Bipolar high frequency dual output energy control method and system based on tissue impedance feedback

Abstract

This invention belongs to the field of energy control technology, specifically a bipolar high-frequency dual-output energy control method and system based on tissue impedance feedback. The method includes: acquiring dual-channel raw impedance time-series sequences; separating slowly varying and high-frequency noise components through singular spectrum decomposition; retaining the core slowly varying components to obtain a preliminarily purified impedance core component sequence; extracting phase features; constructing a high-dimensional amplitude-phase joint feature matrix by combining the dual-channel high-frequency excitation frequencies; mapping this matrix using a trained manifold embedding dimensionality reduction model to output a low-dimensional manifold distance metric sequence; constructing a sparse Bayesian dynamic model and optimizing unmixing; selecting the target feature component sequence; correcting dual-channel phase drift to obtain the phase feature sequence; allocating dual-channel high-frequency parameters; and outputting precise bipolar high-frequency dual-output energy control commands. This invention achieves high-precision, high-stability, and safe energy control through end-to-end impedance signal processing and algorithm optimization, adaptively regulating the dual-channel outputs.

Description

Technical Field

[0001] This invention belongs to the field of energy control technology, specifically a bipolar high-frequency dual-output energy control method and system based on tissue impedance feedback. Background Technology

[0002] Traditional control methods only perform simple threshold determination and linear calculations based on the original impedance time-domain data. They do not perform layered demixing and high-dimensional feature mining on impedance signals with mixed noise, phase shift and superimposed interference. They lack refined feature separation and nonlinear dimensionality reduction analysis, and cannot accurately extract the effective core features of the corresponding tissue coagulation state. They are extremely susceptible to electromagnetic interference and tissue impedance mixed signals, resulting in insufficient matching degree of dual-path energy distribution, which has become the core defect restricting the control accuracy. Summary of the Invention

[0003] To overcome the shortcomings of existing technologies, this invention proposes a bipolar high-frequency dual-output energy control method based on tissue impedance feedback. This invention is primarily used to solve the problem of imbalanced distribution of dual-channel high-frequency energy.

[0004] The present invention provides a bipolar high-frequency dual-output energy control method based on tissue impedance feedback, comprising: S1: Collect dual-channel raw impedance time series and perform singular spectrum decomposition on each channel to construct eigenvalue spectrum and singular vector space. Separate the slowly varying components and high-frequency noise components, screen and retain the core slowly varying components, and output the preliminary purified impedance core component sequence.

[0005] S2: Extract the phase features of the dual-path impedance core component sequence, construct a set of amplitude and phase joint features based on the dual-path high-frequency excitation frequency parameters, and form a high-dimensional joint feature matrix.

[0006] S3: Train the manifold embedding dimensionality reduction model using multiple sets of measured impedance samples, input the high-dimensional joint feature matrix into the manifold embedding dimensionality reduction model for mapping, calculate the distance between the two paths in the manifold space, and output a low-dimensional manifold distance metric sequence.

[0007] S4: Using the low-dimensional manifold distance metric sequence as the observation value, a sparse Bayesian dynamic model is constructed by combining the sparse prior distribution. The model hyperparameters are optimized and modeled and peaked, the target feature components are screened, and the target feature component sequence is output.

[0008] S5: Extract the phase information of the dual target feature components in the target feature component sequence, establish the phase and mapping model of the dual signals, correct the phase drift error between the two channels, and output the phase feature sequence.

[0009] S6: Low-amplitude chaotic phase modulation of the phase feature sequence, removing residual interference through chaos and discrimination to obtain a pure feature sequence, allocating dual-channel high-frequency output parameters, and outputting a dual-polar high-frequency dual-output energy precision control command.

[0010] According to the bipolar high-frequency dual-output energy control method based on tissue impedance feedback provided by the present invention, the specific steps in step S1 of outputting the preliminary purified impedance core component sequence are as follows: S11: Acquire dual-channel original impedance time series, and reconstruct the dual-channel original impedance time series data. Construct single-channel trajectory matrices according to fixed delay window lengths to form two-channel trajectory matrices.

[0011] S12: Perform singular value decomposition on the two trajectory matrices respectively to obtain the singular value sequence and left and right singular vectors of the corresponding matrices, which constitute the eigenvalue spectrum and singular vector space of the two signals respectively.

[0012] S13: Based on the numerical distribution law of the eigenvalue spectrum and the temporal component shape corresponding to the singular vector, the two signals are decomposed into a slow-changing component with a gradual amplitude change and a high-frequency disturbance component with rapid fluctuations.

[0013] S14: Based on the slow-changing component and the high-frequency disturbance component, select the singular vector associated with the large eigenvalue for time series reconstruction, and remove the disturbance component to obtain the impedance core component sequence.

[0014] According to the bipolar high-frequency dual-output energy control method based on tissue impedance feedback provided by the present invention, the specific steps for forming the high-dimensional joint feature matrix in step S2 are as follows: S21: Decompose the time-domain signal based on the core component sequence of the dual impedance, calculate the phase shift difference and dynamic fluctuation value of the two sequences for each sampling point, and output the basic feature set of the dual impedance phase.

[0015] S22: Calculate the amplitude correlation mapping based on the dual-path impedance phase basic feature set and the preset dual-path high-frequency excitation frequency parameters, use frequency constraints to gather and integrate effective data, and arrange multi-dimensional feature combinations to form an amplitude-phase coupling dataset.

[0016] S23: Expand the row and column dimensions and stitch features based on the amplitude-phase coupling dataset. Organize the feature arrangement according to the signal time-series correlation logic and frequency matching rules. Integrate the coupling information to build a multi-dimensional data array and obtain a standardized high-dimensional joint feature matrix.

[0017] According to the bipolar high-frequency dual-output energy control method based on tissue impedance feedback provided by the present invention, the specific steps in step S3 for outputting the low-dimensional manifold distance metric sequence are as follows: S31: The model is iteratively trained based on the measured impedance samples corresponding to multiple organizational states. The manifold mapping constraint relationship is fitted according to the distribution law of sample characteristics, and the solidified manifold embedding operation model is output.

[0018] S32: Input the high-dimensional joint feature matrix into the manifold embedding operation model, perform spatial dimension mapping transformation of the dual-path impedance phase trajectory, and output low-dimensional dual-path impedance trajectory characterization data.

[0019] S33: Calculate the quantized distance index between two trajectories based on low-dimensional trajectory representation data and manifold space metric rules, perform continuous solution and regular sorting of global time series data, and output a low-dimensional manifold distance metric sequence.

[0020] According to the bipolar high-frequency dual-output energy control method based on tissue impedance feedback provided by the present invention, the specific steps of outputting the solidified manifold embedding computation model in step S31 are as follows: Measured impedance samples corresponding to different tissue states are collected and their values ​​are normalized and outliers are removed. The dimensions and value ranges of the sample data are unified. Statistical analysis and clustering of the sample feature distribution are performed, and an impedance sample feature dataset is output.

[0021] Based on the impedance sample feature dataset, construct iterative training constraints, continuously update the model weight parameters, fit the exclusive manifold mapping association rules according to the data distribution pattern, gradually converge and correct the mapping boundary relationship, and output the initial model architecture.

[0022] The initial model architecture is compared with the convergence threshold. Based on the error back-correction and generalization ability calibration, the mapping operation logic adapted to bioimpedance characteristics is solidified, the internal operation parameters and manifold constraint ratio of the model are locked, and the solidified manifold embedding operation model is output.

[0023] According to the bipolar high-frequency dual-output energy control method based on tissue impedance feedback provided by the present invention, the specific steps for outputting the target feature component sequence in step S4 are as follows: S41: Using the low-dimensional manifold distance metric sequence as the global observation data source, and combining it with sparse prior distribution constraints for probability mapping fitting, a basic sparse Bayesian computation framework adapted to the dynamic changes of tissue impedance is built, and a sparse Bayesian dynamic model is output.

[0024] S42: Input a small number of tissue calibration samples into the initial sparse Bayesian dynamic model, iteratively correct the hyperparameter values ​​inside the model, weaken the parameter bias caused by environmental interference, and output the corrected computational model.

[0025] S43: Based on the modified operation model, perform dynamic fitting operation on the low-dimensional manifold distance metric sequence, separate the mixed superimposed multi-class feature components, and output the full set of impedance feature components.

[0026] S44: Based on the response characteristics of the dual-path solidification state matched with the set of full impedance characteristic components, irrelevant interference components are screened out and the target characteristic component sequence is output.

[0027] According to the bipolar high-frequency dual-output energy control method based on tissue impedance feedback provided by the present invention, the specific steps for outputting the sparse Bayesian dynamic model in step S41 are as follows: Low-dimensional manifold distance metric sequences under bipolar high-frequency dual-output conditions are collected, and sparse prior distribution intervals are defined by combining tissue impedance characteristics to perform probability mapping fitting operations and generate constraint configuration parameters.

[0028] Based on the constraint configuration parameters, a computational framework adapted to the dynamic changes in tissue impedance is built, and Bayesian probability operation logic is embedded to form a basic sparse Bayesian operation model.

[0029] The dual-channel tissue impedance time series data are imported into the basic sparse Bayesian operation model for dynamic fitting calculation, the data distribution pattern is sorted out, and a low-dimensional feature dataset is output.

[0030] Based on the low-dimensional feature dataset, the bipolar high-frequency energy control requirements are matched, the model iteration threshold and weight allocation rules are optimized, and a sparse Bayesian dynamic model is output.

[0031] According to the bipolar high-frequency dual-output energy control method based on tissue impedance feedback provided by the present invention, the specific steps of outputting the phase characteristic sequence in step S5 are as follows: S51: Extract the phase information of the dual-path target feature components in the target feature component sequence, and analyze the phase value at each time step through Hilbert transform to obtain dual-path phase time series data.

[0032] S52: Construct a phase mapping model based on dual-channel phase timing data, quantify the correlation and parallel rules of the dual-channel phases through model calculations, and output dual-channel phase correlation parameters.

[0033] S53: Locate the phase drift error between the two channels based on the dual-channel phase correlation parameters, perform precise correction through a targeted algorithm, and output corrected dual-channel phase data.

[0034] S54: Based on the corrected dual-channel phase data, integrate the data according to the dual-channel parallel matching requirements and output the phase feature sequence.

[0035] According to the bipolar high-frequency dual-output energy control method based on tissue impedance feedback provided by the present invention, the specific steps in step S6 for outputting the precise control command for bipolar high-frequency dual-output energy are as follows: S61: Obtain the original phase feature sequence data, filter the low amplitude data segments in the sequence according to the set amplitude threshold, correct the phase distribution characteristics of the sequence, and output the modulated phase sequence.

[0036] S62: Based on the chaos and the determination of the constraints, compare the coupling correlation between the modulated phase sequence and the standard chaotic reference signal, identify the residual clutter in the sequence, and output the pure feature sequence.

[0037] S63: Based on the phase and amplitude correlation quantization values ​​of the pure feature sequence, differentiate the high-frequency operating parameters of the dual channels and generate high-frequency output configuration data.

[0038] S64: Based on the high-frequency output configuration data and bipolar level switching logic, perform precise allocation of energy weights for dual output channels and quantization encoding of instructions to generate precise bipolar high-frequency dual-output energy control instructions.

[0039] This invention also provides a bipolar high-frequency dual-output energy control system based on tissue impedance feedback, comprising: The singular decomposition module constructs the eigenvalue spectrum and singular vector space, decomposes them to obtain the slowly varying components and high-frequency noise components, and outputs a preliminary purified impedance core component sequence.

[0040] The amplitude-phase array module is used to extract the phase features of the dual-path impedance core component sequence, construct a joint amplitude-phase feature set, and form a high-dimensional joint feature matrix.

[0041] The manifold dimensionality reduction module is used to train the manifold embedding dimensionality reduction model. It inputs the high-dimensional joint feature matrix into the manifold embedding dimensionality reduction model for mapping, calculates the distance between the two paths in the manifold space, and outputs a low-dimensional manifold distance metric sequence.

[0042] The sparse unmixing module is used to construct a sparse Bayesian dynamic model by using a low-dimensional manifold distance metric sequence as the observation value and combining it with a sparse prior distribution, to filter the target feature components and output the target feature component sequence.

[0043] The phase-parallel module is used to extract the phase information of the dual target feature components in the target feature component sequence, establish a dual-signal phase-parallel mapping model, and output the phase feature sequence.

[0044] The chaotic energy control module is used to perform low-amplitude chaotic phase modulation on the phase feature sequence to obtain a pure feature sequence, allocate dual-channel high-frequency output parameters, and output bipolar high-frequency dual-output energy precise control commands.

[0045] The present invention provides a bipolar high-frequency dual-output energy control method based on tissue impedance feedback, which adaptively regulates the dual outputs through end-to-end impedance signal processing and algorithm optimization to achieve high-precision, high-stability, and safe energy control.

[0046] The beneficial effects of this invention are as follows: 1. This invention introduces a multi-level digital processing architecture combining singular spectrum decomposition, manifold embedding dimensionality reduction, and sparse Bayesian peak unmixing. Using dual-channel tissue impedance time-series signals as feedback, it sequentially performs operations such as time-series reconstruction, feature space construction, high-dimensional feature fusion, manifold distance measurement, sparse modeling unmixing, and phase error correction. This process performs layered purification, feature compression, component separation, and phase correction on the original impedance signal. Based on the end-to-end digital signal analysis logic, it accurately removes invalid impurity components from the impedance signal, captures the subtle gradual impedance changes during tissue solidification in real time, establishes a phase coordination constraint relationship between the two signals, and dynamically allocates the differentiated ratio of high-frequency parameters and energy weights based on the dynamic changes in tissue impedance. It effectively breaks free from the limitations of traditional fixed parameter output modes, achieving precise control of bipolar dual-channel high-frequency energy. It eliminates energy imbalance caused by dual-channel output phase shift and load mismatch, accurately adapts to the impedance differences of tissues in different regions, avoids local energy overload burns or incomplete effects due to insufficient energy, significantly improves the uniformity of tissue thermal action, enhances the dynamic adaptation capability and control precision of bipolar energy output, and ensures stable and controllable tissue action effects during surgery.

[0047] 2. This invention constructs a comprehensive anti-interference processing mechanism from front-end signal acquisition to back-end command output. It combines multiple noise reduction strategies, including singular component screening, manifold feature screening, sparse prior constraints, chaotic phase modulation, and chaotic discriminant analysis, to decompose, identify, and filter out interference from dual-path impedance sequences segment by segment, point by point, and layer by layer. First, it distinguishes effective slowly varying components from high-frequency disturbance noise based on the contribution rate of singular values. Then, it uses a trained manifold model to remove redundant information from high-dimensional features. Next, it separates mixed interference components using a sparse Bayesian model. Finally, it employs low-amplitude chaotic modulation and co-verification to remove residual electromagnetic clutter, loop noise, environmental power frequency interference, and residual errors caused by electrode contact fluctuations. This continuous, step-by-step purification and deviation correction of the signal preserves core effective features strongly correlated with the actual physiological state of the tissue throughout the process, suppressing the coupling effects of multiple types of interference in complex intraoperative environments. It significantly improves the fidelity and timing stability of impedance feedback signals, solves the defects of signal distortion and drift under high-frequency conditions, reduces the interference of external disturbances on energy calculation logic, ensures that the dual-channel energy control calculation is always in line with the actual working conditions of the tissue, guarantees the stability of the bipolar high-frequency system for long-term continuous operation, and improves the operational reliability in extremely complex surgical environments. Attached Figure Description

[0048] The invention will now be further described with reference to the accompanying drawings.

[0049] Figure 1 This is a flowchart illustrating the steps of the bipolar high-frequency dual-output energy control method based on tissue impedance feedback provided in this embodiment of the invention.

[0050] Figure 2This is a flowchart of a bipolar high-frequency dual-output energy control method based on tissue impedance feedback provided in an embodiment of the present invention.

[0051] Figure 3 This is a block diagram of a bipolar high-frequency dual-output energy control system based on tissue impedance feedback provided in an embodiment of the present invention. Detailed Implementation

[0052] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below according to specific embodiments.

[0053] like Figures 1 to 3 As shown in the embodiment of the present invention, the bipolar high-frequency dual-output energy control method based on tissue impedance feedback includes: S1: Collect dual-channel raw impedance time series and perform singular spectrum decomposition on each channel to construct eigenvalue spectrum and singular vector space. Separate the slowly varying components and high-frequency noise components, screen and retain the core slowly varying components, and output the preliminary purified impedance core component sequence.

[0054] S11: Acquire dual-channel raw impedance time-series data using an impedance acquisition device, ensuring consistent sampling frequencies and complete timestamps for both channels. Retain all timing information from the raw data during acquisition without any preprocessing, resulting in two raw impedance time-series sequences, each of length N, denoted as... and Where t = 1, 2, ..., N, and N is the number of sampling points. Subsequently, time-series reconstruction is performed on the two original impedance time-series data streams, selecting a fixed delay window length τ (τ is a positive integer, preset according to the frequency characteristics of the impedance signal) and an embedding dimension m (m is set according to the singular spectrum decomposition requirements). Constructing the trajectory matrix ,matrix The number of rows is m, and the number of columns is Its elements are Where i = 1, 2, ..., m, j = 1, 2, ..., , This is the initial sampling time. Using the same delay window length τ and embedding dimension m, for... Perform the same trajectory matrix construction operation to obtain the trajectory matrix. , Dimensions and They are completely identical, forming two trajectory matrices with the same structure and dimensions.

[0055] S12: Based on the two trajectory matrices and ,right The core of singular value decomposition is to perform singular value decomposition. It can be decomposed into the product of three matrices, i.e. ,in, Let m×m be a left singular vector matrix, and its column vectors be... The eigenvectors are such that the column vectors are mutually orthogonal. for A singular value diagonal matrix, where the elements on the diagonal are the singular values ​​of X1, arranged in descending order to form a singular value sequence. , for The transpose of the right singular vector matrix, with column vectors as The eigenvectors are such that the column vectors are also mutually orthogonal. The same singular value decomposition method is used to analyze the trajectory matrix. Decompose to obtain ,in, for The left singular vector matrix, for The singular value diagonal matrix, the singular value sequence is , for The right singular vector matrix. (The rest is missing from the original text.) Singular value sequences As its eigenvalue spectrum, the left singular vector matrix and right singular vector matrix constitute The corresponding singular vector space. Similarly, Singular value sequences As its eigenvalue spectrum and constitute The corresponding singular vector space is used to construct the eigenvalue spectra of the two signals and the singular vector space.

[0056] S13: Based on the eigenvalue spectra and singular vector spaces of the two signals, analyze the numerical distribution patterns of the eigenvalue spectra of each signal, calculate the contribution rate of each singular value to the sum of all singular values, and distinguish between dominant and secondary singular values ​​through the contribution rate distribution. The contribution rate corresponding to the dominant singular value is significantly higher than that of the secondary singular value. Reconstruct the corresponding time-series components using the singular vector matrix. Combine each column of the left singular vector matrix with the corresponding columns of the singular values ​​and the right singular vector matrix to reconstruct m independent time-series components. Observe the shape of each time-series component. The time-series component corresponding to the dominant singular value has a smooth amplitude change and a low fluctuation frequency, while the time-series component corresponding to the secondary singular value has a violent amplitude fluctuation and a high frequency. Classify the time-series component corresponding to the dominant singular value as a slow-change component and the time-series component corresponding to the secondary singular value as a high-frequency disturbance component, and decompose the two signals into slow-change components and high-frequency disturbance components respectively.

[0057] S14: Combining the slow-change component and high-frequency disturbance component corresponding to the two signals respectively, for each signal, select the top k singular values ​​in the eigenvalue spectrum whose cumulative contribution rate reaches a preset threshold, and extract the first k columns of the left singular vector matrix of that signal to form a matrix. Extract the first k columns of the right singular vector matrix of this signal to form a matrix. And construct a k×k diagonal matrix consisting of the first k singular values. Reconstruction formula based on reduced-rank singular value decomposition Perform the calculation, where, The first k columns of the left singular vector matrix corresponding to the path signal. It is a diagonal matrix composed of the first k singular values. Let k be the first k columns of the right singular vector matrix corresponding to the path signal. The dimensions of each matrix in the formula satisfy... The dimensions are consistent with the original trajectory matrix, and the units are unified with the original impedance matrix, with the left and right sides of the equality being symmetrical. This formula is used to reconstruct the trajectory matrix that retains only the slow-transform component. The high-frequency disturbance components are completely eliminated, and the reconstruction matrix is ​​converted into a one-dimensional time sequence to obtain a single-channel impedance core component sequence. The same operation is performed on the two signals to obtain two-channel impedance core component sequences.

[0058] S2: Extract the phase features of the dual-path impedance core component sequence, construct a set of amplitude and phase joint features based on the dual-path high-frequency excitation frequency parameters, and form a high-dimensional joint feature matrix.

[0059] S21: The acquired dual-channel impedance core component sequences are subjected to refined time-domain signal decomposition processing. Using continuous sampling points as the basic computational unit, point-by-point analytical calculations are performed on the two impedance time-series signals. Based on the phase change characteristics of the two component sequences, the phase offset difference is accurately calculated point-by-point, quantifying the phase difference between different signal channels and capturing the dynamic phase fluctuation values ​​during signal transmission, objectively representing the continuous phase change pattern over time. After numerical calculation, a unified phase data quantization standard and normalization conversion rule are set to eliminate numerical errors caused by signal acquisition deviations and environmental interference. The discretely distributed phase difference and fluctuation data are then regularized, sorted, formatted, and dimensionally calibrated.

[0060] Data is structured and integrated according to the time sequence, standardizing the data format, value range and arrangement logic of single features, establishing a one-to-one mapping relationship between phase features and the original impedance component sequence, eliminating abnormal discrete phase data, and performing orderly collection and standardization of basic phase dimension information. Based on the phase feature data after full calculation and correction, a complete, dimensionally unified and time-corresponding dual-path impedance phase basic feature set is generated.

[0061] S22: Using the dual-channel impedance phase feature set as the core calculation data source, and combining it with the device's preset dual-channel high-frequency excitation frequency parameters, an associated calculation model is constructed. Precise amplitude correlation mapping calculations are performed based on frequency matching constraints. Combining the frequency band characteristics and signal response patterns of the high-frequency excitation, a quantitative mapping relationship between phase characteristics and impedance amplitude parameters is established. Based on the time-series data in the phase feature set, the amplitude information of the two signals is extrapolated and numerically converted, realizing the correlation extension from phase parameters to amplitude characteristics. Constraints such as frequency thresholds and bandwidth ranges corresponding to the high-frequency excitation are introduced. The mapped amplitude data and original phase data are filtered and selected, eliminating invalid data that exceeds the frequency adaptation range or does not conform to the high-frequency excitation response characteristics, and consolidating and integrating the effective feature data. Based on the inherent relationship between the two core parameters of phase and amplitude, multi-dimensional feature combination and arrangement are carried out according to the data dimension division rules and time sequence correspondence logic. This breaks the data limitations of single phase feature, cross-integrates and orderly splices time sequence phase feature and frequency-related amplitude feature, unifies the data arrangement rules and storage structure of multiple features, and performs complementary combination and integrated collection of features of different dimensions to form an amplitude-phase coupled dataset with strong parameter coupling, high data correlation and adaptability to high frequency excitation scenarios.

[0062] S23: To address the issues of single data dimension and insufficient feature fusion in the amplitude-phase coupling dataset, we perform row and column dimension expansion and cross-dimensional feature stitching processing on the entire dataset. Combining the timing acquisition rules for impedance signal detection, we clarify the time-series correlation logic of each set of amplitude-phase coupling data, define the data correspondence under different sampling nodes and excitation frequencies, match the frequency adaptation rules for dual-channel high-frequency excitation, and perform full feature arrangement correction and orderly regularization according to the rules. This corrects problems such as disordered feature arrangement and misaligned dimension correspondence, unifies the arrangement standards of data in each row and column, and ensures accurate binding of phase features, amplitude features, frequency parameters, and timing information. After feature regularization, we layer-by-layer superimpose the core information of multiple sets of amplitude-phase coupling data, performing horizontal dimension expansion and vertical feature stacking to gradually build a multi-level, multi-dimensional composite data array, achieving the systematic integration of scattered coupled features. Based on the array structure optimization algorithm, the data arrangement is improved, and the normalization calibration, dimension unification and format standardization of the global features are performed to eliminate the computational barriers caused by dimensional differences. The constructed multidimensional data array is globally verified and optimized to output a standardized high-dimensional joint feature matrix with a standard structure, high feature density and comprehensive information dimensions.

[0063] S3: The manifold embedding dimensionality reduction model is trained offline using multiple sets of measured impedance samples under different organizational states. The high-dimensional joint feature matrix is ​​input into the manifold embedding dimensionality reduction model to map the dual-path impedance phase trajectory to the low-dimensional manifold space. The distance metric of the dual-path trajectory in the manifold space is calculated, and the low-dimensional manifold distance metric sequence is output.

[0064] S31: Collect measured impedance samples corresponding to various tissue states, covering all typical states that tissues may experience during bipolar high-frequency energy treatment, including normal soft tissue, coagulated tissue of different degrees, edematous tissue, and hemorrhagic tissue. These samples directly correspond to the relationship between bipolar high-frequency energy output parameters and tissue physiological state, and are the core foundation for the model to adapt to actual control scenarios and achieve accurate feedback.

[0065] After data acquisition, all measured impedance samples underwent systematic preprocessing. Outliers caused by high-frequency excitation interference and poor sensor contact during acquisition were eliminated using the 3σ criterion. Then, the min-max normalization method was used to calibrate the sample data to a uniform value range, eliminating dimensional differences between different tissue samples and ensuring the authenticity, consistency, and comparability of sample characteristics. Based on the preprocessed, clean impedance samples, the manifold embedding dimensionality reduction model was trained offline iteratively. During training, the characteristic distribution patterns of impedance samples under different tissue states were combined to specifically fit a dedicated manifold mapping constraint relationship, breaking the limitations of general dimensionality reduction models and focusing on enhancing the model's ability to capture the intrinsic relationship between tissue impedance characteristics and the effects of bipolar high-frequency energy. Through multiple iterations, the model weight parameters were updated, gradually converging and correcting the mapping boundary. After each iteration, the model's fitting error and generalization error were calculated until the error dropped below a preset threshold and the generalization ability met the real-time and accuracy requirements of bipolar high-frequency dual-output energy control. A solidified manifold embedding computation model was then output. This model can efficiently adapt to tissue impedance feedback scenarios and can quickly and accurately achieve dimensionality reduction mapping of high-dimensional tissue impedance characteristics.

[0066] S32: The high-dimensional joint feature matrix integrates the phase characteristics of the dual-path impedance core components and the dual-path high-frequency excitation parameters, comprehensively covering the correlation information between the dual-path impedance signals and excitation parameters during the bipolar high-frequency energy output process. However, it has a high dimension and large data redundancy, and cannot be directly used for real-time energy control decisions.

[0067] The solidified manifold embedding computation model, which is trained by inputting a high-dimensional joint feature matrix, has been trained offline with multiple types of tissue impedance samples. It has built-in manifold mapping rules adapted to tissue impedance feedback scenarios, which can accurately identify and retain core features related to tissue state and high-frequency energy effects, while eliminating irrelevant and redundant information.

[0068] During model operation, the dual-impedance phase trajectories in the high-dimensional joint feature matrix undergo spatial dimension mapping transformation according to built-in mapping rules. Through nonlinear compression projection, the dual-impedance phase trajectories in high-dimensional space are transformed into feature trajectories in low-dimensional manifold space. The entire mapping process strictly matches the bipolar high-frequency dual-output energy control logic, ensuring that the mapped low-dimensional features can accurately represent the dynamic changes of the dual-impedance phases and form a one-to-one correspondence with tissue state and high-frequency energy output parameters, effectively avoiding feature distortion problems during the mapping process. After mapping, the low-dimensional feature data is regularized and calibrated to eliminate minor errors generated during the mapping process, outputting low-dimensional dual-impedance trajectory representation data. This data has low dimensionality, high computational efficiency, and can accurately reflect the correlation between dual-impedance and tissue state, providing reliable data support for subsequent trajectory distance calculation and real-time adjustment of energy control parameters.

[0069] S33: The low-dimensional dual-path impedance trajectory characterization data has been freed from redundant interference and compressed in dimension, accurately reflecting the dynamic changes of the dual-path impedance signal with tissue state, as well as the balance of the dual-path high-frequency energy output. Combined with manifold space metric rules adapted to tissue impedance characteristics, the quantized distance index of the dual-path impedance phase trajectory in the low-dimensional manifold space is calculated time-by-time. The smaller the distance index, the more balanced the dual-path high-frequency energy output and the more uniform the state of the tissue under the influence of high-frequency energy. The larger the distance index, the more likely there is a deviation in the dual-path energy output, requiring timely adjustment of the energy output amplitude and duty cycle to avoid localized overheating or insufficient effect on the tissue.

[0070] During the calculation, the impedance trajectory data in the global time-series dimension is continuously solved to capture the distance change trend of the dual-path trajectories at different times. The data is then regularized and sorted to eliminate errors caused by time-series disorder and data jumps, ensuring the continuity, stability, and accuracy of the distance measurement. Abnormal distance values ​​appearing during the calculation are corrected based on the tissue impedance variation pattern to avoid measurement deviations caused by local interference. A low-dimensional manifold distance measurement sequence is output, which reflects the dynamic differences in the phase of the dual-path impedance in real time and continuously, clearly showing the change pattern of tissue state with the action of bipolar high-frequency energy. This provides a core quantitative basis for the precise adjustment of the amplitude and duty cycle of the bipolar high-frequency dual-output energy in subsequent steps, helping to achieve closed-loop energy control of tissue impedance feedback and ensuring precise matching between energy output and tissue state.

[0071] S4: Using the low-dimensional manifold distance metric sequence as the observation value, a sparse Bayesian dynamic model is constructed by combining the sparse prior distribution. The model hyperparameters are optimized by a small number of calibration samples. The sequence is dynamically modeled and peak unmixed. The target feature components corresponding to the coagulation state of the dual-path tissue are separated and screened, and the purified target feature component sequence is output.

[0072] S41: Based on the dynamic monitoring of tissue impedance under bipolar high-frequency dual-output energy control scenarios, a low-dimensional manifold distance metric sequence obtained from previous dimensionality reduction processing is extracted. This sequence is determined as the data source for global observation, fully carrying multi-dimensional dynamic information such as real-time impedance fluctuations of biological tissues under bipolar electrode action, high-frequency energy output coupling interference, and impedance shifts caused by tissue microenvironment deformation. Combining the naturally sparse distribution characteristics of biological tissue impedance signals, sparse prior distribution constraints adapted to physiological electrophysiological signals are introduced. Addressing the technical characteristics of impedance signal hybridization, discrete distribution of effective features, and irregular instantaneous fluctuations during dual-channel high-frequency energy output, sparse weight thresholds and probability distribution boundaries are defined to avoid computational biases caused by redundant stacking of global data. Based on the probabilistic constraint logic of sparse priors, a probabilistic mapping fitting is performed between the low-dimensional manifold distance metric data and the impedance change pattern, establishing a correlation between data distribution and dynamic changes in tissue impedance. By combining a bipolar high-frequency dual-output energy control operation mechanism, and taking into account the synergy and differences between the two energy outputs as well as the lag in real-time feedback of tissue impedance, a basic sparse Bayesian computational framework adapted to the dynamic changes in tissue impedance is constructed. The framework clarifies the data input criteria, probability iteration rules, impedance variable computation logic, and the linkage mechanism of the dual-channel energy parameters. Based on the framework, variable modeling and relationship deduction are performed to quantify the continuous variation characteristics of tissue impedance under the action of a high-frequency electric field, outputting a standardized sparse Bayesian dynamic model.

[0073] S42: During bipolar high-frequency dual-output energy control, external interference factors such as ambient temperature and humidity, high-frequency electromagnetic interference, bipolar electrode contact impedance shift, and intraoperative tissue fluid infiltration can cause model parameter inaccuracies. A small number of precise tissue calibration samples covering different coagulation stages, energy output levels, and tissue types are selected as the core calibration data source for parameter optimization. These calibration samples are then input into the initial sparse Bayesian dynamic model. The sample data accurately matches the real tissue impedance response characteristics under independent and synergistic effects of dual-channel high-frequency energy output, closely reflecting the actual clinical application conditions of bipolar energy regulation.

[0074] Multiple rounds of closed-loop iterative correction calculations were performed around the model's internal hyperparameters, focusing on dynamic optimization of core hyperparameters such as impedance attenuation coefficient, sparse constraint weights, probability mapping correction coefficients, and dual-path energy correlation coupling parameters. The deviation between the model's output data and the actual tissue impedance feedback was compared using the standard impedance true values ​​of calibration samples. A gradient iterative convergence algorithm was used to gradually compress the parameter error range, specifically mitigating parameter shifts caused by high-frequency electromagnetic radiation, external environmental disturbances, and differences in electrode operating conditions, correcting impedance calculation deviations caused by unbalanced dual-path energy output. During the iteration process, the core operational logic of the sparse Bayesian algorithm was retained to maintain the model's ability to analyze low-dimensional manifold data and enhance its anti-interference performance under complex bipolar high-frequency conditions, avoiding feature loss due to over-correction. Iteration continued until the hyperparameters converged to a stable range and the impedance fitting error reached a preset standard, completing full-dimensional parameter correction and model optimization. This resulted in a corrected operational model with stronger anti-interference capabilities, adapted to the requirements of bipolar high-frequency dual-output energy control, improving the accuracy of subsequent impedance signal analysis and energy regulation.

[0075] S43: The modified computational model inherits the low-dimensional manifold distance metric sequence generated throughout the bipolar high-frequency dual-output energy control process, and performs continuous, real-time dynamic fitting calculations to accurately track the instantaneous changes and slow drift patterns of biological tissue impedance during the alternating output and power adjustment of dual-channel high-frequency energy. Combining the sparse constraint rules built into the modified model, and leveraging the high-order analytical capabilities of the sparse Bayesian algorithm, it performs signal peak demixing and deconstruction calculations to address the core issue of impedance signal superposition and mixing. When bipolar electrodes exert dual-channel energy in parallel, multiple effects such as tissue coagulation deformation, cell electrolyte migration, thermal damage accumulation, and high-frequency electric field coupling generate superimposed impedance signals. Different dimensional feature components are intermingled and coupled, making direct differentiation difficult; therefore, demixing and deconstruction are necessary to achieve signal layer separation.

[0076] Based on sparse prior constraints, the distribution range of effective impedance characteristics is defined, and the physiological response mechanism and energy action mechanism corresponding to different peaks in the mixed signal are identified. Multiple independent characteristic components are separated, including tissue thermal damage impedance components, electrode contact impedance components, high-frequency electromagnetic interference components, and tissue intrinsic physiological impedance components. During the decomposition operation, the energy timing characteristics of the bipolar high-frequency dual-output system are strictly matched to distinguish the impedance signal differences under single-path energy action and dual-path synergistic action, preserving the timing correlation and numerical variation characteristics of each component, and preventing the loss of effective information caused by excessive demixing.

[0077] By multi-dimensional hierarchical decomposition, independent component modeling, and data orthogonalization processing, the mixed superposition of multiple signals is completely eliminated, the impedance change information of the entire tissue solidification cycle is fully covered, and all the decomposed independent feature data are integrated and systematically summarized to form a full-coverage, multi-dimensional, and distinguishable set of full impedance feature components.

[0078] S44: Based on the full set of impedance characteristic components, and deeply integrated with the core control objective of bipolar high-frequency dual-output energy control, this system anchors the core laws of differentiated response and combined control of dual-path tissue coagulation states, establishing exclusive feature matching standards and screening rules. Focusing on the physiological evolution mechanism of tissue coagulation under bipolar energy, it clarifies the impedance response characteristics corresponding to different stages of coagulation (initial, middle, and late stages), and analyzes the core laws of tissue impedance fluctuation trends, amplitude ranges, and rates of change when dual-path energy is independently output and linked for adjustment. This serves as the core reference for feature matching. The full set of impedance characteristic components is compared and matched one by one to accurately identify the signal source and functional correlation of various components, distinguishing effective features directly related to the tissue coagulation process from unrelated interference signals and redundant noise data. Irrelevant interference components caused by intraoperative instrument vibration, environmental electromagnetic noise, and instantaneous fluid flow are selectively screened and eliminated. Simultaneously, redundant signal acquisition, residual computational fitting, and invalid noise components derived from high-frequency harmonics are removed to avoid the negative impact of mixed signals on the precise control of bipolar energy. It retains the core effective components that can intuitively reflect the degree of tissue coagulation, dynamic evolution of impedance, and the effect of dual-path energy coupling, maintaining the continuity, temporal sequence, and correlation of feature data, and adapting to the dynamic adjustment requirements of bipolar high-frequency dual-output energy. It optimizes the feature arrangement logic and data output format, completing component sorting and integration according to the energy regulation timing and impedance change gradient, eliminating component data fragmentation, enhancing the adaptability of feature sequences to dual-path energy closed-loop control, and refining and simplifying to generate highly recognizable, highly correlated, and low-noise target feature component sequences, providing standardized feature data support for dynamic adjustment of bipolar high-frequency output power and precise control of tissue coagulation.

[0079] S5: Extract the phase information of the dual-channel target feature components, establish a phase mapping model for the dual-channel signals, correct the phase drift error between the two channels, and output a phase feature sequence that matches the dual-channel tissue state.

[0080] S51: In the bipolar high-frequency dual-output energy control scenario, the target feature component sequence is the core effective data obtained after preprocessing the tissue impedance feedback signal, directly carrying the impedance dynamic response characteristics during tissue coagulation under the action of dual-channel high-frequency energy. The core of this step is to extract the phase information of the dual-channel target feature components in this sequence, because the phase parameter can accurately reflect the synergy between the dual-channel high-frequency energy output and the tissue impedance feedback, and is a key indicator for judging the parallelism of dual-channel energy and avoiding tissue damage. Combining the energy characteristics of bipolar high-frequency dual output, the tissue impedance feedback signal has the characteristics of high-frequency oscillation and instantaneous fluctuation. Traditional phase extraction methods are easily interfered with, so Hilbert transform is used for analysis—Hilbert transform can accurately capture the instantaneous phase of the signal and adapt to the dynamic changes of high-frequency signals. Fourier transform can clearly analyze the phase characteristics of the signal in the frequency domain, while taking into account the stability of the phase analysis. Through the above transformations, phase analysis is performed on each moment of the dual-channel signal in the target feature component sequence, and phase anomalies are eliminated to obtain dual-channel phase timing data. The data corresponds in real time to the timing of dual-channel high-frequency energy output and tissue impedance feedback.

[0081] S52: Dual-channel phase timing data directly reflects the phase change pattern of the dual-channel tissue impedance feedback signal under the action of bipolar high-frequency dual-output energy. Its parallelism directly determines the synergy of the dual-channel energy output, thus affecting the uniformity and safety of tissue coagulation. Using dual-channel phase timing data as the sole input, and closely adhering to the core requirements of bipolar high-frequency dual-output energy control, a dedicated phase parallel mapping model is constructed. During model construction, the dynamic characteristics of tissue impedance feedback are combined with key parameters such as the power linkage rules of dual-channel high-frequency energy output and the hysteresis characteristics of tissue impedance phase response. Through iterative model calculations, core indicators such as the correlation strength between the dual-channel phases, the parallel error range, and the phase change synergy coefficient are quantified, outputting dual-channel phase correlation parameters. These parameters accurately characterize the parallelism of the dual-channel phases, clarify the potential patterns and influencing factors of dual-channel phase offset, ensure the dynamic adaptability of the model to bipolar high-frequency dual-output energy control, and avoid parallel control deviations caused by the model's disconnect from actual working conditions.

[0082] S53: In practical applications of bipolar high-frequency dual-output energy control, factors such as intraoperative environmental electromagnetic interference, fluctuations in the contact impedance between the bipolar electrodes and tissue, changes in the microenvironment during tissue coagulation, and equipment precision errors can easily lead to phase drift errors in both channels. This results in an imbalance in the high-frequency energy output of the two channels, affecting the accuracy of the tissue impedance feedback signal, leading to poor tissue coagulation and even surgical complications. Based on the dual-channel phase correlation parameters, through in-depth analysis of these parameters, the location, amplitude, and temporal distribution of phase drift errors between the two channels are precisely identified, clarifying the sources of error (such as instantaneous drift caused by electromagnetic interference and continuous drift caused by contact impedance fluctuations) and their degree of impact. Combining the real-time and precision requirements of bipolar high-frequency energy control, a targeted phase correction algorithm is employed. Based on the correlation parameters' correlation patterns and deviation thresholds, drift errors are precisely compensated and corrected. This targeted offsets phase deviations caused by environmental interference, equipment errors, and changes in the tissue microenvironment, ensuring that the phase changes of the two channels always conform to the true pattern of tissue impedance feedback and maintain synergy with the dual-channel high-frequency energy output. During the calibration process, the inherent correlation between the dual-channel phase and tissue impedance state and high-frequency energy output parameters is strictly preserved to avoid over-calibration leading to phase distortion. The calibrated dual-channel phase data is output to ensure the synergy and accuracy of the dual-channel high-frequency energy output, thus facilitating precise control of the tissue coagulation process.

[0083] S54: The corrected dual-channel phase data has completely eliminated phase drift errors caused by various interference factors, accurately reflecting the true phase state of tissue impedance feedback under dual-channel high-frequency energy. However, the data format and timing arrangement are not yet fully adapted to the closed-loop control requirements of bipolar high-frequency dual-output energy control, and cannot be directly used for subsequent energy output parameter adjustment and real-time monitoring of tissue coagulation state. Combining the timing characteristics of tissue impedance feedback and the real-time requirements of energy control, the corrected dual-channel phase data is systematically integrated. During the integration process, firstly, the data format is standardized, and the data sampling frequency, phase unit, and timing reference are unified to ensure the timing consistency and comparability of the dual-channel phase data. Secondly, according to the collaborative logic of dual-channel high-frequency energy output, the phase data is time-sequentially arranged to strengthen the parallel correlation of the dual-channel phases, eliminating redundant data and outliers generated during the integration process to ensure data simplicity and effectiveness. Finally, the integrated data is verified based on the dynamic changes in tissue impedance feedback to ensure that the data accurately reflects the relationship between dual-channel energy output and tissue coagulation state. Through the above integration process, a phase characteristic sequence is output. This sequence can be directly connected to a bipolar high-frequency dual-output energy control closed-loop system, providing core data support for real-time dynamic adjustment of energy output power and accurate monitoring of tissue coagulation state, effectively ensuring the uniformity, safety and accuracy of the tissue coagulation process.

[0084] S6: Apply low-amplitude chaotic phase modulation to the phase feature sequence, and remove residual interference through chaos and discrimination to obtain the pure feature sequence corresponding to the true state of the dual-channel tissue. Based on the sequence, dynamically allocate the amplitude and duty cycle of the dual-channel high-frequency output energy to perform precise feedback control of the dual-polar high-frequency dual-output energy.

[0085] S61: Obtain the original phase feature sequence data, filter the low amplitude data segments in the sequence according to the set amplitude threshold, use the chaotic phase mapping rule to perform global low amplitude chaotic phase modulation, correct the phase distribution characteristics of the sequence, and output the modulated phase sequence.

[0086] Continuous time-series data is acquired based on tissue impedance feedback. Time-domain signal decomposition and phase correlation calculations are performed to generate original phase feature sequence data. A dynamic amplitude threshold is set based on the real-time impedance fluctuation characteristics of biological tissues, conforming to the signal amplitude judgment criteria for impedance differences in different tissue regions. The original phase feature sequence is then screened segment by segment across the entire domain based on this fixed threshold, accurately identifying low-amplitude fluctuation data segments within the sequence. These data segments often correspond to signal distortion regions caused by weak changes in tissue impedance. Based on chaotic phase mapping rules adapted to biological impedance signal characteristics, the fixed phase adjustment mode is abandoned. Low-amplitude chaotic phase modulation is applied to the low-amplitude data segments and the entire sequence domain obtained from the screening. Due to the irregular and weak fluctuation characteristics of chaotic variables, the phase shift problem caused by tissue impedance drift is weakened. The overall sequence phase distribution pattern is corrected to offset phase disorder caused by biological tissue contact impedance and environmental electromagnetic disturbances, optimizing the phase continuity and parallelism of the time-series signal, and outputting a modulated phase sequence with regular parameters and stable phase characteristics.

[0087] S62: Focusing on the stable transmission requirements of biological tissue impedance feedback signals, this paper establishes a chaotic model adapted to high-frequency medical signals and determines the constraints, clarifying three core judgment indicators: phase coupling error, timing matching degree, and signal correlation coefficient. Using the modulated phase sequence as the comparison subject, a standard chaotic reference signal conforming to the characteristics of human bioelectrical signals is imported. Quantitative calculations of the coupling correlation degree between the two sets of signals are performed at each timing node to accurately identify abnormal features such as phase shift, frequency distortion, and amplitude jumps. Considering the characteristics of interference sources in the tissue impedance detection scenario, residual interference components such as electromagnetic clutter, impedance detection loop noise, and environmental power frequency interference are disassembled point by point in the modulated sequence. Through the linked calculations of data removal, deviation correction, and signal compensation, invalid interference data is stripped away, retaining the effective signal components that highly match the changes in tissue impedance. This eliminates the impact of residual interference on signal accuracy, compresses the signal distortion error range, and outputs a pure feature sequence with smooth waveform, pure characteristics, and unified impedance correlation.

[0088] S63: Based on the time-series analysis algorithm, core quantitative values ​​such as phase shift and amplitude fluctuation within the pure feature sequence are decomposed. All data correspond to the dynamic changes in real-time impedance feedback of biological tissues, which can intuitively reflect tissue impedance differences and load fluctuation states. Referring to the medical control logic of bipolar high-frequency energy output, a dedicated preset energy matching constraint logic is built. Combined with tissue tolerance threshold, impedance load range, and dual-output balance requirements, parameter adjustment boundaries are defined to avoid abnormal operating conditions such as energy overload and output imbalance.

[0089] Based on the phase and amplitude quantization data obtained from the decomposition, the dual-channel high-frequency operating parameters are differentiated according to load adaptation requirements. The frequency reference, power base, and response rate of the two output channels are independently adjusted to match the differentiated load requirements of different regions with varying impedances. Through multi-dimensional parameter linkage calibration, the parameter calculation logic and impedance feedback change rhythm are unified to integrate and arrange the two sets of channel-specific high-frequency data, generating dual-channel high-frequency output configuration data with closed-loop logic, independent parameters, and adaptability to dynamic impedance changes. This provides the core calculation basis for bipolar energy distribution calculations.

[0090] S64: Based on the dual-channel high-frequency output configuration data and combined with the bipolar level switching logic specific to the tissue impedance feedback control scenario, the timing, amplitude range, and switching interval rules for forward and reverse bipolar level switching are clearly defined to adapt to the safety control standards for high-frequency effects on biological tissues. Based on the parameter differences between the two configuration data channels and combined with real-time tissue impedance load data, a refined energy weighting calculation is performed for the dual output channels. The energy ratio of the two outputs is dynamically allocated according to the impedance matching principle, achieving a dynamic control mode of low energy output in high impedance regions and adaptive energy replenishment in low impedance regions. Quantitative encoding of control commands is performed, converting frequency parameters, power parameters, level switching commands, and energy ratio data into standardized digital control codes, unifying the command transmission format and calculation rules. Encoding deviations are corrected using a closed-loop correction mechanism, and secondary verification of commands is performed using real-time impedance feedback data to eliminate parameter conversion errors and output delay issues. This generates precise timing, controllable energy, and dual-channel collaborative adaptation bipolar high-frequency dual-output energy precision control commands, achieving dynamic bipolar energy regulation based on tissue impedance feedback.

[0091] like Figure 3 As shown, the present invention also provides a bipolar high-frequency dual-output energy control system based on tissue impedance feedback, comprising: The singular decomposition module constructs the eigenvalue spectrum and singular vector space, decomposes them to obtain the slowly varying components and high-frequency noise components, and outputs a preliminary purified impedance core component sequence.

[0092] The amplitude-phase array module is used to extract the phase features of the dual-path impedance core component sequence, construct a joint amplitude-phase feature set, and form a high-dimensional joint feature matrix.

[0093] The manifold dimensionality reduction module is used to train the manifold embedding dimensionality reduction model. It inputs the high-dimensional joint feature matrix into the manifold embedding dimensionality reduction model for mapping, calculates the distance between the two paths in the manifold space, and outputs a low-dimensional manifold distance metric sequence.

[0094] The sparse unmixing module is used to construct a sparse Bayesian dynamic model by using a low-dimensional manifold distance metric sequence as the observation value and combining it with a sparse prior distribution, to filter the target feature components and output the target feature component sequence.

[0095] The phase-parallel module is used to extract the phase information of the dual target feature components in the target feature component sequence, establish a dual-signal phase-parallel mapping model, and output the phase feature sequence.

[0096] The chaotic energy control module is used to perform low-amplitude chaotic phase modulation on the phase feature sequence to obtain a pure feature sequence, allocate dual-channel high-frequency output parameters, and output bipolar high-frequency dual-output energy precise control commands.

[0097] In summary, this embodiment provides a bipolar high-frequency dual-output energy control method and system based on tissue impedance feedback. It utilizes multi-state measured impedance sample training modeling, dual-channel phase mapping, and bipolar level linkage control logic. By incorporating impedance samples from different tissue structures, coagulation stages, and physiological states, it achieves iterative model optimization. Combined with real-time phase correction, amplitude-phase joint feature mining, and energy closed-loop matching constraints, it establishes a one-to-one mapping relationship between tissue impedance characteristics and dual-channel high-frequency output parameters. Based on real-time analyzed impedance phase and amplitude quantization data, it dynamically adjusts the frequency reference, power base, level switching timing, and energy ratio of the two output channels. Differential command encoding and quantization output are completed according to dual-channel collaborative control rules. This system can adapt to impedance fluctuations in various soft tissues, diseased tissues, and different intraoperative physiological microenvironments, strictly adhering to tissue tolerance thresholds to achieve safety margin constraints and realizing flexible energy output on demand. It effectively avoids the compatibility shortcomings caused by uniform output parameters, reduces the risk of tissue damage caused by the incompatibility of dual-path energy, narrows the range of intraoperative thermal damage, optimizes the safety boundary of bipolar high-frequency energy action, and at the same time broadens the types of tissues and clinical application scenarios applicable to the equipment, taking into account the control safety, clinical applicability and operational versatility.

[0098] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0099] 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 it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. These modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A bipolar high-frequency dual-output energy control method based on tissue impedance feedback, characterized in that, include: S1: Collect dual-channel raw impedance time series and perform singular spectrum decomposition on each channel to construct eigenvalue spectrum and singular vector space. Separate the slowly varying components and high-frequency noise components, screen and retain the core slowly varying components, and output the preliminary purified impedance core component sequence. S2: Extract the phase features of the dual-path impedance core component sequence, construct a set of amplitude and phase joint features based on the dual-path high-frequency excitation frequency parameters, and form a high-dimensional joint feature matrix; S3: Train the manifold embedding dimensionality reduction model using multiple sets of measured impedance samples, input the high-dimensional joint feature matrix into the manifold embedding dimensionality reduction model for mapping, calculate the distance between the two-path trajectories in the manifold space, and output a low-dimensional manifold distance metric sequence. S4: Using the low-dimensional manifold distance metric sequence as the observation, construct a sparse Bayesian dynamic model by combining the sparse prior distribution, optimize the model hyperparameters and model and unmix peaks, screen the target feature components, and output the target feature component sequence. S5: Extract the phase information of the dual target feature components in the target feature component sequence, establish a dual signal phase and mapping model, correct the phase drift error between the two channels, and output the phase feature sequence; S6: Perform low-amplitude chaotic phase modulation on the phase feature sequence, and obtain a pure feature sequence by eliminating residual interference through chaos and discrimination. Allocate dual-channel high-frequency output parameters and output a dual-polar high-frequency dual-output energy precision control command.

2. The bipolar high-frequency dual-output energy control method based on tissue impedance feedback according to claim 1, characterized in that: In step S1, the specific steps for outputting the preliminary purified impedance core component sequence are as follows: S11: Acquire dual-channel raw impedance time series, and reconstruct the dual-channel raw impedance time series data. Construct single-channel trajectory matrices according to fixed delay window lengths to form two-channel trajectory matrices. S12: Perform singular value decomposition on the two trajectory matrices respectively to obtain the singular value sequence and left and right singular vectors of the corresponding matrices, thus forming the eigenvalue spectrum and singular vector space of the two signals respectively. S13: Based on the numerical distribution law of the eigenvalue spectrum and the temporal component shape corresponding to the singular vector, the two signals are decomposed into a slow-changing component with a smooth amplitude change and a high-frequency disturbance component with rapid fluctuations. S14: Based on the slow-change component and the high-frequency disturbance component, select the singular vector associated with the large eigenvalue for time-series reconstruction, and remove the disturbance component to obtain the impedance core component sequence.

3. The bipolar high-frequency dual-output energy control method based on tissue impedance feedback according to claim 1, characterized in that: In step S2, the specific steps for forming the high-dimensional joint feature matrix are as follows: S21: Decompose the time-domain signal based on the core component sequence of the dual impedance, calculate the phase shift difference and dynamic fluctuation value of the two sequences point by point, and output the basic feature set of the dual impedance phase. S22: Calculate the amplitude correlation mapping based on the dual-path impedance phase basic feature set and the preset dual-path high-frequency excitation frequency parameters, use frequency constraints to gather and integrate effective data, perform multi-dimensional feature combination and arrangement, and form an amplitude-phase coupling dataset. S23: Expand the row and column dimensions and stitch features according to the amplitude-phase coupling dataset, organize the feature arrangement according to the signal time sequence correlation logic and frequency matching rules, fuse the coupling information to build a multi-dimensional data array, and obtain a standardized high-dimensional joint feature matrix.

4. The bipolar high-frequency dual-output energy control method based on tissue impedance feedback according to claim 1, characterized in that: In step S3, the specific steps for outputting the low-dimensional manifold distance metric sequence are as follows: S31: Perform iterative training of the model based on the measured impedance samples corresponding to multiple tissue states, fit the manifold mapping constraint relationship according to the sample feature distribution law, and output the solidified manifold embedding operation model. S32: Input the high-dimensional joint feature matrix into the manifold embedding operation model, perform spatial dimension mapping transformation of the dual-path impedance phase trajectory, and output low-dimensional dual-path impedance trajectory characterization data. S33: Calculate the quantized distance index between the two trajectories based on the low-dimensional trajectory representation data and manifold space metric rules, perform continuous solution and regular sorting of the global time series data, and output the low-dimensional manifold distance metric sequence.

5. The bipolar high-frequency dual-output energy control method based on tissue impedance feedback according to claim 4, characterized in that: In step S31, the specific steps for outputting the solidified manifold embedding computation model are as follows: Collect measured impedance samples corresponding to different tissue states and perform numerical normalization and outlier removal. Unify the sample data dimension and value range, perform statistical analysis and clustering characterization of sample feature distribution, and output impedance sample feature dataset. Based on the impedance sample feature dataset, construct iterative training constraints, continuously update the model weight parameters iteratively, fit the exclusive manifold mapping association rule according to the data distribution law, gradually converge and correct the mapping boundary relationship, and output the initial model architecture. The initial model architecture is compared with the convergence threshold. Based on the error back-correction and generalization ability calibration, the mapping operation logic adapted to bioimpedance characteristics is solidified, the internal operation parameters and manifold constraint ratio of the model are locked, and the solidified manifold embedding operation model is output.

6. The bipolar high-frequency dual-output energy control method based on tissue impedance feedback according to claim 1, characterized in that: In step S4, the specific steps for outputting the target feature component sequence are as follows: S41: Using the low-dimensional manifold distance metric sequence as the global observation data source, and combining it with the sparse prior distribution constraint to perform probability mapping fitting, a basic sparse Bayesian operation framework adapted to the dynamic changes of tissue impedance is built, and a sparse Bayesian dynamic model is output. S42: Input a small number of tissue calibration samples into the sparse Bayesian dynamic model, iteratively correct the hyperparameter values ​​inside the model, weaken the parameter deviation caused by environmental interference, and output the corrected calculation model. S43: Perform dynamic fitting operation on the low-dimensional manifold distance metric sequence according to the modified operation model, demix and deconstruct the signal peaks according to the sparse constraints, separate the mixed and superimposed multi-class feature components, and output the full set of impedance feature components. S44: Based on the response characteristic law of the dual-path solidification state matched with the full impedance characteristic component set, irrelevant interference components and redundant noise components are screened out and the target characteristic component sequence is output.

7. The bipolar high-frequency dual-output energy control method based on tissue impedance feedback according to claim 6, characterized in that: In step S41, the specific steps for outputting the sparse Bayesian dynamic model are as follows: Low-dimensional manifold distance metric sequences under bipolar high-frequency dual-output are collected, and sparse prior distribution intervals are defined by combining tissue impedance characteristics to perform probability mapping fitting operations and generate constraint configuration parameters. Based on the aforementioned constraint configuration parameters, a computational framework adapted to the dynamic changes in tissue impedance is constructed, and Bayesian probability computation logic is embedded to form a basic sparse Bayesian computation model. The dual-path organization impedance time series data is imported into the basic sparse Bayesian operation model for dynamic fitting, the data distribution pattern is sorted out, and a low-dimensional feature dataset is output. Based on the low-dimensional feature dataset, the bipolar high-frequency energy control requirements are matched, the model iteration threshold and weight allocation rules are optimized, and a sparse Bayesian dynamic model is output.

8. The bipolar high-frequency dual-output energy control method based on tissue impedance feedback according to claim 1, characterized in that: In step S5, the specific steps for outputting the phase feature sequence are as follows: S51: Extract the phase information of the dual-path target feature components in the target feature component sequence, analyze the phase value at each time step through Hilbert transform, and obtain the dual-path phase time series data; S52: Construct a phase mapping model based on the dual-channel phase timing data, quantify the correlation and parallel rules of the dual-channel phases through model calculation, and output the dual-channel phase correlation parameters; S53: Locate the phase drift error between the two channels based on the dual-channel phase correlation parameters, perform precise correction through a targeted algorithm, offset the phase deviation caused by environmental and equipment interference, and output corrected dual-channel phase data; S54: Integrate the corrected dual-channel phase data and output a phase feature sequence.

9. The bipolar high-frequency dual-output energy control method based on tissue impedance feedback according to claim 1, characterized in that: In step S6, the specific steps for outputting the bipolar high-frequency dual-output energy precision control command are as follows: S61: Obtain the original phase feature sequence data, filter the low amplitude data segment in the sequence according to the set amplitude threshold, correct the phase distribution characteristics of the sequence, and output the modulated phase sequence. S62: Based on the chaos and the determination of the constraints, compare the coupling correlation between the modulated phase sequence and the standard chaotic reference signal, identify the residual clutter in the sequence, and output the pure feature sequence; S63: Based on the phase and amplitude correlation quantization values ​​of the pure feature sequence, differentiated values ​​are assigned to the dual-channel high-frequency operating parameters to generate high-frequency output configuration data; S64: Based on the high-frequency output configuration data, perform precise allocation of energy weights for dual output channels and quantization encoding of instructions to generate precise control instructions for dual-polar high-frequency dual-output energy.

10. A bipolar high-frequency dual-output energy control system based on tissue impedance feedback, which employs the bipolar high-frequency dual-output energy control method based on tissue impedance feedback as described in any one of claims 1 to 9, characterized in that, The control system includes: The singular decomposition module constructs the eigenvalue spectrum and singular vector space, decomposes them to obtain the slowly varying components and high-frequency noise components, and outputs a preliminary purified impedance core component sequence. The amplitude and phase array module is used to extract the phase features of the initially purified impedance core component sequence, construct the amplitude and phase joint feature set, and form a high-dimensional joint feature matrix. The manifold dimensionality reduction module is used to train the manifold embedding dimensionality reduction model. The high-dimensional joint feature matrix is ​​input into the manifold embedding dimensionality reduction model for mapping, the distance between the two-path trajectories in the manifold space is calculated, and a low-dimensional manifold distance metric sequence is output. The sparse unmixing module is used to construct a sparse Bayesian dynamic model by combining the low-dimensional manifold distance metric sequence with the sparse prior distribution, using the low-dimensional manifold distance metric sequence as the observation value, filtering the target feature components, and outputting the target feature component sequence. The phase-parallel module is used to extract the phase information of the dual-path target feature components in the target feature component sequence, establish a dual-path signal phase-parallel mapping model, and output the phase feature sequence. The chaotic energy control module is used to perform low-amplitude chaotic phase modulation on the phase feature sequence to obtain a pure feature sequence, allocate dual-channel high-frequency output parameters, and output a dual-polar high-frequency dual-output energy precision control command.

Images