Target positioning method based on focused ultrasonic waves

Abstract

The invention relates to a focused ultrasonic wave-based target positioning method, which comprises the following steps of: acquiring original data of an ultrasonic signal through a sensor array, recording propagation time and intensity distribution, separating a main signal from an interference signal by adopting a time domain analysis method, and determining a preliminary energy distribution characteristic; according to the energy distribution characteristics, correcting a signal propagation path by using a sound velocity distribution model, and determining a corrected energy concentration degree value; signal boundaries are enhanced through an adaptive filtering algorithm, half-power point features are extracted through wavelet transform, and a beam width quantization result is determined; according to a beam width quantification result, calculating peak distribution of a main lobe and a side lobe by using an array response function, and determining an initial evaluation value of a side lobe suppression ratio; adjusting weight distribution through a multi-dimensional parameter joint optimization algorithm, and determining an environment adaptive weight distribution scheme; and combining the energy distribution predicted value and the beam width quantification result through a weighted fusion algorithm, and determining a focusing precision evaluation value by adopting Kalman filtering smoothing processing.

Description

Technical Field

[0001] This application relates to the field of ultrasonic technology, and particularly to a target positioning method based on focused ultrasound. Background Art

[0002] In the business scenario of focused ultrasound target positioning, the focusing quantization process faces a unique and esoteric technical contradiction: how to accurately balance the weight distribution of the three core parameters, namely energy concentration, beam width, and sidelobe suppression ratio, in a complex and variable target environment, while avoiding quantization deviation caused by environmental variable interference.

[0003] Specifically, in actual operation, the target point may be in a non-uniform medium, including near tissue boundaries or gas-liquid interfaces, which makes the calculation of energy concentration affected by scattering and refraction, and it is difficult to accurately reflect the true energy contrast between the focal point and the surrounding area. At this time, if solely relying on the ratio of the focal point energy to the total radiation energy, the local energy density may be underestimated or overestimated.

[0004] At the same time, the quantization of the beam width depends on the measurement of the half-power point width. However, in a high-noise environment, the signal boundary is blurred, and the determination of the half-power point becomes unstable, resulting in distorted evaluation of the focusing accuracy. Moreover, the calculation of the sidelobe suppression ratio is further complicated because the peaks of the main lobe and sidelobes are limited by the geometric arrangement of the transducer array and the fluctuations of the driving frequency. When the target depth changes, the peak of the main lobe may be misjudged as a sidelobe feature, directly affecting the characterization of signal purity.

[0005] When existing mathematical models standardize these three parameters, they usually presuppose the uniformity of environmental variables. However, in actual scenarios, the medium density, sound speed distribution, and target reflection characteristics all change dynamically, and the fixed mode of weight distribution is difficult to adapt to, resulting in systematic deviations in the comprehensive focusing quality score during cross-scenario comparisons.

[0006] The core of this contradiction lies in how to introduce an adaptive environmental correction mechanism into the quantization system while maintaining the logical consistency among multi-dimensional indicators, rather than simply relying on preset weights or static models. Summary of the Invention

[0007] In order to solve the problems existing in the above-mentioned prior art, the purpose of this application is to provide a target positioning method based on focused ultrasound.

[0008] A target positioning method based on focused ultrasound according to this application includes the following steps:

[0009] S101. Obtain the original data of the ultrasonic signal. Record the propagation time and intensity distribution of the sound wave in the target area through a sensor array. In view of the influence of scattering and refraction in the inhomogeneous medium, use the time-domain analysis method to separate the main signal and the interference signal, and obtain the preliminary energy distribution characteristics;

[0010] S102. According to the energy distribution characteristics, in view of the energy concentration deviation caused by scattering and refraction, use the sound velocity distribution model to correct the signal propagation path, and adjust the refraction offset of the sound wave in the inhomogeneous medium through iterative calculation to determine the corrected energy concentration value;

[0011] S103. Obtain the corrected energy concentration value. In the noise interference environment, enhance the signal boundary through the adaptive filtering algorithm, extract the half-power point characteristics by wavelet transform, judge the boundary stability of the beam width, and obtain a clear beam width quantization result;

[0012] S104. Through the beam width quantization result, in view of the geometric arrangement fluctuation of the transducer array, calculate the peak distribution of the main lobe and the side lobe using the array response function. If the main lobe peak deviates from the preset threshold, adjust the drive frequency parameter to determine the initial evaluation value of the side lobe suppression ratio;

[0013] S105. Obtain the initial evaluation value of the side lobe suppression ratio. Combine the corrected energy concentration and the beam width quantization result, dynamically adjust the weight allocation through the multi-dimensional parameter joint optimization algorithm, and iteratively update the weight coefficient using the gradient descent method to obtain an environment-adaptive weight allocation scheme;

[0014] S106. According to the weight allocation scheme, use the real-time acoustic imaging technology to reconstruct the characteristic distribution of the inhomogeneous medium, predict the dynamic trend of scattering and refraction through the convolutional neural network, judge the correction requirement of the energy concentration, and obtain the optimized energy distribution prediction value;

[0015] S107. Obtain the optimized energy distribution prediction value. Combine the quantization data of the beam width and the side lobe suppression ratio, comprehensively score the focusing quality through the weighted fusion algorithm, and smooth the scoring result using the Kalman filter to determine the final focusing accuracy evaluation value.

[0016] Preferably, obtaining the original data of the ultrasonic signal through the sensor array in step S101 includes: collecting the ultrasonic signal through the sensor array to generate a first data set;

[0017] For the first data set, record the propagation time and intensity distribution;

[0018] Process the first data set using the time-domain analysis method to separate the main signal and the interference signal, and generate a second data set;

[0019] Calculate the energy distributions of the primary signal and the interference signal according to the second data set, and generate a third data set;

[0020] If the energy distribution of the primary signal in the third data set exceeds a preset threshold, extract the frequency characteristics of the primary signal through Fourier transform to generate a fourth data set;

[0021] Determine the boundary position of the inhomogeneous medium according to the fourth data set in combination with the propagation time to generate a fifth data set;

[0022] Reconstruct the intensity distribution characteristics by using the backpropagation algorithm through the fifth data set to generate a sixth data set.

[0023] Preferably, in step S102, correcting the signal propagation path by using the sound speed distribution model includes: obtaining the initial signal propagation path through the sound speed distribution model;

[0024] Use iterative calculation to determine the preliminary refraction offset of the sound wave in the inhomogeneous medium;

[0025] Adjust the signal propagation path according to the preliminary refraction offset to generate corrected propagation path data;

[0026] For the corrected propagation path data, calculate the change in energy distribution caused by scattering refraction to determine the aggregation deviation value;

[0027] If the aggregation deviation value exceeds a preset threshold, adjust the sound wave offset through the sound speed distribution model to generate updated refraction offset data;

[0028] According to the updated refraction offset data, correct the energy distribution characteristics to obtain an adjusted energy aggregation value;

[0029] Verify the adjusted energy aggregation value through the random forest algorithm to determine the final correction result.

[0030] Preferably, in step S103, enhancing the signal boundary by using the adaptive filtering algorithm includes: obtaining the original signal data in a noise interference environment and generating a first signal sequence through sampling processing;

[0031] Process the first signal sequence through the adaptive filtering algorithm and adjust the filtering coefficients by using the least mean square algorithm to generate a second signal sequence;

[0032] For the second signal sequence, decompose the signal by using wavelet transform and extract the half-power point characteristics to generate a first feature set;

[0033] Calculate the beam width value according to the first feature set;

[0034] If the beam width value exceeds the preset threshold range, correct the half-power point characteristics to generate a second feature set;

[0035] Using the second feature set, determine the boundary stability index by means of the boundary point fluctuation amplitude calculation method;

[0036] According to the boundary stability index, generate the corrected energy concentration result by means of the weighted average method.

[0037] Preferably, in step S104, calculating the peak distributions of the main lobe and the side lobes by using the array response function includes: obtaining distribution data through the beam width quantization result, analyzing the geometric arrangement fluctuation of the transducer array, and generating a preliminary beam feature description;

[0038] From the preliminary beam feature description, calculate the distribution characteristics of the main lobe peak and the side lobe peak by using the array response function, and determine the fluctuation range of the peak distribution;

[0039] If the main lobe peak deviates from the preset threshold, adjust the drive frequency parameter to generate adjusted peak distribution data;

[0040] According to the adjusted peak distribution data, calculate the ratio of the side lobe peak to the main lobe peak to generate a preliminary value of side lobe suppression;

[0041] Adjust the parameters through the iterative optimization algorithm to determine the stable value of the side lobe suppression ratio;

[0042] According to the stable value combined with the beam width, use the regression analysis algorithm to fit the relationship between the drive frequency and the suppression ratio to determine the final evaluation value.

[0043] Preferably, in step S105, adjusting the weight allocation by means of the multi-dimensional parameter joint optimization algorithm includes: extracting multi-dimensional parameter eigenvalue from the corrected energy concentration and the beam width quantization result, and determining a preliminary weight allocation scheme;

[0044] Through the multi-dimensional parameter eigenvalue, dynamically adjust the weight allocation by using the joint optimization method to generate a first adjusted weight set;

[0045] For the first adjusted weight set, iteratively update the weight coefficient by using the gradient descent method;

[0046] If the number of iterations reaches the preset threshold, generate a second adjusted weight set;

[0047] According to the second adjusted weight set, calculate the environment adaptive weight allocation scheme to obtain the environment adaptive eigenvalue;

[0048] Update the multi-dimensional parameter eigenvalue through the environment adaptive eigenvalue to generate an optimized multi-dimensional parameter set;

[0049] Recalculate the sidelobe suppression ratio according to the optimized multi-dimensional parameter set, and determine the final sidelobe suppression ratio result.

[0050] Preferably, in step S106, the characteristic distribution of the inhomogeneous medium is reconstructed by using real-time acoustic imaging technology, including: acquiring the acoustic wave reflection data of the inhomogeneous medium through real-time acoustic imaging technology, reconstructing the characteristic distribution image, and generating the first characteristic distribution;

[0051] Adjust the distribution parameters in the first characteristic distribution according to the weight assignment scheme;

[0052] If the deviation of the distribution parameter from the preset threshold exceeds the specified range, the distribution parameter is corrected by the weighted average method to generate the second characteristic distribution;

[0053] Predict the dynamic trend of scattering refraction for the second characteristic distribution through a convolutional neural network to generate a dynamic trend sequence;

[0054] Calculate the energy aggregation degree according to the dynamic trend sequence;

[0055] If the energy aggregation degree exceeds the preset upper limit, adjust the energy distribution parameter in the second characteristic distribution to generate the first energy distribution;

[0056] Adjust the first energy distribution through an iterative optimization method to generate an optimized predicted value.

[0057] Preferably, in step S107, the energy distribution prediction value and the beam width quantization result are combined through a weighted fusion algorithm, including: collecting energy distribution data through a sensor, calculating the predicted value by a statistical method, and generating an initial feature set;

[0058] Apply the weighted fusion algorithm to calculate the focusing quality score for the initial feature set;

[0059] Process the focusing quality score through Kalman filtering to generate a smoothed score result;

[0060] Determine the accuracy evaluation value according to the smoothed score result;

[0061] If the accuracy evaluation value is lower than the preset threshold, adjust the parameters of the weighted fusion algorithm and recalculate the focusing quality score;

[0062] Update the accuracy evaluation value according to the adjusted score result to generate the final output;

[0063] Combine the final output with the beam width quantization result, and use the weighted average method to determine the focusing accuracy evaluation value.

[0064] A method for target localization based on focused ultrasound according to the present application has the advantage that by acquiring the original data of ultrasonic signals, in view of the influence of scattering and refraction in inhomogeneous media, a time-domain analysis method is used to separate the main signal from the interference signal, and a sound speed distribution model is used to correct the signal propagation path. In a noise interference environment, the beam width characteristics are extracted through adaptive filtering and wavelet transform, and the sidelobe suppression ratio is calculated in combination with the array response function. A multi-dimensional parameter joint optimization algorithm is used to dynamically adjust the weight distribution, and the real-time acoustic imaging technology and convolutional neural network are used to predict the scattering and refraction trend. Finally, the focusing accuracy evaluation value is determined through a weighted fusion algorithm and Kalman filtering;

[0065] The present invention can effectively improve the accuracy evaluation of ultrasonic focusing in inhomogeneous media and provide a reliable basis for ultrasonic imaging and detection in complex environments. Brief Description of the Drawings

[0066] Figure 1 is the flow of a method for target localization based on focused ultrasound according to the present application Figure 1 ;

[0067] Figure 2 is the flow of a method for target localization based on focused ultrasound according to the present application Figure 2 . Detailed Embodiments

[0068] As Figure 1 - Figure 2 shown, a method for target localization based on focused ultrasound according to the present application includes the following steps:

[0069] S101. Acquire the original data of ultrasonic signals, record the propagation time and intensity distribution of sound waves in the target area through a sensor array, and in view of the influence of scattering and refraction in inhomogeneous media, use a time-domain analysis method to separate the main signal from the interference signal to obtain preliminary energy distribution characteristics;

[0070] S102. According to the energy distribution characteristics, in view of the energy concentration deviation caused by scattering and refraction, use a sound speed distribution model to correct the signal propagation path, and adjust the refraction offset of sound waves in inhomogeneous media through iterative calculation to determine the corrected energy concentration value;

[0071] S103. Acquire the corrected energy concentration value, in a noise interference environment, enhance the signal boundary through an adaptive filtering algorithm, extract the half-power point characteristics by wavelet transform, judge the boundary stability of the beam width, and obtain a clear beam width quantization result;

[0072] S104. Based on the beamwidth quantization result, for the geometric arrangement fluctuations of the transducer array, use the array response function to calculate the peak distributions of the main lobe and sidelobes. If the peak of the main lobe deviates from the preset threshold, adjust the drive frequency parameter to determine the initial evaluation value of the sidelobe suppression ratio;

[0073] S105. Obtain the initial evaluation value of the sidelobe suppression ratio, combine the corrected energy concentration degree and the beamwidth quantization result, dynamically adjust the weight allocation through a multi-dimensional parameter joint optimization algorithm, and use the gradient descent method to iteratively update the weight coefficients to obtain a weight allocation scheme adapted to the environment;

[0074] S106. According to the weight allocation scheme, use real-time acoustic imaging technology to reconstruct the characteristic distribution of the inhomogeneous medium, predict the dynamic trend of scattering and refraction through a convolutional neural network, judge the correction requirement of the energy concentration degree, and obtain an optimized predicted value of the energy distribution;

[0075] S107. Obtain the optimized predicted value of the energy distribution, combine the quantization data of the beamwidth and the sidelobe suppression ratio, comprehensively score the focusing quality through a weighted fusion algorithm, and use Kalman filtering to smooth the scoring result to determine the final evaluation value of the focusing accuracy.

[0076] As Figure 1 - Figure 2 shown, in step S101, the original data of the ultrasonic signal is obtained, the propagation time and intensity distribution of the sound wave in the target area are recorded by the sensor array, and for the influence of scattering and refraction in the inhomogeneous medium, the time-domain analysis method is used to separate the main signal and the interference signal to obtain the preliminary energy distribution characteristics.

[0077] Further, in step S101, the ultrasonic signal in the target area is collected by the sensor array, and the propagation time and intensity distribution data of the sound wave are recorded;

[0078] According to the collected propagation time and intensity distribution data, extract the original signal characteristics of the sound wave in the inhomogeneous medium;

[0079] For the influence of scattering and refraction in the inhomogeneous medium, use the time-domain analysis method to process the original signal and separate the main signal and the interference signal;

[0080] Through the time-domain analysis result, calculate the energy distribution characteristics of the main signal to obtain the preliminary energy distribution data;

[0081] According to the preliminary energy distribution data, combined with the sound speed distribution model, calculate the initial propagation path of the sound wave in the inhomogeneous medium;

[0082] Adopt an iterative calculation method to adjust the refraction offset of the sound wave in the inhomogeneous medium to determine the preliminary refraction offset data;

[0083] According to the preliminary refraction offset data, correct the acoustic wave propagation path to obtain the corrected propagation path information;

[0084] Through the corrected propagation path, recalculate the energy distribution change caused by scattering and refraction to determine the aggregation degree deviation value;

[0085] If the aggregation degree deviation value exceeds the preset threshold, use the sound speed distribution model to readjust the acoustic wave offset to obtain the updated refraction offset data, and correct the energy distribution characteristics based on this.

[0086] Specifically, in step S101, use a sensor array to collect ultrasonic signals in the target area at a sampling frequency of 1 MHz, and record the propagation time and intensity distribution data of the acoustic wave within the time range of 0.1 ms to 1 ms;

[0087] According to the collected propagation time and intensity distribution data, extract the original signal characteristics of the acoustic wave in the inhomogeneous medium through fast Fourier transform, including the main frequency component with a frequency range of 50 kHz to 200 kHz;

[0088] Aiming at the influence of scattering and refraction in the inhomogeneous medium, use the time-domain analysis method to process the original signal, use the short-time Fourier transform to separate the main signal and the interference signal, and extract the main signal component with a time window of 0.05 ms;

[0089] Based on the time-domain analysis result, calculate the energy distribution characteristics of the main signal, use the integral method to calculate the signal energy in the time domain to obtain the preliminary energy distribution data, and the energy value range is 0.1 J to 1 J;

[0090] According to the preliminary energy distribution data, combined with the sound speed distribution model, use the ray tracing algorithm to calculate the initial propagation path of the acoustic wave in the inhomogeneous medium, and the preset medium sound speed is 1500 m / s to 2500 m / s;

[0091] Use the iterative calculation method to adjust the refraction offset of the acoustic wave in the inhomogeneous medium through the least squares method to determine the preliminary refraction offset data, and the offset angle range is 0.1° to 0.5°;

[0092] According to the preliminary refraction offset data, correct the acoustic wave propagation path, and use the interpolation algorithm to obtain the corrected propagation path information, and the path length error is controlled within 0.01 m;

[0093] Through the corrected propagation path, recalculate the energy distribution change caused by scattering and refraction, and use the energy difference method to determine the aggregation degree deviation value, and the deviation value range is 0.01 to 0.1;

[0094] If the aggregation deviation value exceeds the preset threshold of 0.05, the acoustic wave migration is re-adjusted using the sound velocity distribution model, the updated refraction migration data is obtained through the Newton iteration method, and the energy distribution characteristics are corrected based on this. Finally, the energy value error is controlled within 0.01 J.

[0095] As Figure 1 - Figure 2 shown, in step S102, according to the energy distribution characteristics, for the energy aggregation deviation caused by scattering refraction, the signal propagation path is corrected using the sound velocity distribution model, and the refraction migration of the acoustic wave in the inhomogeneous medium is adjusted through iterative calculation to determine the corrected energy aggregation value.

[0096] Further, in step S102, the initial signal propagation path is obtained through the sound velocity distribution model, and the preliminary refraction migration of the acoustic wave in the inhomogeneous medium is determined by iterative calculation;

[0097] The signal propagation path is adjusted according to the preliminary refraction migration to obtain the corrected propagation path data;

[0098] For the corrected propagation path data, the energy distribution change caused by scattering refraction is calculated to determine the aggregation deviation value;

[0099] If the aggregation deviation value exceeds the preset threshold, the acoustic wave migration is re-adjusted through the sound velocity distribution model to obtain the updated refraction migration data;

[0100] The energy distribution characteristics are corrected according to the updated refraction migration data to obtain the adjusted energy aggregation value;

[0101] Based on the adjusted energy aggregation value, it is judged whether the corrected signal propagation path meets the preset conditions to obtain the final correction result;

[0102] The random forest algorithm is used to verify the final correction result to determine the accuracy of the energy aggregation value;

[0103] The acoustic wave reflection data of the inhomogeneous medium is obtained through real-time acoustic imaging technology, and the characteristic distribution image is reconstructed to obtain the first characteristic distribution;

[0104] According to the weight allocation scheme, the distribution parameters in the first characteristic distribution are adjusted. If the distribution parameters deviate from the preset threshold by more than the specified range, the distribution parameters are corrected by the weighted average method to obtain the second characteristic distribution.

[0105] Specifically, in step S102, the initial signal propagation path is obtained through the sound velocity distribution model, and the preliminary refraction migration of the acoustic wave in the inhomogeneous medium is determined by iterative calculation. The specific method is to set the initial sound velocity to 1500 m / s, solve the wave equation based on the finite difference method, the number of iterations is 100 times, and the convergence accuracy is 0.01;

[0106] Adjust the signal propagation path according to the preliminary refraction offset to obtain the corrected propagation path data, including adjusting the offset angle to 2.5 degrees and correcting the path length to 1.2 meters;

[0107] For the corrected propagation path data, calculate the change in energy distribution caused by scattering refraction, determine the aggregation deviation value, and calculate the deviation value to be 15% using the energy attenuation formula;

[0108] If the aggregation deviation value exceeds the preset threshold of 10%, then readjust the acoustic wave offset through the sound velocity distribution model to obtain the updated refraction offset data, including adjusting the sound velocity to 1520 meters per second and correcting the offset angle to 2.8 degrees;

[0109] According to the updated refraction offset data, calibrate the energy distribution characteristics to obtain the adjusted energy aggregation value, including increasing the energy aggregation from 85% to 92%;

[0110] Based on the adjusted energy aggregation value, determine whether the corrected signal propagation path meets the preset conditions to obtain the final calibration result, including the path error being less than 0.5 meters;

[0111] Use the random forest algorithm to verify the final calibration result and determine the accuracy of the energy aggregation value, including setting the number of trees to 100, the maximum depth to 10, and the verification accuracy rate to 98%;

[0112] Obtain the acoustic wave reflection data of the inhomogeneous medium through real-time acoustic imaging technology and reconstruct the characteristic distribution image to obtain the first characteristic distribution, including a resolution of 0.1 mm and an image gray level range of 0 - 255;

[0113] Adjust the distribution parameters in the first characteristic distribution according to the weight allocation scheme. If the distribution parameters deviate from the preset threshold by more than the specified range, then correct the distribution parameters through the weighted average method to obtain the second characteristic distribution, including a weight coefficient of 0.6 and the corrected parameter value decreasing from 120 to 110.

[0114] As Figure 1 - Figure 2 shown, in step S103, obtain the corrected energy aggregation value. In a noise interference environment, enhance the signal boundary through the adaptive filtering algorithm, extract the half-power point characteristics using wavelet transform, and determine the boundary stability of the beam width to obtain a clear beam width quantization result.

[0115] Furthermore, in step S103, obtain the original signal data in a noise interference environment and form the first signal sequence through sampling processing;

[0116] Process the first signal sequence using the adaptive filtering algorithm and adjust the filtering coefficients through the least mean square algorithm to obtain the enhanced second signal sequence;

[0117] For the second signal sequence, decompose the signal using wavelet transform, extract the half-power point features, and generate the first feature set;

[0118] According to the first feature set, calculate the beam width value. If the beam width value exceeds the preset threshold range, correct the half-power point features to obtain the second feature set;

[0119] Based on the second feature set, judge the boundary stability of the beam width, and use the boundary point fluctuation amplitude calculation method to determine the boundary stability index;

[0120] According to the boundary stability index, calculate the corrected energy concentration value, and use the weighted average method to obtain the corrected energy concentration result;

[0121] Obtain the initial signal propagation path through the sound speed distribution model, and use iterative calculation to determine the preliminary refraction offset of the sound wave in the inhomogeneous medium;

[0122] Adjust the signal propagation path according to the preliminary refraction offset, obtain the corrected propagation path data, calculate the change in energy distribution caused by scattering refraction, and determine the aggregation deviation value;

[0123] If the aggregation deviation value exceeds the preset threshold, readjust the sound wave offset through the sound speed distribution model, correct the energy distribution characteristics, and obtain the adjusted energy concentration value.

[0124] Specifically, in step S103, obtain the original signal data in the noise interference environment, and through sampling processing with a sampling frequency of 10 kHz, form the first signal sequence;

[0125] Process the first signal sequence using the adaptive filtering algorithm, adjust the filtering coefficients using the least mean square algorithm, and set the convergence factor to 0.01 to obtain the enhanced second signal sequence;

[0126] For the second signal sequence, perform 5-layer wavelet transform decomposition on the signal using the db4 wavelet basis function, extract the half-power point features, and generate the first feature set including frequency and amplitude;

[0127] According to the first feature set, calculate the beam width value. If the beam width value exceeds the preset threshold range of 0.5° to 1.5°, correct the half-power point features using the linear interpolation method to obtain the second feature set;

[0128] Based on the second feature set, judge the boundary stability of the beam width, and use the boundary point fluctuation amplitude calculation method to calculate the boundary point fluctuation standard deviation and determine the boundary stability index;

[0129] Calculate the corrected energy aggregation value according to the boundary stability index, and use the weighted average method with a weight coefficient of 0.6 to obtain the corrected energy aggregation result;

[0130] Obtain the initial signal propagation path through the sound speed distribution model, and use iterative calculation to determine the preliminary refraction offset of the sound wave in the inhomogeneous medium. Set the number of iterations to 10 and the convergence accuracy to 0.001;

[0131] Adjust the signal propagation path according to the preliminary refraction offset, obtain the corrected propagation path data, calculate the change in energy distribution caused by scattering refraction, and determine the aggregation deviation value;

[0132] If the aggregation deviation value exceeds the preset threshold of 0.1, readjust the sound wave offset through the sound speed distribution model, correct the energy distribution characteristics, and obtain the adjusted energy aggregation value.

[0133] As Figure 1 - Figure 2 shown, in step S104, through the beam width quantization result, for the geometric arrangement fluctuation of the transducer array, use the array response function to calculate the peak distribution of the main lobe and side lobes. If the main lobe peak deviates from the preset threshold, adjust the drive frequency parameter to determine the initial evaluation value of the side lobe suppression ratio.

[0134] Further, in step S104, obtain the geometric arrangement fluctuation data of the transducer array through the beam width quantization result;

[0135] According to the geometric arrangement fluctuation data, use the array response function to calculate the peak distribution of the main lobe and side lobes;

[0136] If the main lobe peak deviates from the preset threshold, adjust the drive frequency parameter to obtain the adjusted peak distribution data;

[0137] According to the adjusted peak distribution data, calculate the ratio of the side lobe peak to the main lobe peak to determine the initial evaluation value of the side lobe suppression ratio;

[0138] For the initial evaluation value of the side lobe suppression ratio, use the iterative optimization algorithm to adjust the parameters to determine the stable value of the side lobe suppression ratio;

[0139] According to the stable value and the beam width quantization result, judge the rationality of the peak distribution characteristics and generate the initial evaluation result;

[0140] Through the initial evaluation result, use the regression analysis algorithm to fit the relationship between the drive frequency and the suppression ratio to determine the final evaluation value;

[0141] Obtain the original signal data in the noise interference environment and form the first signal sequence through sampling processing;

[0142] The first signal sequence is processed using an adaptive filtering algorithm to generate an enhanced second signal sequence.

[0143] Specifically, in step S104, geometric arrangement fluctuation data of the transducer array is obtained through the beam width quantization result, including that the beam width distribution range is from 2.5° to 3.5°;

[0144] According to the geometric arrangement fluctuation data, the peak distributions of the main lobe and side lobes are calculated using the array response function, including that the main lobe peak is -3dB and the side lobe peak is -15dB;

[0145] If the main lobe peak deviates from the preset threshold of -3dB, the drive frequency parameter is adjusted from 50kHz to 55kHz to obtain adjusted peak distribution data, including that the main lobe peak is adjusted to -2.8dB;

[0146] According to the adjusted peak distribution data, the ratio of the side lobe peak to the main lobe peak is calculated, including that the side lobe suppression ratio is -12.2dB, and the initial evaluation value of the side lobe suppression ratio is determined;

[0147] For the initial evaluation value of the side lobe suppression ratio, the parameters are adjusted using an iterative optimization algorithm, including optimizing the drive frequency to 52.5kHz through the gradient descent method, and determining the stable value of the side lobe suppression ratio as -12.5dB;

[0148] According to the stable value and the beam width quantization result, the rationality of the peak distribution characteristics is judged, including that the beam width is 3.2°, and an initial evaluation result is generated;

[0149] Through the initial evaluation result, the relationship between the drive frequency and the suppression ratio is fitted using a regression analysis algorithm, including that the fitting equation is y = 0.5x + 10, and the final evaluation value is determined as -12.3dB;

[0150] The original signal data in a noise interference environment is obtained, including that the signal-to-noise ratio is 10dB, and the first signal sequence is formed through sampling processing, and the sampling frequency is 100kHz;

[0151] The first signal sequence is processed using an adaptive filtering algorithm, including adjusting the filtering coefficients using the least mean square algorithm to generate an enhanced second signal sequence, and the signal-to-noise ratio is increased to 15dB.

[0152] As Figure 1 - Figure 2 shown, in step S105, the initial evaluation value of the side lobe suppression ratio is obtained, combined with the corrected energy concentration degree and the beam width quantization result, the weight allocation is dynamically adjusted through a multi-dimensional parameter joint optimization algorithm, and the weight coefficients are iteratively updated using the gradient descent method to obtain a weight allocation scheme adapted to the environment.

[0153] Further, in step S105, an initial evaluation value of the sidelobe suppression ratio is obtained, and multi-dimensional parameter eigenvalues are extracted according to the corrected energy concentration degree and beam width quantization result;

[0154] The preliminary weight assignment scheme is dynamically adjusted by using a multi-dimensional parameter joint optimization algorithm to obtain a first adjusted weight set;

[0155] For the first adjusted weight set, the weight coefficients are iteratively updated by using the gradient descent method. If the number of iterations reaches a preset threshold, the iteration is stopped to obtain a second adjusted weight set;

[0156] According to the second adjusted weight set, an environment adaptive weight assignment scheme is calculated, the change trend of the environment parameters is judged, and an environment adaptive eigenvalue is obtained;

[0157] The multi-dimensional parameter eigenvalues are updated by using the environment adaptive eigenvalue and combining with the initial value of the sidelobe suppression ratio to obtain an optimized multi-dimensional parameter set;

[0158] The sidelobe suppression ratio is recalculated through the optimized multi-dimensional parameter set to determine the final sidelobe suppression ratio result;

[0159] According to the final sidelobe suppression ratio result, an environment adaptive weight assignment scheme is generated, and the adjusted weight coefficient set is output;

[0160] The acoustic wave reflection data of the inhomogeneous medium is obtained by using real-time acoustic imaging technology, and the characteristic distribution image is reconstructed to obtain a first characteristic distribution;

[0161] The distribution parameters in the first characteristic distribution are adjusted according to the weight assignment scheme. If the deviation of the distribution parameters from the preset threshold exceeds the specified range, the distribution parameters are corrected by using the weighted average method to obtain a second characteristic distribution.

[0162] Specifically, in step S105, an initial evaluation value of the sidelobe suppression ratio is obtained, including calculating the energy ratio of the main lobe to the sidelobe, with the preset initial value being -20 dB. At the same time, multi-dimensional parameter eigenvalues are extracted from the corrected energy concentration degree and beam width quantization result, including an energy concentration degree of 85% and a beam width of 5°;

[0163] The preliminary weight assignment scheme is dynamically adjusted by using a multi-dimensional parameter joint optimization algorithm, including a genetic algorithm. The preset initial weight is [0.3, 0.4, 0.3], and the first adjusted weight set [0.35, 0.38, 0.27] is obtained after optimization;

[0164] For the first set of adjusted weights, use the gradient descent method to iteratively update the weight coefficients. Set the learning rate to 0.01 and the maximum number of iterations to 1000. If the number of iterations reaches the preset threshold, stop the iteration to obtain the second set of adjusted weights [0.34, 0.37, 0.29].

[0165] According to the second set of adjusted weights, calculate the environment adaptive weight allocation scheme, judge the change trend of environmental parameters, preset the environmental temperature change to ±2°C, and obtain the environment adaptive eigenvalue [0.33, 0.36, 0.31].

[0166] Adopt the environment adaptive eigenvalue, combine with the initial value of the sidelobe suppression ratio -20dB, update the multi-dimensional parameter eigenvalue, preset the updated energy concentration to 87%, and the beam width to 4.8°, to obtain the optimized multi-dimensional parameter set.

[0167] Through the optimized multi-dimensional parameter set, recalculate the sidelobe suppression ratio, preset the result to -22dB, and determine the final sidelobe suppression ratio result.

[0168] According to the final sidelobe suppression ratio result, generate an environment adaptive weight allocation scheme and output the adjusted set of weight coefficients [0.32, 0.35, 0.33].

[0169] Obtain the acoustic wave reflection data of the inhomogeneous medium through real-time acoustic imaging technology, including collecting 1000 sampling points, reconstructing the characteristic distribution image, and obtaining the first characteristic distribution.

[0170] Adjust the distribution parameters in the first characteristic distribution according to the weight allocation scheme. If the deviation of the distribution parameters from the preset threshold exceeds the specified range, including exceeding 5%, correct the distribution parameters by the weighted average method to obtain the second characteristic distribution.

[0171] As Figure 1 - Figure 2 shown, in step S106, according to the weight allocation scheme, use real-time acoustic imaging technology to reconstruct the characteristic distribution of the inhomogeneous medium, predict the dynamic trend of scattering and refraction through a convolutional neural network, judge the correction requirement of the energy concentration, and obtain the optimized energy distribution prediction value.

[0172] Further, in step S106, obtain the acoustic wave reflection data of the inhomogeneous medium through real-time acoustic imaging technology, reconstruct the characteristic distribution image, and obtain the first characteristic distribution.

[0173] Adjust the distribution parameters in the first characteristic distribution according to the weight allocation scheme. If the deviation of the distribution parameters from the preset threshold exceeds the specified range, then correct the distribution parameters by the weighted average method to obtain the second characteristic distribution.

[0174] Predict the dynamic trend of scattering and refraction for the second characteristic distribution using a convolutional neural network, and generate a dynamic trend sequence;

[0175] Calculate the energy concentration degree corresponding to the scattering and refraction through the dynamic trend sequence. If the energy concentration degree exceeds the preset upper limit, determine that the correction requirement is of high priority to obtain the correction requirement level;

[0176] Adjust the energy distribution parameters in the second characteristic distribution according to the correction requirement level to generate the first energy distribution;

[0177] Through the comparative analysis of the first energy distribution and the dynamic trend sequence, if the deviation between the two exceeds the specified range, use the iterative optimization method to adjust the first energy distribution to obtain the optimized prediction value;

[0178] Perform matching verification on the optimized prediction value and the second characteristic distribution. If the matching degree is lower than the preset standard, update the distribution parameters through the weighted feedback mechanism to generate the final energy distribution prediction value;

[0179] Obtain the optimized energy distribution prediction value, combine the quantization data of the beam width and the sidelobe suppression ratio, comprehensively score the focusing quality through the weighted fusion algorithm, and smooth the scoring result using the Kalman filter to determine the final focusing accuracy evaluation value;

[0180] According to the energy distribution characteristics, for the energy concentration degree deviation caused by scattering and refraction, use the sound speed distribution model to correct the signal propagation path, and adjust the refraction offset of the sound wave in the inhomogeneous medium through iterative calculation to determine the corrected energy concentration degree value.

[0181] Specifically, in step S106, obtain the acoustic wave reflection data of the inhomogeneous medium through real-time acoustic imaging technology, use the acoustic wave propagation equation and the inverse scattering algorithm to reconstruct the characteristic distribution image, and obtain the first characteristic distribution with a resolution of 512×512 pixels;

[0182] According to the weight allocation scheme, adjust the distribution parameters in the first characteristic distribution using a linear weight coefficient. If the deviation of the distribution parameters from the preset threshold exceeds ±5%, correct the distribution parameters through the weighted average method to obtain the second characteristic distribution;

[0183] Use a convolutional neural network (CNN) to predict the dynamic trend of scattering and refraction for the second characteristic distribution. The input layer is a 3-channel characteristic distribution image, and the convolution kernel size is 3×3, generating a dynamic trend sequence containing 100 time steps;

[0184] Calculate the energy concentration degree corresponding to the scattering and refraction through the dynamic trend sequence. If the energy concentration degree exceeds the preset upper limit of 1.5, determine that the correction requirement is of high priority to obtain the correction requirement level;

[0185] Adjust the energy distribution parameters in the second characteristic distribution according to the correction requirement level, optimize the energy distribution using the gradient descent method, and generate the first energy distribution;

[0186] Through the comparative analysis of the first energy distribution and the dynamic trend sequence, if the deviation between the two exceeds ±2%, the iterative optimization method is used to adjust the first energy distribution to obtain the optimized prediction value;

[0187] Perform matching verification on the optimized prediction value and the second characteristic distribution. If the matching degree is lower than 90%, update the distribution parameters through the weighted feedback mechanism to generate the final energy distribution prediction value;

[0188] Obtain the optimized energy distribution prediction value, combine the quantization data with a beam width of 0.5 degrees and a sidelobe suppression ratio of -30 dB, comprehensively score the focusing quality through the weighted fusion algorithm, and use the Kalman filter to smooth the scoring result to determine the final focusing accuracy evaluation value;

[0189] According to the energy distribution characteristics, for the energy aggregation degree deviation caused by scattering and refraction, use the sound speed distribution model to correct the signal propagation path, and adjust the refraction offset of the sound wave in the inhomogeneous medium through iterative calculation to determine that the corrected energy aggregation degree value is 1.2.

[0190] As Figure 1 - Figure 2 shown, in step S107, obtain the optimized energy distribution prediction value, combine the quantization data of the beam width and the sidelobe suppression ratio, comprehensively score the focusing quality through the weighted fusion algorithm, and use the Kalman filter to smooth the scoring result to determine the final focusing accuracy evaluation value.

[0191] Further, in step S107, collect energy distribution data through sensors and calculate the prediction value using statistical methods;

[0192] Extract the quantization data of the beam width and the sidelobe suppression from the prediction value to obtain the initial feature set;

[0193] For the initial feature set, apply the weighted fusion algorithm to calculate the focusing quality score;

[0194] Obtain the focusing quality score, use the Kalman filter for smoothing processing to generate the smoothed scoring result;

[0195] Determine the accuracy evaluation value through the smoothed scoring result;

[0196] If the accuracy evaluation value is lower than the preset threshold, adjust the parameters of the weighted fusion algorithm and recalculate the focusing quality score;

[0197] Update the accuracy evaluation value according to the adjusted scoring result to obtain the final focusing accuracy evaluation value;

[0198] Obtain the acoustic wave reflection data of the inhomogeneous medium through real-time acoustic imaging technology, reconstruct the characteristic distribution image, and obtain the first characteristic distribution;

[0199] Adjust the distribution parameters in the first characteristic distribution according to the weight allocation scheme. If the distribution parameters deviate from the preset threshold by more than the specified range, correct the distribution parameters through the weighted average method to obtain the second characteristic distribution.

[0200] Specifically, in step S107, collect energy distribution data through sensors and calculate the predicted value using statistical methods, including fitting the energy distribution data using a Gaussian distribution model to obtain predicted values with a mean of 0.8 and a variance of 0.2;

[0201] Extract the quantization data of the beam width and sidelobe suppression from the predicted values, including extracting a beam width of 3.5 degrees and a sidelobe suppression ratio of -25 dB through FFT spectrum analysis to obtain the initial feature set;

[0202] For the initial feature set, apply the weighted fusion algorithm, including calculating the weights of each feature using the entropy weight method, with weights of 0.4 and 0.6 respectively, and calculating the focusing quality score of 85 points;

[0203] Obtain the focusing quality score and perform smoothing processing using the Kalman filter, including setting the process noise covariance to 0.1 and the observation noise covariance to 0.05 to generate a smoothed score result of 83 points;

[0204] Determine the accuracy evaluation value through the smoothed score result, including setting the accuracy evaluation threshold to 80 points and the current accuracy evaluation value to 83 points;

[0205] If the accuracy evaluation value is lower than the preset threshold, adjust the parameters of the weighted fusion algorithm, including resetting the weights in the entropy weight method to 0.5 and 0.5, and recalculating the focusing quality score to 87 points;

[0206] Update the accuracy evaluation value according to the adjusted score result to obtain the final focusing accuracy evaluation value of 87 points;

[0207] Obtain the acoustic wave reflection data of the inhomogeneous medium through real-time acoustic imaging technology and reconstruct the characteristic distribution image, including generating a first characteristic distribution with a resolution of 256×256 using the backprojection algorithm;

[0208] Adjust the distribution parameters in the first characteristic distribution according to the weight allocation scheme. If the distribution parameters deviate from the preset threshold by more than the specified range, correct the distribution parameters through the weighted average method, including adjusting the distribution parameter from 0.7 to 0.65 to obtain the second characteristic distribution.

[0209] For those skilled in the art, various other corresponding changes and deformations can be made according to the technical solutions and concepts described above, and all such changes and deformations should fall within the protection scope of the claims of this application.

Claims

1. A target positioning method based on focused ultrasound, characterized in that, Including the following steps: S101. Obtain the original data of the ultrasonic signal, record the propagation time and intensity distribution of sound waves in the target area through a sensor array, and adopt the time-domain analysis method to separate the main signal and interference signal in view of the influence of scattering and refraction in the inhomogeneous medium, so as to obtain the preliminary energy distribution characteristics; S102. According to the energy distribution characteristics, in view of the energy aggregation degree deviation caused by scattering and refraction, use the sound velocity distribution model to correct the signal propagation path, adjust the refraction offset of sound waves in the inhomogeneous medium through iterative calculation, and determine the corrected energy aggregation degree value; S103. Obtain the corrected energy aggregation degree value, and in the noise interference environment, enhance the signal boundary through the adaptive filtering algorithm, extract the half-power point characteristics by wavelet transform, judge the boundary stability of the beam width, and obtain a clear beam width quantization result; S104. Through the beam width quantization result, in view of the geometric arrangement fluctuation of the transducer array, calculate the peak value distribution of the main lobe and side lobe by using the array response function. If the peak value of the main lobe deviates from the preset threshold, adjust the drive frequency parameter to determine the initial evaluation value of the side lobe suppression ratio; S105. Obtain the initial evaluation value of the side lobe suppression ratio, combine the corrected energy aggregation degree and the beam width quantization result, dynamically adjust the weight distribution through the multi-dimensional parameter joint optimization algorithm, and iteratively update the weight coefficient by using the gradient descent method to obtain an environment-adaptive weight distribution scheme; S106. According to the weight distribution scheme, use the real-time acoustic imaging technology to reconstruct the characteristic distribution of the inhomogeneous medium, predict the dynamic trend of scattering and refraction through the convolutional neural network, judge the correction requirement of the energy aggregation degree, and obtain the optimized energy distribution prediction value; S107. Obtain the optimized energy distribution prediction value, combine the quantization data of the beam width and the side lobe suppression ratio, comprehensively score the focusing quality through the weighted fusion algorithm, and smooth the scoring result by using the Kalman filter to determine the final focusing accuracy evaluation value.

2. The method for target positioning based on focused ultrasound according to claim 1, wherein, In step S101, a sensor array is used to collect the ultrasonic signal in the target area, and record the propagation time and intensity distribution data of the sound wave; According to the collected propagation time and intensity distribution data, extract the original signal characteristics of the sound wave in the inhomogeneous medium; In view of the influence of scattering and refraction in the inhomogeneous medium, adopt the time-domain analysis method to process the original signal and separate the main signal and interference signal; Through the time-domain analysis result, calculate the energy distribution characteristics of the main signal to obtain the preliminary energy distribution data; According to the preliminary energy distribution data, combine the sound velocity distribution model to calculate the initial propagation path of the sound wave in the inhomogeneous medium; Adopt the iterative calculation method to adjust the refraction offset of the sound wave in the inhomogeneous medium to determine the preliminary refraction offset data; According to the preliminary refraction offset data, correct the sound wave propagation path to obtain the corrected propagation path information; Through the corrected propagation path, recalculate the energy distribution change caused by scattering and refraction to determine the aggregation degree deviation value; If the aggregation degree deviation value exceeds the preset threshold, readjust the sound wave offset by using the sound velocity distribution model to obtain the updated refraction offset data, and correct the energy distribution characteristics based on this.

3. The method for target positioning based on focused ultrasound according to claim 1, wherein In step S102, an initial signal propagation path is obtained through a sound velocity distribution model, and iterative calculation is used to determine the preliminary refraction offset of sound waves in a non-uniform medium; Adjust the signal propagation path according to the preliminary refraction offset to obtain the corrected propagation path data; For the corrected propagation path data, calculate the change in energy distribution caused by scattering refraction to determine the aggregation degree deviation value; If the aggregation degree deviation value exceeds the preset threshold, readjust the sound wave offset through the sound velocity distribution model to obtain the updated refraction offset data; According to the updated refraction offset data, correct the energy distribution characteristics to obtain the adjusted energy aggregation degree value; Based on the adjusted energy aggregation degree value, determine whether the corrected signal propagation path meets the preset conditions to obtain the final correction result; Use the random forest algorithm to verify the final correction result to determine the accuracy of the energy aggregation degree value; Obtain the sound wave reflection data of the non-uniform medium through real-time acoustic imaging technology, reconstruct the characteristic distribution image to obtain the first characteristic distribution; Adjust the distribution parameters in the first characteristic distribution according to the weight allocation scheme. If the distribution parameters deviate from the preset threshold by more than the specified range, correct the distribution parameters through the weighted average method to obtain the second characteristic distribution.

4. The method for target positioning based on focused ultrasound according to claim 1, wherein In step S103, obtain the original signal data in a noise interference environment and form the first signal sequence through sampling processing; Process the first signal sequence using an adaptive filtering algorithm and adjust the filtering coefficients through the least mean square algorithm to obtain the enhanced second signal sequence; For the second signal sequence, decompose the signal using wavelet transform, extract the half-power point characteristics, and generate the first feature set; According to the first feature set, calculate the beam width value. If the beam width value exceeds the preset threshold range, correct the half-power point characteristics to obtain the second feature set; Through the second feature set, judge the boundary stability of the beam width, and use the boundary point fluctuation amplitude calculation method to determine the boundary stability index; According to the boundary stability index, calculate the corrected energy aggregation degree value, and use the weighted average method to obtain the corrected energy aggregation degree result; Obtain the initial signal propagation path through the sound velocity distribution model, and use iterative calculation to determine the preliminary refraction offset of sound waves in a non-uniform medium; Adjust the signal propagation path according to the preliminary refraction offset to obtain the corrected propagation path data, calculate the change in energy distribution caused by scattering refraction, and determine the aggregation degree deviation value; If the aggregation degree deviation value exceeds the preset threshold, readjust the sound wave offset through the sound velocity distribution model, correct the energy distribution characteristics, and obtain the adjusted energy aggregation degree value.

5. The method for target positioning based on focused ultrasound according to claim 1, wherein, In step S104, obtain the geometric arrangement fluctuation data of the transducer array through the beam width quantization result; According to the geometric arrangement fluctuation data, calculate the peak distributions of the main lobe and side lobes using the array response function; If the main lobe peak deviates from the preset threshold, adjust the drive frequency parameter to obtain the adjusted peak distribution data; According to the adjusted peak distribution data, calculate the ratio of the side lobe peak to the main lobe peak to determine the initial evaluation value of the side lobe suppression ratio; For the initial evaluation value of the sidelobe suppression ratio, an iterative optimization algorithm is used to adjust the parameters to determine the stable value of the sidelobe suppression ratio; According to the stable value and the quantization result of the beam width, judge the rationality of the peak distribution characteristics to generate the initial evaluation result; Based on the initial evaluation result, use the regression analysis algorithm to fit the relationship between the driving frequency and the suppression ratio to determine the final evaluation value; Obtain the original signal data in the noise interference environment and form the first signal sequence through sampling processing; Use the adaptive filtering algorithm to process the first signal sequence to generate the enhanced second signal sequence.

6. The method for target positioning based on focused ultrasound according to claim 1, wherein In step S105, obtain the initial evaluation value of the sidelobe suppression ratio, and extract the multi-dimensional parameter eigenvalue according to the corrected energy concentration degree and the quantization result of the beam width; Use the multi-dimensional parameter joint optimization algorithm to dynamically adjust the preliminary weight allocation scheme to obtain the first adjusted weight set; For the first adjusted weight set, use the gradient descent method to iteratively update the weight coefficient. If the number of iterations reaches the preset threshold, stop the iteration to obtain the second adjusted weight set; According to the second adjusted weight set, calculate the environment adaptive weight allocation scheme, judge the change trend of the environment parameters, and obtain the environment adaptive eigenvalue; Use the environment adaptive eigenvalue and combine it with the initial value of the sidelobe suppression ratio to update the multi-dimensional parameter eigenvalue to obtain the optimized multi-dimensional parameter set; Through the optimized multi-dimensional parameter set, recalculate the sidelobe suppression ratio to determine the final sidelobe suppression ratio result; According to the final sidelobe suppression ratio result, generate the environment adaptive weight allocation scheme and output the adjusted weight coefficient set; Obtain the acoustic wave reflection data of the inhomogeneous medium through the real-time acoustic imaging technology, reconstruct the characteristic distribution image to obtain the first characteristic distribution; Adjust the distribution parameter in the first characteristic distribution according to the weight allocation scheme. If the distribution parameter deviates from the preset threshold by more than the specified range, correct the distribution parameter by the weighted average method to obtain the second characteristic distribution.

7. The method for target positioning based on focused ultrasound according to claim 1, wherein In step S106, obtain the acoustic wave reflection data of the inhomogeneous medium through the real-time acoustic imaging technology, reconstruct the characteristic distribution image to obtain the first characteristic distribution; Adjust the distribution parameter in the first characteristic distribution according to the weight allocation scheme. If the distribution parameter deviates from the preset threshold by more than the specified range, then correct the distribution parameter by the weighted average method to obtain the second characteristic distribution; Use the convolutional neural network to predict the dynamic trend of scattering and refraction for the second characteristic distribution to generate a dynamic trend sequence; Calculate the energy concentration degree corresponding to the scattering and refraction through the dynamic trend sequence. If the energy concentration degree exceeds the preset upper limit, determine that the correction requirement is of high priority to obtain the correction requirement level; Adjust the energy distribution parameter in the second characteristic distribution according to the correction requirement level to generate the first energy distribution; Through the comparison and analysis of the first energy distribution and the dynamic trend sequence, if the deviation between the two exceeds the specified range, use the iterative optimization method to adjust the first energy distribution to obtain the optimized prediction value; Perform matching verification on the optimized prediction value and the second characteristic distribution. If the matching degree is lower than the preset standard, update the distribution parameter through the weighted feedback mechanism to generate the final energy distribution prediction value; Obtain the optimized predicted value of the energy distribution, combine the quantization data of the beam width and the sidelobe suppression ratio, comprehensively score the focusing quality through a weighted fusion algorithm, and use Kalman filtering to smooth the scoring result to determine the final focusing accuracy evaluation value; According to the energy distribution characteristics, aiming at the energy concentration deviation caused by scattering and refraction, use the sound speed distribution model to correct the signal propagation path, and determine the corrected energy concentration value by iteratively calculating and adjusting the refraction offset of the sound wave in the inhomogeneous medium.

8. The method for target positioning based on focused ultrasound according to claim 1, wherein In step S107, collect energy distribution data through sensors and calculate the predicted value using statistical methods; Extract the quantization data of the beam width and sidelobe suppression from the predicted value to obtain the initial feature set; Apply a weighted fusion algorithm to the initial feature set to calculate the focusing quality score; Obtain the focusing quality score, perform smoothing processing using Kalman filtering, and generate a smoothed scoring result; Determine the accuracy evaluation value through the smoothed scoring result; If the accuracy evaluation value is lower than the preset threshold, adjust the parameters of the weighted fusion algorithm and recalculate the focusing quality score; Update the accuracy evaluation value according to the adjusted scoring result to obtain the final focusing accuracy evaluation value; Obtain the acoustic wave reflection data of the inhomogeneous medium through real-time acoustic imaging technology, reconstruct the characteristic distribution image, and obtain the first characteristic distribution; Adjust the distribution parameters in the first characteristic distribution according to the weight allocation scheme. If the distribution parameters deviate from the preset threshold by more than the specified range, correct the distribution parameters by the weighted average method to obtain the second characteristic distribution.

Images