Requirements assessment method for a dynamic observation experimental platform for ocean nonlinear acoustic scattering sound field
By constructing a meta-model library, we have achieved rapid quantitative correlation and bidirectional evaluation of the performance indicators and physical resources of the marine nonlinear acoustic scattering sound field experimental platform. This solves the problem of quantification under resource constraints in existing technologies and improves design efficiency and scheme optimization capabilities.
Patent Information
- Application Number
- CN202611194194.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-07
- Publication Date
- 2026-11-10
AI Technical Summary
Existing technologies cannot quickly quantify the correlation between the performance indicators of marine nonlinear acoustic scattering sound field experimental platforms and physical resources, and cannot conduct bidirectional evaluation under resource constraints. This leads to designs relying on empirical estimations or high-cost simulations, resulting in long iteration cycles and an inability to support rapid scheme demonstration and iteration.
A meta-model library is constructed, which maps the performance index parameters of functional components to physical resource consumption. Numerical optimization algorithms are used to perform sensitivity analysis and index inverse solution under resource constraints, so as to achieve rapid bidirectional quantitative evaluation of performance index and physical resources.
It enables rapid quantitative correlation and two-way evaluation of performance indicators and physical resources under resource constraints, improves design efficiency, supports rapid solution iteration and optimization, and provides a scientific resource planning tool.
Smart Images

Figure CN123069867A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the technical field of underwater acoustic engineering experimental platform design and simulation evaluation, and specifically relates to a requirement evaluation method for a dynamic observation experimental platform for marine nonlinear acoustic scattering sound field. Background Technology
[0002] The marine nonlinear acoustic scattering sound field dynamic observation and control experimental platform is a comprehensive underwater acoustic experimental device used to study the propagation, scattering, and interaction mechanisms of nonlinear sound waves in the marine environment. This type of platform typically consists of multiple functional modules, including a flow field monitoring subsystem, a sound field emission subsystem, a signal sampling subsystem, and a data processing and control subsystem. These modules are highly coupled, and adjustments to the performance indicators of any module can have a cascading effect on the resource consumption of other modules. During the platform's design phase, designers need to decompose the top-level experimental requirements into specific system functions and performance indicators (such as emission source level, number of acquisition channels, and number of processing passbands). Based on this, they must further quantify and evaluate the demands of these performance indicators on the overall platform resources (such as total power consumption, rack installation space, data bus bandwidth, and real-time computing load) to determine the optimal performance indicator configuration within a limited resource budget.
[0003] Traditional platform design often relies on expert experience-based estimations or complex multiphysics simulations. The former lacks quantitative basis and struggles to handle the coupling relationships between multiple indicators, while the latter is computationally expensive and has a long iteration cycle, making it unsuitable for rapid scheme demonstration and iteration in the early stages. Currently, there is a lack of evaluation tools and methods specifically designed for the complex system characteristics of nonlinear experimental platforms, capable of rapidly correlating performance indicators with physical resources and supporting sensitivity analysis and inverse kinematics under resource constraints.
[0004] Therefore, how to provide a demand assessment method for a dynamic observation experimental platform for ocean nonlinear acoustic scattering that can overcome the shortcomings of existing technologies in rapidly quantifying and correlating performance indicators with physical resources and in conducting bidirectional assessments under resource constraints has become an urgent technical problem to be solved. Summary of the Invention
[0005] This invention provides a requirement assessment method for a dynamic observation experimental platform for marine nonlinear acoustic scattering sound field. It provides assessment tools and methods for the complex nonlinear system characteristics of such platforms, overcoming the shortcomings of existing technologies that cannot quickly quantify and correlate performance indicators with physical resources and cannot conduct bidirectional assessments under resource constraints.
[0006] In one embodiment of the present invention, a requirement assessment method for a dynamic observation experimental platform for ocean nonlinear acoustic scattering sound field is provided, comprising: Obtain the performance index parameter set corresponding to each of the multiple functional components of the experimental platform; Construct multiple meta-models that correspond one-to-one with each of the aforementioned functional components. Each meta-model is used to map the performance index parameters of the corresponding functional component to the amount of physical resources occupied by the functional component. In response to the input current design scheme, the corresponding meta-model is called for each performance index parameter in the current design scheme, the single resource consumption of each functional component is calculated, and a total resource requirement list is generated by summarizing according to resource category. The total resource requirement list is compared with the preset platform resource constraint upper limit to identify bottleneck resources. When a target adjustment amount for at least one key performance indicator (i.e., the main indicator) is received, the first optimization process is executed: the total resource requirement after the adjustment of the main indicator is recalculated based on the corresponding meta-model. If the updated requirement of at least one type of resource exceeds its constraint limit, the reduction task is allocated according to the resource occupancy weight of each meta-model that consumes that type of resource before the adjustment. The numerical optimization algorithm is used to search in reverse within the performance indicator parameter space of each meta-model to find new parameter values that satisfy the reduced resource quota. The new parameter values of each meta-model are summarized and the quantitative value of the impact of the change of the main indicator on other sub-indicators is calculated to complete the sensitivity analysis. When a total constraint change for at least one type of platform resource is received, a second optimization process is executed: resource increment or decrement quotas are allocated according to the resource occupancy weights of each meta-model consuming that type of resource before the change, and the numerical optimization algorithm is used to search backwards in the performance index parameter space of each meta-model for one or more sets of candidate parameter values that satisfy the changed resource quotas, and all candidate parameter values are integrated to evaluate the impact range of the total constraint change on the overall platform performance index.
[0007] Furthermore, multiple meta-models corresponding one-to-one with each of the aforementioned functional components are constructed, including: According to the functional composition of the experimental platform, the meta-model is divided into flow field monitoring, sound field transceiver, signal acquisition and data processing and control. Each category further includes sub-models for at least one resource type among power consumption, spatial volume and bandwidth. Each meta-model is encapsulated in the form of a dynamic link library and registered to the meta-model library. The performance index parameters are derived from the functional performance list obtained by capability-function mapping decomposition.
[0008] Furthermore, each of the aforementioned meta-models follows a unified interface specification, including at least: a computation execution interface, used to receive input parameters, i.e., performance index parameters, encapsulated in a structured data format, perform mapping calculations, and return output results, i.e., resource usage, encapsulated in the same structured data format; and a memory release interface, used to release the memory resources occupied by the results returned by the computation execution interface.
[0009] Furthermore, in the first optimization process, the resource reduction quota is allocated as follows: the excess amount of resources exceeding the constraint limit is distributed according to the proportion of each meta-model consuming that type of resource before adjustment, and the reduction quota allocated to each meta-model is positively correlated with its initial occupancy ratio.
[0010] Furthermore, the numerical optimization algorithm performs the search in the following manner: a loss function is constructed based on the difference between the resource occupancy output by the meta-model and the target resource quota allocated to the meta-model; the loss function is converged to below a preset threshold by iteratively adjusting the input performance index parameter values of the meta-model; and the input parameter values corresponding to the convergence are taken as candidate solutions that satisfy the target resource quota.
[0011] Furthermore, when the performance index parameter space corresponding to the meta-model is a non-convex space, the numerical optimization algorithm adopts a strategy of performing gradient descent iterations after multiple random initializations, and selects the set that satisfies the resource quota constraint and has the best comprehensive performance index from the multiple candidate solutions obtained as the final output.
[0012] Furthermore, the input performance parameters of the flow field monitoring meta-model include measurement accuracy and image resolution, and the output resource occupancy includes at least the space occupied by the monitoring module; the input performance parameters of the sound field transceiver meta-model include the transmitting sound source level and the number of transmitting array elements, and the output resource occupancy includes at least the power consumption of the transmitter cabinet and the space occupied by the power amplifier module.
[0013] Furthermore, the input performance parameters of the signal acquisition meta-model include the number of acquisition channels and the sampling rate, and the output resource occupancy includes at least the power consumption of the acquisition system and the data storage bandwidth; the input performance parameters of the data processing and control meta-model include the number of processing passbands and the number of floating-point operations per second, and the output resource occupancy includes at least the power consumption of the computing server and the required memory capacity.
[0014] Furthermore, in the first optimization process, the primary indicator is the number of acquisition channels, and the secondary indicators include the sampling rate and the number of quantization bits; the calculation method for the influence of the quantization value is as follows: calculate the change and rate of change of each secondary indicator before and after the adjustment of the primary indicator, and output the sensitivity correlation curve between the primary indicator and each secondary indicator in a graphical manner.
[0015] In another embodiment of the present invention, a system-level requirements assessment device for an experimental platform for dynamic observation and control of marine nonlinear acoustic scattering sound field is provided, comprising: One or more processors; and a storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the requirement assessment method for a dynamic observation experimental platform for ocean nonlinear acoustic scattering sound field as described above.
[0016] The beneficial effects of this invention are as follows: As can be seen from the above scheme, the embodiments of the present invention provide a method and system for demand assessment of a dynamic observation experimental platform for marine nonlinear acoustic scattering sound field, belonging to the technical field of underwater acoustic engineering experimental platform design and simulation evaluation. The method includes: constructing a meta-model library for the experimental platform, where each meta-model maps the performance index parameters of corresponding functional components to the occupancy of at least one type of physical resource; responding to the input current design scheme, calling each meta-model to calculate the occupancy of individual resources and summarizing them by resource category to generate a total resource demand list, comparing it with a preset resource constraint upper limit to identify bottleneck resources; when adjusting key performance indicators, executing a first optimization process, allocating reduction quotas according to the initial occupancy ratio of each meta-model based on the principle of limited resources, and using numerical optimization algorithms to back-search for the optimal parameter combination that satisfies the new quota, completing sensitivity analysis; when platform resource constraints change, executing a second optimization process, allocating resource increment or decrement quotas according to the initial occupancy ratio of each meta-model, and back-solving the feasible value range of performance indicators. The present invention can achieve rapid bidirectional quantitative evaluation and iterative optimization of performance indicators and physical resources under resource constraints. Attached Figure Description
[0017] Figure 1 This is a flowchart illustrating the requirements assessment method for a dynamic observation experimental platform for marine nonlinear acoustic scattering sound field, as described in an embodiment of the present invention. Detailed Implementation
[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.
[0019] Meta-models are computational models used to calculate resource requirements in rapid system-level requirement assessments. Currently, four types of meta-models have been constructed based on the main functional components of the marine nonlinear acoustic scattering sound field dynamic observation and control experimental platform: process monitoring meta-model, sound field emission meta-model, signal acquisition meta-model, and data processing and control meta-model. These meta-models can be used to calculate the resource consumption requirements of performance indicators under the capability requirements of the marine nonlinear acoustic scattering sound field dynamic observation and control experimental platform.
[0020] Sensitivity is defined as: for any performance index A, calculating the impact of changes in A on other indices, where A represents a certain index, and RA represents the resource requirement corresponding to that index. The currently designed sensitivity analysis algorithm is based on a meta-model. Through the meta-model, we can obtain the resource requirement corresponding to a certain performance index. A key principle here is the finite nature of total resources. When one index increases, the resources it requires increase, which naturally may lead to a decrease in the resources available for another index, thus causing the index to decrease. When index A increases, its resource requirements increase, leading to a decrease in index B, which also requires the same resource.
[0021] like Figure 1 As shown, Figure 1 This is a flowchart illustrating the requirements assessment method for a dynamic observation experimental platform for marine nonlinear acoustic scattering sound field, as described in an embodiment of the present invention.
[0022] Figure 1 A requirement assessment method for a dynamic observation experimental platform for ocean nonlinear acoustic scattering sound field includes: Obtain the performance index parameter set corresponding to each of the multiple functional components of the experimental platform; Construct multiple meta-models that correspond one-to-one with each of the aforementioned functional components. Each meta-model is used to map the performance index parameters of the corresponding functional component to the amount of physical resources occupied by the functional component. In response to the input current design scheme, the corresponding meta-model is called for each performance index parameter in the current design scheme, the single resource consumption of each functional component is calculated, and a total resource requirement list is generated by summarizing according to resource category. The total resource requirement list is compared with the preset platform resource constraint upper limit to identify bottleneck resources. When a target adjustment amount for at least one key performance indicator (i.e., the main indicator) is received, the first optimization process is executed: the total resource requirement after the adjustment of the main indicator is recalculated based on the corresponding meta-model. If the updated requirement of at least one type of resource exceeds its constraint limit, the reduction task is allocated according to the resource occupancy weight of each meta-model that consumes that type of resource before the adjustment. The numerical optimization algorithm is used to search in reverse within the performance indicator parameter space of each meta-model to find new parameter values that satisfy the reduced resource quota. The new parameter values of each meta-model are summarized and the quantitative value of the impact of the change of the main indicator on other sub-indicators is calculated to complete the sensitivity analysis. When a total constraint change for at least one type of platform resource is received, a second optimization process is executed: resource increment or decrement quotas are allocated according to the resource occupancy weights of each meta-model consuming that type of resource before the change, and the numerical optimization algorithm is used to search backwards in the performance index parameter space of each meta-model for one or more sets of candidate parameter values that satisfy the changed resource quotas, and all candidate parameter values are integrated to evaluate the impact range of the total constraint change on the overall platform performance index.
[0023] In this embodiment of the invention, a method for demand assessment of a dynamic observation experimental platform for marine nonlinear acoustic scattering sound fields is proposed. This method constructs a meta-model library that encapsulates the mathematical relationships between performance indicators and resource consumption, transforming subjective estimations that previously relied on personal experience into objective calculations based on mathematical models. By inputting the performance indicator parameters from the design scheme into the meta-model, the total resource requirements of various platforms for the scheme can be quickly obtained, and compared with preset resource constraints to identify bottleneck resources.
[0024] This invention, for the first time, applies meta-modeling technology to the quantitative mapping of "performance indicators → physical resources," filling a gap in existing meta-modeling technologies in this direction. It simultaneously possesses forward solution (resource demand analysis) and backward optimization (sensitivity analysis + indicator inverse solution) capabilities, whereas existing technologies can only process in one direction. The meta-model can encapsulate mathematical relationships of any form (linear, nonlinear, piecewise functions, etc.), naturally supporting resource assessment for nonlinear systems and overcoming the linear assumptions of linear programming methods. It is specifically designed for the complex system characteristics of an experimental platform for dynamic observation and control of marine nonlinear acoustic scattering sound fields.
[0025] In another embodiment of the present invention, constructing multiple meta-models corresponding one-to-one with each of the functional components includes: According to the functional composition of the experimental platform, the meta-model is divided into flow field monitoring, sound field transceiver, signal acquisition and data processing and control. Each category further includes sub-models for at least one resource type among power consumption, spatial volume and bandwidth. Each meta-model is encapsulated in the form of a dynamic link library and registered to the meta-model library. The performance index parameters are derived from the functional performance list obtained by capability-function mapping decomposition.
[0026] Among them, the four types of meta-models correspond one-to-one with the actual functional modules (flow field monitoring, sound field transmission and reception, signal acquisition, and data processing) of the marine nonlinear acoustic scattering experimental platform, making the evaluation results more valuable for engineering reference. Each type can be further subdivided into sub-models for different resource types (power consumption, space, bandwidth, etc.), forming a hierarchical meta-model management system. The models are encapsulated in the form of dynamic link libraries and registered in the meta-model library, ensuring the reusability and manageability of the meta-models.
[0027] In another embodiment of the present invention, each of the meta-models follows a unified interface specification, including at least: a computation execution interface, used to receive input parameters, i.e. performance index parameters, encapsulated in a structured data format, perform mapping calculations, and return output results, i.e. resource usage, encapsulated in the same structured data format; and a memory release interface, used to release the memory resources occupied by the results returned by the computation execution interface.
[0028] In this embodiment of the invention, a unified interface specification enables seamless integration of meta-models written by different categories and developers into the evaluation system without additional adaptation. A dedicated memory release interface prevents memory leaks during dynamic library calls, ensuring the stability of long-term, large-scale evaluation calculations. Encapsulating input and output in a structured data format (such as JSON) ensures the universality and parsability of data exchange.
[0029] In another embodiment of the present invention, in the first optimization process, the resource reduction quota is allocated as follows: the excess amount of resources exceeding the constraint limit is distributed according to the proportion of each meta-model consuming such resources before adjustment, and the reduction quota allocated to each meta-model is positively correlated with its initial proportion of occupancy.
[0030] Existing resource management methods and sensitivity analysis techniques typically employ either globally unified reductions or manually specified reduction amounts when resources exceed limits, lacking a refined allocation mechanism based on the actual resource occupancy ratio of each functional module. In this invention, reduction quotas are allocated according to the initial occupancy ratio of each meta-model, reflecting the fair principle that those who occupy more resources bear a greater responsibility for reductions. The calculation and allocation of reduction quotas are entirely automated by the system, requiring no manual intervention, significantly improving evaluation efficiency. The allocation logic is transparent, allowing designers to clearly trace the basis for allocating specific reduction amounts to each meta-model.
[0031] In another embodiment of the present invention, the numerical optimization algorithm performs the search as follows: a loss function is constructed based on the difference between the resource occupancy output by the meta-model and the target resource quota allocated to the meta-model; the input performance index parameters of the meta-model are iteratively adjusted until the loss function converges to below a preset threshold; and the input parameter values corresponding to the convergence are taken as candidate solutions that satisfy the target resource quota. Here, the resource quota constraint is transformed into an optimization problem of the loss function, and the precise parameter values that satisfy the constraint are obtained through iterative solution, rather than a rough estimate. The algorithm automatically executes iterations until the loss function converges to the preset threshold, without the need for manual adjustments.
[0032] In another embodiment of the present invention, when the performance index parameter space corresponding to the meta-model is a non-convex space, the numerical optimization algorithm adopts a strategy of performing gradient descent iterations after multiple random initializations, and selects the set that satisfies the resource quota constraint and has the best comprehensive performance index from the multiple candidate solutions obtained as the final output.
[0033] By performing multiple random initializations, the gradient descent method effectively avoids getting trapped in local optima in non-convex spaces. Selecting a set of solutions from multiple candidate solutions that satisfies resource quota constraints and has the best overall performance ensures the quality of the optimization results. Even with a complex performance index space in the meta-model, this strategy can still achieve reliable optimization results.
[0034] In another embodiment of the present invention, the input performance parameters of the flow field monitoring meta-model include measurement accuracy and image resolution, and the output resource occupancy includes at least the space occupied by the monitoring module; the input performance parameters of the sound field transceiver meta-model include the transmitting sound source level and the number of transmitting array elements, and the output resource occupancy includes at least the power consumption of the transmitter cabinet and the space occupied by the power amplifier module.
[0035] In another embodiment of the present invention, the input performance parameters of the signal acquisition meta-model include the number of acquisition channels and the sampling rate, and the output resource occupancy includes at least the power consumption of the acquisition system and the data storage bandwidth; the input performance parameters of the data processing and control meta-model include the number of processing passbands and the number of floating-point operations per second, and the output resource occupancy includes at least the power consumption of the computing server and the required memory capacity.
[0036] This invention covers all four major functional modules of the experimental platform (flow field monitoring, sound field transmission and reception, signal acquisition, and data processing), forming a complete evaluation system. It accurately identifies key performance indicators affecting resource consumption in the signal acquisition and data processing functions, such as the number of acquisition channels, sampling rate, processing passbands, and computing power, making the evaluation targeted. The output parameters (power consumption, bandwidth, and memory capacity) are highly matched to the actual resource consumption characteristics of these functional modules.
[0037] In another embodiment of the present invention, during the first optimization process, the primary indicator is the number of acquisition channels, and the secondary indicators include the sampling rate and the number of quantization bits; the method for calculating the influence of the quantization value is as follows: calculate the change and rate of change of each secondary indicator before and after the adjustment of the primary indicator, and output the sensitivity correlation curve between the primary indicator and each secondary indicator in a graphical manner.
[0038] In another embodiment of the present invention, a system-level requirements assessment device for an experimental platform for dynamic observation and control of marine nonlinear acoustic scattering sound field is provided, comprising: One or more processors; and a storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the requirement assessment method for a dynamic observation experimental platform for ocean nonlinear acoustic scattering sound field as described above.
[0039] In another embodiment of the present invention, a method for rapid evaluation of system-level requirements of a nonlinear experimental platform is provided. This method constructs a "meta-model" characterizing the mathematical relationship between platform performance indicators and resource consumption. Based on a meta-model library, it achieves rapid calculation of resource requirements from performance indicators, sensitivity analysis of performance indicators under resource constraints, and inverse solution of performance indicators. This provides a scientific and efficient quantitative analysis tool for resource planning and performance trade-offs in nonlinear experimental platforms.
[0040] A rapid method for evaluating system-level requirements of a nonlinear experimental platform includes the following steps: S1: Construct a meta-model library for nonlinear experimental platforms. The meta-model is a computable model that encapsulates the mathematical relationships between the performance indicators of a specific system function and the required platform resources, integrated in the form of a standardized dynamic link library. The performance indicators are derived from a list of functional performance obtained through capability-function mapping decomposition. The platform resources include, but are not limited to, total platform power consumption, rack installation space and load-bearing capacity, data bus bandwidth, real-time computing load, system synchronization accuracy margin, and maximum continuous test duration. S2: Resource requirement calculation based on meta-model. For a platform design scheme with specific performance index values already decomposed, a corresponding meta-model is matched for each performance index; the amount of various resources consumed by each performance index is calculated by calling the meta-model, and summarized by resource type to obtain the total resource requirement list under the design scheme; this list is compared with the preset platform resource constraints to identify resource bottlenecks; S3: Sensitivity analysis of performance metrics based on resource constraints. When adjusting one or more key performance metrics (primary metrics), perform the following sub-steps: S31: Recalculate the total resource requirements after adjusting the main indicators based on the meta-model; S32: Traverse all resource types. If the demand for a resource exceeds its constraint limit, proceed to the resource reallocation process: First, identify all meta-models consuming the excess resource; second, determine the "resource reduction quota" that each meta-model must bear based on its initial resource occupancy ratio; finally, for each meta-model that needs resource reduction, use a numerical optimization algorithm (gradient descent method) to search its performance index space to find the new performance index value under the reduced resource quota. S33: Summarize all the adjusted performance index values of the meta-model to form a new set of performance indices, and calculate the degree of influence of the main index change on other indices (sub-indices), that is, complete the sensitivity analysis. S4: Inverse Derivation of Performance Metrics Based on Resource Constraints. When a certain total resource constraint of the platform changes, the following sub-steps are executed to inversely deduce the possible range of values for the performance metric: S41: Calculate the adjustment amount for this resource; S42: Identify all meta-models that consume this resource, and allocate the resource adjustment amount to each relevant meta-model according to the initial resource usage ratio; S43: For each meta-model, a numerical optimization algorithm is used to solve one or more sets of performance index values that meet the new resource quota obtained. S44: Integrate the performance index values obtained from the inverse solution of all meta-models to evaluate the impact of changes in resource constraints on the overall performance of the platform.
[0041] In step S1, the meta-model library includes the following categories: flow field monitoring meta-model, nonlinear parameter acquisition meta-model, sound field transceiver meta-model, data acquisition and observation meta-model, and data processing and control meta-model. Each meta-model category can be further subdivided into sub-models for different resource types. For example, for the sound field transceiver function, a "transmission system power consumption meta-model" and a "transmission system space occupancy meta-model" can be constructed.
[0042] Preferably, in step S1, all meta-models follow a unified interface specification, including at least: an external call interface modelExecute, used to receive input parameters (performance metrics) encapsulated in JSON format, perform calculations, and return output results (resource requirements) encapsulated in JSON format; and a memory release interface freeMemory, used to release the memory occupied by the results returned by the modelExecute interface.
[0043] Preferably, in steps S32 and S43, the numerical optimization algorithm is the gradient descent method, which iteratively adjusts the input performance index value of the meta-model so that its output resource demand value approaches the target resource quota.
[0044] In another embodiment of the present invention, a method for rapid evaluation of system-level requirements of a nonlinear experimental platform is provided. Its core implementation relies on the construction of a meta-model library, and on this basis, three core tasks are carried out: resource requirement calculation, sensitivity analysis, and performance index inverse solution.
[0045] The first step: Constructing a meta-model library for the experimental platform for dynamic observation and control of marine nonlinear acoustic scattering sound field serves as a bridge connecting "functional performance" and "experimental platform resources." Each meta-model corresponds to a specific system function (such as the "multi-channel synchronous acquisition function" under the signal acquisition meta-model), and its inputs are several key performance indicators of that function, while its output is the resource requirements of that function for certain platforms.
[0046] Define Meta-model Categories and Examples: For the experimental platform for dynamic observation and control of marine nonlinear acoustic scattering sound fields, we construct four categories of meta-models: Flow Field Monitoring Meta-model: Input indicators may include "measurement accuracy," "image resolution," and "measurement duration," etc.; output resources may include "transmitter cabinet power consumption (W)," "monitoring module space occupied (U)," and "total weight of monitoring system (kg)." Sound Field Emission Meta-model: Input indicators may include "emission source level (dB)," "emission signal bandwidth (Hz)," and "number of transmission array elements," etc.; output resources may include "transmitter cabinet power consumption (W)," "power amplifier module space occupied (U)," and "total weight of transmission system (kg)." Signal Acquisition Meta-model: Input indicators may include "number of acquisition channels," "sampling rate (Hz)," "quantization bits (bit)," and "synchronization accuracy (ns)," etc.; output resources may include "acquisition system power consumption (W)," "data storage bandwidth (GB / s)," and "cabinet space (U)." Data processing metamodel: Input metrics may include "processing passband count", "floating-point operations per second (FLOPS)", "processing latency (ms)", etc.; output resources may include "computing server power consumption (W)", "required memory capacity (GB)" and "number of computing nodes".
[0047] Implement standardized metamodel interfaces: All metamodels are encapsulated as C++ dynamic libraries and implement two standard external interfaces: char modelExecute(const char inputJson): The platform calls this interface, passing in a JSON string containing performance metric values: {"CHANNEL_NUM": 64,"SAMPLE_RATE": 1e6, "SYNC_ACCURACY": 10}. After internal calculation by the metamodel, it returns a JSON string containing resource requirement results: {"POWER": 350.5,"SPACE":4.2}. void freeMemory(char ptr): Used to release the memory occupied by the results returned by modelExecute.
[0048] Metamodel entry management: Upload the compiled dynamic library file to the platform metamodel library, and configure its input parameters (corresponding performance indicators) and output parameters (corresponding resource types) with information such as identifiers, names, and units to complete the registration of the metamodel.
[0049] Step 2: Resource requirement calculation based on meta-model. In the platform solution design software, the user has obtained a list of functional performance containing specific values through capability-function mapping (e.g., emission source level = 220 dB, number of acquisition channels = 256, number of real-time processing passbands = 32, etc.).
[0050] Associate Meta-Models: For each metric in the performance list, select an appropriate meta-model for association. For example, the metrics "Number of Acquisition Channels" and "Sampling Rate" are associated with "Signal Acquisition Meta-Model_Power Consumption Sub-Model" and "Signal Acquisition Meta-Model_Spatial Sub-Model," respectively. Set Resource Constraints: Set the overall resource constraint limits for the platform in the software, such as: total power consumption ≤ 5000W, total rack space ≤ 42U, total weight ≤ 2000kg, etc. Perform Calculation: When the user clicks the "Resource Calculation" button, the software automatically iterates through all associated meta-models, calls their modelExecute interfaces, inputs the current performance metric values, obtains the resource requirements for each item, and categorizes and summarizes them. Result Presentation and Alarms: The software displays the total resource requirements in a table or dashboard format and compares them with the preset constraints. If the requirement for a certain resource (such as power consumption) exceeds the constraint value, an alarm will be triggered for that resource requirement data, indicating that there is a resource bottleneck in the design.
[0051] Step 3: Sensitivity analysis of performance indicators based on resource constraints. When designers want to understand the impact of improving a key indicator (such as increasing the number of acquisition channels from 256 to 512), they can perform sensitivity analysis.
[0052] Triggered Analysis: The user adjusts the "Number of Acquisition Channels" (main indicator) from 256 to 512 by dragging the corresponding indicator slider to the desired value or directly inputting the desired value. Resource Recalculation: The software calls the relevant acquisition meta-models to calculate the new power consumption and space requirements after the increase in the number of channels. Assume the calculation finds that the total power consumption has increased from 4800W to 5800W, exceeding the 5000W constraint. Resource Reallocation and Indicator Backpropagation: The system identifies that the "power consumption" resource exceeds the limit. It identifies all meta-models consuming "power consumption" resources, assuming they include: Acquisition Meta-model_A (current power consumption 3500W), Processing Meta-model_B (current power consumption 1000W), and Transmission Meta-model_C (current power consumption 300W). The system needs to reduce power consumption by 800W (5800-5000). Based on the "limited resource principle" and the initial ratio, the reduction quota that each meta-model should bear is calculated. For example, according to the initial power consumption ratio, Acquisition Meta-model_A needs to reduce (3500 / 4800). 800 ≈ 583W, processing the meta-model_B requires a reduction of (1000 / 4800). 800 ≈ 167W, the emitter meta-model_C needs to be reduced by (300 / 4800) 800 ≈ 50W. For each metamodel, gradient descent is used to search its performance metric space for a combination of performance metrics that meets the new power consumption quota (e.g., the new quota for acquisition metamodel_A is 3500-583=2917W). For example, to reduce power consumption, acquisition metamodel_A might automatically lower its associated "sampling rate" or "quantization bits". All adjusted performance metric values for all metamodels are then aggregated to form a new performance metric scheme that meets the total resource constraints.
[0053] Output sensitivity results: The software compares the performance index schemes before and after the adjustment, calculates the specific values that the increase in the number of acquisition channels leads to the forced decrease in other indicators such as the sampling rate and quantization bits, and displays this impact relationship in the form of charts, thereby quantitatively revealing the trade-off relationship between indicators.
[0054] Step 4: Inverse analysis of performance metrics based on resource constraints. When the platform resource budget changes (e.g., the overall height of the experimental platform is increased, and the space resources of a single rack increase from 42U to 48U), the designers want to know how the performance can be improved.
[0055] Triggering Reverse Engineering: The user adjusts the "Rack Space" constraint value from 42U to 48U in the software. Allocating New Resources: The system identifies all meta-models consuming "Rack Space" (such as transmit, acquisition, and processing meta-models) and allocates the additional 6U of space according to their current space occupancy ratio. Reverse Solving Performance Metrics: For each meta-model that receives an additional space quota, gradient descent is used again to reverse-engineer the associated performance metrics (such as "number of transmit elements" and "number of processing passbands") under the increased resources. Outputting Reverse Engineering Results: The software displays the theoretically achievable new range of values for each performance metric under the new 48U space constraint, providing a clear direction for design optimization.
[0056] By applying the above four steps in a cyclical manner, designers can quickly and quantitatively evaluate the impact of different design choices on resources during the scheme design phase of nonlinear experimental platforms, and find the optimal balance point under strict resource constraints, significantly improving design efficiency and scheme scientificity.
[0057] This invention's technical solution encapsulates complex physical relationships into "black box" functions through a pre-built meta-model library, enabling rapid conversion of performance indicators into resource requirements and significantly improving the speed of solution iteration and trade-off analysis. The provided sensitivity analysis function clearly and quantitatively demonstrates the cascading impact of any performance indicator change on all other relevant platform indicators, allowing designers to intuitively understand the coupling and competitive relationships between indicators. The performance indicator inverse derivation function supports deriving feasible performance indicator ranges from resource constraints, providing a direct tool for optimizing platform capabilities within given resource boundaries.
[0058] Through the two analyses described above, bottlenecks and irrationalities in resource allocation can be identified early in the design process, guiding designers to scientifically adjust performance indicators or re-plan resources to achieve optimal platform resource allocation. The methodology developed in this invention, centered on a meta-model and incorporating resource constraints for both forward and reverse analysis, is not only applicable to underwater acoustic nonlinear experimental platforms but can also be extended to the early conceptual design and requirements demonstration stages of other complex experimental systems or equipment after domain adaptation.
[0059] This invention provides a method and system for demand assessment of a dynamic observation experimental platform for marine nonlinear acoustic scattering sound fields, belonging to the technical field of underwater acoustic engineering experimental platform design and simulation evaluation. The method includes: constructing a meta-model library for the experimental platform, where each meta-model maps the performance index parameters of corresponding functional components to the occupancy of at least one type of physical resource; responding to the input current design scheme, calling each meta-model to calculate the occupancy of individual resources and summarizing them by resource category to generate a total resource demand list, comparing it with a preset resource constraint upper limit to identify bottleneck resources; when adjusting key performance indicators, executing a first optimization process, allocating reduction quotas according to the initial occupancy ratio of each meta-model based on the principle of limited resources, and using numerical optimization algorithms to back-search for the optimal parameter combination that satisfies the new quota, completing sensitivity analysis; when platform resource constraints change, executing a second optimization process, allocating resource increment or decrement quotas according to the initial occupancy ratio of each meta-model, and back-solving the feasible value range of performance indicators. The present invention provides a requirement assessment method for a dynamic observation experimental platform for marine nonlinear acoustic scattering sound field. This method is designed to assess the complex nonlinear system characteristics of such platforms, overcoming the shortcomings of existing technologies that cannot quickly quantify and correlate performance indicators with physical resources and cannot conduct bidirectional assessments under resource constraints.
[0060] The above are preferred embodiments of the present invention. It should be noted that, for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.
Claims
1. A requirement assessment method for a dynamic observation experimental platform for marine nonlinear acoustic scattering sound field, characterized in that, The method includes: Obtain the performance index parameter set corresponding to each of the multiple functional components of the experimental platform; Construct multiple meta-models that correspond one-to-one with each of the aforementioned functional components. Each meta-model is used to map the performance index parameters of the corresponding functional component to the amount of physical resources occupied by the functional component. In response to the input current design scheme, the corresponding meta-model is called for each performance index parameter in the current design scheme, the single resource consumption of each functional component is calculated, and a total resource requirement list is generated by summarizing according to resource category. The total resource requirement list is compared with the preset platform resource constraint upper limit to identify bottleneck resources. When a target adjustment amount for at least one key performance indicator (i.e., the main indicator) is received, the first optimization process is executed: the total resource requirement after the adjustment of the main indicator is recalculated based on the corresponding meta-model. If the updated requirement of at least one type of resource exceeds its constraint limit, the reduction task is allocated according to the resource occupancy weight of each meta-model that consumes that type of resource before the adjustment. The numerical optimization algorithm is used to search in reverse within the performance indicator parameter space of each meta-model to find new parameter values that satisfy the reduced resource quota. The new parameter values of each meta-model are summarized and the quantitative value of the impact of the change of the main indicator on other sub-indicators is calculated to complete the sensitivity analysis. When a total constraint change for at least one type of platform resource is received, a second optimization process is executed: resource increment or decrement quotas are allocated according to the resource occupancy weights of each meta-model consuming that type of resource before the change, and the numerical optimization algorithm is used to search backwards in the performance index parameter space of each meta-model for one or more sets of candidate parameter values that satisfy the changed resource quotas, and all candidate parameter values are integrated to evaluate the impact range of the total constraint change on the overall platform performance index.
2. The requirement assessment method for a dynamic observation experimental platform for marine nonlinear acoustic scattering sound field according to claim 1, characterized in that, Construct multiple meta-models corresponding one-to-one with each of the aforementioned functional components, including: According to the functional composition of the experimental platform, the meta-model is divided into flow field monitoring, sound field transceiver, signal acquisition and data processing and control. Each category further includes sub-models for at least one resource type among power consumption, spatial volume and bandwidth. Each meta-model is encapsulated in the form of a dynamic link library and registered to the meta-model library. The performance index parameters are derived from the functional performance list obtained by capability-function mapping decomposition.
3. The requirement assessment method for a dynamic observation experimental platform for marine nonlinear acoustic scattering sound field according to claim 1, characterized in that, Each of the aforementioned meta-models follows a unified interface specification, including at least: a computation execution interface, used to receive input parameters (i.e., performance index parameters) encapsulated in a structured data format, perform mapping calculations, and return output results (i.e., resource usage) encapsulated in the same structured data format; and a memory release interface, used to release the memory resources occupied by the results returned by the computation execution interface.
4. The requirement assessment method for a dynamic observation experimental platform for marine nonlinear acoustic scattering sound field according to claim 1, characterized in that, In the first optimization process, the resource reduction quota is allocated as follows: the excess amount of resources exceeding the constraint limit is distributed according to the proportion of each meta-model that consumes this type of resource before adjustment. The reduction quota allocated to each meta-model is positively correlated with its initial proportion of occupancy.
5. The requirement assessment method for a dynamic observation experimental platform for marine nonlinear acoustic scattering sound field according to claim 1, characterized in that, The numerical optimization algorithm performs the search as follows: a loss function is constructed based on the difference between the resource occupancy output by the meta-model and the target resource quota allocated to the meta-model. The loss function is converged to below a preset threshold by iteratively adjusting the input performance index parameter values of the meta-model. The input parameter values corresponding to the convergence are taken as candidate solutions that satisfy the target resource quota.
6. The requirement assessment method for a dynamic observation experimental platform for marine nonlinear acoustic scattering sound field according to claim 5, characterized in that, When the performance index parameter space corresponding to the meta-model is a non-convex space, the numerical optimization algorithm adopts a strategy of performing gradient descent iterations after multiple random initializations, and selects the set that satisfies the resource quota constraint and has the best comprehensive performance index from the multiple candidate solutions obtained as the final output.
7. The requirement assessment method for a dynamic observation experimental platform for marine nonlinear acoustic scattering sound field according to claim 1, characterized in that, The input performance parameters of the flow field monitoring meta-model include measurement accuracy and image resolution, and the output resource occupancy includes at least the space occupied by the monitoring module; the input performance parameters of the sound field transceiver meta-model include the transmitting sound source level and the number of transmitting array elements, and the output resource occupancy includes at least the power consumption of the transmitter cabinet and the space occupied by the power amplifier module.
8. The requirement assessment method for a dynamic observation experimental platform for marine nonlinear acoustic scattering sound field according to claim 1, characterized in that, The input performance parameters of the signal acquisition meta-model include the number of acquisition channels and the sampling rate, and the output resource usage includes at least the power consumption of the acquisition system and the data storage bandwidth; the input performance parameters of the data processing and control meta-model include the number of processing passbands and the number of floating-point operations per second, and the output resource usage includes at least the power consumption of the computing server and the required memory capacity.
9. The requirement assessment method for a dynamic observation experimental platform for marine nonlinear acoustic scattering sound field according to claim 1, characterized in that, In the first optimization process, the primary indicator is the number of acquisition channels, and the secondary indicators include the sampling rate and the number of quantization bits. The calculation method for the influence of the quantization value is as follows: calculate the change and rate of change of each secondary indicator before and after the adjustment of the primary indicator, and output the sensitivity correlation curve between the primary indicator and each secondary indicator in a graphical manner.
10. A system-level requirements assessment device for an experimental platform for dynamic observation and control of marine nonlinear acoustic scattering sound field, characterized in that, include: One or more processors; and storage devices for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the requirements assessment method for a dynamic observation experimental platform for marine nonlinear acoustic scattering sound field as described in any one of claims 1 to 9.
