), where the dynamic weight vector W d is determined according to the historical operation data of the exoskeleton and the evaluation of the current operation environment complexity. Through the forward kinematic formula P = f k (θ) and the inverse dynamic formula in combination with the dynamic weights (the dynamic weight vector W d ), the adjustment motion parameters (joint angle adjustment amount) are calculated. The calculation formula for the joint angle adjustment amount Δθ is:

[0095]

[0096] where J is the Jacobian matrix; λ and μ are control gains, which are determined by experiment and simulation optimization. In this embodiment, the kinematic and dynamic fusion algorithm based on dynamic weights is used, which can adjust the motion trajectory more flexibly and accurately according to the motion state of the insulator and the actual situation of the exoskeleton.

[0097] In practical applications, the upper limb exoskeleton is used to operate in coordination with the movement of the insulator, and the desired position P of the end effector of the exoskeleton d usually needs to be determined according to the movement state information of the insulator. The current position of the insulator can be used as the desired position P of the end effector of the exoskeleton d , or considering the delay of the exoskeleton movement and the movement trend of the insulator, the position of the insulator can also be predicted to obtain a more appropriate desired position P d . The dynamic weight vector W d 0 is used to adjust the importance of each joint in motion control. The movement state information such as the swing amplitude, speed, and direction of the insulator will change the complexity of the operation environment of the exoskeleton. An environment complexity evaluation model can be established, and the environment complexity evaluation value can be obtained according to the movement state information of the insulator for this evaluation model. The dynamic weight vector W is determined according to the environment complexity evaluation value and the historical operation data of the exoskeleton d . For example, first obtain the basic weights of each joint under different environment complexities according to the historical operation data, and then adjust according to the current environment complexity.

[0098] It should be noted that after obtaining the movement state information of the insulator through deep learning analysis, based on this information and the current position and attitude of the exoskeleton, using kinematic and dynamic algorithms, various motion parameters required to adjust the exoskeleton motion trajectory are calculated (adjustment motion parameters). In a complex environment, the weights of the joints in the direction of the operation target will increase, thereby enhancing the influence of the joints in the adjustment of the motion trajectory. Finally, according to the real-time motion state of the insulator and the current situation of the exoskeleton, the joint angle adjustment amount required to adjust the exoskeleton motion trajectory is flexibly and accurately calculated as the key motion parameter (adjustment motion parameter). According to the calculated adjustment motion parameters, such as joint angle adjustment amount, speed, and acceleration, etc., the adjustment motion parameters are converted into control instructions that can be recognized by the exoskeleton drive device. These control instructions include key information such as the target angle and target speed of each joint. Subsequently, the control instructions are transmitted to the exoskeleton drive device through the communication interface, and the drive device adjusts the motion of each joint of the exoskeleton according to the control instructions, realizing the adjustment of the exoskeleton motion trajectory, ensuring that the moving insulator can be accurately tracked, and completing the operation of replacing the insulator.

[0099] In one embodiment, the method further includes: establishing a mapping model between environmental parameters and insulator motion interference; real-time monitoring of environmental parameters at the operation site; performing real-time analysis according to the environmental parameters to obtain the change trend of the environmental parameters, and judging whether it interferes with the insulator motion according to the change trend of the environmental parameters; if so, generating a compensation instruction according to the mapping model and the environmental parameters; and performing motion trajectory adjustment and correction on the upper limb exoskeleton based on the compensation instruction.

[0100] In one embodiment, a mapping model between environmental parameters and insulator motion interference is established, including: generating original environmental parameters according to wind speed, temperature, humidity, and equipment vibration frequency; generating data samples based on experimental observations of insulator motion under different combinations of original environmental parameters; training a multiple linear regression model based on particle swarm optimization according to the data samples to obtain a mapping model between environmental parameters and insulator motion interference.

[0101] Specifically, as Figure 3 shown, environmental parameter perception and compensation: real-time monitor the environmental parameters of the operation site, judge the influence of environmental factors on insulator motion according to the pre-established mapping model, and send compensation instructions to the upper limb exoskeleton for motion trajectory adjustment when necessary. Among them, to construct a mapping model between environmental parameters and insulator motion interference, a multiple linear regression model based on particle swarm optimization can be used. Under different combinations of environmental parameters, a large number of experimental observations on insulator motion are carried out to obtain data samples. With the help of the particle swarm optimization algorithm, the optimal regression coefficients that minimize the objective function value are found, so as to determine the mapping relationship between environmental parameters and the amount of insulator motion interference, and form a mapping model.

[0102] Exemplarily, during the environmental parameter perception and compensation process, when establishing a mapping model between environmental parameters and insulator motion interference, a multiple linear regression model based on particle swarm optimization is used. Let the environmental parameters (original environmental parameters) be X = (x1, x2,..., x m ), including wind speed, temperature, humidity, and equipment vibration frequency, and the amount of insulator motion interference be Y. Through a large number of experimental observations on insulator motion under different combinations of environmental parameters, data samples (X i , Y i ) are obtained, where i = 1, 2,..., N. The particle swarm optimization algorithm is used to find the optimal regression coefficients β = (β0, β1,..., β m ) to minimize the objective function . Each particle in the particle swarm represents a set of regression coefficients, and its position and velocity are continuously updated in the search space. After multiple iterations of optimization, the final regression coefficients are determined to form a mapping model This model (mapping model) can more accurately reflect the influence of environmental parameters on insulator motion and provide a reliable basis for generating compensation instructions.

[0103] In one embodiment, control instructions are sent to the upper limb exoskeleton according to the adjusted motion parameters.

[0104] Specifically, motion smoothing can also be performed between the motion trajectory adjustment and the exoskeleton drive device to smooth the control commands output by the motion trajectory adjustment and avoid mutations and impacts during the exoskeleton movement. Exemplarily, a smoothing algorithm based on Bezier curves can be adopted. Let the control command sequence be C = (c1, c2,..., c n ), by determining the control points of the Bezier curve, the control commands are converted into a smooth curve. The determination of the Bezier curve control points is based on the motion characteristics and operation requirements of the exoskeleton. According to the maximum acceleration and speed limits of the exoskeleton joints and the smoothness requirements for the operation of replacing insulators, the positions of the control points are adjusted. For the time parameter t ∈ [0, 1], the smoothed control command C smooth (t) is calculated through the Bezier curve formula:

[0105]

[0106] where B i,n (t) is the Bernstein polynomial. The control commands after this motion smoothing can make the exoskeleton move more smoothly, extend the service life of the device, and improve the comfort of the operator at the same time.

[0107] It can be understood that environmental parameter perception and compensation: It also has an environmental risk warning function. When it detects that the environmental parameters exceed the safe threshold range, it not only sends compensation commands for motion trajectory adjustment, but also issues a warning signal to the operator. Among them, the safe threshold can be determined by analyzing a large amount of power operation accident data and the performance test data of the exoskeleton in different environments. The warning signal includes but is not limited to audible and visual alarms. Alarms are issued by the flashing of the indicator lights on the exoskeleton and the sounding of alarm tones through the speaker to remind the operator. At the same time, the current environmental parameters and warning information can be uploaded to the remote monitoring center in real time, facilitating the management personnel to timely grasp the situation of the power equipment maintenance operation site and take corresponding measures to further ensure the operation safety.

[0108] Exemplarily, various sensors such as wind speed sensors, temperature and humidity sensors, and vibration sensors are deployed in the working environment to collect environmental parameters such as wind force, temperature, humidity, and the vibration frequency and amplitude of equipment at the job site in real time. The data collected by the sensors can be transmitted to the processor for environmental parameter perception and compensation through wired or wireless communication methods. The processor analyzes the collected environmental parameters in real time, calculates the change trend of the environmental parameters, and the deviation between the current value and the preset safety threshold, so as to judge the environmental change situation. When the processor for environmental parameter perception and compensation determines that environmental factors have a significant interference on the movement of the insulator and may affect the operation accuracy of the exoskeleton, a correction suggestion for the movement parameters is calculated according to the mapping model, and a compensation instruction is generated. The compensation instruction may include correction values for joint angle adjustment amount, speed, and acceleration movement parameters. Then, the compensation instruction is sent through the communication interface, and the calculated movement parameters (adjusted movement parameters) are corrected according to the compensation instruction to ensure that the exoskeleton can still stably and accurately track and operate the insulator in a complex and changeable environment.

[0109] It should be noted that the method described in this embodiment will be described below in combination with two specific implementation examples:

[0110] Implementation Example 1: On a transmission line in a mountainous area, due to the long-term erosion of the insulator by the natural environment, it needs to be replaced. During the operation, the wind speed is relatively high, reaching 8 m / s, the wind direction is unstable and there are gusts, which have a significant impact on the stability of the insulator and the operation accuracy of the exoskeleton. The terrain of the mountainous area where the transmission line is located is complex, and there are trees and rocks around, further increasing the difficulty of the operation.

[0111] The high-definition camera on the exoskeleton collects the image sequence of the insulator and the surrounding environment at a frame rate of 30 frames per second. Since the light in the mountainous environment is complex, the collected images (original images) may have noise. Therefore, the original images are first preprocessed with Gaussian filtering, and the high-frequency noise in the images is effectively filtered through convolution operations to make the images smoother, obtaining image data, which lays a stable data foundation for subsequent optical flow calculation. Calculate the motion vector based on the improved gradient method. In a certain local area, the gradient I of the image in the x direction x = 0.3, the gradient I of the image in the y direction y = 0.2, the gradient I of the image in the time direction t = 0.1, the number of pixel points n participating in the calculation is 10, and the direction weight matrix W is constructed by performing principal component analysis on a large number of insulator image samples in different mountainous area scenes, extracting key feature vectors, and according to the correlation between the feature vectors and the directions.

[0112] Taking a certain pixel point as an example, assuming that its weight w in the x direction i,x = 0.6, the weight w in the y direction i,y = 0.4, substituting into the formula:

[0113]

[0114] Calculate the motion vector V=(V x , V y ) of the pixels in this area, providing basic data for subsequent analysis.

[0115] Input the motion vector information (motion vector V) obtained by the optical flow method into a multi-level cascaded attention network. The network structure includes multiple convolutional layers, pooling layers, and attention mechanism modules. An attention mechanism module is introduced after each convolutional operation. Assume the input feature map F l of the l-th layer. Through the formula:

[0116] A l = sigmoid(W1·(W2·F l + b2) + b1)

[0117] Calculate the attention weight A l , where the weight matrices W1, W2 and the bias terms b1, b2 are determined during the model training process. When the deep learning model processes the motion vector information, it can automatically focus on the key motion features of the insulator, suppress irrelevant interferences such as the swaying of trees in the mountain background, and identify the motion state information that the insulator swings with a large amplitude and high speed under strong wind, and the swinging direction changes with the wind direction.

[0118] Given the expected position P d of the end effector of the exoskeleton, the current position P, and the joint angles θ=(θ1, θ2,..., θ n ), according to the historical operation data of the exoskeleton in the strong wind environment in the mountains, and the assessment of the complexity of the operation environment caused by the frequent changes of the current wind direction and wind speed, determine the dynamic weight vector W d = (w d1 , w d2 ,..., w dn ). The joint weights close to the swinging direction of the insulator will increase, enhancing their influence in the motion trajectory adjustment.

[0119] Use the forward kinematic formula P = f k (θ) and the inverse dynamic formula to calculate the joint angle adjustment amount Δθ in combination with the dynamic weight. The formula is:

[0120]

[0121] Among them, the Jacobian matrix J is determined according to the mechanical structure of the exoskeleton, and the control gains λ, μ are optimized and set through multiple experiments. Calculate the key motion parameters such as the joint angle adjustment amount required to adjust the exoskeleton motion trajectory, so that the exoskeleton motion trajectory adapts to the insulator motion.

[0122] The wind speed sensor collects wind speed data in real time. When the processor determines that the wind speed exceeds the safety threshold of 8 m / s, through the multiple linear regression model based on particle swarm optimization, let the wind speed \(x_1 = 8\) m / s, the wind direction \(x_2\) (assuming it is expressed in angles), and other environmental parameters (such as temperature, humidity, equipment vibration frequency, etc.) be \(x_3,\cdots,x\) m , according to a large number of experimental observation data samples of insulator movement under different combinations of wind speed, wind direction and other environmental parameters, use the particle swarm optimization algorithm to find the optimal regression coefficients \(\beta=(\beta_0,\beta_1,\cdots,\beta\) m ), to minimize the objective function , determine the mapping relationship between environmental parameters and the interference amount of insulator movement, and form a mapping model

[0123] According to the mapping model, calculate the interference amount \(Y\) of insulator movement, send a compensation instruction to the motion trajectory adjustment module to correct the motion parameters. At the same time, the indicator light on the exoskeleton flashes quickly, and the speaker emits a sharp alarm sound to remind the operator to pay attention to safety, and upload the current environmental parameters (wind speed, wind direction) and warning information to the remote monitoring center in real time, so that the monitoring personnel can timely grasp the on-site situation.

[0124] Example 2: Some insulators in a substation are aging and need to be replaced. During the operation, it is a high-temperature and high-humidity weather in summer, the temperature reaches 38 °C, the relative humidity reaches 80%, and the heat generated by the operation of electrical equipment in the substation further aggravates the high temperature of the local environment. This high-temperature and high-humidity environment may not only affect the performance of insulators, but also cause the exoskeleton operator to be prone to fatigue and the operation accuracy to decline.

[0125] The camera captures images at 25 frames per second. Since the high-temperature and high-humidity environment may cause the images to be blurred or noisy, first use Gaussian filtering to preprocess the captured images (original images), perform convolution processing on each frame of the image one by one, effectively filter the high-frequency noise in the images, make the images smoother, obtain image data, and provide stable data for subsequent optical flow calculation.

[0126] When calculating the motion vector, assume that in a specific area \(I\) x = 0.25, \(I\) y = 0.18, \(I\) t = 0.08, the number of pixel points \(n\) participating in the calculation is 8, and the direction weight matrix \(W\) is constructed based on the principal component analysis of a large number of insulator image samples in different substation scenarios. Taking a certain pixel point as an example, if the weight \(w\) in the \(x\) direction i,x = 0.5, and the weight \(w\) in the \(y\) direction i,y = 0.5, substitute into the formula:

[0127]

[0128] Calculate the motion vector V of the pixels in this area and obtain the dynamic change information of the insulator and the surrounding environment.

[0129] Input the motion vector information into the multi-level cascaded attention network. Each layer in the model uses the formula:

[0130] A l = sigmoid(W1·(W2·F l + b2)+ b1)

[0131] Calculate the attention weight A l , where the weight matrices W1, W2 and the bias terms b1, b2 are determined during the training of the deep learning model. The deep learning model can automatically focus on capturing the key motion features related to the insulator, identify the slow displacement of the insulator due to possible condensation on the surface caused by high temperature and high humidity, and the motion state change information caused by the small deformation due to thermal expansion and contraction.

[0132] Determine the desired position P of the end effector of the exoskeleton d , the current position P, and the joint angles θ = (θ1, θ2,..., θ n ). According to the historical operation data of the exoskeleton in high temperature and high humidity environments, and the assessment of the complexity of the operation environment caused by factors such as the heat dissipation of equipment in the current substation, determine the dynamic weight vector W d =(w d1 , w d2 ,..., w dn ). The joint weights in the direction of the insulator displacement will be appropriately increased.

[0133] Use the forward kinematic formula P = f k (θ) and the inverse dynamic formula Combine the dynamic weights to calculate the joint angle adjustment amount Δθ. The formula is:

[0134]

[0135] Among them, the Jacobian matrix J is determined according to the mechanical structure of the exoskeleton, and the control gains λ, μ are set through experimental optimization. Calculate the key motion parameters required to adjust the exoskeleton motion trajectory to ensure that the exoskeleton can accurately approach and operate the insulator.

[0136] The temperature and humidity sensors collect temperature and humidity data. The processor determines that both the temperature of 38°C and the relative humidity of 80% exceed the safety threshold. Using the environmental parameters of temperature x1 = 38°C and humidity x2 = 80% as inputs, through the multiple linear regression model based on particle swarm optimization Calculate the motion interference amount of the insulator. Among them, the regression coefficients β = (β0, β1,..., βm ) is determined by using the particle swarm optimization algorithm to find the optimal value through a large number of experimental observation data samples of the insulator movement under different combinations of temperature, humidity and other environmental parameters.

[0137] When it is calculated that the environmental factors cause obvious interference to the insulator movement, a compensation instruction is sent to the movement trajectory adjustment module to correct the movement parameters. At the same time, the indicator lights on the exoskeleton flash at a specific frequency, and the speaker emits a soft but continuous alarm sound to remind the operator to pay attention to the working environment, and the current environmental parameters and early warning information are uploaded to the remote monitoring center in real time, so that the monitoring personnel can pay attention to the on-site situation at any time and make corresponding decisions.

[0138] Compared with the prior art, the upper limb exoskeleton control method for assisting the upper limb exoskeleton to replace the insulator in this embodiment has the following beneficial effects: By using the optical flow method to monitor the dynamic changes of the insulator and the surrounding environment in real time, and combining deep learning analysis to accurately identify and analyze the moving insulator, the movement trajectory of the exoskeleton can be adjusted in time to ensure that the exoskeleton can stably and accurately track and operate the insulator, which not only reduces the physical burden of the operator, but also improves the safety and efficiency of the operation. In addition, through environmental parameter perception and compensation, the environmental parameters of the operation site, such as wind speed, temperature, humidity and equipment vibration frequency, can be monitored in real time, and the influence of environmental factors on the insulator movement can be judged according to the pre-established mapping model. When the environmental factors cause obvious interference to the insulator movement, a compensation instruction can be automatically sent to correct the movement parameters of the movement trajectory adjustment, ensuring that the exoskeleton can still operate stably in a complex and changeable environment. This environmental adaptability and robustness make this control method have a wide application prospect in the fields of power equipment maintenance and so on.

[0139] This embodiment provides an upper limb exoskeleton control method for assisting in replacing the insulator, including: collecting image data of the insulator and the on-site environment, and obtaining motion vector information based on the preset optical flow method according to the image data; analyzing the motion vector information based on the deep learning algorithm to obtain the motion state information of the insulator; obtaining the adjusted motion parameters of the upper limb exoskeleton based on the dynamic fusion algorithm according to the motion state information of the insulator combined with the position information of the upper limb exoskeleton; and sending a control instruction to the upper limb exoskeleton according to the adjusted motion parameters. In this embodiment, the optical flow method is used to monitor the dynamic changes of the insulator and the surrounding environment in real time, and combined with deep learning analysis to accurately identify and analyze the moving insulator, the movement trajectory of the upper limb exoskeleton can be adjusted in time to ensure that the upper limb exoskeleton stably and accurately tracks and operates the insulator, thereby not only reducing the physical burden of the operator, but also improving the operation safety and efficiency.

[0140] In addition, an embodiment of the present invention further provides a storage medium, on which an upper limb exoskeleton control program for assisting in replacing insulators is stored. When the upper limb exoskeleton control program for assisting in replacing insulators is executed by a processor, the steps of the upper limb exoskeleton control method for assisting in replacing insulators as described above are implemented.

[0141] Refer to Figure 5 , Figure 5 which is a structural block diagram of an embodiment of the upper limb exoskeleton control system for assisting in replacing insulators according to the present invention.

[0142] As Figure 5 shown, the upper limb exoskeleton control system for assisting in replacing insulators includes:

[0143] An optical flow monitoring module 10, configured to collect image data of the insulator and the on-site environment, and obtain motion vector information based on the preset optical flow method according to the image data;

[0144] An insulator analysis module 20, configured to analyze the motion vector information based on a deep learning algorithm to obtain the motion state information of the insulator;

[0145] A trajectory adjustment module 30, configured to obtain the adjustment motion parameters of the upper limb exoskeleton based on the dynamic fusion algorithm according to the motion state information of the insulator in combination with the position information of the upper limb exoskeleton;

[0146] A motion control module 40, configured to issue a control instruction to the upper limb exoskeleton according to the adjustment motion parameters.

[0147] Specifically, the optical flow monitoring module 10 includes an optical flow method monitoring sub-module, configured to monitor the dynamic changes of the insulator and the surrounding environment in real time and obtain motion vector information. As Figure 6 shown, in the optical flow method monitoring sub-module, the image sequence captured by the camera is processed through a custom optical flow algorithm, which is based on an improved gradient method and introduces a direction weight matrix. Let the gradients of the image in the x and y directions be I x and I y , the gradient in the time direction be I t , and the direction weight matrix be W. By performing principal component analysis on a large number of insulator image samples in different scenarios, key feature vectors are extracted, and the weight matrix W is constructed according to the correlation between the feature vectors and the directions. The calculation formula of the motion vector V=(V x , V y ) is:

[0148]

[0149] where n is the number of pixel points participating in the calculation in the local area of the image, w i,x and wi,y The weights in the x and y directions at the i-th pixel, respectively.

[0150] Specifically, the insulator analysis module 20 includes a deep learning analysis sub-module, which uses deep learning algorithms to analyze the optical flow data, identify the insulators in motion, and obtain their motion state information. As Figure 6 shown, the deep learning model constructed by the deep learning analysis sub-module is a multi-level cascaded attention network. The input of the model is the motion vector information obtained by the optical flow method. An attention mechanism module is introduced after each convolutional operation. Let the input feature map of the l-th layer be F l , and the attention weight A l is calculated by the following formula:

[0151] A l = sigmoid(W1·(W2·F l + b2)+ b1)

[0152] where W1 and W2 are weight matrices; b1 and b2 are bias terms. This model can automatically focus on the key motion features related to the insulators and suppress the interference of irrelevant backgrounds.

[0153] Specifically, the trajectory adjustment module 30 includes a motion trajectory adjustment sub-module, which adjusts the motion trajectory of the exoskeleton according to the insulator motion state information, combined with the current position and posture of the exoskeleton. As Figure 6 shown, when calculating the adjustment motion parameters, the motion trajectory adjustment sub-module adopts a kinematics and dynamics fusion algorithm based on dynamic weights. Let the expected position of the end effector of the exoskeleton be P d , the current position be P, and the joint angles be θ = (θ1, θ2,..., θ n ), and the dynamic weight vector W d = (w d1 , w d2 ,..., w dn ). According to the historical operation data of the exoskeleton and the evaluation of the complexity of the current operation environment, the dynamic weight vector W d is determined. Through the forward kinematic formula P = f k (θ) and the inverse dynamic formula combined with the dynamic weight W d the adjustment parameters are calculated. The calculation formula for the joint angle adjustment amount Δθ is:

[0154]

[0155] where J is the Jacobian matrix, and λ and μ are control gains.

[0156] Specifically, as Figure 6As shown, the system further includes an environmental parameter perception and compensation module, which is used to monitor the environmental parameters of the operation site in real time, judge the influence of environmental factors on the movement of the insulator according to the pre-established mapping model, and send a compensation instruction to the movement trajectory adjustment module when necessary. In the environmental parameter perception and compensation module, when establishing the mapping model between environmental parameters and insulator movement interference, a multivariate linear regression model based on particle swarm optimization is adopted. Let the environmental parameters be X = (x1, x2,..., x m ), including wind speed, temperature, humidity, and equipment vibration frequency, and the insulator movement interference amount be Y. Through a large number of experimental observations on the movement of the insulator under different combinations of environmental parameters, data samples (X i , Y i ) are obtained, where i = 1, 2,..., N. The particle swarm optimization algorithm is used to find the optimal regression coefficients β = (β0, β1,..., β m ) to minimize the objective function . Each particle in the particle swarm represents a set of regression coefficients, and its position and velocity are continuously updated in the search space. After multiple iterations of optimization, the final regression coefficients are determined to form the mapping model

[0157] Exemplarily, when the processor of the environmental parameter perception and compensation module determines that the environmental factors have a significant interference on the movement of the insulator and may affect the operation accuracy of the exoskeleton, the correction suggestions for the movement parameters are calculated based on the mapping model, and a compensation instruction is generated. The compensation instruction includes the correction values for the joint angle adjustment amount, speed, and acceleration movement parameters. Then, through the communication interface, the compensation instruction is sent to the movement trajectory adjustment sub-module, and the movement trajectory adjustment sub-module corrects the calculated movement parameters according to the compensation instruction.

[0158] Exemplarily, a data preprocessing sub-module is provided between the optical flow method monitoring sub-module and the deep learning analysis sub-module. This sub-module performs noise reduction and normalization processing on the motion vector information obtained by the optical flow method. For noise reduction, an adaptive median filtering algorithm is adopted. According to the statistical characteristics of the motion vectors of the pixel points in the local area, the size of the filtering window is dynamically adjusted. Let the size of the local area window be S. By calculating the mean value and standard deviation σ V of the motion vectors within the window, when σ V is greater than the threshold T, the window S is increased; otherwise, the window S is decreased. The normalization processing adopts the maximum-minimum normalization algorithm to normalize the motion vector V to the interval [0, 1]. The formula is:

[0159]

[0160] where V min and V max are respectively the minimum and maximum values of the motion vector within a period of time.

[0161] Exemplarily, a motion smoothing sub-module is provided between the motion trajectory adjustment sub-module and the exoskeleton driving device. The motion smoothing sub-module is used to smooth the control instructions output by the motion trajectory adjustment sub-module to avoid mutations and impacts during the movement of the exoskeleton. A smoothing algorithm based on Bezier curves is adopted. Let the control instruction sequence be C = (c1, c2,..., c n ), by determining the control points of the Bezier curve, the control instructions are converted into a smooth curve. The determination of the Bezier curve control points is based on the motion characteristics and operation requirements of the exoskeleton. According to the maximum acceleration and speed limits of the exoskeleton joints and the smoothness requirements for the operation of replacing insulators, the positions of the control points are adjusted. For the time parameter t ∈ [0, 1], the smoothed control instruction C smooth (t) is calculated by the Bezier curve formula:

[0162]

[0163] where B i,n (t) is the Bernstein polynomial.

[0164] Exemplarily, the environmental parameter perception and compensation module also has an environmental risk warning function. When it detects that the environmental parameters exceed the safe threshold range, it not only sends compensation instructions to the motion trajectory adjustment sub-module but also issues a warning signal to the operator. The safe threshold is determined by analyzing a large amount of power operation accident data and the performance test data of the exoskeleton in different environments. The warning signal includes audible and visual alarms. The indicator light on the exoskeleton flashes and the speaker emits an alarm sound to remind the operator. At the same time, the system uploads the current environmental parameters and warning information to the remote monitoring center in real time.

[0165] This embodiment provides an upper limb exoskeleton control system for assisting in replacing insulators. By monitoring the dynamic changes of the insulator and the surrounding environment through the optical flow method and combining deep learning analysis to accurately identify and analyze the moving insulator, it can timely adjust the motion trajectory of the upper limb exoskeleton to ensure that the upper limb exoskeleton stably and accurately tracks and operates the insulator, thereby not only reducing the physical burden of the operator but also improving the operation safety and efficiency.

[0166] It should be noted that for the technical details not described in detail in this embodiment of the upper limb exoskeleton control system for assisting in replacing insulators, reference can be made to the upper limb exoskeleton control method for assisting in replacing insulators as described above provided in any embodiment of the present invention, which will not be elaborated here.

[0167] It should be understood that the above is only for illustration and does not constitute any limitation to the technical solution of the present invention. In specific applications, those skilled in the art can set it according to needs, and the present invention does not make any restrictions in this regard.

[0168] It should be noted that the above-described workflow is only illustrative and does not limit the protection scope of the present invention. In actual applications, those skilled in the art can select some or all of them according to actual needs to achieve the purpose of the solution of this embodiment, and no restrictions are made here.

[0169] In addition, it should be noted that in this article, the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such a process, method, article or system. Without further limitations, an element defined by the statement "including one..." does not exclude the existence of another identical element in the process, method, article or system including that element.

[0170] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0171] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-described embodiment methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as a read-only memory (ROM) / RAM, magnetic disk, optical disk), and includes several instructions for causing a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in various embodiments of the present invention.

[0172] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. An upper limb exoskeleton control method for assisting in replacing insulators, characterized in that, Including: Collecting image data of insulators and the on-site environment, and obtaining motion vector information based on the preset optical flow method according to the image data; Analyzing the motion vector information based on a deep learning algorithm to obtain the motion state information of the insulators; Based on the dynamic fusion algorithm, combining the motion state information of the insulators with the position information of the upper limb exoskeleton to obtain the adjustment motion parameters of the upper limb exoskeleton; Sending a control command to the upper limb exoskeleton according to the adjustment motion parameters.

2. The method according to claim 1, characterized in that, The step of collecting image data of insulators and the on-site environment, and obtaining motion vector information based on the preset optical flow method according to the image data includes: Real-time collecting the original images of insulators and the on-site environment, and preprocessing the original images to obtain image data; Performing principal component analysis on several insulator image samples to extract key feature vectors; Constructing a direction weight matrix according to the correlation between the key feature vectors and directions; Improving the gradient method in the preset optical flow method based on the direction weight matrix to obtain an improved gradient method; Obtaining motion vector information according to the improved gradient method and the image data.

3. The method according to claim 2, wherein The step of obtaining motion vector information according to the improved gradient method and the image data includes: Calculating the first direction gradient, the second direction gradient, and the time direction gradient according to the improved gradient method and the image data; Calculating the first direction motion vector and the second direction motion vector respectively according to the first direction gradient, the second direction gradient, the time direction gradient, and the direction weight matrix; Obtaining motion vector information according to the first direction motion vector and the second direction motion vector.

4. The method according to claim 1, characterized in that, The step of analyzing the motion vector information based on a deep learning algorithm to obtain the motion state information of the insulators includes: Constructing a deep learning model based on a multi-level cascaded attention network; Inputting the motion vector information into the deep learning model to identify the moving insulators and obtain the key motion features related to the insulators; wherein, the key motion features include the motion direction and speed information of the insulators; Generating the motion state information of the insulators according to the key motion features.

5. The method according to claim 1, characterized in that The step of based on the dynamic fusion algorithm, combining the motion state information of the insulators with the position information of the upper limb exoskeleton to obtain the adjustment motion parameters of the upper limb exoskeleton includes: Setting the expected position of the end effector of the upper limb exoskeleton, and obtaining the current position and the joint angles of the end effector of the upper limb exoskeleton; Obtaining a dynamic weight vector according to the motion state information of the insulators combined with the historical operation data of the upper limb exoskeleton; Based on the dynamic fusion algorithm, obtaining the joint angle adjustment amount according to the expected position, the current position, the joint angles, and the dynamic weight vector; Taking the joint angle adjustment amount as the adjustment motion parameters of the upper limb exoskeleton.

6. The method according to claim 5, wherein The step of obtaining a dynamic weight vector according to the motion state information of the insulators combined with the historical operation data of the upper limb exoskeleton includes: Obtaining the historical operation data of the upper limb exoskeleton in a preset environment; Real-time obtaining the current wind direction and wind speed information; Evaluating the complexity of the operation environment according to the motion state information of the insulators combined with the historical operation data and the current wind direction and wind speed information to obtain a dynamic weight vector.

7. The method according to any one of claims 1 to 6, characterized in that, The method further includes: Establish a mapping model between environmental parameters and insulator motion interference; Real-time monitor the environmental parameters at the operation site; Conduct real-time analysis based on the environmental parameters to obtain the change trend of the environmental parameters, and determine whether it interferes with the insulator motion according to the change trend of the environmental parameters; If so, generate a compensation instruction according to the mapping model and the environmental parameters; Based on the compensation instruction, adjust and correct the motion trajectory of the upper limb exoskeleton.

8. The method according to claim 7, wherein The establishment of the mapping model between environmental parameters and insulator motion interference includes: Generate original environmental parameters according to wind speed, temperature, humidity, and equipment vibration frequency; Generate data samples based on experimental observations of insulator motion under different combinations of original environmental parameters; Train a multiple linear regression model based on particle swarm optimization according to the data samples to obtain a mapping model between environmental parameters and insulator motion interference.

9. An upper limb exoskeleton control system for assisting in replacing insulators, characterized in that, Include: An optical flow monitoring module, configured to collect image data of the insulator and the on-site environment, and obtain motion vector information based on the preset optical flow method according to the image data; An insulator analysis module, configured to analyze the motion vector information based on a deep learning algorithm to obtain the motion state information of the insulator; A trajectory adjustment module, configured to obtain the adjusted motion parameters of the upper limb exoskeleton based on a dynamics fusion algorithm according to the motion state information of the insulator combined with the position information of the upper limb exoskeleton; A motion control module, configured to issue a control instruction to the upper limb exoskeleton according to the adjusted motion parameters.

10. An electronic device, characterized in that, The electronic device includes: a memory, a processor, and a control program for the upper limb exoskeleton for assisting in replacing insulators stored on the memory and executable on the processor. The control program for the upper limb exoskeleton for assisting in replacing insulators is configured to implement the method for controlling the upper limb exoskeleton for assisting in replacing insulators according to any one of claims 1 to 8.

Images