Ultrasonic operation equipment frequency locking control method based on reinforcement learning

Abstract

The invention discloses an ultrasonic surgical equipment frequency locking control method based on reinforcement learning, which comprises the following frequency locking control steps: acquiring the working state information of an ultrasonic knife, the state information comprising the phase difference between driving current and driving voltage, knife head impedance and working frequency; a phase compensation value can be obtained by inputting the state information into the trained reinforcement learning model; the state control module of the ultrasonic knife can determine the working state of the next moment according to the phase compensation value. According to the method, the phase difference of the control current and voltage of the ultrasonic knife can be compensated in real time, so that the problem that the phase zero point and the impedance lowest point are not matched due to the change of the working temperature of the transducer is solved, the ultrasonic knife works in the optimal resonance state all the time, heating of the transducer is reduced, and the working efficiency of the ultrasonic knife is improved.

Description

Technical Field

[0001] The present invention relates to an ultrasonic surgical device, and in particular to an ultrasonic surgical device frequency locking control method used in the ultrasonic surgical device. Background Art

[0002] Ultrasonic surgical equipment is mainly composed of a host, a transducer and a blade. The working principle of the ultrasonic scalpel is to use the electrostrictive effect of the transducer to convert electrical energy into mechanical energy, drive the blade to work through the amplification and coupling of the amplitude transformer, and radiate energy to local tissues of the human body, thereby performing surgical treatment. When the frequency of the driving current of the transducer is consistent with the natural frequency of the scalpel vibration system, the ultrasonic scalpel works in a resonant state. At this time, the blade equivalent impedance is the lowest, the blade amplitude is the largest, the energy will be transferred to the blade to the greatest extent, and the working efficiency is the highest. If it does not work in a resonant state, most of the energy will be consumed inside the transducer, causing the transducer to heat up, resulting in a decrease in the output power and efficiency of the vibration system, a decrease in the output amplitude, and a decrease in the quality of the surgery.

[0003] Therefore, frequency locking control is extremely important in ultrasonic surgical equipment. Existing frequency locking control methods are usually based on real-time adjustment of the frequency of the driving current to make the phase difference Δφ between the driving current and the driving voltage of the transducer equal to 0. However, due to the existence of the static capacitance C0 in the transducer equivalent model, when ΔΦ is 0, the series resonant branch is not in a perfect resonant state, that is, the blade equivalent impedance is not at the minimum value. Therefore, in the frequency locking control method of ultrasonic surgical equipment, a certain phase compensation is usually performed based on the static capacitance C0 value tested before use. However, since the static capacitance C0 value will change with the change of the transducer temperature, the traditional fixed compensation method cannot ensure that the ultrasonic surgical equipment always maintains a resonant state during long-term use, resulting in a decrease in the cutting efficiency of the blade and detuning. Summary of the invention

[0004] The present invention aims to solve the problem that the existing ultrasonic surgical equipment has the possibility that the ultrasonic surgical knife does not work in a resonant state, resulting in most of the energy being consumed inside the transducer, thereby causing the transducer to heat up, resulting in reduced output power and efficiency of the vibration system, reduced output amplitude, and even reduced surgical quality. The present invention provides a frequency locking control method for ultrasonic surgical equipment based on reinforcement learning, which can enable the ultrasonic scalpel to always work in a resonant state, prevent the reduction of cutting efficiency and detuning during long-term use, and can intelligently and real-time perform phase compensation during the frequency locking control process.

[0005] The specific technical solution adopted by the present invention to solve the above technical problems is: a frequency locking control method of ultrasonic surgical equipment based on reinforcement learning, characterized in that it includes the following frequency locking control steps: S1. Obtaining ultrasonic scalpel working status information S: the working status information includes the phase difference ΔΦ between the driving current and the driving voltage of the ultrasonic scalpel, the blade impedance Z and the operating frequency F; S2. Calculate the phase compensation value: input the working state information obtained in the above step S1 into the trained reinforcement learning model to obtain the phase compensation value Φc; S3. Update the working state of the ultrasonic scalpel, and input the phase compensation value Φc into the state control module to determine the working state of the ultrasonic scalpel at the next moment; S4. In the above steps S1 to S3, the phase difference of the ultrasonic knife control current and voltage is compensated in real time to make up for the mismatch between the phase zero point and the lowest impedance point caused by the change of the transducer working temperature, so that the ultrasonic knife can always lock the frequency control and work in the best resonant state, reduce the heating of the transducer, and keep the ultrasonic knife system in a resonant state.

[0006] Preferably, in the above steps S1 to S2, the working status information S of the ultrasonic knife includes at least one of the historical working statuses [St, St-1, St-2, ..., St-n] (n≥0), wherein St represents the working status at the current moment, and St-n represents the working status at the previous n moments. The more moments included in the working status information, the more accurate the phase compensation calculation value is, but the longer the calculation time is. Users can choose the number according to actual needs.

[0007] Preferably, the reinforcement learning model includes a state space, an action space, a learning goal and a training process, wherein the state space includes the phase difference between the driving current and the driving voltage of the ultrasonic knife, the blade impedance and the operating frequency; the action space includes increasing the phase difference, reducing the phase difference and maintaining the phase difference unchanged; the learning goal is to make the impedance value in the working state of the ultrasonic knife be at the minimum point under the current working environment; the training process includes: outputting an action in the action space according to the state space of the ultrasonic knife at the current moment, and the ultrasonic knife adjusts the working state according to the action instruction. If the impedance decreases or remains unchanged at the next moment, a positive reward is given to the model; if the impedance increases at the next moment, a negative reward is given to the model. If the model's action can always keep the ultrasonic knife working at the lowest impedance point, it indicates that the model training is completed, and the trained reinforcement learning model is output.

[0008] Preferably, the state control module adjusts the working state of the ultrasonic scalpel by controlling the driving current frequency, driving current magnitude or driving voltage magnitude of the ultrasonic scalpel to achieve the purpose of changing the phase difference.

[0009] The beneficial effects of the present invention are: it can keep the ultrasonic knife working in a resonant state at all times, prevent the occurrence of a decrease in cutting efficiency and detuning during long-term use, and can intelligently and real-timely perform phase compensation during the frequency locking control process. The present invention can compensate for the phase difference of the ultrasonic knife control current and voltage in real time by combining the use of a reinforcement learning model in the frequency locking control algorithm to make up for the mismatch between the phase zero point and the lowest impedance point caused by the change in the operating temperature of the transducer, so that the ultrasonic knife always works in the best resonant state, reduces the heating of the transducer, and improves the working efficiency of the ultrasonic knife. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0011] Figure 1 It is a flow chart of a frequency locking control method of ultrasonic surgical equipment based on reinforcement learning of the present invention. DETAILED DESCRIPTION

[0012] Figure 1 In the embodiment shown, a frequency locking control method of ultrasonic surgical equipment based on reinforcement learning includes the following frequency locking control steps: S1. Obtaining ultrasonic scalpel working status information S: the working status information includes the phase difference ΔΦ between the driving current and the driving voltage of the ultrasonic scalpel, the blade impedance Z and the operating frequency F; S2. Calculate the phase compensation value: input the working state information obtained in the above step S1 into the trained reinforcement learning model to obtain the phase compensation value Φc; S3. Update the working state of the ultrasonic scalpel, and input the phase compensation value Φc into the state control module to determine the working state of the ultrasonic scalpel at the next moment; S4. In the above steps S1 to S3, the phase difference of the ultrasonic knife control current and voltage is compensated in real time to make up for the mismatch between the phase zero point and the lowest impedance point caused by the change of the transducer working temperature, so that the ultrasonic knife can always lock the frequency control and work in the best resonant state, reduce the heating of the transducer, and keep the ultrasonic knife system in a resonant state.

[0013] In the above steps S1 to S2, the working status information S of the ultrasonic knife includes at least one of the historical working statuses [St, St-1, St-2, ..., St-n] (n≥0), where St represents the working status at the current moment, and St-n represents the working status at the previous n moments. The more moments included in the working status information, the more accurate the calculation of the phase compensation value, but the longer the calculation time. Users can choose the number according to actual needs. The reinforcement learning model includes a state space, an action space, a learning goal and a training process. The state space includes the phase difference between the driving current and the driving voltage of the ultrasonic knife, the blade impedance and the working frequency; the action space includes increasing the phase difference, reducing the phase difference and maintaining the phase difference unchanged; the learning goal is to make the impedance value in the working state of the ultrasonic knife at the minimum point under the current working environment (our goal is to make our ultrasonic knife system always in a resonant state, and the LCR series resonant circuit in the resonant state has one phase difference of 0 and another circuit equivalent impedance of the lowest, because there will be additional capacitive reactance or inductive reactance in the non-resonant state, and since the phase difference cannot be directly measured, the impedance can be used as a learning goal and a basis for judging the end of training); the training process includes: outputting an action in the action space according to the state space of the ultrasonic knife at the current moment, the ultrasonic knife adjusts the working state according to the action instruction, if the impedance decreases or remains unchanged at the next moment, the model is given a positive reward, if the impedance increases at the next moment, the model is given a negative reward, if the model action can always keep the ultrasonic knife working at the lowest impedance point, it indicates that the model training is completed, and the reinforcement learning model after training is output. The state control module adjusts the working state of the ultrasonic knife by controlling the driving current frequency, driving current size or driving voltage size of the ultrasonic knife to achieve the purpose of changing the phase difference. It can keep the ultrasonic knife working in a resonant state at all times, prevent the reduction of cutting efficiency and detuning during long-term use, and can perform phase compensation intelligently and in real time during the frequency locking control process.

[0014] Since the static capacitance C0 is inside the transducer and cannot be measured directly, the current signal we actually collect is the sum of the currents of the two parallel branches of the static capacitance C0 and the equivalent resistance R1 including the load and loss. Therefore, when the phase difference of the measured voltage and current signals is 0, the phase difference between the current and voltage in the series branch of the dynamic inductance L1 and the parallel static capacitance C1 in the equivalent circuit when the piezoelectric transducer is working is not 0, which means that this branch is not at the theoretical resonant frequency point. The "phase compensation" in the above description of the manual means that we calculate a phase compensation value through an algorithm to keep the phase difference of the measured voltage and current signals at this phase compensation value point, rather than the phase 0 point.

[0015] The above content and structure describe the basic principle, main features and advantages of the product of the present invention, which should be understood by those skilled in the art. The above examples and descriptions are only for explaining the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which are within the scope of the present invention. The scope of the present invention is defined by the attached claims and their equivalents.

Claims

1. A frequency locking control method for ultrasonic surgical equipment based on reinforcement learning, characterized in that: The frequency locking control steps are as follows: S1. Obtaining ultrasonic scalpel working status information S: the working status information includes the phase difference ΔΦ between the driving current and the driving voltage of the ultrasonic scalpel, the blade impedance Z and the operating frequency F; S2. Calculate the phase compensation value: input the working state information obtained in the above step S1 into the trained reinforcement learning model to obtain the phase compensation value Φc; S3. Update the working state of the ultrasonic scalpel: input the phase compensation value Φc into the state control module to determine the working state of the ultrasonic scalpel at the next moment; S4. In the above steps S1 to S3, the phase difference of the ultrasonic knife control current and voltage is compensated in real time to make up for the mismatch between the phase zero point and the lowest impedance point caused by the change of the transducer working temperature, so that the ultrasonic knife can always lock the frequency control and work in the best resonant state, reduce the heating of the transducer, and keep the ultrasonic knife system in a resonant state.

2. The method for frequency locking control of ultrasonic surgical equipment based on reinforcement learning according to claim 1, characterized in that: In the above steps S1 to S2, the working status information S of the ultrasonic knife includes at least one of the historical working statuses [St, St-1, St-2, ..., St-n] (n≥0), where St represents the working status at the current moment, and St-n represents the working status at the previous n moments. The more moments included in the working status information, the more accurate the calculation of the phase compensation value, but the longer the calculation time. Users can choose the number according to actual needs.

3. The method for frequency locking control of ultrasonic surgical equipment based on reinforcement learning according to claim 1, characterized in that: The reinforcement learning model includes a state space, an action space, a learning goal and a training process. The state space includes the phase difference between the driving current and the driving voltage of the ultrasonic knife, the blade impedance and the working frequency; the action space includes increasing the phase difference, reducing the phase difference and maintaining the phase difference unchanged; The learning goal is to make the impedance value in the working state of the ultrasonic knife at the minimum point under the current working environment; the training process includes: outputting an action in an action space according to the state space of the ultrasonic knife at the current moment, and the ultrasonic knife adjusts the working state according to the action instruction. If the impedance decreases or remains unchanged at the next moment, the model is given a positive reward. If the impedance increases at the next moment, the model is given a negative reward. If the model's action can always keep the ultrasonic knife working at the lowest impedance point, it indicates that the model training is completed, and the trained reinforcement learning model is output.

4. The method for frequency locking control of ultrasonic surgical equipment based on reinforcement learning according to claim 1, characterized in that: The state control module adjusts the working state of the ultrasonic scalpel by controlling the driving current frequency, driving current size or driving voltage size of the ultrasonic scalpel to achieve the purpose of changing the phase difference.

Images