A temperature prediction based ultrasonic blade protection method and apparatus

Abstract

The application relates to the technical field of ultrasonic equipment, and provides an ultrasonic knife protection method and device based on temperature prediction. The temperature extreme value and average temperature trend of a knife head are predicted, and adaptive discrimination is performed based on the temperature extreme value and average temperature trend. When shearing has been completed or the temperature of the knife head area exceeds the melting point of a gasket, the output power can be reduced in time, the ultrasonic knife wear can be reduced, the ultrasonic knife use time can be prolonged, and therefore the ultrasonic knife can be effectively protected. When continuous excitation or continuous shearing of the ultrasonic knife is performed, the shearing efficiency of the ultrasonic knife can be further improved on the basis of reliable protection of the gasket, adaptive switching is performed between ultrasonic knife protection and high-efficiency shearing states, the operation mode of an operator is adaptively adjusted, adaptive discrimination is performed by using the temperature extreme value and average temperature trend, compared with a mode of protecting the knife head based on single temperature prediction, the technical scheme of the application can improve the accuracy and reliability of knife head protection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ultrasonic equipment, and particularly relates to an ultrasonic knife protection method and device based on temperature estimation. BACKGROUND

[0002] The ultrasonic knife transmits electric energy to a piezoelectric transducer through an energy generator, converts the electric energy into ultrasonic mechanical energy by the piezoelectric transducer, thereby obtains ultrasonic vibration, and amplifies the ultrasonic vibration by a knife head of the ultrasonic knife, for shearing the cut object.

[0003] When shearing the cut object, the operator doctor habitually uses continuous excitation of the ultrasonic knife head to shear the cut object and the free object in order to improve the shearing operation efficiency of the cut object and the free object. Since the operation is continuous excitation, when the shearing is completed, the high energy output is still maintained, which causes the jaw of the ultrasonic knife to always wear the pad, thereby causing the knife head of the ultrasonic knife to generate a high temperature, easily causing the melting and loss of the pad, and reducing the service life of the ultrasonic knife. SUMMARY

[0004] The present application provides an ultrasonic knife protection method and device based on temperature estimation, which avoids the jaw of the ultrasonic knife always wearing the pad when the operator doctor uses the ultrasonic knife to shear the cut object, thereby avoiding the knife head of the ultrasonic knife generating a high temperature, causing the melting and loss of the pad, and reducing the service life of the ultrasonic knife.

[0005] By predicting the temperature extreme value and average temperature trend of the knife head, and adaptively discriminating based on the temperature extreme value and average temperature trend, when the shearing is completed or the temperature of the knife head region exceeds the melting point of the pad, the output power can be reduced in time to reduce the wear of the ultrasonic knife, prolong the use time of the ultrasonic knife, thereby effectively protecting the ultrasonic knife. When the ultrasonic knife is continuously excited or continuously sheared, the shearing efficiency of the ultrasonic knife can be further improved on the basis of reliably protecting the pad, adaptively switches between the ultrasonic knife protection and the high-efficiency shearing state, adaptively operates the operation mode of the operator, and adaptively discriminates by using the temperature extreme value and average temperature trend. Compared with the mode of protecting the knife head based on single temperature prediction, the technical scheme of the present application can improve the accuracy and reliability of the knife head protection.

[0006] The present application provides an ultrasonic knife protection method based on temperature estimation, comprising:

[0007] In the case that the ultrasonic knife shears the cut object, the temperature extreme value and average temperature trend of the knife head are predicted;

[0008] If the temperature extreme value is greater than or equal to a first temperature threshold, the output current of the ultrasonic knife generator is adjusted to a first current;

[0009] if the temperature extreme value is less than the first temperature threshold and greater than or equal to the second temperature threshold, obtaining a shear-off point prediction result of the ultrasonic knife, and if the shear-off point prediction result is true, adjusting the output current of the ultrasonic knife generator to the first current, the first temperature threshold being greater than the second temperature threshold;

[0010] after the output current of the ultrasonic knife generator is adjusted to the first current, if the average temperature trend indicates that the ultrasonic knife is in a continuous shearing state, adjusting the output current of the ultrasonic knife generator to a second current, the second current being greater than the first current.

[0011] In a possible implementation, in the step of determining, according to the average temperature trend, that the ultrasonic knife is in a continuous shearing state, the following steps are included:

[0012] if, within a first preset time, the average temperature of the knife head decreases by more than a first preset value, and within a second preset time, the average temperature of the knife head increases to be greater than a second preset value, it is determined that the ultrasonic knife is in a continuous shearing state, wherein the second preset time is later than the first preset time.

[0013] In a possible implementation, in the step of predicting the temperature extreme value and the average temperature trend of the ultrasonic knife, the following steps are included:

[0014] in the case that the ultrasonic knife is shearing a cut object, obtaining the output voltage, the output current, the first derivative of the resonant frequency and the voltage-current phase difference of the ultrasonic knife generator;

[0015] inputting the output voltage, the output current, the first derivative of the resonant frequency and the voltage-current phase difference into a pre-established extreme temperature prediction model to obtain a temperature extreme value output by the extreme temperature prediction model.

[0016] In a possible implementation, before the step of predicting the temperature extreme value and the average temperature trend of the ultrasonic knife, the following steps are included:

[0017] obtaining the output voltage, the output current, the first derivative of the resonant frequency and the voltage-current phase difference of the ultrasonic knife generator at the same historical moment, and the temperature extreme value corresponding to the same historical moment;

[0018] establishing a first initial model, taking the output voltage, the output current, the first derivative of the resonant frequency and the voltage-current phase difference at the same historical moment as the input of the first initial model, taking the temperature extreme value corresponding to the same historical moment as the output of the first initial model, and training the first initial model to obtain an extreme temperature prediction model.

[0019] In a possible implementation, in the case that the ultrasonic knife shears the cutting object, in the step of predicting the temperature extreme value and the average temperature trend of the ultrasonic knife, the step comprises the following steps of:

[0020] In the case that the ultrasonic knife shears the cutting object, the phase difference between the output current and the voltage current of the ultrasonic knife generator is acquired;

[0021] The phase difference between the output current and the voltage current is input into the pre-established temperature trend prediction model to obtain the average temperature trend output by the temperature trend prediction model.

[0022] In a possible implementation, before predicting the temperature extreme value and the average temperature trend of the ultrasonic knife in the case that the ultrasonic knife shears the cutting object, the method further comprises the following steps of:

[0023] The phase difference between the output current and the voltage current of the ultrasonic knife generator at the same historical moment is acquired, and the average temperature corresponding to the same historical moment is acquired;

[0024] A second initial model is established, the phase difference between the output current and the voltage current at the same historical moment is input into the second initial model as an input of the second initial model, the average temperature corresponding to the same historical moment is output as an output of the second initial model, and the second initial model is trained to obtain the temperature trend prediction model.

[0025] In a possible implementation, in the step of acquiring the shear point prediction result of the ultrasonic knife, the step comprises the following steps of:

[0026] Variable parameters of a third historical electric signal of the ultrasonic knife generator in a preset time period are acquired, and the variable parameters of the third historical electric signal are preprocessed, and the shear point of the ultrasonic knife in shearing the cutting object is in the preset time period;

[0027] The variable parameters of the third historical electric signal in a first time period are marked as true to obtain a first training set, and the variable parameters of the third historical electric signal in a second time period are marked as true to obtain a second training set, the first time period and the second time period are in the preset time period, and the first time period is earlier than the second time period;

[0028] A third initial model and a fourth initial model are constructed, the third initial model is trained by using the first training set to obtain a first model, and the fourth initial model is trained by using the second training set to obtain a second model;

[0029] In the case that the ultrasonic knife shears the cutting object, variable parameters of a real-time electric signal of the ultrasonic knife generator are acquired;

[0030] The variable parameters of the real-time electric signal are input into the first model and the second model as inputs of the first model and the second model to obtain a first output of the first model and a second output of the second model;

[0031] The predicted demand is obtained, and a predicted result at the cutting point in time is obtained according to the predicted demand, the first output and the second output.

[0032] In a possible implementation, the step of obtaining the predicted demand comprises:

[0033] The thickness of the object to be cut is obtained.

[0034] According to the thickness of the object to be cut and a preset corresponding relationship, a weight corresponding to the predicted demand is determined, and the preset corresponding relationship comprises a weight relationship between the thickness of the object to be cut and the predicted demand.

[0035] In a possible implementation, in the step of preprocessing the variable parameters of the third historical electric signal, the following is included:

[0036] The average value of each parameter in the variable parameters of the third historical electric signal in a third time period is calculated, and is marked as a reference parameter of each parameter, and the third time period is a starting time period in a preset time period.

[0037] Based on the reference parameter of each parameter, the value of each parameter in the variable parameters of the third historical electric signal is adjusted.

[0038] In a possible implementation, in the step of preprocessing the variable parameters of the historical electric signal, the following is included:

[0039] Based on a preset rule, the abnormal values in the variable parameters of the historical electric signal are removed.

[0040] In a possible implementation, in the step of preprocessing the variable parameters of the third historical electric signal, the following is included:

[0041] The variable parameters of the third historical electric signal are processed by using an exponential average sliding model, and the exponential average sliding model is:

[0042] X t =βX t-1 +(1-β)θ t ;

[0043] Wherein, β is a weight parameter, θ t is a weight parameter obtained at the tth update, and X t is a moving average of the variable parameters of the third historical electric signal obtained at the tth update.

[0044] In a possible implementation, in the step of preprocessing the variable parameters of the third historical electric signal, the following is included:

[0045] The variable parameters of the third historical electric signal are processed by scale scaling and standardization according to a preset ratio, and a standardization processing model is:

[0046]

[0047] wherein μ is the mean of the variable parameters of the third historical electrical signal, and σ is the standard deviation of the variable parameters of the third historical electrical signal.

[0048] The application also provides an ultrasonic knife protection device based on temperature prediction, comprising:

[0049] a temperature prediction module configured to predict a temperature extreme value and an average temperature trend of the knife head in a case where the ultrasonic knife is shearing a cut object;

[0050] an adjustment module configured to adjust an output current of an ultrasonic knife generator to a first current if the temperature extreme value is greater than or equal to a first temperature threshold value;

[0051] and configured to obtain a shear break point prediction result of the ultrasonic knife if the temperature extreme value is less than the first temperature threshold value and greater than or equal to a second temperature threshold value, and adjust the output current of the ultrasonic knife generator to the first current if the shear break point prediction result is true, the first temperature threshold value being greater than the second temperature threshold value;

[0052] and further configured to adjust the output current of the ultrasonic knife generator to a second current if the average temperature trend indicates that the ultrasonic knife is in a continuous shearing state after the output current of the ultrasonic knife generator is adjusted to the first current, the second current being greater than the first current.

[0053] The ultrasonic knife protection method and device based on temperature prediction provided by the embodiments of the application can reduce the wear of the ultrasonic knife and prolong the use time of the ultrasonic knife by predicting the temperature extreme value and the average temperature trend of the knife head and adaptively judging based on the temperature extreme value and the average temperature trend, so that the output power can be reduced in time when the shearing has been completed or the temperature of the knife head region exceeds the melting point of the pad, thereby effectively protecting the ultrasonic knife, improving the shearing efficiency of the ultrasonic knife on the basis of reliably protecting the pad, adaptively switching between the ultrasonic knife protection and the high-efficiency shearing state, and adaptively operating the operation mode of the operator, and the adaptive judgment based on the temperature extreme value and the average temperature trend can improve the accuracy and reliability of the protection of the knife head compared with the mode of protecting the knife head based on a single temperature prediction. BRIEF DESCRIPTION OF DRAWINGS

[0054] In order to more clearly illustrate the technical solutions of the application, the drawings required in the embodiments will be briefly introduced as follows. Obviously, for those skilled in the art, other drawings can also be obtained based on these drawings without any creative effort.

[0055] Figure 1A temperature prediction-based ultrasonic knife protection method provided by an embodiment of the present application Figure 1 ;

[0056] Figure 2 The temperature extreme curve and the average temperature trend curve of the knife head region of the ultrasonic knife provided by an embodiment of the present application are collected

[0057] Figure 3 The structure diagram of a combination of a full connection neural network and a recurrent neural network in a series structure provided by an embodiment of the present application

[0058] Figure 4 The structure diagram of a full connection neural network model provided by an embodiment of the present application

[0059] Figure 5 The structure diagram of a recurrent neural network model provided by an embodiment of the present application

[0060] Figure 6 The diagram of the average temperature trend in a continuous shearing process provided by an embodiment of the present application

[0061] Figure 7 The knife head structure diagram provided by an embodiment of the present application Figure 1 ;

[0062] Figure 8 The knife head structure diagram provided by an embodiment of the present application Figure 2 ;

[0063] Figure 9 The diagram of the resonance frequency and the phase difference between the current and the voltage varying with shearing provided by an embodiment of the present application

[0064] Figure 10 The diagram of the output current and the output voltage varying with shearing provided by an embodiment of the present application

[0065] Figure 11 The flowchart of a temperature prediction-based ultrasonic knife protection method provided by an embodiment of the present application Figure 2 ;

[0066] Figure 12 The combination structure diagram of the first model and the second model provided by an embodiment of the present application DETAILED DESCRIPTION

[0067] When an operator shears a cut object by using an ultrasonic knife, in order to avoid that the temperature of the knife head of the ultrasonic knife is too high, which can cause the knife head of the ultrasonic knife to generate a high temperature and reduce the service life of the ultrasonic knife, an embodiment of the present application provides a temperature prediction-based ultrasonic knife protection method and device.

[0068] The method for protecting an ultrasonic knife based on temperature prediction provided in the embodiments of the present application comprises steps S110 to S140, as shown in the figure. Figure 1

[0069] S110, in the case where the ultrasonic knife is cutting a cutting object, the temperature extreme value and average temperature trend of the blade head of the ultrasonic knife are predicted.

[0070] As shown in the figure, the temperature extreme value curve and average temperature trend curve of the blade head region of the ultrasonic knife are obtained by the method for protecting an ultrasonic knife based on temperature prediction provided in the embodiments of the present application. Obviously, the temperature extreme value curve is characterized by a large change range, and has a larger fluctuation compared with the average temperature trend curve, while the average temperature trend curve is more stable and can better represent the temperature trend change. Figure 2

[0071] In the step of predicting the temperature extreme value, the output voltage, output current, first derivative of the resonant frequency and voltage-current phase difference of the ultrasonic knife generator are obtained in the case where the ultrasonic knife is cutting a cutting object, and the output voltage, output current, first derivative of the resonant frequency and voltage-current phase difference are input into a pre-established extreme temperature prediction model to obtain the temperature extreme value output by the extreme temperature prediction model.

[0072] In the step of predicting the average temperature trend, the output current and voltage-current phase difference of the ultrasonic knife generator are obtained in the case where the ultrasonic knife is cutting a cutting object, and the output current and voltage-current phase difference are input into a pre-established temperature trend prediction model to obtain the average temperature trend output by the temperature trend prediction model.

[0073]

[0074] ​​​The temperature trend prediction model uses second historical electric signal data as training data. Specifically, first, the output current and the phase difference between the voltage and the current of the ultrasonic knife generator at the same historical time are obtained, and the average temperature corresponding to the same historical time is obtained. The second historical electric signal data includes the output current and the phase difference between the voltage and the current of the ultrasonic knife generator at the same historical time, and also includes the average temperature of the blade head at the same time. Then, a second initial model is established, the output current and the phase difference between the voltage and the current at the same historical time are taken as the input of the second initial model, and the average temperature corresponding to the same historical time is taken as the output of the second initial model, the second initial model is trained to obtain the temperature trend prediction model.

[0075] In the process of training the first initial model and the second initial model, when the training data of the blade head temperature is obtained, the initial temperature of the blade head is affected by many states such as the working state of the blade head, the material of the blade head, the shape of the blade head and the room temperature environment. In view of this, when collecting the training data, the average value of the trend temperature in the first 0.5 seconds is taken as the reference temperature, which is taken as the initial temperature of the blade head. In this way, the obtained blade head temperature has good representativeness and anti-interference.

[0076] In some embodiments, the extreme temperature prediction model and the temperature trend prediction model provided by the present application are two prediction models with similar structures and different weights. For example, as shown in Figure 3 The series structure is used to combine the fully connected neural network and the recurrent neural network, the recurrent neural network is used as the first input network, the fully connected neural network is used as the second input network, and the first initial model or the second initial model is constructed. In this way, the output of the fully connected neural network is taken as the prediction result of the extreme temperature or the average temperature trend.

[0077] More specifically, the neural network algorithm is a mathematical model inspired by neurons, which is composed of multiple nodes connected to each other and can be used to model the complex relationship between data. The structure of the fully connected neural network is a multi-layer structure, and the neurons in each layer are fully connected to each other. As shown in Figure 4 As shown in some embodiments, the fully connected neural network model used by the first initial model or the second initial model includes an input layer, a hidden layer and an output layer. The fitting ability of the fully connected neural network model is related to the number of layers and the number of neurons. For example, Figure 4 In the first initial model or the second initial model, two hidden layers are set, and the number of hidden units in each layer corresponds to the number of output units of the recurrent neural network model. The number of neurons can be set to 8, 16, 32 or 64. The multi-layer neural network model used by the present application has 4 layers, and the number of nodes in each layer is set to 40, 16, 4 and 1. The number of nodes in the output layer is 1, and the output of the output layer is the prediction result of the extreme temperature or the average temperature trend.

[0078] In some embodiments, considering that the collected ultrasonic knife electric signal is a time series signal, in order to better process the time series information of the electric signal, a recurrent neural network model with memory capability is used. As shown in Figure 5 The first input network of the first initial model or the second initial model is constructed by using a recurrent neural network model. The recurrent neural network model can memorize the previous signal, and has stronger feature extraction capability for time series signals. In this way, the first initial model or the second initial model is constructed by using a recurrent neural network model and a fully connected neural network model.

[0079] Since the electric signal of the ultrasonic generator has stronger correlation in a short time, in some embodiments, during the process of predicting the temperature extreme value and the average temperature trend of the knife head, the electric signal collection time interval of the ultrasonic generator is set to 10 milliseconds, and the number of recurrent nodes of the recurrent neural network is set to 40. In this way, the recurrent neural network can process the electric signals within a time interval of 400 milliseconds at the same time, thereby strengthening the correlation of the temperature extreme value and the average temperature trend in the time scale. It should be noted that the variants of the recurrent neural network model with memory capability are also within the optional range of the embodiments of the present application, for example: long short-term memory network or gated neural unit.

[0080] In some embodiments, the first initial model or the second initial model can also be a combination of one or more of machine learning algorithms or deep learning algorithm models. More specifically, the machine learning algorithm can include a linear regression algorithm, a support vector machine algorithm, a nearest neighbor algorithm, a decision tree algorithm, a random forest algorithm, or a naive Bayes algorithm, and the deep learning algorithm can include a fully connected neural network, a convolutional neural network, a generative adversarial network, or a recurrent neural network.

[0081] In some embodiments, the directly collected ultrasonic knife generator signal is: the output voltage, the output current, the resonance frequency, the current phase value and the voltage phase value of the ultrasonic knife generator. Then, the first derivative of the resonance frequency and the voltage-current phase difference are calculated.

[0082] After collecting the output voltage, the output current, the resonance frequency, the current phase value and the voltage phase value, the data needs to be preprocessed. First, the pre-processing considers the jitter generated by the ultrasonic knife shearing process and the influence of the transmission process of the high-speed circuit on the electric signal, which causes abnormal values of the collected electric signal. Therefore, the electric signal with significant abnormal data is eliminated, for example: negative values of current and voltage, resonance frequency lower than 40KHz, current and voltage phase difference when frequency locking is not successful. After eliminating such abnormal data, it is beneficial to the training and inference result of the extreme temperature prediction model and the temperature trend prediction model.

[0083] In some embodiments, for the extreme temperature prediction model, the output voltage, output current, first derivative of the resonant frequency, and voltage-current phase difference are collected with greater relevance. For the temperature trend prediction model, the output current and output voltage are collected with greater relevance, and the output current and voltage-current phase difference are obtained. However, in actual application, it is not limited to only collecting the above-mentioned electrical signals, but also collecting multi-dimensional variables such as sensor-collected temperature, resistance, ultrasonic knife parameters, or generator parameters with relatively small relevance.

[0084] In order to enable the first initial model or the second initial model to better learn the deep features of the electrical signals, in some embodiments, the first historical electrical signals or the second historical electrical signals are standardized. Since the electrical signals obtained by each ultrasonic knife generator excitation have a certain offset, the first historical electrical signals or the second historical electrical signals collected by the present application are calculated for the standard mean and standard deviation, and the input electrical signals need to be standardized by using the standard mean and standard deviation, thereby obtaining new input variables. The corresponding first standardization processing model is:

[0085]

[0086] The first standardization processing model is: x is each parameter in the first historical electrical signals or the second historical electrical signals, μ is the mean of each parameter in the first historical electrical signals or the second historical electrical signals, and σ is the standard deviation of each parameter in the first historical electrical signals or the second historical electrical signals.

[0087] When the collected data is preprocessed, the adjustment rate of the data is much greater than the collection rate of the electrical signals, resulting in that the fluctuation amplitude of the collected electrical signals is still large. In some embodiments, an exponential moving average method is used to process the electrical signals, and the corresponding first exponential moving average model is:

[0088] X t = βX t-1 + (1-β)θ t .

[0089] θ t is the weight parameter of each parameter in the first historical electrical signals or the second historical electrical signals obtained at the tth update, X t is the moving average of each parameter in the first historical electrical signals or the second historical electrical signals at the tth update. β is the weight parameter, and the greater β is, the more relevant the value obtained by the moving average is to the historical value. In the present embodiment, the β weight parameter is set to 0.90, and a signal curve with good smoothing effect can be obtained. It should be noted that the weight parameter is only used in the present embodiment, and other reasonable weight parameters are also within the scope of selection.

[0090] In the process of predicting the temperature extreme value and average temperature trend of the blade head of the ultrasonic knife, in the embodiments of the present application, the influence of the delay of the data electrical signal on the electrical signal generated in the first 0.5 seconds is considered, the average value of each parameter in the electrical signal collected in the first 0.5 seconds is calculated, and the average value is set as a reference, and each parameter in the electrical signal collected after 0.5 seconds is subtracted from the reference as effective data.

[0091] Since the temperature of the blade head is an accumulated variable and has continuity in time and strong correlation in time, the present application sets the data amount of 5 seconds as the training data length for training the first initial model and the second initial model, and generally takes 2 n The device for training the model can have better operation parallelism. Specifically, the present application adopts 512 as the training input data length.

[0092] In some embodiments, the training parameters of the first initial model and the second initial model are set as follows: the input data length of the training data is set to 512, the batch size of the training data input is set to 128, and the optimizer with a decay parameter of 0.01 is set for learning, so as to improve the adaptability of the first initial model and the second initial model to the temperature prediction under different generator input signals and prevent the model from overfitting to the training set. The present application considers that the prediction temperature error of the first initial model and the second initial model in the initial stage is large, and sets a large learning rate of 0.05 for learning in the initial training, appropriately accelerates the learning rate, and gradually reduces the learning rate after the training reaches a certain accuracy, and gradually performs fine-tuning training of the model.

[0093] In some embodiments, in order to optimize the training effect of the first initial model and the second initial model, it is considered that the essence of temperature prediction is regression prediction, and there is a large value in the prediction feature, and SoothL1Loss is adopted as the loss function of model training. The model of SoothL1Loss is as follows:

[0094]

[0095] When the difference between the predicted temperature and the true temperature is large, the training process is not easy to gradient explosion because of the segmented loss, so that the training of the first initial model and the second initial model is stable and convergent.

[0096] S120, if the temperature extreme value is greater than or equal to the first temperature threshold, the output current of the ultrasonic knife generator is adjusted to the first current.

[0097] The first temperature threshold is greater than the gasket melting point temperature, for example. The first temperature threshold is set to 280 degrees Celsius. In the normal shearing state of the ultrasonic knife, the temperature of the ultrasonic knife head will not exceed the first temperature threshold. When the ultrasonic knife head shears the gasket, the temperature of the ultrasonic knife head will rapidly increase, which will cause the temperature to exceed the first temperature threshold. In the embodiment of the present application, when the temperature extreme is greater than or equal to the first temperature threshold, the output current of the ultrasonic knife generator is adjusted to the first current, thereby effectively reducing the temperature of the ultrasonic knife head and avoiding continuous damage to the gasket of the ultrasonic knife.

[0098] In S130, if the temperature extreme is less than the first temperature threshold and greater than or equal to a second temperature threshold, a shearing point prediction result of the ultrasonic knife is obtained. If the shearing point prediction result is true, the output current of the ultrasonic knife generator is adjusted to the first current. The first temperature threshold is greater than the second temperature threshold.

[0099] The second temperature threshold is greater than the temperature of the object to be cut, for example. The second temperature threshold is set to 210 degrees Celsius. When the temperature extreme is greater than or equal to the second temperature threshold, it is necessary to determine whether the ultrasonic knife has completed the shearing operation on the object to be cut.

[0100] Specifically, the shearing point prediction result of the ultrasonic knife is obtained. If the shearing point prediction result is false, it indicates that the object to be cut does not have a shearing point. At this time, the output current of the ultrasonic knife is not adjusted to ensure the continuous operation of the ultrasonic knife. If the shearing point prediction result is true, it indicates that the object to be cut has a shearing point. At this time, in order to avoid the continuous increase of the temperature of the ultrasonic knife head, the output current of the ultrasonic knife generator is adjusted to the first current to prevent continuous damage to the gasket.

[0101] In S140, after the output current of the ultrasonic knife generator is adjusted to the first current, if the average temperature trend indicates that the ultrasonic knife is in a continuous shearing state, the output current of the ultrasonic knife generator is adjusted to a second current. The second current is greater than the first current.

[0102] Specifically, in order to meet the gasket protection requirement, in the case where the shearing point prediction result is true, the ultrasonic knife outputs a lower first current. In actual application, if the object to be cut is a free object, multiple shearing points will appear during the shearing process of the ultrasonic knife. When the shearing point appears, it does not mean that the shearing of the object to be cut is completed, and further shearing operation is still required.

[0103] Specifically, whether the ultrasonic knife is in continuous shearing state is judged according to the average temperature trend. If the ultrasonic knife is not in continuous shearing state, it indicates that the cut object is completely cut off, that is, the shearing operation is completed, at this time, the ultrasonic knife is controlled to stop the cutting operation. If the ultrasonic knife is in continuous shearing state, the ultrasonic knife continues to output the first current, and the shearing operation cannot be completed at a faster speed, at this time, the output current of the ultrasonic knife generator is adjusted to the second current, so that the temperature of the blade head of the ultrasonic knife is increased, thereby the ultrasonic knife continues to complete the shearing operation.

[0104] In order to realize the continuous shearing operation, in the embodiment of the application, the characteristics of the average temperature trend can be used to better distinguish the temperature change of the shearing gap. As shown in Figure 6 the average temperature trend in the continuous shearing process is shown. The trend temperature drops after the ultrasonic knife leaves the tissue and the gasket, and the trend temperature rises again due to continuous shearing.

[0105] In the step of judging whether the ultrasonic knife is in continuous shearing state according to the average temperature trend, if the average temperature of the blade head decreases by more than a first preset value within a first preset time, and the average temperature of the blade head rises to more than a second preset value within a second preset time, it is determined that the ultrasonic knife is in continuous shearing state, wherein the second preset time is later than the first preset time.

[0106] That is, in the technical scheme of the embodiment of the application, whether the ultrasonic knife is in continuous shearing state is judged according to two characteristics. The first characteristic is the average temperature drop feature. Specifically, the average temperature occurs within the first preset time when the blade head leaves the cut object and the gasket, for example, the first preset time is set to 0.5 seconds, and the first preset value is set to 40 degrees Celsius, that is, the average temperature of the blade head decreases by more than 40 degrees Celsius within 0.5 seconds, that is, it is determined that the ultrasonic knife head stops clamping and opens. The second characteristic is the average temperature continuous rise feature. With the second shearing, the average temperature of the application rises again, for example, the average temperature occurs within the second preset time when the blade head contacts the cut object and the gasket, the second preset time is set to 0.5 seconds, that is, the first preset time is 0.5 seconds to 1 second after the blade head leaves the cut object and the gasket, and the first preset value is set to 50 degrees Celsius, that is, the average temperature of the blade head rises by more than 50 degrees Celsius within 0.5 seconds to 1 second after the blade head leaves the cut object and the gasket, that is, it is determined that the ultrasonic knife shears the cut object. If the above two characteristics are met, it is determined that the ultrasonic knife is in continuous shearing state.

[0107] As shown in Figure 7 the blade head applying the technical scheme of the embodiment of the application, the gasket after 300 shearing operations shows that the gasket is almost undamaged. As shown in Figure 8As shown, the results of the control group, which did not implement any protective measures, showed that the gasket was already severely damaged after the cutter head performed 300 shearing cycles.

[0108] This application provides a temperature-prediction-based ultrasonic scalpel protection method. By predicting the extreme and average temperature trends of the scalpel tip and adaptively judging based on these trends, the method can reduce the output power in a timely manner when shearing is completed or the temperature in the scalpel tip area exceeds the melting point of the gasket. This reduces ultrasonic scalpel wear and extends its service life, effectively protecting the ultrasonic scalpel. During continuous ultrasonic excitation or continuous shearing, the method can further improve the shearing efficiency of the ultrasonic scalpel while reliably protecting the gasket. It adaptively switches between ultrasonic scalpel protection and high-efficiency shearing states, adapting to the operator's operating methods. Compared to methods that protect the scalpel tip based on a single temperature prediction, this application's technical solution can improve the accuracy and reliability of scalpel tip protection.

[0109] In some embodiments, when identifying whether the ultrasonic scalpel is in a connected shearing state, it is not limited to determining it by the average temperature trend, but can also be determined by relevant variables such as voltage values ​​and resonant frequencies that change due to load variations.

[0110] In some embodiments, the first current value is set to 0.7 times the second current value. When the ultrasonic scalpel outputs the first current value, it can successfully complete the shearing operation on the object being cut, and provide better protection for the gasket.

[0111] When an ultrasonic scalpel performs shearing operations, the electrical signal parameters of the ultrasonic scalpel generator exhibit distinct characteristics: the output current is constant, while the output voltage changes with the shearing of the object being cut. For example... Figure 9 The figure shows the curves of resonant frequency and current-voltage phase difference as a function of shearing. To ensure the ultrasonic scalpel operates at its highest efficiency, the current-voltage phase difference is typically set to 0. This positions the ultrasonic scalpel's operating circuit at its resonant point. As the shearing process continues, as... Figure 10 The curves shown are output current and output voltage as a function of shear. The actual impedance changes with the load, and the piezoelectric crystal also changes in real time with the operating temperature.

[0112] In some embodiments, such as Figure 11 As shown, the steps for obtaining the predicted shear point of the ultrasonic scalpel include: S1110 to S1160.

[0113] S1110: Obtain the variable parameters of the historical electrical signal of the ultrasonic scalpel generator within a preset time period, and preprocess the variable parameters of the historical electrical signal to ensure that the cutting point of the ultrasonic scalpel cutting the object is within the preset time period.

[0114] Because the output voltage value changes correspondingly with the different substances contacted by the ultrasonic blade head in the shearing process, specifically, the output voltage value is relatively low when the ultrasonic blade shears the cutting object. As the shearing proceeds, the blade head contacts the pad, and the output voltage value increases.

[0115] The current phase value, the voltage phase value, the target phase value and the resonance frequency of the electric parameters used in the present application are closely related to the changes of the cutting object. Specifically, because the difference between the detected current voltage phase value and the target phase value is approximately equal to 0, and the change rate of the resonance frequency is relatively constant during the stable shearing of the ultrasonic blade on the cutting object. As the shearing proceeds, the blade head contacts the pad, and the difference between the current voltage phase value and the target phase value deviates from 0, and the change rate of the resonance frequency becomes larger, and the waveform produces larger fluctuations. In this way, the resonance frequency as the collected electric signal parameter has more obvious distinguishing characteristics.

[0116] The third historical electric signal refers to the output current value, the output voltage value, the current phase value, the voltage phase value, the target phase value and the resonance frequency. In order to obtain training data with higher information content and better feature expression, the third historical electric signal can be further processed to obtain the variable parameters of the third historical electric signal. For example, the current phase value and the voltage phase value are processed to obtain the current voltage phase difference; the current voltage phase difference and the target phase value are processed to obtain the difference between the current voltage phase difference and the target phase difference; and the first derivative of the resonance frequency is calculated. In this way, the variable parameters of the historical electric signal are: the output current value, the output voltage value, the current voltage phase difference, the difference between the current voltage phase difference and the target phase difference, and the first derivative of the resonance frequency.

[0117] It should be noted that the current phase value, the voltage phase value, the target phase value and the resonance frequency of the ultrasonic blade generator are collected because the above-mentioned electric signal parameters are closely related to the changes of the cutting object. In actual application, it is not limited to only collecting the above-mentioned electric signal parameters. For example, due to the difference in the circuit properties of the ultrasonic blade generator, if there are other electric signal parameters that are closely related to the changes of the cutting object, they can also be collected as electric signal parameters. For example, the collected electric signal parameters can also include multi-dimensional variables such as sensor collected temperature, resistance, ultrasonic blade parameters or generator parameters.

[0118] In addition, when determining the variable parameters of the third historical electric signal, the ratio of the voltage and current can be calculated to obtain the impedance value, and the product of the voltage and current can be calculated to obtain the power value, and the impedance value and the power value can be used as the variable parameters of the third historical electric signal.

[0119] In the subsequent training processes of the third initial model and the fourth initial model, in order to enable the third initial model and the fourth initial model to better learn the deep features of the training data, the input training data is scaled to scale the variable parameters of the historical electrical signals to similar scales.

[0120] For example, the resonant frequency is near 55000 Hz, and the current-voltage phase difference is within ±180°, if the first derivative of the resonant frequency and the current-voltage phase difference are directly input into the third initial model and the fourth initial model for training, it is difficult for the third initial model and the fourth initial model to learn effective features.

[0121] To this end, the third historical electrical signal or the variable parameters of the third historical electrical signal are processed by using the scaling method, for example, the resonant frequency is subtracted by 55000 Hz, and then multiplied by a reduction coefficient 0.1 to obtain a new frequency variable, so that the new frequency variable has the same order of magnitude as the phase difference variable.

[0122] In addition, during data collection, there is a certain offset in the collected electrical signal parameters of the ultrasonic knife generator, in order to avoid the influence of the offset on the data features. In some embodiments, first, the standard mean and standard deviation of each parameter in the third historical electrical signal are calculated, and then the standard mean and standard deviation are used to standardize each parameter in the third historical electrical signal. The corresponding second standardization processing model is as follows:

[0123]

[0124] The second standardization processing model is: x' is each parameter in the third historical electrical signal, μ' is the mean of each parameter in the third historical electrical signal, and σ' is the standard deviation of each parameter in the third historical electrical signal.

[0125] In some embodiments, the variable parameters of the third historical electrical signal are used for model training, and a certain amount of data will be obtained in the actual application process. For example, during the multiple cutting processes of the ultrasonic knife, the related parameters are collected. The preset time needs to include the time when the cut object is cut off (the cutting point moment). For example, the time interval of data collection is 10 ms, and the data collected by the preset time is the data points from 400 data points before the cutting point to 50 data points after the cutting point.

[0126] In some embodiments, the electrical signal of the ultrasonic knife generator is collected with a frequency locking process, so that the collected electrical signal of the ultrasonic knife generator has instability in the initial stage of collection. To solve this problem, the application includes the following in the step of preprocessing the variable parameters of the third historical electrical signal: calculating the average value of each parameter in the variable parameters of the third historical electrical signal in the third time period, and marking the average value as the reference parameter of each parameter, the third time period being the starting time period in the preset time period; and adjusting each parameter in the variable parameters of the historical electrical signal based on the reference parameter of each parameter. For example, the average value of the parameters collected in the first 0.5 seconds is calculated, and the average value of the parameters collected in the first 0.5 seconds is set as the reference, and the parameters collected after 0.5 seconds are subtracted from the reference as effective data.

[0127] Because the electrical signal of the ultrasonic knife generator is easily affected by temperature, load and other uncertain factors, the waveform of the electrical signal is unstable. To this end, in some embodiments, the step of preprocessing the variable parameters of the third historical electrical signal includes: using a second exponential moving average model to process the variable parameters of the historical electrical signal, and the corresponding second exponential moving average model is:

[0128] X t ′=β′X t-1 ′+(1-β′)θ t ′。

[0129] Where β' is a weight parameter, θ t ′ is the weight parameter of each parameter in the third historical electrical signal obtained in the tth update, X t ′ is the moving average of the variable parameters of the third historical electrical signal obtained in the tth update. For example, a 30-window average moving window is used to smooth the occurrence signal, and the processed signal does not change the original change trend, but is smoother after removing high-frequency small-amplitude fluctuations. In addition, the variable parameters of the third historical electrical signal can also be processed by moving average method, low-pass filtering method, polynomial fitting method, local weighted scatter smoothing method or Kalman filtering method.

[0130] S1120, marking the result corresponding to the variable parameters of the historical electrical signal in the first time period as true to obtain a first training set; and marking the result corresponding to the variable parameters of the third historical electrical signal in the second time period as true to obtain a second training set, the first time period and the second time period being in the preset time period, and the first time period being earlier than the second time period.

[0131] In some embodiments, the data collection time of the first training set is earlier, for example, the variable parameter of the third historical electrical signal in the first time period is the data point 200 before the cutting point to the data point 100 before the cutting point. In this way, the model trained by the first training set has good performance in predicting in advance. The data collection time of the second training set is later, for example, the variable parameter of the third historical electrical signal in the second time period is the data point 50 before the cutting point to the data point 50 after the cutting point, so that the model trained by the second training set has the performance of more reliable prediction results.

[0132] It should be noted that after marking the result corresponding to the variable parameter of the third historical electrical signal as true, the result corresponding to the variable parameter of the third historical electrical signal in other stages is also marked as false. Wherein, the result corresponding to the variable parameter of the third historical electrical signal is true, which means that the cutting object corresponding to the variable parameter of the third historical electrical signal is cut off, and the result corresponding to the variable parameter of the third historical electrical signal is false, which means that the cutting object corresponding to the variable parameter of the third historical electrical signal is not cut off. It should be noted that after marking the result corresponding to the variable parameter as true, the corresponding variable parameter is a true sample, and after marking the result corresponding to the variable parameter as false, the corresponding variable parameter is a false sample.

[0133] In the embodiments of the present application, a large amount of collected real and effective data needs to be cut, and the data after cutting is used to train the model. In order to take advantage of the batch training of the training device, the training input data needs to have a uniform length. In the step of preprocessing the variable parameter of the historical electrical signal, the time value of the early warning cutting point is obtained, and the training data in the first training set and the second training set is cut based on the time value.

[0134] For example, based on the characteristics of the cutting object, the time effectiveness of the cutting warning is required to be less than 1 second, and the electrical signal of the ultrasonic knife generator has strong correlation in time. The cutting time of the training data of the first training set and the second training set is set to 1 second, that is, the data amount of the cutting time of 1 second is set as the uniform length of the training data. And the length of the training data is taken as 2 n When the device for training the model has better parallelism, the present application adopts 128 as the uniform length of the training input data. On the other hand, the present application takes the fragment with a length of 128 in the original data as the training input of the subsequent third initial model and fourth initial model with a step length of 16, which can effectively increase the number of data samples in the training set and make the fusion model learn deep data features sufficiently.

[0135] S1130, constructing a third initial model and a fourth initial model, and training the third initial model with the first training set to obtain a first model; training the fourth initial model with the second training set to obtain a second model.

[0136] Wherein, the first model and the second model involved in the present application are two models with similar structures and different weights, in order to obtain the first model and the second model with different weights. After constructing the first initial model and the second initial model, first, the variable parameters of the third historical electric signal are used to pre-train the first initial model and the second initial model to obtain the basic weight, wherein during the pre-training process, the batch size of the training input is set to 1024, and the learning rate is set to 0.05. Then, the first initial model loaded with the basic weight is trained with the training data in the first training set. Wherein, during the process of training the first initial model loaded with the basic weight, the batch size of the training input is set to 254, the decay parameter is set to 0.01, the learning rate is set to 0.005, and the cross-entropy loss function is set to: Wherein, is the label value of the true sample in the first training set, p t is the prediction probability result of the true cutting output by the first initial model, is the label value of the false sample in the first training set, p f is the prediction probability result of the false cutting output by the first initial model, and a is the influence factor of the label value of the true sample in the first training set.

[0137] And, the second initial model loaded with the basic weight is trained with the training data in the second training set, wherein during the process of training the second initial model loaded with the basic weight, the batch size of the training input is set to 254, the decay parameter is set to 0.01, the learning rate is set to 0.005, and the cross-entropy loss function is set to: Wherein, is the label value of the true sample in the second training set, p t is the prediction probability result of the true cutting output by the second initial model, is the label value of the false sample in the second training set, p f is the prediction probability result of the false cutting output by the second initial model, and a' is the influence factor of the true sample in the second training set.

[0138] Wherein, the cross-entropy loss function with the influence factor can increase the loss of the label value of the true sample, affect the learning of the model on the true sample, improve the overall accuracy, and avoid the situation that the true sample occupies a very small part of the training data, resulting in unbalanced training labels.

[0139] In the embodiments of the present application, when a=1.2 and a'=1.2, the third initial model and the fourth initial model have better training effects. In addition, the loss function is not limited to the cross-entropy loss function, and the mean square error loss function, the square loss function or the absolute value loss function can also be used.

[0140] In S1140, when the ultrasonic knife shears the cutting object, a variable parameter of a real-time electrical signal of the ultrasonic knife generator is obtained.

[0141] It should be noted that before the variable parameter of the real-time electrical signal is input into the first model and the second model, the variable parameter of the real-time electrical signal needs to be preprocessed as necessary, and the preprocessing process is the same as the process of the variable parameter of the third historical electrical signal when the third initial model and the fourth initial model are trained, that is, referring to the preprocessing process of the variable parameter of the third historical electrical signal, the same preprocessing process is performed on the variable parameter of the real-time electrical signal before it is input into the first model or the second model.

[0142] In S1150, the variable parameter of the real-time electrical signal is input into the first model and the second model to obtain a first output of the first model and a second output of the second model.

[0143] The first output and the second output are probability values representing the prediction result of the shear point of the cutting object, and the probability value representing the true value can be a probability value ranging from 0 to 1 or a probability value set with a total sum of 1.

[0144] The first output output by the first model has better prediction advancement, that is, the prediction result of the shear point can be given earlier before the shear point occurs. The second output output by the second model has higher reliability, that is, the shear point result is more reliable.

[0145] In S1160, a prediction requirement is obtained, and a prediction result of the shear point is obtained according to the prediction requirement, the first output and the second output.

[0146] The prediction requirement mainly reflects the expectation of the operator for the prediction result, for example, the prediction requirement of the operator is to know the prediction result of the shear point earlier in order to control the parameters of the ultrasonic knife in advance. For another example, the requirement of the operator is that the prediction result of the shear point is more reliable to ensure more accurate operation of the ultrasonic knife.

[0147] In actual application, after taking the variable parameter of the real-time electrical signal as the input of the first model and the second model to obtain the first output of the first model and the second output of the second model, the result of the first output can be first judged. For example, when the first output represents that the signal feature recognition at the cutting point is successful, the prediction result at the cutting point is given, and the operator is prompted through sound, light, electricity and the like, so as to facilitate the operator to implement the protection measures such as limiting the output power of the ultrasonic knife generator.

[0148] Alternatively, after the first output represents that the signal feature recognition at the cutting point is successful, the result of the second output is judged. When the second output represents that the signal feature recognition at the cutting point is successful, the prediction result at the cutting point is given in a weighted manner according to the prediction requirement.

[0149] The weighted model is Y = a'' x y1 + (1-a'') x y2, and the weight a'' is a set of values that can be adjusted according to application scenarios. The operator can determine the weight value reflecting the prediction requirement according to the characteristics of the cut object, for example, set a'' = 0.9, so that the judgment result has stronger advance nature when dealing with thicker cut objects. For example, set a'' = 0.7, so that the judgment result has stronger reliability when dealing with thinner cut objects.

[0150] In other words, the weight of the prediction requirement can be determined by the thickness of the cut object, for example, first, the thickness of the cut object is obtained, then according to the thickness of the cut object and the preset corresponding relationship, the weight corresponding to the prediction requirement is determined, that is, the value of the prediction requirement a'' is determined. The preset corresponding relationship includes the thickness of the cut object and the weight relationship of the prediction requirement.

[0151] The first model or the second model provided by the embodiments of the present application can be one or a combination of more than one of the models based on machine learning algorithm or deep learning algorithm. More specifically, the machine learning algorithm can include linear regression algorithm, support vector machine algorithm, nearest neighbor algorithm, decision tree algorithm, random forest algorithm or naive Bayes algorithm, and the deep learning algorithm can include fully connected neural network, convolutional neural network, generative adversarial network or recurrent neural network.

[0152] For example, the first model adopts a combination of the fully connected neural network model formed by the fully connected neural network in the optional range and the recurrent neural network model formed by the recurrent neural network.

[0153] The fully connected neural network model is a mathematical model inspired by neurons, which is connected by multiple nodes and can be used to model complex relationships between data. The fully connected neural network model is a multi-layer structure, and the neurons in each layer are fully connected. For example, in some embodiments, the fully connected neural network model used includes an input layer, a hidden layer, and an output layer.

[0154] Further, the fitting ability of the fully connected neural network model is highly related to the number of layers and neurons. The more layers and the more neurons, the stronger the mathematical fitting ability of the fully connected neural network model. In the embodiments of the present application, two hidden layers are set, and each hidden unit corresponds to the number of output units of the recurrent neural network. Generally, the number of neurons is set to 8, 16, 32, 64, or 128. In the embodiments of the present application, the number of neurons is set to 16.

[0155] In some embodiments, a recurrent neural network model with memory capability can also be used to construct the third initial model or the fourth initial model. The recurrent neural network model can remember the previous signals and has stronger feature extraction capability for time series signals. Therefore, the first initial model or the second initial model is constructed by using the recurrent neural network model and the fully connected neural network model.

[0156] For example, considering the time continuity of the ultrasonic generator signal in practical applications, the ultrasonic generator signal has stronger correlation in a short time. In some embodiments, the time interval for collecting the ultrasonic generator signal is set to 10 milliseconds. Therefore, the number of recurrent nodes of the recurrent neural network is designed to be 40. In this way, the electrical signals within the time interval of 400 milliseconds before and after can be processed at the same time, and the correlation of the output results in the time scale is strengthened.

[0157] It should be noted that the first model and the second model of the present application can be various combinations of machine learning algorithm models and deep learning algorithm models, or combinations of the same type of algorithm models. The combination modes are parallel, series, or cross. For example, as shown in FIG. 1, a specific mode adopted in the embodiments of the present application is a model structure in which the first model and the second model constructed above are fused in parallel. In particular, the two basic model structures are the same, but the training data is different, the weight parameters are different, and the discriminant results obtained by forward reasoning are different. Figure 12

[0158] ​The technical scheme of the application establishes a first model with better prediction advancement and a second model with better reliability through the third historical electric signal parameter, obtains the variable parameter of the real-time electric signal of the ultrasonic knife generator in the case that the ultrasonic knife is cutting the cut object, obtains the first output of the first model and the second output of the second model, and obtains the prediction result of the cutting point moment according to the prediction demand, the first output and the second output based on the prediction demand of the operator, so as to give the cutting point prediction result meeting the demand and avoid the case that the operator observes that the cut object has been cut, resulting in that the head of the ultrasonic knife generates a high temperature and the service life of the ultrasonic knife is reduced.

[0159] The application further provides an ultrasonic knife protection device based on temperature estimation, comprising a temperature prediction module and an adjustment module. The temperature prediction module is configured to predict the temperature extreme value and average temperature trend of the head of the ultrasonic knife in the case that the ultrasonic knife is cutting the cut object. The adjustment module is configured to adjust the output current of the ultrasonic knife generator to the first current if the temperature extreme value is greater than or equal to the first temperature threshold, and to obtain the cutting point prediction result of the ultrasonic knife if the temperature extreme value is less than the first temperature threshold and greater than or equal to the second temperature threshold, and adjust the output current of the ultrasonic knife generator to the first current if the cutting point prediction result is true, the first temperature threshold being greater than the second temperature threshold. The adjustment module is further configured to adjust the output current of the ultrasonic knife generator to the second current if the average temperature trend indicates that the ultrasonic knife is in the continuous cutting state after the output current of the ultrasonic knife generator is adjusted to the first current, the second current being greater than the first current.

[0160] The ultrasonic knife protection method and device based on temperature estimation provided by the application can reduce the wear of the ultrasonic knife and prolong the use time of the ultrasonic knife by reducing the output power in time when the cutting has been completed or the temperature of the head region exceeds the melting point of the pad, thereby effectively protecting the ultrasonic knife. In the continuous excitation or continuous cutting of the ultrasonic knife, the cutting efficiency of the ultrasonic knife can be further improved on the basis of reliable protection of the pad, the ultrasonic knife protection and high-efficiency cutting state are adaptively switched, the operation mode of the operator is adaptively adjusted, and the temperature extreme value and average temperature trend are used for adaptive discrimination. Compared with the mode of protecting the head of the ultrasonic knife based on a single temperature prediction, the technical scheme of the application can improve the accuracy and reliability of the head protection.

[0161] The above detailed description is merely descriptive of the application and specific embodiments thereof. It is not intended as a limitation on the scope of the application. Changes, equivalent substitutions, improvements, combinations, and the like, which are apparent to a skilled artisan, are covered by the following claims.

Claims

1. A temperature prediction based ultrasonic blade protection device, comprising: The temperature prediction-based ultrasonic knife protection device comprises: a temperature prediction module configured to predict a temperature extreme value and an average temperature trend of a blade head in a case where the ultrasonic knife is shearing a cutting object; an adjustment module configured to adjust an output current of an ultrasonic knife generator to a first current if the temperature extreme value is greater than or equal to a first temperature threshold value; and configured to obtain a shear break point prediction result of the ultrasonic knife if the temperature extreme value is less than the first temperature threshold value and greater than or equal to a second temperature threshold value, and adjust the output current of the ultrasonic knife generator to the first current if the shear break point prediction result is true, the first temperature threshold value being greater than the second temperature threshold value; and further configured to adjust the output current of the ultrasonic knife generator to a second current if the average temperature trend indicates that the ultrasonic knife is in a continuous shearing state after the output current of the ultrasonic knife generator is adjusted to the first current, the second current being greater than the first current. The adjustment module is further configured to determine that the ultrasonic knife is in the continuous shearing state if, within a first preset time, an average temperature drop of the blade head is greater than a first preset value, and within a second preset time, the average temperature of the blade head rises to be greater than a second preset value, the second preset time being later than the first preset time.

2. The temperature prediction based ultrasonic blade protection device of claim 1, wherein, The temperature prediction module is further configured to: obtain a first derivative of an output voltage, an output current, a resonant frequency and a voltage-current phase difference of the ultrasonic knife generator in a case where the ultrasonic knife is shearing a cutting object; input the output voltage, the output current, the first derivative of the resonant frequency and the voltage-current phase difference into a pre-established extreme temperature prediction model to obtain a temperature extreme value output by the extreme temperature prediction model.

3. The temperature prediction based ultrasonic blade protection device of claim 2, wherein, The temperature prediction module is further configured to: obtain the output voltage, the output current, the first derivative of the resonant frequency and the voltage-current phase difference of the ultrasonic knife generator at a same historical time, and a temperature extreme value corresponding to the same historical time; establish a first initial model, take the output voltage, the output current, the first derivative of the resonant frequency and the voltage-current phase difference at the same historical time as inputs of the first initial model, take the temperature extreme value corresponding to the same historical time as an output of the first initial model, and train the first initial model to obtain the extreme temperature prediction model.

4. The temperature prediction based ultrasonic blade protection device of claim 1, wherein, The temperature prediction module is further configured to: obtain an output current and a voltage-current phase difference of the ultrasonic knife generator in a case where the ultrasonic knife is shearing a cutting object; input the output current and the voltage-current phase difference into a pre-established temperature trend prediction model to obtain an average temperature trend output by the temperature trend prediction model.

5. The temperature prediction based ultrasonic blade protection device of claim 4, wherein, The temperature prediction module is further configured to: obtain the output current and the voltage-current phase difference of the ultrasonic knife generator at a same historical time, and an average temperature corresponding to the same historical time; establish a second initial model, take the output current and the voltage-current phase difference at the same historical time as inputs of the second initial model, take the average temperature corresponding to the same historical time as an output of the second initial model, and train the second initial model to obtain the temperature trend prediction model.

6. The temperature prediction based ultrasonic blade protection device of claim 1, wherein, The adjustment module is further configured to: acquire variable parameters of a third historical electrical signal of the ultrasonic knife generator in a preset time period, and pre-process the variable parameters of the third historical electrical signal, wherein a shearing point of the ultrasonic knife in shearing the cutting object is in the preset time period; label a result corresponding to the variable parameters of the third historical electrical signal in a first time period as true to obtain a first training set, and label a result corresponding to the variable parameters of the third historical electrical signal in a second time period as true to obtain a second training set, the first time period and the second time period being in the preset time period, and the first time period being earlier than the second time period; construct a third initial model and a fourth initial model, train the third initial model with the first training set to obtain a first model, and train the fourth initial model with the second training set to obtain a second model; acquire variable parameters of a real-time electrical signal of the ultrasonic knife generator in a case where the ultrasonic knife is shearing the cutting object; input the variable parameters of the real-time electrical signal as inputs of the first model and the second model to obtain a first output of the first model and a second output of the second model; acquire a prediction requirement, and obtain a prediction result of the shearing point according to the prediction requirement, the first output and the second output.

7. The temperature prediction based ultrasonic blade protection device of claim 5, wherein, The temperature prediction module is further configured to: acquire a thickness of the sheared object; determine a weight corresponding to the prediction requirement according to the thickness of the cut object and a preset corresponding relationship, the preset corresponding relationship containing a weight relationship between the thickness of the cut object and the prediction requirement.

8. The temperature prediction based ultrasonic blade protection device of claim 6, wherein, The adjustment module is further configured to: calculate average values of each parameter in the variable parameters of the third historical electrical signal in a third time period, and label the average values as reference parameters of the each parameter, the third time period being a starting time period in the preset time period; adjust values of the each parameter in the variable parameters of the third historical electrical signal based on the reference parameters of the each parameter.

Images