A phased array element optimization method and system for sound field regulation in complex acoustic environments

Abstract

The application discloses a phased array element optimization method and system for sound field regulation of a complex sound environment, and belongs to the field of medical devices and ultrasonic focusing technology. The method comprises the following steps: acquiring ideal driving phases and amplitudes of each array element, and determining amplitude weights of the array elements based on physical parameters of the complex sound environment; clustering the ideal phases of all effective array elements into a preset number of initial subarrays, each subarray corresponding to a driving channel; constructing a multi-target comprehensive evaluation function taking focal point sound pressure as a main target and phase difference and channel load difference in a group as side targets; performing directional multi-target fine tuning, and through amplitude target search and phase hard constraint test, whether the array element is transferred to a candidate channel is determined based on whether the score of the multi-target evaluation function is improved; iterating until convergence, and outputting a final driving channel and array element mapping relationship. The application significantly reduces the number of driving channels, guarantees focal point sound pressure, phase consistency and power balance, eliminates algorithm randomness, and reduces system hardware cost.

Description

Technical Field

[0001] This invention relates to the fields of medical devices and ultrasound focusing technology, and in particular to a phased array element optimization method and system for sound field control in complex acoustic environments. Background Technology

[0002] Focused ultrasound has been widely used in biological and medical fields such as thrombolysis and targeted drug delivery. In transcranial focused ultrasound (tFUS) treatment, the complex acoustic environment of the skull (non-uniform spatial distribution of geometry and physical parameters) can cause transducer focus shifts or even distortions, increasing treatment safety risks. To improve acoustic focusing performance in complex acoustic environments and achieve high gain and precise focusing of acoustic energy, phased-array focusing or time-reversal techniques are typically combined with a phased array composed of multiple elements (usually more than 1000) to obtain ideal acoustic focusing effects.

[0003] Ultrasonic phased array technology controls the emission phase and amplitude of each independent element in a transducer array, utilizing the principle of wave interference to focus or deflect the sound beam at specific locations in space. Traditional phased arrays often utilize the high concurrency capabilities of FPGAs to achieve nanosecond-level delay and phase control across multiple channels. In the power drive circuit, impedance matching between the drive power supply and the transducer load is required to maximize energy transfer, and high consistency of the load across each channel of the hardware drive circuit is also essential. Skull density ratio (SDR) is defined as the density ratio of the diploic layer to the dense cortical bone, used to evaluate the efficiency of acoustic energy penetration through the skull. The K-means algorithm is an unsupervised clustering analysis algorithm that maps the ideal phase of high-dimensional physical array elements to a feature space, clustering to reduce the dimensionality to a finite number of target electronic channels, ensuring that elements merged into the same channel have extremely high physical compensation phase consistency.

[0004] However, existing dimensionality reduction grouping methods and control systems still face the following technical challenges in practical applications: (1) Sound field divergence caused by dimensionality reduction: If the channels are directly connected in hard parallel (e.g., compressed to 24 independent channels) and traditional uniform dimensionality reduction or random grouping is used, it often leads to a decrease in spatial sampling rate, resulting in a cliff-like drop in focusing ability and high-intensity grating lobes (severe side lobes and energy dispersion).

[0005] (2) Hardware impedance matching problem and power imbalance: The hardware drive circuit has extremely high requirements for load consistency. In order to ensure the safe operation of the hardware power amplifier module, the power of each array element connected in parallel in the same electronic channel should be as balanced as possible. Traditional algorithms often only consider the focusing effect and ignore the impedance matching problem caused by amplitude differences, resulting in power imbalance of the piezoelectric elements controlled by some channels.

[0006] (3) Existing algorithms have “multi-objective conflict” and “randomness defects”: Existing global optimization algorithms (such as genetic algorithm GA) have blind random mutation, which not only consumes huge computing power, but also results inconsistency in each run (lack of stability). More seriously, when forcibly leveling the power load of each group, it is very easy to destroy the existing physical compensation phase consistency within the group (leading to destructive interference), causing irregular attenuation of the effective sound pressure at the focal point. Summary of the Invention

[0007] In view of this, the purpose of the present invention is to provide a phased array element optimization method and system for sound field control in complex acoustic environments. This method significantly reduces the number of phased array driving channels (thereby reducing hardware costs) while stably achieving high-gain, distortion-free sound field focusing in complex acoustic environments, and ensuring load balancing of each channel and determinism of algorithm results.

[0008] To achieve the above objectives, the present invention provides the following technical solution: The phased array element optimization method for sound field control in complex acoustic environments provided by this invention includes the following steps: Step 1: Obtain the ideal driving phase and amplitude required for each element in the entire array to achieve focusing at the target focus, and determine the amplitude weight of each element based on the physical parameters of the complex acoustic environment; Step 2: Using the ideal phase of all effective array elements as clustering features, cluster them into a preset number of initial subarray sets using a clustering algorithm. Each subarray corresponds to a driving channel, and the equivalent center phase and total amplitude weight load of each driving channel are calculated. Step 3: Construct a multi-objective comprehensive evaluation function with the total effective sound pressure at the focal point as the main objective and the intra-group phase difference and inter-channel total amplitude weight load difference of all driving channels as secondary objectives; Step 4: Based on the homogeneous driving characteristics of the driving channels, perform directional multi-objective fine-tuning. This fine-tuning process includes: Array elements whose amplitudes significantly deviate from the channel average value in each driving channel are extracted as free array elements; A candidate driving channel is targeted for the free array element, such that the average amplitude of the candidate driving channel matches the amplitude of the free array element. A phase hard constraint test is performed on the free array element and the candidate driving channel. If the phase difference between the two exceeds a preset tolerance threshold, the transfer is rejected. If the phase hard constraint test is passed, then based on the multi-objective comprehensive evaluation function, the global score before and after transferring the free array element to the candidate driving channel is evaluated. If the score is improved, then the transfer is confirmed. Step 5: Repeat Step 4 until all free array elements fail to meet the phase hard constraint or fail to improve the global score. Output the final mapping relationship between the driving channel and the array elements to complete the optimization.

[0009] Furthermore, the determination of the amplitude weight of each array element based on the physical parameters of the complex acoustic environment in step one specifically includes: The three-dimensional physical space is discretized into a computational grid. For each array element, the set of grid points it covers is extracted and spatial downsampling is performed to obtain multiple emission points. For each emission point, a spatial ray path to the target focus is constructed, and the density profile on this path is extracted from the 3D CT image; The density profile is segmented, and the maximum density value is searched in the front and rear segments respectively as the density of the outer and inner plates of the skull. The low quantile value is extracted in the middle segment as the equivalent density of the diploic plate. The transmittance index of a single ray is calculated based on the density of the outer plate, the density of the inner plate, and the equivalent density of the barrier. Calculate the mathematical average of the transmittance indices of all rays corresponding to an array element, and use it as the global equivalent transmittance of that array element. The global equivalent transmittance is converted into the amplitude weight of the array element through a nonlinear mapping function.

[0010] Furthermore, when converting the global equivalent transmittance into the amplitude weight of the array element using a nonlinear mapping function, high and low thresholds are set: When the global equivalent transmittance is lower than the first preset threshold, the amplitude weight of the array element is set to 0. When the global equivalent transmittance is higher than the second preset threshold, the amplitude weight of the array element is set to 1. When the global equivalent transmittance is between the first preset threshold and the second preset threshold, the amplitude weight is determined using a linear transition method.

[0011] Furthermore, in step two, the ideal phases of all effective array elements are used as clustering features for clustering, specifically as follows: The ideal phase of each effective array element is mapped to the complex plane to construct a clustering feature vector; The K-means clustering algorithm is used to cluster the entire array into an initial set of subarrays for the target number of driving channels.

[0012] Furthermore, the multi-objective comprehensive evaluation function constructed in step three has the following specific form:

[0013] in, This represents the global score of the multi-objective comprehensive evaluation function; This is the sum of the effective sound pressure at the focal point; It is the sum of the variances of the phases of the array elements within the K channels and the central phase; Total amplitude load for K channels The variance between them; α, β, γ are preset constant weighting coefficients.

[0014] Furthermore, in step four, before extracting the free array elements, a core array element freezing step is also included: Calculate the channel center phase and the average amplitude requirement of the array elements within each driving channel; Set a phase core tolerance limit and an amplitude core tolerance limit; For each element in the driving channel, if the difference between its phase and the phase at the center of the channel is less than the phase core tolerance limit, and the difference between its amplitude and the average amplitude requirement is less than the amplitude core tolerance limit, then the element is determined to be a core anchoring element, and it is frozen in subsequent iterations to not participate in cross-group transfer.

[0015] Furthermore, in step four, the targeted search for candidate driving channels for the free array elements specifically involves: Among all driving channels globally, one or more channels whose average amplitude is closest to the amplitude of the free array element itself are selected as candidate channels.

[0016] Furthermore, the preset phase tolerance threshold in the phase hard constraint test is λ / 4, where λ is the operating wavelength of the ultrasonic phased array.

[0017] Furthermore, the loop termination condition for the directional multi-objective fine-tuning is: all unfrozen free array elements are unable to perform transitions due to triggering phase hard constraints, or all hypothetical transitions fail to make the global score of the multi-objective comprehensive evaluation function exceed the current score.

[0018] This invention provides a phased array element optimization system for sound field manipulation in complex acoustic environments, comprising: The ideal parameter acquisition module is used to acquire the ideal driving phase and amplitude required for each element in the entire array to achieve focusing at the target focus. The amplitude weight determination module is used to determine the amplitude weight of each array element based on the physical parameters of the complex acoustic environment. The initial dimensionality reduction and clustering module is used to take the ideal phase of all effective array elements as clustering features and use a clustering algorithm to cluster them into a preset number of initial subarray sets, with each subarray corresponding to a driving channel; The evaluation function construction module is used to construct a multi-objective comprehensive evaluation function with the total effective sound pressure at the focal point as the main objective and the differences in intra-group phase difference and inter-channel total amplitude weight load of all driving channels as secondary objectives. The directional multi-objective fine-tuning module is used to extract free array elements based on the homogeneous driving characteristics of the driving channel, perform amplitude targeted search and phase hard constraint verification, and evaluate the transfer gain according to the multi-objective comprehensive evaluation function to determine the final mapping relationship between the driving channel and the array elements.

[0019] The beneficial effects of this invention are as follows: This invention provides a phased array element optimization method for sound field control in complex acoustic environments, belonging to the fields of medical devices and ultrasound focusing technology. The method includes: obtaining the ideal driving phase and amplitude of each array element, and determining the amplitude weight of the array elements based on the physical parameters of the complex acoustic environment; clustering the ideal phases of all effective array elements into a preset number of initial subarrays, each subarray corresponding to a driving channel; constructing a multi-objective comprehensive evaluation function with focal sound pressure as the primary objective and intra-group phase difference and channel load difference as secondary objectives; performing directional multi-objective fine-tuning, and determining whether an array element is transferred to a candidate channel based on the improvement of the multi-objective evaluation function score through amplitude-targeted search and phase hard constraint verification; iterating until convergence, and outputting the final mapping relationship between driving channels and array elements. This invention significantly reduces the number of driving channels while ensuring focal sound pressure, phase consistency, and power balance, eliminating algorithm randomness, and reducing system hardware costs. Compared with existing technologies, this invention has the following beneficial effects: (1) Eliminate algorithm randomness and ensure stable and reliable results. This invention abandons the probabilistic mutation mechanism of heuristic algorithms such as genetic algorithms, and adopts a deterministic fine-tuning algorithm with a pure rule-based architecture. Given the same input data, the results of multiple runs are consistent, exhibiting extremely high clinical safety and engineering stability, and avoiding the treatment uncertainty risks caused by traditional stochastic algorithms.

[0020] (2) Coordinate multiple conflicting objectives to achieve dimensionality reduction, power equalization and sound pressure maximization. This invention constructs a multi-objective comprehensive evaluation function with focal sound pressure as the primary objective and intra-group phase consistency and channel load balancing as constraints, combined with phase hard constraint testing and hypothesis transfer evaluation mechanisms. While significantly compressing a large number of physical array elements into a limited number of drive channels (e.g., 8-24 channels), it effectively avoids focusing capability degradation and high-intensity grating lobes caused by a decrease in spatial sampling rate. Furthermore, while ensuring intra-group phase consistency, it balances the power of piezoelectric elements associated with each channel, solving the hardware impedance matching problem and automatically maintaining the baseline of no attenuation of focal sound pressure.

[0021] (3) Achieve high-quality focal spot control under extreme channel conditions and reduce hardware costs Even with the maximum number of undersampled channels, this invention can still achieve focal spot resolution close to the traditional geometric limit, improve the half-width at half-maximum (WHM) index, and provide a feasible technical solution for phased array channel optimization in complex acoustic environments such as low-cost transcranial focused ultrasound therapy. It significantly reduces the difficulty of system integration and the cost of instrument development, which is conducive to the mass production and clinical promotion of the equipment.

[0022] The above and other objects, advantages, and features of the present invention will be more fully set forth and demonstrated through the following detailed description of specific embodiments in conjunction with the accompanying drawings. Those skilled in the art, upon referring to the following detailed description and the accompanying drawings, will be able to better understand and realize the above advantages of the present invention. Other objects, features, and advantages of the present invention will become clearer after being described in detail in the detailed description section in conjunction with the accompanying drawings. Attached Figure Description

[0023] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the following drawings are provided for illustration.

[0024] Figure 1 A flowchart for array element optimization based on amplitude weighting and phase classification; Figure 2 For accurate SDR extraction and soft-weighted preprocessing based on full-element dense ray tracing; Figure 3 For initial dimensionality reduction clustering of K-means based on ideal phase; Figure 4 For directional multi-objective fine-tuning based on intra-group dual-feature consistency and phase hard constraints; Figure 5 This is a comparison image of the focal spot in the sound field simulation. Detailed Implementation

[0025] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0026] Example 1

[0027] To optimize the sound field in complex acoustic environments and minimize the number of phased array elements and reduce system development costs, this invention proposes a sparse array optimization strategy for hardware load-Balancing. This strategy includes a dimensionality reduction preprocessing method based on skull density ratio (SDR) weighting and physical phase clustering, as well as a multi-objective directed fine-tuning algorithm based on phase hard constraints.

[0028] like Figure 1 As shown, Figure 1 This is a flowchart of a phased array element optimization method for sound field control in complex acoustic environments. The process starts from the beginning and executes four core steps sequentially: Step S1 performs precise SDR extraction and soft-weighted preprocessing based on dense ray tracing of all array elements, calculating the amplitude weight for each element; Step S2 performs initial K-means dimensionality reduction clustering based on ideal phase, initially merging physical array elements into the target number of electronic channels; Step S3 constructs a multi-objective comprehensive evaluation function, i.e., Score = α·(P_focus) - β·Var(Φ_group) - γ·Var(W_group), used to quantify focus sound pressure, intra-group phase consistency, and channel load balance; Step S4 performs directional multi-objective fine-tuning based on intra-group dual-feature consistency and phase hard constraints, iteratively optimizing the element-channel mapping. After each fine-tuning, convergence is checked. If convergence fails, the process returns to Step S4 for further fine-tuning; if convergence occurs, the optimal dimensionality reduction grouping table for hardware-driven efficiency is output, and the process ends. This figure summarizes the entire process from raw data to the final optimal channel allocation, and embodies the core technical route of "preprocessing-clustering-scoring-iterative fine-tuning".

[0029] This embodiment is based on an array element optimization process using amplitude weighting and phase classification. To achieve the core dimensionality reduction and optimization, the algorithm is a "maximum gradient hill-climbing fine-tuning control method based on SDR amplitude weighting - physical phase clustering - multi-objective evaluation function". The specific algorithm flow is as follows: Step S1: Accurate SDR Extraction and Soft Weighted Preprocessing Based on Full-Element Dense Ray Tracing like Figure 2 As shown, Figure 2To achieve accurate SDR extraction and soft-weighted preprocessing based on full-element dense ray tracing, the ideal driving phase of each element is first obtained through acoustic simulation backpropagation. Secondly, each element is discretized into dense emission points, and rays are constructed from each emission point to the focal point. Interpolation along these rays in the CT image yields continuous skull density profiles. Then, effective bone segments are extracted from the density profiles, and the densities of the anterior outer plate, posterior inner plate, and mid-section barrier are extracted. Single-ray transmittance is calculated. Finally, the transmittance of all rays from the same element is averaged to obtain the global equivalent density. The signal is then transformed into an amplitude weight Wi for the array element via a double-threshold nonlinear mapping, achieving soft-weighted preprocessing. This figure illustrates how dense ray tracing can be used to assess the impact of skull inhomogeneity on sound penetration and assign appropriate weights to each array element. The specific process is as follows: (1) Obtaining single-point acoustic features and ideal phase: A virtual point sound source is set at the target focus, and the complex acoustic transfer operators reaching the N array elements are obtained through acoustic simulation and backpropagation calculation. Extract the ideal driving phase and amplitude required for perfect focusing of the i-th array element.

[0030] (2) CT density profile extraction based on dense mesh and ray tracing: The three-dimensional physical space is discretized into a computational grid. For the i-th transducer element, the set of all effective grid points it covers in space is extracted. To balance computational efficiency and spatial sampling rate, this set of grid points is uniformly downsampled (e.g., with a step size of 2) to obtain the dense emission point cloud corresponding to the element (a single element typically contains hundreds to thousands of emission points). For each emission point, a spatial ray path from that point to the physical focal point is constructed, and linear interpolation is performed along this ray in the 3D CT image to obtain a continuous skull density profile along the ray path.

[0031] (3) Characteristic peak finding and single-ray SDR solution: For each extracted skull density profile, adaptive threshold segmentation is performed to extract effective bone segments. The effective bone segments are divided into anterior (e.g., the first 30%), middle, and posterior (e.g., the last 30%) segments based on length.

[0032] First, the maximum density value is searched in both the anterior and posterior segments, and these values ​​are defined as the density of the outer table of the skull penetrated by the ray. Density of Inner Table .

[0033] Secondly, in the middle section (diploe layer region), the density values ​​are sorted, and the values ​​of the specified low quantiles (e.g., the values ​​of the top 10% quantiles after sorting) are extracted as the equivalent density of the diploe. This allows for precise characterization of the acoustic barrier properties of the porous structure of the barrier layer.

[0034] Then, the transmittance of the single ray r is calculated: ; in, Indicates transmittance; This indicates the density of the inner plate being cut off; This indicates the density of the cut outer panel; Indicates the equivalent density of the barrier; (4) Array element-level SDR mean fusion and soft weighted mapping: Iterate through all dense rays corresponding to the i-th element and calculate the effective rays. The mathematical average value is used as the global equivalent skull transmittance of this large-area array element. .

[0035] Subsequently, a nonlinear mapping function is constructed to... Converted into the amplitude weight of the array element And set high and low thresholds. and : when hour, = 0 (Disable inefficient array elements); when hour, (Linear transition); when hour, = 1 (fully open high-efficiency array element).

[0036] Step S2: Initial Dimensionality Reduction Clustering of K-means Based on Ideal Phase like Figure 3 As shown, Figure 3 This paper presents the initial dimensionality reduction clustering process of K-means based on ideal phase. First, inefficient array elements with zero amplitude weights are removed. The ideal phases of the remaining effective array elements are mapped to the complex plane to construct a non-jump feature space, where each array element corresponds to a two-dimensional feature vector. Second, the K-means clustering algorithm is used to cluster all effective array elements into K initial subarray sets representing the target number of channels, achieving initial dimensionality reduction from physical array elements to a finite number of electronic channels. Finally, the equivalent center phase and the total amplitude weight load of each subarray (i.e., each driving channel) are extracted. The specific process is as follows: (1) Constructing a non-jump feature space: All effective array elements ( Ideal phase >0) Mapping to the complex plane to construct cluster feature vectors .

[0037] (2) Generate initial good solution: Use K-means clustering algorithm to cluster the whole array into an initial subarray set with the target number of channels K (e.g., K = 24).

[0038] (3) Extracting channel features: Calculate the channel equivalent center phase of each initial subarray k. Total amplitude weighted load of the current channel (within the group) (The sum of these factors) ensures extremely high "intra-group phase consistency".

[0039] Step S3: Construct a multi-objective comprehensive evaluation function (Score Function) Construct a global multi-objective scoring function for the primary objective (focus sound pressure level) and secondary objectives (intra-group phase difference, channel load range): (1) Here, Score represents the global score of the multi-objective comprehensive evaluation function, which is used to quantify the overall performance of the current array element-channel mapping scheme. This is the sum of the effective sound pressure at the focal point; It is the sum of the variances of the phases of the array elements within the K channels and the central phase; Total amplitude load for K channels The variance between them. α, β, γ are preset constant weighting coefficients (e.g., set α=10, β=2, γ=5).

[0040] Step S4: Oriented multi-objective fine-tuning based on intra-group dual-feature consistency and phase hard constraints like Figure 4 As shown, Figure 4To implement a directional multi-objective fine-tuning process based on intra-group dual-feature consistency and phase hard constraints, the following steps are taken: First, core anchoring elements satisfying dual-feature consistency are frozen according to phase and amplitude tolerance limits, while elements with significantly deviated amplitudes are extracted as edge-free elements. Then, it is determined whether any unprocessed free elements exist: if so, amplitude-targeted search is performed to find the candidate channel B with the closest average amplitude in the global channels; next, a phase veto check is performed: if the phase difference between the element and the center of channel B exceeds a preset red line threshold, the element is abandoned and the process returns to process the next free element; if the phase difference is acceptable, a multi-objective comprehensive evaluation pre-run is performed, calculating the global score before and after the assumed transfer. If the score improves after the transfer, the transfer is confirmed and the phase and load feature parameters of the channel are updated; otherwise, the transfer is abandoned. This process is repeated until all free elements are processed, at which point the fine-tuning terminates, the algorithm converges, and the hardware-driven optimal dimensionality-reduced grouping table is output.

[0041] By leveraging the parallel-driven, same-source voltage characteristics of ultrasonic phased arrays, a "group-wide amplitude-targeted matching" and "hard phase constraint" mechanism is introduced. Based on ensuring both phase and amplitude physical boundaries within the group, precise multi-target directional fine-tuning is performed, as follows: (1) Core element freezing and edge free element extraction: After the initial clustering is completed, the channel center phase of each of the K channels (groups) is calculated. Average amplitude requirement of array elements within the channel Set the "phase core tolerance limit". "and "amplitude core tolerance limit" ".

[0042] For each array element i in channel k, if it satisfies and If the array element is identified as a "core anchoring array element", it will be frozen by default in subsequent iterations and will not participate in cross-group transfer calculations, thus greatly compressing the invalid calculation space.

[0043] If an array element deviates significantly from the average value of its channel in terms of amplitude, it is extracted as an "amplitude-free array element" and included in the queue to be optimized.

[0044] (2) Amplitude-targeted search and phase hard constraint test: Following the hardware physical principle of "same group, same source driving", for a certain high-amplitude free array element belonging to channel A in the queue to be optimized. (its own amplitude is) And much larger Perform amplitude-targeted search: Instead of blindly searching for the channel with the lowest total load, we target and search for the average amplitude of candidate channels among the global K channels. ; and the amplitude of the free array element The closest one or more candidate channels (e.g., channel B, satisfying) ).

[0045] Phase hard constraint test: After finding the amplitude-matching channel B, force the calculation of array elements. Ideal phase Center phase with candidate channel B absolute short-range phase difference Set strict phase tolerance thresholds. (For example ).like If the condition is not met, it is considered a "severe phase mismatch," and the Hard Constraint Rejection mechanism is immediately triggered. The system directly terminates the operation of moving the array element into channel B and instead searches for the next candidate channel or retains it in the original channel.

[0046] (3) Multi-objective comprehensive evaluation for decisive victory: The system performs a multi-objective scoring game only if the free array element i successfully matches the amplitude of channel B and fully satisfies the "hard phase constraint" condition of channel B. To maintain a global multi-objective evaluation system, the focus sound pressure is the main objective, and the phase variance within each global group is the secondary objective, as shown in equation (1).

[0047] Calculate the global score before and after the transfer. And the hypothetical global score .like This demonstrates that the transfer, while satisfying strict phase consistency and hardware amplitude consistency, further improves the overall sound field efficiency. The algorithm officially confirms the transfer and updates the characteristic parameters in real time.

[0048] (4) Fast convergence and termination mechanism: The system iteratively examines the free elements in each channel and attempts directional transfers. If all unfrozen free elements fail to perform a transfer due to triggering "hard phase constraints," or if all transfer assumption scores fail to exceed the current score: When the algorithm determines that the physical limits of the current sound field and hardware have been reached, the fine-tuning process is quickly terminated, and the optimal dimensionality reduction parameters are output.

[0049] Example 2

[0050] This embodiment provides a system that can be implemented using software, hardware, or a combination of both, for executing the optimization method described in Embodiment 1. The system includes the following functional modules: The ideal parameter acquisition module is used to obtain the ideal driving phase and amplitude required for each array element in the entire array to achieve focusing at the target focus. This module is realized through acoustic simulation back propagation calculation: a virtual point sound source is set at the target focus, and the sound field propagation is solved by the time-domain pseudospectral method or the angular spectrum method to obtain the complex acoustic transfer operator reaching the surface of each array element, and then the ideal phase and ideal amplitude of each array element are extracted.

[0051] The amplitude weight determination module is used to determine the amplitude weight of each array element based on physical parameters of complex acoustic environments (such as cranial CT data in transcranial therapy). This module performs the following operations: discretizing the surface of each array element into a dense emission point cloud; constructing a spatial ray path from each emission point to the focal point; interpolating along the path in the 3D CT image to obtain a density profile; segmenting the density profile to extract the outer plate density, inner plate density, and equivalent density of the barrier; calculating the single-ray transmittance index; averaging the transmittance of all effective rays of the same array element to obtain the global equivalent transmittance; and converting the global equivalent transmittance into the amplitude weight of the array element through a nonlinear mapping function with preset high and low thresholds.

[0052] The initial dimensionality reduction and clustering module uses the ideal phase of all effective array elements (with amplitude weights greater than zero) as clustering features. It employs the K-means clustering algorithm to cluster them into a preset number of initial subarray sets (i.e., the target number of driving channels), with each subarray corresponding to one driving channel. This module maps the ideal phase of each array element to the complex plane to construct a feature vector, performs K-means initialization to avoid local optima, and outputs the equivalent center phase and total amplitude weight load of each channel after iterative convergence.

[0053] The evaluation function construction module is used to construct a multi-objective comprehensive evaluation function with the sum of effective sound pressure levels at the focal point as the primary objective and the differences in intra-group phase differences and inter-channel total amplitude weighted load differences as secondary objectives. This function uses a weighted summation method, with the sum of intra-group phase variances and inter-channel total amplitude weighted load variances as secondary objectives.

[0054] in, This is the sum of the effective sound pressure at the focal point; It is the sum of the variances of the phases of the array elements within the K channels and the central phase; Total amplitude load for K channels The variance between them; α, β, and γ are preset constant weighting coefficients, which respectively reflect the importance attached to maximizing sound pressure, phase consistency, and power balance. For example, α=10, β=2, and γ=5.

[0055] The directional multi-objective fine-tuning module is used to extract free array elements based on the homogeneous driving characteristics of the driving channel, perform amplitude targeted search and phase hard constraint verification, and evaluate the transfer gain according to the multi-objective comprehensive evaluation function to determine the final mapping relationship between the driving channel and the array elements.

[0056] The module first freezes the core anchoring elements that satisfy the dual-feature consistency based on the phase tolerance limit and the amplitude tolerance limit, and marks the elements with significant amplitude deviations as free elements. For each free element, it searches for the candidate channel with the closest average amplitude in the global channels. Then, it performs a phase hard constraint test: if the short-range phase difference between the phase of the free element and the center phase of the candidate channel exceeds a preset red line threshold, the transfer is rejected. If the test is passed, the evaluation function construction module is called to calculate the global score before the transfer and after the assumed transfer. Only when the score increases is the transfer confirmed and the phase and load parameters of the channel updated. The process is iterated until all free elements fail to meet the transfer conditions, and finally, the dimensionality reduction grouping table with the optimal hardware driving efficiency is output.

[0057] The modules described above work together to automate the entire process from raw acoustic and imaging data to optimal channel mapping. This system can be deployed on general-purpose computing platforms or embedded hardware (such as FPGA accelerator cards) for generating phased array drive configurations in complex acoustic environments such as transcranial focused ultrasound therapy.

[0058] The above-described embodiments are merely preferred embodiments provided to fully illustrate the present invention, and the scope of protection of the present invention is not limited thereto. Equivalent substitutions or modifications made by those skilled in the art based on the present invention are all within the scope of protection of the present invention. The scope of protection of the present invention is defined by the claims.

Claims

1. A phased array element optimization method for sound field control in complex acoustic environments, characterized in that, Includes the following steps: Step 1: Obtain the ideal driving phase and amplitude required for each element in the entire array to achieve focusing at the target focus, and determine the amplitude weight of each element based on the physical parameters of the complex acoustic environment; Step 2: Using the ideal phase of all effective array elements as clustering features, cluster them into a preset number of initial subarray sets using a clustering algorithm. Each subarray corresponds to a driving channel, and the equivalent center phase and total amplitude weight load of each driving channel are calculated. Step 3: Construct a multi-objective comprehensive evaluation function with the total effective sound pressure at the focal point as the main objective and the intra-group phase difference and inter-channel total amplitude weight load difference of all driving channels as secondary objectives; Step 4: Based on the homogeneous driving characteristics of the driving channels, perform directional multi-objective fine-tuning. This fine-tuning process includes: Array elements whose amplitudes significantly deviate from the channel average value in each driving channel are extracted as free array elements; A candidate driving channel is targeted for the free array element, such that the average amplitude of the candidate driving channel matches the amplitude of the free array element. A phase hard constraint test is performed on the free array element and the candidate driving channel. If the phase difference between the two exceeds a preset tolerance threshold, the transfer is rejected. If the phase hard constraint test is passed, then based on the multi-objective comprehensive evaluation function, the global score before and after transferring the free array element to the candidate driving channel is evaluated. If the score is improved, then the transfer is confirmed. Step 5: Repeat Step 4 until all free array elements fail to meet the phase hard constraint or fail to improve the global score. Output the final mapping relationship between the driving channel and the array elements to complete the optimization.

2. The phased array element optimization method for sound field control in complex acoustic environments according to claim 1, characterized in that, Step one, which determines the amplitude weight of each array element based on the physical parameters of the complex acoustic environment, specifically includes: The three-dimensional physical space is discretized into a computational grid. For each array element, the set of grid points it covers is extracted and spatial downsampling is performed to obtain multiple emission points. For each emission point, a spatial ray path to the target focus is constructed, and the density profile on this path is extracted from the 3D CT image; The density profile is segmented, and the maximum density value is searched in the front and rear segments respectively as the density of the outer and inner plates of the skull. The low quantile value is extracted in the middle segment as the equivalent density of the diploic plate. The transmittance index of a single ray is calculated based on the density of the outer plate, the density of the inner plate, and the equivalent density of the barrier. Calculate the mathematical average of the transmittance indices of all rays corresponding to an array element, and use it as the global equivalent transmittance of that array element. The global equivalent transmittance is converted into the amplitude weight of the array element through a nonlinear mapping function.

3. The phased array element optimization method for sound field control in complex acoustic environments according to claim 2, characterized in that, When converting the global equivalent transmittance into the amplitude weight of the array element using a nonlinear mapping function, high and low thresholds are set: When the global equivalent transmittance is lower than the first preset threshold, the amplitude weight of the array element is set to 0. When the global equivalent transmittance is higher than the second preset threshold, the amplitude weight of the array element is set to 1. When the global equivalent transmittance is between the first preset threshold and the second preset threshold, the amplitude weight is determined using a linear transition method.

4. The phased array element optimization method for sound field control in complex acoustic environments according to claim 1, characterized in that, In step two, the ideal phases of all effective array elements are used as clustering features for clustering, specifically as follows: The ideal phase of each effective array element is mapped to the complex plane to construct a clustering feature vector; The K-means clustering algorithm is used to cluster the entire array into an initial set of subarrays for the target number of driving channels.

5. The phased array element optimization method for sound field control in complex acoustic environments according to claim 1, characterized in that, The multi-objective comprehensive evaluation function constructed in step three has the following specific form: in, This represents the global score of the multi-objective comprehensive evaluation function; This is the sum of the effective sound pressure at the focal point; It is the sum of the variances of the phases of the array elements within the K channels and the central phase; Total amplitude load for K channels The variance between them; α, β, γ are preset constant weighting coefficients.

6. The phased array element optimization method for sound field control in complex acoustic environments according to claim 1, characterized in that, In step four, before extracting the free array elements, a core array element freezing step is also included: Calculate the channel center phase and the average amplitude requirement of the array elements within each driving channel; Set a phase core tolerance limit and an amplitude core tolerance limit; For each element in the driving channel, if the difference between its phase and the phase at the center of the channel is less than the phase core tolerance limit, and the difference between its amplitude and the average amplitude requirement is less than the amplitude core tolerance limit, then the element is determined to be a core anchoring element, and it is frozen in subsequent iterations to not participate in cross-group transfer.

7. The phased array element optimization method for sound field control in complex acoustic environments according to claim 1, characterized in that, In step four, the free array elements are targeted to search for candidate driving channels, specifically as follows: Among all driving channels globally, one or more channels whose average amplitude is closest to the amplitude of the free array element itself are selected as candidate channels.

8. The phased array element optimization method for sound field control in complex acoustic environments according to claim 1, characterized in that, The preset phase tolerance threshold in the phase hard constraint test is λ / 4, where λ is the working wavelength of the ultrasonic phased array.

9. The phased array element optimization method for sound field control in complex acoustic environments according to claim 1, characterized in that, The loop termination condition for the directional multi-objective fine-tuning is: all unfrozen free array elements are unable to perform the transition due to triggering phase hard constraints, or all hypothetical transitions fail to make the global score of the multi-objective comprehensive evaluation function exceed the current score.

10. A phased array element optimization system for sound field control in complex acoustic environments, characterized in that, include: The ideal parameter acquisition module is used to acquire the ideal driving phase and amplitude required for each element in the entire array to achieve focusing at the target focus. The amplitude weight determination module is used to determine the amplitude weight of each array element based on the physical parameters of the complex acoustic environment. The initial dimensionality reduction and clustering module is used to take the ideal phase of all effective array elements as clustering features and use a clustering algorithm to cluster them into a preset number of initial subarray sets, with each subarray corresponding to a driving channel; The evaluation function construction module is used to construct a multi-objective comprehensive evaluation function with the total effective sound pressure at the focal point as the main objective and the differences in intra-group phase difference and inter-channel total amplitude weight load of all driving channels as secondary objectives. The directional multi-objective fine-tuning module is used to extract free array elements based on the homogeneous driving characteristics of the driving channel, perform amplitude targeted search and phase hard constraint verification, and evaluate the transfer gain according to the multi-objective comprehensive evaluation function to determine the final mapping relationship between the driving channel and the array elements.

Images