Network data transmission method and system for network virtual surgery

Abstract

The invention discloses a network data transmission method and system for a network virtual surgery, and the method comprises the steps: obtaining a work online feature of the network virtual surgery, and inputting the work online feature into a pre-trained back-propagation neural network to obtain four monitoring output features; calculating to obtain a data compression rate, a prediction compensation window size, redundancy retransmission times and a local interpolation smoothing factor according to the four monitoring output characteristics; data transmission is executed after working data of the network virtual surgery is compressed according to the data compression rate; in the data transmission process of the network virtual surgery, delay is compensated based on a Kalman predictor predicting the size of a compensation window, redundant retransmission is triggered according to the number of redundant retransmission times to resist packet loss, and a local interpolation smoothing factor is substituted into an interpolation filter to smooth network jitter; a back propagation neural network is combined with a virtual operation force feedback system, and adaptive adjustment of delay, jitter and packet loss problems in a networked virtual operation is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of network data transmission, and particularly relates to a network data transmission method and system for network virtual surgery. Background Art

[0002] The virtual surgery system pursues high authenticity and stability. By simulating multi-modal feedback in real surgery, it provides a more realistic operation experience for virtual surgery. Therefore, a network data transmission method for network virtual surgery is of great significance for improving the surgical training effect and surgical safety.

[0003] Regarding the problem of aortic tumor growth, the prior art uses the finite element method (FEM) to establish a multi-scale fluid-structure interaction model. Such methods achieve haptic rendering through physical modeling, but have problems of large data volume and long calculation time, and it is difficult to meet the real-time requirements. The computational complexity of multi-scale fluid-structure interaction leads to a significant delay in system response. Especially in scenarios with frequent interactions, the local lag effect will reduce the real-time performance and realism of force feedback.

[0004] The multi-scale fluid-structure interaction model lacks an adaptive compensation mechanism for dynamic lag error and cannot adjust the output weight according to the real-time actions of the operator. In addition, the stability and refresh rate of the platform also need to be further improved to prevent signal interruption or data loss. Summary of the Invention

[0005] The present invention provides a network data transmission method and system for network virtual surgery, which combines a backpropagation neural network with a virtual surgery force feedback system to achieve adaptive adjustment of delay, jitter, and packet loss problems in networked virtual surgery.

[0006] To achieve the above object, the technical solution adopted by the present invention is:

[0007] The first aspect of the present invention provides a network data transmission method for network virtual surgery, including:

[0008] Obtain the working online characteristics of network virtual surgery, input the working online characteristics into a pre-trained backpropagation neural network to obtain four monitored output characteristics; calculate the data compression ratio, prediction compensation window size, redundant retransmission times, and local interpolation smoothing factor according to the four monitored output characteristics;

[0009] Perform data transmission after compressing the working data of network virtual surgery according to the data compression ratio; during the data transmission process of network virtual surgery, compensate for the delay with a Kalman predictor based on the prediction compensation window size, trigger redundant retransmission to counter packet loss according to the redundant retransmission times, and substitute the local interpolation smoothing factor into an interpolation filter to smooth network jitter.

[0010] Further, the backpropagation neural network includes an input layer, a first hidden layer, a second hidden layer, and an output layer; the activation functions in the first hidden layer and the second hidden layer are set to function; the activation function of the output layer is set to function.

[0011] Further, the training process of the backpropagation neural network includes:

[0012] Obtain the working training features of the network virtual surgery and input them into the input layer of the backpropagation neural network. Feature extraction is performed on the working training features through the first hidden layer to obtain the first intermediate features, and the first intermediate features are input into the second hidden layer to obtain the second intermediate features;

[0013] Cluster the second intermediate features through the output layer to obtain the training output features, and calculate the first training loss value of the backpropagation neural network according to the training output features; use the first training loss value to optimize the weights of the backpropagation neural network,

[0014] Set the transmission parameters of the network virtual surgery based on the training output features and perform data transmission, and record the working monitoring features during the data transmission process of the network virtual surgery; calculate the second training loss value according to the working monitoring features recorded twice at intervals of the set time, and use the second training loss value to optimize the weights of the backpropagation neural network;

[0015] Use the working monitoring features as the working training features to retrain the backpropagation neural network, and repeat the training process of the backpropagation neural network iteratively until the trained backpropagation neural network is output when the preset threshold is reached.

[0016] Further, the working online features, working monitoring features, and working training features include the end-to-end delay , packet loss rate , bandwidth utilization , network jitter , instrument movement speed , the proportion of the key tissue deformation area , and force feedback error .

[0017] Further, feature extraction is performed on the working training features through the first hidden layer to obtain the first intermediate features, which specifically includes:

[0018]

[0019] In the formula, represents the first intermediate feature output by the th node of the first hidden layer, is the activation function; is the number of nodes in the input layer; is the node number in the input layer; represents the th node in the input layer to the th node in the first hidden layer, represents the working training feature received by the th node in the input layer, represents the bias term of the th node in the first hidden layer.

[0020] Furthermore, input the first intermediate feature into the second hidden layer to obtain the second intermediate feature, which specifically includes:

[0021]

[0022] In the formula, represents the second intermediate feature output by the th node in the second hidden layer, is the number of nodes in the first hidden layer; is the node number in the first hidden layer; represents the th node in the first hidden layer to the th node in the second hidden layer, represents the first intermediate feature output by the th node in the first hidden layer, represents the th node in the second hidden layer.

[0023] Furthermore, cluster the second intermediate feature through the output layer to obtain the training output feature, which specifically includes:

[0024]

[0025]

[0026]

[0027]

[0028] In the formula, , , and are respectively the training output features output by each node in the output layer; is the activation function; is the number of nodes in the second hidden layer; is the node number in the second hidden layer; is the The connection weight from the -th node of the second hidden layer to the first node of the output layer, The connection weight from the -th node of the second hidden layer to the second node of the output layer, The connection weight from the -th node of the second hidden layer to the third node of the output layer, The connection weight from the -th node of the second hidden layer to the fourth node of the output layer; The second intermediate feature output by the -th node of the second hidden layer; The bias term of the first node of the output layer, The bias term of the second node of the output layer, The bias term of the third node of the output layer,

[0029]

[0030]

[0031]

[0032]

[0033]

[0034] Among them, represents the loss function, represents the weight coefficient of the effective delay, represents the effective delay, represents the weight coefficient of the force feedback error, represents the predicted force feedback value, represents the actual force feedback value, represents the weight coefficient of the compression rate fluctuation term, represents the change rate of the compression rate over time, represents the data compression rate, represents the time, represents the weight coefficient of the penalty term for the number of redundant retransmissions, represents the penalty term for the number of redundant retransmissions, is the number of redundant retransmissions; is the end-to-end delay in the network virtual surgery process, is the local interpolation smoothing factor, is the weight of the predicted compensation window size,​​ is the predicted compensation window size; , , and are the training output features output by each node of the output layer respectively.

[0035] Furthermore, the second training loss value is calculated every set time interval based on the working monitoring features recorded in two adjacent times, specifically including:

[0036] Obtain the working monitoring features recorded in two adjacent times currently every set time interval, and use the working monitoring features as training monitoring samples. The training monitoring samples include the actual lag time of the network virtual surgery, the data compression rate, the force feedback value, and the number of redundant retransmissions;

[0037] Calculate the second training loss value according to the working monitoring features, and the expression formula is:

[0038] ;

[0039] ;

[0040] In the formula, , , are the weights of the backpropagation neural network after the , , th sliding sampling window training respectively; represents the learning rate, is the number of training monitoring samples in the sliding supplementary window; is the change gradient of the loss value in the th training iteration, represents the loss value in the th training iteration; represents the L2 regularization coefficient; is the set weight; is the actual lag time after the adjustment strategy of the backpropagation neural network in the th training iteration; represents the predicted force feedback value in the th training iteration, represents the actual force feedback value in the th training iteration; is the unit time of each training iteration of the backpropagation neural network; and are the data compression rates in the , th training iterations respectively; is the number of redundant retransmissions in the th training iteration.

[0041] Further, it further includes: when the proportion of the key tissue deformation area in the network virtual surgery is greater than a certain value, restricting the data compression rate to be greater than a certain value; when the instrument movement speed in the network virtual surgery exceeds the set speed threshold, increasing the size of the prediction compensation window according to a set ratio ; when the force feedback error in the network virtual surgery exceeds the set error threshold, increasing the number of redundant retransmissions according to a set ratio.

[0042] The second aspect of the present invention provides a network data transmission system for network virtual surgery, including:

[0043] An acquisition module, configured to obtain the online working characteristics of the network virtual surgery;

[0044] A prediction module, configured to input the online working characteristics into a pre-trained backpropagation neural network to obtain four monitored output characteristics; calculate the data compression rate, the size of the prediction compensation window, the number of redundant retransmissions, and the local interpolation smoothing factor according to the four monitored output characteristics;

[0045] An execution module, which compresses the working data of the network virtual surgery according to the data compression rate and then performs data transmission; during the data transmission process of the network virtual surgery, compensates for the delay based on the Kalman predictor with the size of the prediction compensation window, triggers redundant retransmissions to counter packet loss according to the number of redundant retransmissions, and substitutes the local interpolation smoothing factor into the interpolation filter to smooth network jitter.

[0046] Further, it further includes a training module, and the training module trains the backpropagation neural network, specifically including:

[0047] Obtaining the working training characteristics of the network virtual surgery and inputting them into the input layer of the backpropagation neural network, extracting features from the working training characteristics through the first hidden layer to obtain the first intermediate feature, and inputting the first intermediate feature into the second hidden layer to obtain the second intermediate feature;

[0048] Performing clustering on the second intermediate feature through the output layer to obtain the training output feature, calculating the first training loss value of the backpropagation neural network according to the training output feature; optimizing the weights of the backpropagation neural network using the first training loss value,

[0049] Setting the transmission parameters of the network virtual surgery based on the training output feature and performing data transmission, recording the working monitoring characteristics during the data transmission process of the network virtual surgery; calculating the second training loss value according to the adjacent two recorded working monitoring characteristics every set time interval, and optimizing the weights of the backpropagation neural network using the second training loss value;

[0050] Retrain the backpropagation neural network with the work monitoring features as work training features, and repeat the training process of the backpropagation neural network until the preset threshold is reached to output the trained backpropagation neural network.

[0051] The third aspect of the present invention provides an electronic device, including a storage medium and a processor; the storage medium is used to store instructions; the processor is used to operate according to the instructions to execute the network data transmission method for network virtual surgery described in the first aspect.

[0052] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0053] The present invention combines the backpropagation neural network with the virtual surgery force feedback system, realizing the adaptive adjustment of the problems of delay, jitter, and packet loss in networked virtual surgery, thereby maintaining the best performance under different network conditions and surgical scenarios and enhancing the robustness of the system.

[0054] Through the Kalman predictor based on the predicted compensation window size, the present invention can effectively compensate for the delay in network transmission and ensure the real-time nature of the force feedback signal; substituting the local interpolation smoothing factor into the interpolation filter to smooth the influence of network jitter on the force feedback signal further improves the stability and accuracy of the force feedback; triggering redundant retransmission according to the number of redundant retransmission times can effectively combat the problem of network packet loss.

[0055] By monitoring the network status and surgical operation changes in real time, the present invention dynamically adjusts the neural network weights and transmission parameters, significantly enhancing the adaptability and stability of the system. Using the work monitoring features as training data to continuously iterate and optimize the network enables the system to better adapt to complex network environments and surgical requirements, improving the authenticity of the force feedback and the overall performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 is a flowchart of a network data transmission method for network virtual surgery provided in Embodiment 1 of the present invention;

[0057] Figure 2 is a structural diagram of the backpropagation neural network provided in Embodiment 1 of the present invention. DETAILED DESCRIPTION

[0058] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and should not be used to limit the protection scope of the present invention.

[0059] Embodiment 1

[0060] As Figures 1 to 2 shown, this embodiment provides a network data transmission method for network virtual surgery, including:

[0061] Construct the input layer, the first hidden layer, the second hidden layer and the output layer of the backpropagation neural network; in this embodiment, the input layer has 7 nodes, the first hidden layer and the second hidden layer have 12 nodes, and the output layer has 4 nodes; the activation functions in the first hidden layer and the second hidden layer are set to function; the activation function of the output layer is set to A mixed activation of function and linear activation.

[0062] Train the backpropagation neural network, specifically including:

[0063] Obtain the working training features, which include the end-to-end delay , packet loss rate , bandwidth utilization , network jitter , instrument movement speed , the proportion of the key area of tissue deformation and force feedback error

[0064] Input the working training features into the input layer of the backpropagation neural network, and extract the features of the working training features through the first hidden layer to obtain the first intermediate feature. The expression formula is:

[0065]

[0066] In the formula, represents the first intermediate feature output by the th node of the first hidden layer, is the activation function; is the number of nodes in the input layer. In this embodiment, ; is the node serial number in the input layer; represents the th node of the input layer to the th node of the first hidden layer. The connection weight, represents the working training feature received by the th node of the input layer, represents the th node of the first hidden layer. The bias term.

[0067] Input the first intermediate feature into the second hidden layer to obtain the second intermediate feature; the expression formula is:

[0068]

[0069] In the formula, represents the The second intermediate feature output by the node, is the number of nodes in the first hidden layer. In this embodiment, ; is the node serial number in the first hidden layer; represents the connection weight from the th node in the first hidden layer to the th node in the second hidden layer, represents the first intermediate feature output by the th node in the first hidden layer, represents the bias term of the th node in the second hidden layer.

[0070] The training output feature is obtained by clustering the second intermediate feature through the output layer, specifically including:

[0071]

[0072]

[0073]

[0074]

[0075] In the formula, is the training output feature output by the first node in the output layer, is the training output feature output by the second node in the output layer, is the training output feature output by the third node in the output layer, is the training output feature output by the fourth node in the output layer; is the activation function; is the number of nodes in the second hidden layer. In this embodiment, ; is the node serial number in the second hidden layer; is the connection weight from the th node in the second hidden layer to the first node in the output layer, is the connection weight from the th node in the second hidden layer to the second node in the output layer, is the connection weight from the th node in the second hidden layer to the third node in the output layer, is the connection weight from the th node in the second hidden layer to the fourth node in the output layer; represents the second intermediate feature output by the th node in the second hidden layer; is the bias term of the first node in the output layer, is the bias term of the second node in the output layer, is the bias term of the 3rd node in the output layer, is the bias term of the 4th node in the output layer.

[0076] Calculate the first training loss value of the backpropagation neural network according to the training output features; the expression formula is:

[0077]

[0078]

[0079] , ,

[0080]

[0081] where, represents the loss function, represents the weight coefficient of the effective delay, represents the effective delay, represents the weight coefficient of the force feedback error, represents the predicted force feedback value, represents the actual force feedback value, represents the weight coefficient of the compression rate fluctuation term, represents the change rate of the compression rate over time, represents the data compression rate, represents time, represents the weight coefficient of the penalty term for the number of redundant retransmissions, represents the penalty term for the number of redundant retransmissions, is the number of redundant retransmissions; is the end-to-end delay in the network virtual surgery process, is the local interpolation smoothing factor, is the weight of the prediction compensation window size, is the prediction compensation window size; is the training output feature output by the 3rd node in the output layer, is the training output feature output by the 1st node in the output layer, is the training output feature output by the 4th node in the output layer, is the training output feature output by the 2nd node in the output layer.

[0082] Optimize the weights of the backpropagation neural network using the first training loss value, specifically including:

[0083]

[0084]

[0085]

[0086]

[0087] In the formula, represents the gradient of the data compression ratio; represents the gradient of the prediction compensation window size, represents the learning rate; represents the gradient of the redundant retransmission times; represents the gradient of the local interpolation smoothing factor;

[0088] Calculate the gradients of the nodes in the second hidden layer and the first hidden layer, specifically including:

[0089]

[0090]

[0091] In the formula, represents the gradient of the th node in the second hidden layer, represents the gradient of the th node in the output layer, represents the connection weight from the th node in the second hidden layer to the th node in the output layer, represents the indicator function, which takes 1 when and 0 otherwise;

[0092] represents the gradient of the th node in the first hidden layer, represents the gradient of the th node in the second hidden layer, represents the weight from the th node in the first hidden layer to the th node in the second hidden layer; represents the indicator function, which takes 1 when and 0 otherwise;

[0093]

[0094]

[0095]

[0096] In the formula, represents the learning rate, represents the L2 regularization coefficient.

[0097] Set the transmission parameters of the network virtual surgery based on the training output features and perform data transmission, and record the working monitoring features in the data transmission process of the network virtual surgery;

[0098] Calculate the second training loss value according to the working monitoring features recorded twice adjacent to each other at set time intervals, specifically including:

[0099] Obtain the working monitoring features recorded twice adjacent to each other at set time intervals, and use the working monitoring features as training monitoring samples. The training monitoring samples include the actual lag time, data compression rate, force feedback value, and redundant retransmission times of the network virtual surgery;

[0100] Calculate the second training loss value according to the working monitoring features, and the expression formula is:

[0101] ;

[0102] ;

[0103] In the formula, , , are the weights of the backpropagation neural network after the , , rd sliding sampling window training respectively; represents the learning rate, is the number of training monitoring samples in the sliding supplementary window; is the change gradient of the loss value in the th training iteration, represents the loss value in the th training iteration; represents the L2 regularization coefficient; is the set weight; is the actual lag time after the adjustment strategy of the backpropagation neural network in the th training iteration; represents the predicted force feedback value in the th training iteration, represents the actual force feedback value in the th training iteration; and are the data compression rates in the , th training iterations respectively; is the unit time for each training iteration of the backpropagation neural network; is the redundant retransmission times in the th training iteration.

[0104] Optimize the weights of the backpropagation neural network using the second training loss value; retrain the backpropagation neural network with the working monitoring features as the working training features, and repeat the training process of the backpropagation neural network iteratively until the trained backpropagation neural network is output when the preset threshold is reached.

[0105] In this embodiment, by monitoring the network status and surgical operation changes in real time, the neural network weights and transmission parameters are dynamically adjusted, significantly enhancing the adaptability and stability of the system. Using the working monitoring features as training data and continuously iteratively optimizing the network enables the system to better adapt to complex network environments and surgical requirements, improving the authenticity of force feedback and the overall performance.

[0106] Obtain the working online features of network virtual surgery. The working training features include the end-to-end delay , packet loss rate , bandwidth utilization , network jitter , instrument movement speed , the proportion of the key area of tissue deformation , and force feedback error .

[0107] Input the working online features into the pre-trained backpropagation neural network to obtain four monitoring output features; calculate the data compression rate, prediction compensation window size, redundant retransmission times, and local interpolation smoothing factor based on the four monitoring output features;

[0108] Perform data transmission after compressing the working data of network virtual surgery according to the data compression rate; during the data transmission of network virtual surgery, compensate for the delay with a Kalman predictor based on the prediction compensation window size, trigger redundant retransmission to counter packet loss according to the redundant retransmission times, and substitute the local interpolation smoothing factor into the interpolation filter to smooth the network jitter.

[0109] When the proportion of the key area of tissue deformation in network virtual surgery is greater than , limit the data compression rate to be greater than ; when the instrument movement speed in network virtual surgery exceeds the set speed threshold, increase the prediction compensation window size according to the set ratio ; when the force feedback error in network virtual surgery exceeds the set error threshold, increase the redundant retransmission times according to the set ratio.

[0110] The Kalman predictor based on the predicted compensation window size can effectively compensate for the delay in network transmission and ensure the real-time performance of the force feedback signal. Substituting the local interpolation smoothing factor into the interpolation filter can smooth the influence of network jitter on the force feedback signal, further improving the stability and accuracy of the force feedback. Triggering redundant retransmission according to the number of redundant retransmissions can effectively combat the problem of network packet loss.

[0111] In this embodiment, the backpropagation neural network is combined with the virtual surgery force feedback system to achieve adaptive adjustment of the problems of delay, jitter, and packet loss in networked virtual surgery, so as to maintain the best performance under different network conditions and surgical scenarios and enhance the robustness of the system.

[0112] Embodiment 2

[0113] This embodiment provides a network data transmission system for network virtual surgery. The network data transmission system can execute the network data transmission method described in Embodiment 1. The network data transmission system includes:

[0114] An acquisition module for obtaining the online working characteristics of network virtual surgery;

[0115] A prediction module for inputting the online working characteristics into the pre-trained backpropagation neural network to obtain four monitored output characteristics; calculating the data compression rate, the predicted compensation window size, the number of redundant retransmissions, and the local interpolation smoothing factor according to the four monitored output characteristics;

[0116] An execution module for performing data transmission after compressing the working data of network virtual surgery according to the data compression rate; during the data transmission of network virtual surgery, compensating for the delay with a Kalman predictor based on the predicted compensation window size, triggering redundant retransmission to combat packet loss according to the number of redundant retransmissions, and substituting the local interpolation smoothing factor into the interpolation filter to smooth network jitter;

[0117] A training module for training the backpropagation neural network.

[0118] The training module trains the backpropagation neural network, specifically including:

[0119] Obtaining the working training characteristics of network virtual surgery and inputting them into the input layer of the backpropagation neural network, extracting features from the working training characteristics through the first hidden layer to obtain the first intermediate feature, and inputting the first intermediate feature into the second hidden layer to obtain the second intermediate feature;

[0120] Performing clustering on the second intermediate feature through the output layer to obtain the training output feature, calculating the first training loss value of the backpropagation neural network according to the training output feature; optimizing the weights of the backpropagation neural network using the first training loss value,

[0121] Set the transmission parameters of the network virtual surgery based on the training output features and perform data transmission, and record the working monitoring features of the data transmission process in the network virtual surgery; calculate the second training loss value according to the working monitoring features recorded twice adjacent to each other at intervals of a set time, and use the second training loss value to optimize the weights of the backpropagation neural network;

[0122] Use the working monitoring features as working training features to retrain the backpropagation neural network, and repeat the training process of the backpropagation neural network iteratively until the trained backpropagation neural network is output when a preset threshold is reached.

[0123] Embodiment 3

[0124] This embodiment provides an electronic device, including a storage medium and a processor; the storage medium is used for storing instructions; the processor is used for operating according to the instructions to execute the network data transmission method for network virtual surgery described in Embodiment 1.

[0125] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0126] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram can be realized by computer program instructions, and the combination of the flows and / or blocks in the flowchart and / or block diagram can also be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0127] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured product including instruction means, and the instruction means realizes the functions specified in Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0128] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are executed on the computer or other programmable apparatus to produce a computer-implemented process, thereby the instructions executed on the computer or other programmable apparatus provide steps for realizing the functions specified in one process Figure 1 one process or a plurality of processes and / or blocks Figure 1 or a plurality of blocks.

[0129] The above are only the preferred embodiments of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and modifications can be made, and these improvements and modifications should also be regarded as the protection scope of the present invention.

Claims

1. A network data transmission method for network virtual surgery, characterized in that, Including: Obtain the online working characteristics of the network virtual surgery, and input the online working characteristics into the pre-trained backpropagation neural network to obtain four monitored output characteristics; Calculate the data compression rate, prediction compensation window size, redundant retransmission times, and local interpolation smoothing factor based on the four monitored output characteristics; Perform data transmission after compressing the working data of the network virtual surgery according to the data compression rate; during the data transmission process of the network virtual surgery, compensate for the delay with a Kalman predictor based on the prediction compensation window size, trigger redundant retransmission to counter packet loss according to the redundant retransmission times, and substitute the local interpolation smoothing factor into the interpolation filter to smooth the network jitter.

2. The network data transmission method according to claim 1, wherein The backpropagation neural network includes an input layer, a first hidden layer, a second hidden layer, and an output layer; The training process of the backpropagation neural network includes: Obtain the working training characteristics of the network virtual surgery and input them into the input layer of the backpropagation neural network. Extract features from the working training characteristics through the first hidden layer to obtain the first intermediate feature, and input the first intermediate feature into the second hidden layer to obtain the second intermediate feature; Cluster the second intermediate feature through the output layer to obtain the training output feature, and calculate the first training loss value of the backpropagation neural network according to the training output feature; use the first training loss value to optimize the weights of the backpropagation neural network, Set the transmission parameters of the network virtual surgery based on the training output feature and perform data transmission, and record the working monitoring characteristics during the data transmission process of the network virtual surgery; calculate the second training loss value according to the adjacent two recorded working monitoring characteristics every set time interval, and use the second training loss value to optimize the weights of the backpropagation neural network; Use the working monitoring characteristics as the working training characteristics to retrain the backpropagation neural network, and repeat the training process of the backpropagation neural network until the preset threshold is reached to output the trained backpropagation neural network.

3. The network data transmission method according to claim 2, wherein The working online features, working monitoring features, and working training features include the end-to-end delay during the network virtual surgery process , packet loss rate , bandwidth utilization , network jitter , instrument movement speed , the proportion of the key area of tissue deformation , and force feedback error .

4. The network data transmission method according to claim 2, wherein Extract features from the working training characteristics through the first hidden layer to obtain the first intermediate feature, and input the first intermediate feature into the second hidden layer to obtain the second intermediate feature, specifically including: ； In the formula, represents the first intermediate feature output by the th node of the first hidden layer, is the activation function; is the number of nodes in the input layer; is the node serial number in the input layer; represents the connection weight from the th node of the input layer to the th node of the first hidden layer, represents the working training feature received by the th node of the input layer, represents the bias term of the th node of the first hidden layer; ； In the formula, represents the second intermediate feature output by the th node of the second hidden layer, is the number of nodes in the first hidden layer; is the node number in the first hidden layer; represents the connection weight from the th node of the first hidden layer to the th node of the second hidden layer, represents the bias term of the th node of the second hidden layer.

5. The network data transmission method according to claim 2, wherein Cluster the second intermediate feature through the output layer to obtain the training output feature, specifically including: ； ； ； ； In the formula, , , and are the training output features output by each node in the output layer respectively; is the activation function; is the number of nodes in the second hidden layer; is the node serial number in the second hidden layer; , , and are the connection weights from the -th node in the second hidden layer to each node in the output layer respectively; represents the second intermediate feature output by the -th node in the second hidden layer; , , and are the bias terms of each node in the output layer.

6. The network data transmission method according to claim 5, wherein Calculate the first training loss value of the backpropagation neural network according to the training output feature, specifically including: ； ； ， ， ； ； Among them, represents the loss function, represents the weight coefficient of the effective delay, represents the effective delay, represents the weight coefficient of the force feedback error, represents the predicted force feedback value, represents the actual force feedback value, represents the weight coefficient of the compression rate fluctuation term, represents the change rate of the compression rate over time, represents the data compression rate, represents time, represents the weight coefficient of the penalty term for the number of redundant retransmissions, represents the penalty term for the number of redundant retransmissions, is the number of redundant retransmissions; is the end-to-end delay in the network virtual surgery process, is the local interpolation smoothing factor, is the weight of the prediction compensation window size, is the prediction compensation window size; , , and are the training output features output by each node in the output layer, respectively.

7. The network data transmission method according to claim 6, wherein Calculate the second training loss value according to the adjacent two recorded working monitoring characteristics every set time interval, specifically including: Obtain the current adjacent two recorded working monitoring characteristics every set time interval, and use the working monitoring characteristics as the training monitoring samples. The training monitoring samples include the actual lag time, data compression rate, force feedback value, and redundant retransmission times of the network virtual surgery; Calculate the second training loss value according to the working monitoring characteristics, and the expression formula is: ； ； In the formula, , , are respectively the weights of the backpropagation neural network after the , , -th sliding sampling window training; represents the learning rate, is the number of training monitoring samples within the sliding sampling window; is the change gradient of the loss value in the -th training iteration, represents the loss value in the -th training iteration; represents the L2 regularization coefficient; is the set weight; is the actual lag time after the adjustment strategy of the backpropagation neural network in the -th training iteration; represents the predictive force feedback value in the -th training iteration, represents the actual force feedback value in the -th training iteration; and are respectively the data compression ratios in the , -th training iteration; is the unit time for each training iteration of the backpropagation neural network; is the number of redundant retransmissions in the -th training iteration.

8. The network data transmission method according to claim 1, wherein Also including: When the proportion of the key area of tissue deformation in network virtual surgery is greater than , the data compression ratio is restricted to be greater than ; when the instrument movement speed in network virtual surgery exceeds the set speed threshold, the size of the prediction compensation window is increased according to the set ratio; when the force feedback error in network virtual surgery exceeds the set error threshold, the number of redundant retransmissions is increased according to the set ratio.

9. A network data transmission system for network virtual surgery, characterized in that, Including: An acquisition module for obtaining the online working characteristics of the network virtual surgery; A prediction module for inputting the online working characteristics into the pre-trained backpropagation neural network to obtain four monitored output characteristics; Calculate the data compression rate, prediction compensation window size, redundant retransmission times, and local interpolation smoothing factor based on the four monitored output characteristics; The execution module compresses the working data of the network virtual surgery according to the data compression ratio and then performs data transmission; during the data transmission of the network virtual surgery, the Kalman predictor based on the predicted compensation window size compensates for the delay, triggers redundant retransmission to counter packet loss according to the number of redundant retransmission times, and substitutes the local interpolation smoothing factor into the interpolation filter to smooth the network jitter.

10. An electronic device, comprising a storage medium and a processor; the storage medium is used for storing instructions; characterized in that, The processor is configured to operate according to the instruction to execute the network data transmission method for network virtual surgery according to any one of claims 1 to 8.

Images