An adaptive radio frequency energy control simulation method based on multi-source data fusion

Abstract

This invention belongs to the field of medical device control technology and discloses an adaptive radio frequency energy control simulation method based on multi-source data fusion. The method includes the following steps: acquiring simulated tissue data through an integrated sensor module; preprocessing the acquired multi-source raw data to obtain a standardized dataset; using a weighted fusion algorithm to perform feature fusion on the preprocessed multi-source data to obtain a fused feature vector; calculating the optimal radio frequency energy output parameters based on the fused feature vector using a preset control model, and feeding this parameter back to the radio frequency device in real time to dynamically adjust the energy output; acquiring the adjusted power data in real time and comparing it with a preset threshold; if the threshold is exceeded, repeating the above process until the radio frequency process meets the preset standard. This invention employs the above-mentioned adaptive radio frequency energy control simulation method based on multi-source data fusion to achieve precise and dynamic control of radio frequency energy, adapting to the characteristics and dynamic changes of different tissue types.

Description

Technical Field

[0001] This invention relates to the field of medical device control technology, and in particular to an adaptive radio frequency energy control simulation method based on multi-source data fusion. Background Technology

[0002] In existing technologies, radio frequency (RF) energy control mostly adopts a fixed parameter mode, which presets a single energy output value based on experience and keeps it unchanged throughout the process. However, the tissue characteristics of RF vary significantly, and during the RF process, the temperature and impedance of the tissue will change dynamically with the process, while the environment of the RF region will also change. The fixed energy control mode cannot adapt to tissue differences and dynamic changes.

[0003] In existing technologies, some radio frequency energy control methods attempt to introduce a single parameter for feedback adjustment. However, relying solely on a single data point cannot fully reflect the true state of the radio frequency process, resulting in limited control accuracy. These methods still cannot solve the core challenges of individual adaptation and dynamic adjustment.

[0004] Therefore, developing a radio frequency energy control method that can integrate multi-source data and achieve adaptive adjustment to adapt to tissue differences and dynamic changes, thereby improving radio frequency accuracy and stability, has become an urgent technical problem to be solved. Summary of the Invention

[0005] The purpose of this invention is to provide an adaptive radio frequency energy control simulation method based on multi-source data fusion, which dynamically adjusts the radio frequency energy output by fusing real-time data from multiple sources, thereby achieving precise and stable control of radio frequency energy.

[0006] To achieve the above objectives, this invention provides an adaptive radio frequency energy control simulation method based on multi-source data fusion, comprising the following steps: Step S1: Multi-source data acquisition. Simulated tissue data is acquired through an integrated sensor module, including radio frequency tissue parameters, radio frequency area environmental parameters, and radio frequency equipment operating parameters. Step S2: Data preprocessing. Outlier removal, missing value completion, and standardization are performed on the collected multi-source raw data to obtain a standardized dataset. Step S3: Multi-source data fusion. A weighted fusion algorithm is used to fuse the features of the preprocessed multi-source data to obtain a fused feature vector. Step S4: Adaptive energy control decision-making. Based on the fused feature vector, the optimal radio frequency energy output parameters are calculated through a preset control model and fed back to the radio frequency device in real time to dynamically adjust the energy output. Step S5: Closed-loop verification and iterative optimization. Real-time data after power adjustment is collected and compared with the preset threshold. If the threshold is exceeded, steps S2-S4 are repeated until the RF process meets the preset standard.

[0007] Preferably, in step S1, the specific process of multi-source data acquisition is as follows: Simulated tissue data is collected through an integrated sensor module. All data are collected synchronously at a sampling frequency of 0.8-1.25MHz to ensure the real-time performance and synchronization of the data. Step S11: Radiofrequency tissue parameters include tissue morphology and tissue impedance Z; wherein, tissue morphology includes tissue stiffness D and tissue thickness H; Step S12: Environmental parameters of the radio frequency area include: real-time temperature T of the radio frequency area and ambient humidity RH of the radio frequency area; Step S13, Radio frequency equipment operating parameters: Collected through the monitoring module built into the radio frequency equipment, including the current radio frequency output power P, radio frequency, and output time t; wherein, the radio frequency is fixed at 4MHz, and the power adjustment range is 0-200W.

[0008] Preferably, in step S2, the specific process of data preprocessing is as follows: The collected multi-source raw data is preprocessed to eliminate the effects of noise, outliers and missing values, resulting in a standardized dataset. Step S21, Outlier Removal: Use the 3σ criterion to remove outlier data. For each type of data, calculate its mean μ and standard deviation σ, and remove data that exceed the range of [μ-3σ, μ+3σ]. Step S22, Missing Value Imputation: For missing data that appears during the data collection process, linear interpolation is used to impute the missing data, as shown below: ; in, These are the padding values ​​for missing data points; The previous valid data point before the missing data point; This represents the next valid data point after the missing data point; i is the index of the missing data point. Step S23, Standardization: Using the min-max standardization method, all preprocessed data are mapped to the [0,1] interval to eliminate the influence of dimensions. The standardization formula is: ; in, x represents the standardized data; x represents the preprocessed raw data. The minimum value of the data; The maximum value of the data; Step S24: After preprocessing, the standardized dataset is obtained, as shown below: ; in, A standardized subset of radio frequency tissue parameters; A standardized subset of environmental parameters for the radio frequency region; A standardized subset of operating parameters for radio frequency equipment.

[0009] Preferably, in step S3, the specific process of multi-source data fusion is as follows: A weighted fusion algorithm is used to fuse the features of the preprocessed multi-source data, and the fused feature vector is obtained by combining the influence weights of each data type on radio frequency energy control. Step S31: Determine the weighting coefficients: Determine the weights of various data types using the analytic hierarchy process (AHP), and combine this with expert evaluation to determine the weights of the radiofrequency tissue parameters. RF region environmental parameter weights RF equipment operating parameter weights And satisfy ; Step S32, Feature Fusion Calculation: The standardized data at each sampling time are weighted and summed to obtain the fused feature value. As shown below: ; in, , and These are the standardized values ​​for tissue impedance, tissue stiffness, and tissue thickness, respectively. and These are the standardized values ​​for temperature and humidity in the radio frequency region, respectively. and These are the standardized values ​​for RF output power and output time, respectively. Step S33: Construct the fused feature vector: Combine the fused feature values ​​at each sampling time. Arranged according to time series, the fused feature vector is obtained. Where n is the number of samples.

[0010] Preferably, in step S4, the specific process of adaptive energy control decision-making is as follows: Based on the fused feature vector, an adaptive energy control model is constructed to calculate the optimal RF energy output parameters, including output power P and RF frequency f, which remains constant at 4MHz. The parameters are fed back to the RF device in real time to dynamically adjust the energy output. Step S41: Establish the control model: Construct an adaptive energy control model using a BP neural network, with the input being a fused feature vector. The output is the optimal RF output power. ; Step S42, Model Training and Calibration: Train the BP neural network. During training, the gradient descent method is used to optimize the loss function. The mean squared error is used, as shown below: ; Where m is the number of training samples; This represents the optimal power output by the model. The optimal power verified in practice; Training is stopped when the loss function value is less than 0.001, and the calibrated adaptive energy control model is obtained. Step S43, Real-time Energy Adjustment: Input the fused feature vector into the calibrated control model to calculate the optimal RF output power at the current moment. and optimize RF output power The data is transmitted in real time to the radio frequency (RF) device, which then adjusts the output power based on the optimal RF output power. Adjust the energy output.

[0011] Preferably, the structure of the BP neural network is as follows: the number of neurons in the input layer is n, consistent with the dimension of the fused feature vector; the number of neurons in the hidden layer is 2n+1; and the number of neurons in the output layer is 1; the activation function is... The Sigmoid function is used, as shown below: ; in, This is the input value for the Sigmoid function.

[0012] Preferably, the power adjustment constraints are set as follows: the single adjustment range does not exceed 5W to avoid tissue damage caused by sudden power changes; and the power output range is controlled within 0-200W.

[0013] Preferably, in step S5, the specific process of closed-loop verification and iterative optimization is as follows: Multi-source data of simulated tissues after real-time power adjustment are acquired, preprocessed, and fused to obtain a new fused feature vector, which is then compared with a preset fused feature threshold range. Compare; If the new fusion eigenvalue F is in Within the range, the current RF energy output parameters are maintained; if F < This indicates insufficient energy. Repeat steps S2-S4 to increase the optimal power output; if F> This indicates that the energy is too high. Repeat steps S2-S4 to reduce the optimal power output. The closed-loop verification process is performed every 10ms until the RF process ends.

[0014] Preferably, the criterion for determining the end of the radiofrequency process is: the tissue impedance Z stabilizes at 500-800Ω and the duration is not less than 3s.

[0015] Therefore, the present invention employs the above-mentioned adaptive radio frequency energy control simulation method based on multi-source data fusion, and the beneficial effects are as follows: (1) Outstanding creativity: This invention innovatively integrates multi-source data and constructs an adaptive control model through a weighted fusion algorithm and a BP neural network, thereby realizing the dynamic and precise adjustment of radio frequency energy.

[0016] (3) High precision: Through the fusion of multi-source data, the true state of the radio frequency process is fully reflected, and the energy output is dynamically adjusted to ensure the stability of the radio frequency energy output.

[0017] (4) Highly practical: It is adaptable to different organizational types and has wide applicability.

[0018] (5) Fast response speed: The response time of the entire control process does not exceed 0.5s, and the closed-loop verification is performed once every 0.5s, which can track the dynamic changes of the radio frequency process in real time and ensure the timeliness and accuracy of energy control.

[0019] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0020] Figure 1 This is a flowchart of an adaptive radio frequency energy control simulation method based on multi-source data fusion according to the present invention. Detailed Implementation

[0021] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0022] like Figure 1 As shown, the present invention provides an adaptive radio frequency energy control simulation method based on multi-source data fusion, comprising the following steps: Step S1: Multi-source data acquisition. Three types of core data of the simulated tissue are acquired through an integrated sensor module, including radio frequency tissue parameters, radio frequency area environmental parameters, and radio frequency equipment operating parameters. Step S2: Data preprocessing. Outlier removal, missing value completion, and standardization are performed on the collected multi-source raw data to obtain a standardized dataset. Step S3: Multi-source data fusion. A weighted fusion algorithm is used to fuse the features of the preprocessed multi-source data to obtain a fused feature vector. Step S4: Adaptive energy control decision-making. Based on the fused feature vector, the optimal radio frequency energy output parameters are calculated through a preset control model and fed back to the radio frequency device in real time to dynamically adjust the energy output. Step S5: Closed-loop verification and iterative optimization. Real-time data after power adjustment is collected and compared with the preset threshold. If the threshold is exceeded, steps S2-S4 are repeated until the RF process meets the preset standard.

[0023] Example 1 Step S1: Multi-source data acquisition.

[0024] Three types of core data from the simulated tissue are collected using an integrated sensor module. All data are collected synchronously at a sampling frequency of 0.8-1.25MHz to ensure the real-time nature and synchronization of the data.

[0025] Step S11: Radiofrequency tissue parameters include tissue morphology and tissue impedance Z (unit: Ω); wherein, tissue morphology includes tissue stiffness D (unit: HA) and tissue thickness H (unit: mm).

[0026] Step S12: Environmental parameters of the radio frequency area include: real-time temperature T of the radio frequency area (unit: °C) and ambient humidity RH of the radio frequency area (unit: %).

[0027] Step S13, Radio frequency equipment operating parameters: Collected through the monitoring module built into the radio frequency equipment, including the current radio frequency output power P (unit: W), radio frequency f (unit: MHz), and output time t (unit: s); wherein, the radio frequency is fixed at 4MHz, and the power adjustment range is 0-200W.

[0028] Step S2: Data preprocessing.

[0029] The collected raw data from multiple sources are preprocessed to eliminate the effects of noise, outliers, and missing values, resulting in a standardized dataset.

[0030] Step S21, Outlier Removal: Outlier data is removed using the 3σ criterion. For each data category, the mean μ and standard deviation σ are calculated, and data exceeding the range of [μ-3σ, μ+3σ] are removed to avoid the impact of extreme values ​​on subsequent fusion and control.

[0031] Step S22, Missing Value Imputation: For missing data that appears during the data collection process, linear interpolation is used to impute the missing data, as shown below: ; in, These are the padding values ​​for missing data points; The previous valid data point before the missing data point; is the next valid data point after the missing data point; i is the index of the missing data point.

[0032] Step S23, Standardization: Using the min-max standardization method, all preprocessed data are mapped to the [0,1] interval to eliminate the influence of dimensions. The standardization formula is: ; in, x represents the standardized data; x represents the preprocessed raw data. The minimum value of the data; This represents the maximum value of the data.

[0033] Step S24: After preprocessing, the standardized dataset is obtained, as shown below: ; in, A standardized subset of radio frequency tissue parameters; A standardized subset of environmental parameters for the radio frequency region; A standardized subset of operating parameters for radio frequency equipment.

[0034] Step S3: Multi-source data fusion.

[0035] A weighted fusion algorithm is used to fuse the features of the preprocessed multi-source data. The fused feature vector is obtained by combining the influence weights of each data type on radio frequency energy control.

[0036] Step S31: Determine weighting coefficients: Determine the weights of various data types using the Analytic Hierarchy Process (AHP), and combine this with expert evaluation to determine the weights of radiofrequency tissue parameters. RF region environmental parameter weights RF equipment operating parameter weights And satisfy .

[0037] Step S32, Feature Fusion Calculation: The standardized data at each sampling time are weighted and summed to obtain the fused feature value. As shown below: ; in, , and These are the standardized values ​​for tissue impedance, tissue stiffness, and tissue thickness, respectively. and These are the standardized values ​​for temperature and humidity in the radio frequency region, respectively. and These are the standardized values ​​for RF output power and output time, respectively.

[0038] Step S33: Construct the fused feature vector: Combine the fused feature values ​​at each sampling time. Arranged according to time series, the fused feature vector is obtained. Where n is the number of samples.

[0039] Step S4: Adaptive energy control decision.

[0040] Based on the fused feature vector, an adaptive energy control model is constructed to calculate the optimal RF energy output parameters, which mainly include the output power P and the RF frequency f, which remains constant at 4MHz. The parameters are fed back to the RF device in real time to dynamically adjust the energy output.

[0041] Step S41: Establish the control model: Construct an adaptive energy control model using a BP neural network, with the input being a fused feature vector. The output is the optimal RF output power. .

[0042] The structure of a BP neural network is as follows: the input layer has n neurons, consistent with the dimension of the fused feature vector; the hidden layer has 2n+1 neurons; and the output layer has 1 neuron. The activation function is... The Sigmoid function is used, as shown below: ; in, This is the input value for the Sigmoid function.

[0043] Step S42, Model Training and Calibration: Using data containing different tissue types, with a sample size of no less than 1000 groups, train the BP neural network. During training, use gradient descent to optimize the loss function. The mean squared error (MSE) is used, as shown below: ; Where m is the number of training samples; This represents the optimal power output by the model. This represents the optimal power as verified in practice.

[0044] Training is stopped when the loss function value is less than 0.001, and the calibrated adaptive energy control model is obtained.

[0045] Step S43, Real-time Energy Adjustment: Input the fused feature vector obtained in step S3 into the calibrated control model to calculate the optimal RF output power at the current moment. and optimize RF output power The data is transmitted in real time to the radio frequency (RF) device, which then adjusts the output power based on the optimal RF output power. Adjust the energy output.

[0046] At the same time, power adjustment constraints are set: the single adjustment range shall not exceed 5W to avoid tissue damage caused by sudden power changes; the power output range shall be strictly controlled within 0-200W.

[0047] Step S5: Closed-loop verification and iterative optimization.

[0048] Real-time acquisition of multi-source data from simulated tissues after power adjustment (repeating step S1), followed by preprocessing and fusion to obtain a new fused feature vector, which is then compared with a preset fused feature threshold range. For comparison; this embodiment is set , This corresponds to the optimal range of radio frequency.

[0049] If the new fusion eigenvalue F is in Within the range, the current RF energy output parameters are maintained; if F < This indicates insufficient energy. Repeat steps S2-S4 to appropriately increase the optimal power output; if F> If the energy is too high, repeat steps S2-S4 and appropriately reduce the optimal power output.

[0050] The above closed-loop verification process is performed every 10ms until the radio frequency process ends; the judgment criterion is that the tissue impedance Z is stable at 500-800Ω for a duration of not less than 3s.

[0051] Example 2 This embodiment provides a more detailed description of the present invention and is only used to explain the present invention. It does not constitute a limitation on the scope of protection of the present invention.

[0052] 1. Parameter settings for the example.

[0053] This embodiment simulates the radio frequency (RF) process of tissue structure. Initially, the tissue stiffness D = 15HA, the tissue thickness H = 1.5mm, and the initial calibrated tissue impedance Z = 1200Ω. The initial output power of the RF device is P = 15W, the RF frequency f = 4MHz, and the sampling frequency = 1MHz. A preset fusion feature threshold range is also included. The single power adjustment range shall not exceed 5W, and the closed-loop verification period shall be 10ms.

[0054] 2. Specific implementation steps.

[0055] (1) Multi-source data acquisition.

[0056] Data is collected synchronously through an integrated sensor module, and some of the sampled data is shown in Table 1.

[0057] Table 1 Partial Sampling Data

[0058] (2) Data preprocessing.

[0059] Outlier removal: Calculate the mean μ and standard deviation σ of various data. For example, for tissue impedance Z, μ=1050Ω and σ=100Ω; [μ-3σ, μ+3σ]=[750, 1350]. The sampled data are all within this range and there are no outliers.

[0060] Missing value completion: There is no missing data in this example, so no completion is required.

[0061] Standardization: Taking tissue impedance Z as an example, its minimum value maximum value The Z-normalized value at sampling time 0.5s is shown below: ; Similarly, calculate the standardized values ​​of other data to obtain a standardized dataset.

[0062] (3) Multi-source data fusion.

[0063] Determine the weighting coefficients: , , ; Feature fusion calculation: Taking a sampling time of 1.0s as an example, the standardized data is as follows: , , , , , , ; Substituting the values ​​into the fusion formula, the fusion feature value F is calculated as follows: ; The fusion feature value F=0.513 is within the preset threshold range [0.2,0.8], so the current power output is temporarily maintained.

[0064] (4) Adaptive energy control decision.

[0065] Control model: A BP neural network is used, with 8 neurons in the input layer (corresponding to 8 standardized data points), 17 neurons in the hidden layer (2×8+1), and 1 neuron in the output layer.

[0066] Model training: The BP neural network was trained using 1000 sets of data until the loss function L < 0.001, thus obtaining the calibrated control model.

[0067] Energy adjustment: The fused feature vector from the sampling time of 1.5s is input into the control model to calculate the optimal power. If the single adjustment range is 1W ≤ 5W, which meets the constraint, the parameter is fed back to the radio frequency device, and the device adjusts the power to 16W.

[0068] (5) Closed-loop verification and iterative optimization.

[0069] After collecting multi-source data with adjusted power, and after preprocessing and fusion, a new fusion feature value F=0.61 is obtained, which is still within the range of [0.2,0.8], maintaining a power output of 16W.

[0070] The closed-loop verification is continuously performed. When the sampling time reaches 15s, the tissue impedance Z stabilizes at 700Ω and the duration reaches 3s. The radio frequency process is then judged to be over, and the energy output is stopped.

[0071] 3. Implementation results.

[0072] In this embodiment, the adaptive radio frequency energy control method of the present invention achieves precise and stable energy control of the radio frequency device.

[0073] Therefore, the present invention adopts the above-mentioned adaptive radio frequency energy control simulation method based on multi-source data fusion, which overcomes the defects of fixed radio frequency energy control, poor adaptability and low accuracy in the prior art; the present invention achieves precise and stable energy control by fusing multi-source real-time data and dynamically adjusting radio frequency energy output.

[0074] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. An adaptive radio frequency energy control simulation method based on multi-source data fusion, characterized in that, Includes the following steps: Step S1: Multi-source data acquisition. Simulated tissue data is acquired through an integrated sensor module, including radio frequency tissue parameters, radio frequency area environmental parameters, and radio frequency equipment operating parameters. Step S2: Data preprocessing. Outlier removal, missing value completion, and standardization are performed on the collected multi-source raw data to obtain a standardized dataset. Step S3: Multi-source data fusion. A weighted fusion algorithm is used to fuse the features of the preprocessed multi-source data to obtain a fused feature vector. Step S4: Adaptive energy control decision-making. Based on the fused feature vector, the optimal radio frequency energy output parameters are calculated through a preset control model and fed back to the radio frequency device in real time to dynamically adjust the energy output. Step S5: Closed-loop verification and iterative optimization. Real-time data after power adjustment is collected and compared with the preset threshold. If the threshold is exceeded, steps S2-S4 are repeated until the RF process meets the preset standard.

2. The adaptive radio frequency energy control simulation method based on multi-source data fusion according to claim 1, characterized in that, In step S1, the specific process of multi-source data acquisition is as follows: Simulated tissue data is collected through an integrated sensor module. All data are collected synchronously at a sampling frequency of 0.8-1.25MHz to ensure the real-time performance and synchronization of the data. Step S11: Radiofrequency tissue parameters include tissue morphology and tissue impedance Z; wherein, tissue morphology includes tissue stiffness D and tissue thickness H; Step S12: Environmental parameters of the radio frequency area include: real-time temperature T of the radio frequency area and ambient humidity RH of the radio frequency area; Step S13, Radio frequency equipment operating parameters: collected by the monitoring module built into the radio frequency equipment, including the current radio frequency output power P, radio frequency, and output time t; wherein, the radio frequency is fixed at 4MHz, and the power adjustment range is 0-200W.

3. The adaptive radio frequency energy control simulation method based on multi-source data fusion according to claim 2, characterized in that, In step S2, the specific process of data preprocessing is as follows: The collected multi-source raw data is preprocessed to eliminate the effects of noise, outliers and missing values, resulting in a standardized dataset. Step S21, Outlier Removal: Use the 3σ criterion to remove outlier data. For each type of data, calculate its mean μ and standard deviation σ, and remove data that exceed the range of [μ-3σ, μ+3σ]. Step S22, Missing Value Imputation: For missing data that appears during the data collection process, linear interpolation is used to impute it, as shown below: ； in, These are the padding values ​​for missing data points; The previous valid data point is the one preceding the missing data point; This represents the next valid data point after the missing data point; i is the index of the missing data point. Step S23, Standardization: Using the min-max standardization method, all preprocessed data are mapped to the [0,1] interval to eliminate the influence of dimensions. The standardization formula is: ； in, x represents the standardized data; x represents the preprocessed raw data. The minimum value of the data; The maximum value of the data; Step S24: After preprocessing, the standardized dataset is obtained, as shown below: ； in, A standardized subset of radio frequency tissue parameters; A standardized subset of environmental parameters for the radio frequency region; A standardized subset of operating parameters for radio frequency equipment.

4. The adaptive radio frequency energy control simulation method based on multi-source data fusion according to claim 3, characterized in that, In step S3, the specific process of multi-source data fusion is as follows: A weighted fusion algorithm is used to fuse the features of the preprocessed multi-source data, and the fused feature vector is obtained by combining the influence weights of each data type on radio frequency energy control. Step S31: Determine the weighting coefficients: Determine the weights of various data types using the analytic hierarchy process (AHP), and combine this with expert evaluation to determine the weights of the radiofrequency tissue parameters. RF region environmental parameter weights RF equipment operating parameter weights And satisfy ; Step S32, Feature Fusion Calculation: The standardized data at each sampling time are weighted and summed to obtain the fused feature value. As shown below: ； in, , and These are the standardized values ​​for tissue impedance, tissue stiffness, and tissue thickness, respectively. and These are the standardized values ​​for temperature and humidity in the radio frequency region, respectively. and These are the standardized values ​​for RF output power and output time, respectively. Step S33: Construct the fused feature vector: Combine the fused feature values ​​at each sampling time. Arranged according to time series, the fused feature vector is obtained. Where n is the number of samples.

5. The adaptive radio frequency energy control simulation method based on multi-source data fusion according to claim 4, characterized in that, In step S4, the specific process of adaptive energy control decision-making is as follows: Based on the fused feature vector, an adaptive energy control model is constructed to calculate the optimal RF energy output parameters, including output power P and RF frequency f, which remains constant at 4MHz. The parameters are fed back to the RF device in real time to dynamically adjust the energy output. Step S41: Establish the control model: Construct an adaptive energy control model using a BP neural network, with the input being a fused feature vector. The output is the optimal RF output power. ; Step S42, Model Training and Calibration: Train the BP neural network. During training, the gradient descent method is used to optimize the loss function. The mean squared error is used, as shown below: ； Where m is the number of training samples; This represents the optimal power output by the model. The optimal power verified in practice; Training is stopped when the loss function value is less than 0.001, and the calibrated adaptive energy control model is obtained. Step S43, Real-time Energy Adjustment: Input the fused feature vector into the calibrated control model to calculate the optimal RF output power at the current moment. and optimize RF output power The data is transmitted in real time to the radio frequency (RF) device, which then adjusts the output power based on the optimal RF output power. Adjust the energy output.

6. The adaptive radio frequency energy control simulation method based on multi-source data fusion according to claim 5, characterized in that, The structure of a BP neural network is as follows: the input layer has n neurons, consistent with the dimension of the fused feature vector; the hidden layer has 2n+1 neurons; and the output layer has 1 neuron. The activation function is... The Sigmoid function is used, as shown below: ； wherein, is an input value for the Sigmoid function.

7. The adaptive radio frequency energy control simulation method based on multi-source data fusion according to claim 5, characterized in that, The power adjustment constraints are set as follows: the single adjustment range shall not exceed 5W to avoid tissue damage caused by sudden power changes; the power output range shall be controlled between 0-200W.

8. The adaptive radio frequency energy control simulation method based on multi-source data fusion according to claim 5, characterized in that, In step S5, the specific process of closed-loop verification and iterative optimization is as follows: The multi-source data of the simulated tissue after power adjustment is collected in real time, and after preprocessing and fusion, a new fusion feature vector is obtained, and a preset fusion feature threshold range is compared comparison; If the new fusion eigenvalue F is in Within the range, the current RF energy output parameters are maintained; if F < This indicates insufficient energy. Repeat steps S2-S4 to increase the optimal power output; if F> This indicates that the energy is too high. Repeat steps S2-S4 to reduce the optimal power output. The closed-loop verification process is performed every 10ms until the RF process ends.

9. The adaptive radio frequency energy control simulation method based on multi-source data fusion according to claim 8, characterized in that, The criteria for determining the end of the radiofrequency procedure are: the tissue impedance Z stabilizes at 500-800Ω and the duration is not less than 3s.

Images