Public parameter extraction and localization parameter adaptation method based on layered architecture

Abstract

The application discloses a public parameter extraction and localized parameter adaptation method based on a layered architecture, comprising the following steps: collecting multi-source heterogeneous scene original parameters through a global parameter fusion intelligent analysis engine, mining multi-level parameter correlation characteristics through a layered nested parameter perception model, completing parameter cross-architecture adaptation by a cross-architecture parameter migration adaptation model, generating an intermediate parameter set by using a heterogeneous parameter fusion deduction algorithm, extracting a public parameter subset based on hierarchical distribution rules and general characteristics, and finally performing localized adaptation adjustment for a target scene. The application optimizes hardware multi-core heterogeneous architecture and software dynamic parameter configuration, constructs a hierarchical and full-process parameter processing system, improves parameter perception depth, migration adaptation accuracy and intelligent level of fusion extraction, effectively adapts to different application architectures and scene requirements, and guarantees parameter reuse efficiency and system operation stability.

Description

Technical Field

[0001] This invention relates to the field of parameter extraction and parameter adaptation technology, and in particular to a method for extracting common parameters and adapting localized parameters based on a hierarchical architecture. Background Technology

[0002] With the increasing prevalence of multi-source heterogeneous application scenarios, parameter types and complex relationships exist across different architectures. The universal extraction of common parameters and precise adaptation to localized scenarios have become critical requirements for ensuring stable system operation. Various industries are continuously demanding higher efficiency and compatibility in parameter processing, necessitating a technical solution that can transcend different application architectures, integrate heterogeneous parameter resources, and balance common attributes with localized needs. This solution aims to address issues such as low parameter reuse rates and delayed adaptation adjustments across multiple scenarios, supporting the collaborative operation and performance optimization of various systems.

[0003] Currently, parameters to be processed are typically obtained through a single-dimensional parameter filtering mechanism, cross-scenario migration is achieved using basic parameter mapping rules, heterogeneous parameters are processed using a simple weighted fusion method, common parameters are separated using fixed extraction criteria, and finally localization is completed based on preset adaptation parameters. Some technologies introduce basic parameter analysis modules to assist in parameter classification and filtering, but overall, there is a lack of hierarchical parameter processing logic. A systematic technical system covering the entire process from acquisition, perception, migration, fusion to extraction and adaptation has not been formed, and the technical correlation and synergy between the various stages are insufficient.

[0004] Existing technologies suffer from two core drawbacks: First, the targeting of parameter awareness and migration adaptation is insufficient. A multi-level nested awareness mechanism has not been established, making it difficult to deeply mine the correlation characteristics of parameters at different levels. Furthermore, the differences between the source and target architectures are not fully considered during cross-architecture migration, resulting in poor adaptation effects after parameter migration and an inability to quickly adapt to the operational needs of different scenarios. Second, the intelligence level of common parameter extraction and localization adaptation is low. The extraction process does not dynamically adjust based on the hierarchical distribution patterns and general characteristics of parameters, and the adaptation process lacks flexible adjustment strategies. It is difficult to optimize parameter configuration according to the real-time needs of the target scenario, and a full-process intelligent control mechanism has not been formed, affecting the overall efficiency and accuracy of parameter processing. Summary of the Invention

[0005] To overcome the shortcomings and deficiencies of existing technologies, this invention provides a method for extracting common parameters and adapting localized parameters based on a layered architecture.

[0006] The technical solution adopted in this invention is a method for extracting common parameters and adapting localized parameters based on a hierarchical architecture, comprising the following steps: S1, using a global parameter fusion intelligent analysis engine to perform hierarchical perception and acquisition of the original parameter set in a multi-source heterogeneous scenario, and performing preliminary classification and screening based on parameter attribute features to obtain a cluster of parameters to be processed with hierarchical correlation characteristics; S2, using a hierarchical nested parameter perception model to perform multi-level nested feature mining on the cluster of parameters to be processed, and identifying the correlation mapping relationship between parameters at different levels by dynamically adjusting the perception dimension and hierarchical depth; S3, using a cross-architecture parameter migration and adaptation model to refine the correlation mapping relationship. The parameters are then migrated across different application architectures to establish a rule system for parameter adaptation. S4: A heterogeneous parameter fusion and deduction algorithm is used to perform multi-dimensional fusion calculations on the migrated and adapted parameters to generate a fused intermediate parameter set. S5: Based on the intermediate parameter set, a common parameter extraction operation is performed. Combining the general characteristics and hierarchical distribution patterns of the parameters, a subset of universally applicable common parameters is obtained. S6: The subset of common parameters is localized for adaptation and adjustment. Based on the architectural characteristics and operational requirements of the target application scenario, the common parameters are optimized for scenario-based adaptation to form a localized parameter configuration result adapted to the target scenario.

[0007] Furthermore, the expression for the hierarchical nested parameter-aware model includes:

[0008]

[0009] The perceived output value of the j-th parameter at level 1. Let be the perceptual weight coefficient of the i-th level. For hierarchical parameter-aware activation functions, The feature scaling factor for the j-th original parameter. The collected value of the j-th original parameter. Let be the correlation coefficient between the i-th level and the level above it. For the first The perceived output value of the j-th parameter at each level. Let be the perceptual bias correction value for the i-th level. The result is a multi-level parameter fusion sensing result. Let be the fusion weight of the parameters at level i. This is the parameter difference adjustment function between levels, where n is the total number of parameter-aware levels.

[0010] Furthermore, the expression for the cross-architecture parameter migration adaptation model includes:

[0011]

[0012] This is the initial value for parameter migration across architectures. For architectural migration coefficients, Adaptation functions for the differences between the source and target architectures. For the characteristic parameters of the target application architecture, These are characteristic parameters of the source application architecture. This is the cross-architecture migration deviation compensation value. The value of the parameter after cross-architecture adaptation. For the target architecture attribute adjustment function, These are the attribute parameters of the target architecture. To adapt and optimize the coefficients, The adaptation weights for the k-th transfer parameter, is the initial migration value for the k-th migration parameter, and m is the total number of parameters participating in the migration.

[0013] Furthermore, the expression for the heterogeneous parameter fusion inference algorithm is as follows:

[0014] in, The results of heterogeneous parameter fusion simulation are as follows. Let p be the fusion weight coefficient of the p-th heterogeneous parameter. For the p-th heterogeneous parameter value after cross-architecture adaptation, For heterogeneous parameter feature enhancement functions, For the heterogeneous parameter difference balance function, q represents the difference between different types of heterogeneous parameters, and q is the total number of heterogeneous parameter types.

[0015] Furthermore, the global parameter fusion intelligent analysis engine adopts a multi-core heterogeneous processor architecture, integrating an FPGA chip and a GPU acceleration module. The FPGA chip is used for parallel acquisition and rapid filtering of raw parameters, while the GPU acceleration module is responsible for parallel processing of multi-level parameters. The parameter perception layer number is set to a range of 3-8 layers, and the parameter mapping threshold for cross-architecture migration adaptation is set to be dynamically adjustable, adjusted according to the architectural complexity of the target scene through an adaptive algorithm. In the process of extracting common parameters, an extraction strategy based on hierarchical density clustering is adopted. The adjustment step size of localized parameter adaptation is set according to the parameter sensitivity level. The adjustment step size of highly sensitive parameters is controlled in the range of 1 / 3-1 / 2 of that of low-sensitivity parameters. During the adaptation process, real-time feedback adjustment is performed through edge computing nodes to ensure the real-time performance and accuracy of parameter adaptation.

[0016] Further, S2 includes the following sub-steps: S21, constructing a multi-level perception framework based on the attribute features of the parameters, determining the perception dimension and parameter selection conditions of each level, allocating the parameter cluster to be processed to the corresponding perception level according to the hierarchical division rules, and establishing a transmission channel for parameters between levels; S22, starting a hierarchical nested parameter perception model in each perception level, extracting features from the parameters allocated to that level, and capturing the local features and global correlation features of the parameters by adjusting the size of the model's perception window; S23, based on the transmission channel for parameters between levels, transmitting the parameter perception results of each level to the next level, while receiving parameter feature information from the next level, and performing cross-level parameter feature fusion; S24, verifying the fused cross-level parameter features, removing abnormal feature data, updating the perception weights and correlation coefficients of each level, and outputting the parameter feature set after multi-level nested perception processing.

[0017] Further, S3 includes the following sub-steps: S31, collecting feature parameters of the source application architecture and the target application architecture, establishing an architecture feature database, and standardizing the architecture feature parameters based on the requirements of the cross-architecture parameter migration adaptation model; S32, inputting the parameter feature set after hierarchical nested parameter awareness processing, determining the migration mapping relationship of parameters between the source architecture and the target architecture through model calculation, and generating migration path planning results; S33, migrating the parameters from the source architecture to the target architecture according to the migration path planning results, monitoring the transmission status and integrity of the parameters in real time during the migration process, and dynamically correcting deviations that occur during the transmission process; S34, performing adaptability testing on the migrated parameters based on the operating requirements of the target architecture, adjusting the architecture migration coefficient and difference adaptation function parameters in the model according to the test results, and outputting the parameter set after cross-architecture adaptation.

[0018] Further, S4 includes the following sub-steps: S41, classifying the parameter set after cross-architecture adaptation by type, distinguishing different heterogeneous parameter types, clarifying the characteristic attributes and data formats of each type of parameter, and establishing a heterogeneous parameter classification index; S42, initializing various parameters of the heterogeneous parameter fusion inference algorithm, including fusion weight coefficients, feature enhancement function parameters, and difference balance function parameters, and setting different initial values ​​according to parameter types; S43, inputting various heterogeneous parameters into the algorithm model for fusion operation, monitoring the rationality of parameter fusion in real time during the operation, and adjusting the fusion ratio of different types of parameters through the difference balance function; S44, performing consistency verification on the parameter results after fusion operation, removing abnormal data generated during the fusion process, generating a set of intermediate parameters that meets the requirements, and establishing attribute labeling and index information for the intermediate parameters.

[0019] Further, S5 includes the following sub-steps: S51, performing hierarchical distribution analysis on the intermediate parameter set, statistically analyzing the frequency of occurrence and correlation strength of parameters at each level, establishing a parameter hierarchical distribution map, and identifying parameter subsets with high-frequency correlation characteristics; S52, based on the universality judgment criteria of common parameters, verifying each parameter subset with high-frequency correlation characteristics, selecting parameters that have effective effects in multiple scenarios, and initially determining the candidate set of common parameters; S53, performing redundancy detection on the candidate set of common parameters, deleting duplicate parameters and parameters with extremely low correlation, and optimizing the combination structure of common parameters through a global parameter fusion intelligent analysis engine; S54, validating the optimized set of common parameters to ensure the universality and compatibility of each common parameter in different application scenarios, and finally outputting a universally applicable subset of common parameters.

[0020] A method for extracting common parameters and adapting localized parameters based on a layered architecture is disclosed. This method is implemented through a system for extracting common parameters and adapting localized parameters based on a layered architecture, comprising: a multi-level parameter sensing and acquisition unit, a cross-architecture migration and adaptation processing unit, a heterogeneous parameter fusion and deduction unit, a global common parameter extraction unit, a localized parameter adaptation and optimization unit, and an intelligent analysis and control unit. The multi-level parameter sensing and acquisition unit and the cross-architecture migration and adaptation processing unit are connected via a high-speed data transmission bus for transmitting the parameter data after hierarchical sensing and acquisition to the cross-architecture migration and adaptation processing unit. The cross-architecture migration and adaptation processing unit and the heterogeneous parameter fusion and deduction unit... The units communicate bidirectionally via a distributed data interface to transmit migration adaptation parameters and provide feedback on fusion results. The heterogeneous parameter fusion and deduction unit is connected to the global common parameter extraction unit through a parameter feature transmission channel, providing intermediate parameters after fusion for the common parameter extraction. The global common parameter extraction unit is connected to the localized parameter adaptation and optimization unit through a scenario-based adaptation interface, transmitting a subset of common parameters to the localized parameter adaptation and optimization unit for adaptation and adjustment. The intelligent analysis and control unit is connected to each unit through control signal lines, dynamically adjusting various parameter configurations based on the operating status data of each unit, and coordinating the working timing and data transmission rate of each unit.

[0021] Beneficial Effects: This invention proposes a method for extracting common parameters and adapting localized parameters based on a layered architecture. It establishes a multi-level nested perception mechanism through a layered nested parameter perception model, deeply mining the correlation features of parameters at different levels. Combined with a cross-architecture parameter migration and adaptation model, it fully considers the differences between the source and target architectures, improving the targeting and adaptation effect of parameter migration and achieving rapid and accurate adaptation in different scenarios. Utilizing a heterogeneous parameter fusion inference algorithm and a global parameter fusion intelligent analysis engine, it constructs a systematic technical system for the entire process, strengthening the technical correlation and synergy of each link. Simultaneously, in the extraction of common parameters, it dynamically adjusts based on the distribution patterns and general characteristics of parameter levels. In the localization adaptation stage, it adopts flexible adjustment strategies and forms full-process control through intelligent regulation, improving the intelligence level of extraction and adaptation and ensuring real-time optimization of parameter configuration. Through a multi-core heterogeneous hardware architecture and dynamic software parameter design, it improves processing efficiency. Through multi-unit collaborative operation, it achieves increased parameter reuse rate and alleviates adaptation lag issues, ensuring system collaborative operation and performance optimization, meeting the needs for high efficiency and compatibility in parameter processing under multi-source heterogeneous scenarios. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating the overall process of the method of the present invention. Figure 2 This is a flowchart of method step S2 of the present invention; Figure 3 This is a flowchart of method step S3 of the present invention; Figure 4 This is a flowchart of method step S4 of the present invention; Figure 5 This is a flowchart of step S5 of the method of the present invention. Detailed Implementation

[0023] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0024] like Figure 1As shown, a method for extracting common parameters and adapting localized parameters based on a hierarchical architecture includes the following steps: S1, using a global parameter fusion intelligent analysis engine to perform hierarchical perception and collection of the original parameter set in a multi-source heterogeneous scenario, and performing preliminary classification and screening based on parameter attribute features to obtain a cluster of parameters to be processed with hierarchical correlation characteristics; S2, using a hierarchical nested parameter perception model to perform multi-level nested feature mining on the cluster of parameters to be processed, and identifying the correlation mapping relationship between parameters at different levels by dynamically adjusting the perception dimension and hierarchical depth; S3, using a cross-architecture parameter migration and adaptation model to adapt the parameters corresponding to the correlation mapping relationship. The system performs cross-scenario architecture migration processing to establish a parameter adaptation rule system under different application architectures; S4, it uses a heterogeneous parameter fusion and inference algorithm to perform multi-dimensional fusion calculations on the migrated and adapted parameters to generate a fused intermediate parameter set; S5, it performs common parameter extraction operations based on the intermediate parameter set, and combines the general characteristics and hierarchical distribution rules of the parameters to separate a subset of universally applicable common parameters; S6, it performs localized parameter adaptation and adjustment on the subset of common parameters, and optimizes the common parameters according to the architectural characteristics and operational requirements of the target application scenario to form a localized parameter configuration result adapted to the target scenario.

[0025] Step S1 completes the hierarchical acquisition and preliminary screening of raw parameters in multi-source heterogeneous scenarios, providing a high-quality cluster of parameters to be processed for subsequent processing. During implementation, the full-domain parameter fusion intelligent analysis engine activates 16 parallel data acquisition channels to synchronously acquire raw parameter sets from different sources such as industrial control, smart terminals, and cloud services. The acquisition frequency is set to 200 times per second to ensure the real-time performance and integrity of the parameter data. During the acquisition process, the engine classifies parameters into three categories based on their attribute characteristics: numerical, logical, and character. Each category of parameters has a dedicated screening threshold. The effective range for numerical parameters is set to 0 to 10000, logical parameters only retain true and false state data, and character parameters must meet the condition of being between 2 and 50 characters in length. Through the above classification and screening mechanism, parameter data that is out of the valid range, has abnormal status, or is not in a standardized format is removed, and a cluster of parameters to be processed with hierarchical association characteristics is finally obtained. The hierarchical division of parameters in this cluster is determined according to their functional priority in the system, and is divided into three levels: core layer, middle layer, and edge layer. The core layer parameters account for no less than 30%, the middle layer parameters account for 40% to 50%, and the edge layer parameters account for no more than 20%. The parameters between levels are associated through index identifiers, providing a clear data foundation for subsequent multi-level nested perception. The implementation of this step directly determines the efficiency and accuracy of subsequent parameter processing and effectively avoids the interference of invalid parameters on subsequent processes.

[0026] Step S2 utilizes a hierarchical nested parameter perception model to achieve multi-level feature mining of the parameter cluster to be processed, accurately identifying the correlation mapping relationship between parameters at different levels. In implementation, based on the hierarchical distribution of the parameter cluster, the model's perception layers are first set to 5 layers, each corresponding to a functional level of the parameter cluster. The perception dimensions of each layer are configured according to the standard configuration of 8 dimensions for the core layer, 6 dimensions for the intermediate layer, and 4 dimensions for the edge layer. The perception window size is dynamically adjusted from 32 to 128, with an adjustment cycle of one update every 1000 parameter data points processed. After the model starts, feature mining begins with the core layer parameters. A sliding window is used to traverse the parameter data, extracting key features such as the rate of change, correlation strength, and distribution density of the parameters. The calculation cycle for the rate of change of the core layer parameters is set to 50 milliseconds, while the calculation cycles for the rates of change of the intermediate and edge layer parameters are 100 milliseconds and 200 milliseconds, respectively. During the mining process, the model dynamically adjusts the perception weights of each level: the core layer parameter perception weight is set to 0.6, the intermediate layer to 0.3, and the edge layer to 0.1, ensuring that the features of the core parameters are fully mined. Simultaneously, through inter-level data interaction channels, feature data from each level is transmitted to the next higher level, achieving cross-level feature fusion. During the fusion process, a weighted summation method is used to integrate feature information from different levels, ultimately generating a parameter feature set that includes hierarchical correlation information. This step enables in-depth mining of the intrinsic relationships between parameters, breaking the limitations of single-level parameter analysis, providing precise feature support for subsequent cross-architecture migration and adaptation, and significantly improving the targeting and effectiveness of parameter processing.

[0027] Step S3 employs a cross-architecture parameter migration and adaptation model to establish migration and adaptation rules for parameters across different application architectures, ensuring cross-scenario availability of parameters. During implementation, core feature data of the source and target application architectures are first collected, including processing capabilities, interface protocols, and resource configurations. Processing capability is measured by the number of parameters processed per second, with thresholds set at 5000 parameters / second for the source architecture and 8000 parameters / second for the target architecture. Interface protocols must support TCP / IP, HTTP, and MQTT, and the maximum memory usage is set to 4GB. Based on the collected architecture feature data, the model establishes a cross-architecture parameter migration mapping table. This table includes key information such as parameter name, data type, value range, and associated parameters. The mapping relationship is established using a similarity matching algorithm, with a similarity threshold set at 0.85. Only parameter mapping relationships with similarities higher than this threshold are retained. During the migration process, the model transfers parameters from the source architecture to the target architecture according to the rules of the mapping table. The transfer uses a fragmented transmission method, with each fragment set to 1024 bytes in size and a transmission timeout of 5 seconds. After the timeout, an automatic retransmission mechanism is initiated, with a maximum of three retransmissions. Simultaneously, the model monitors the adaptation status of the parameters in real time after migration. By comparing the functional effectiveness and data integrity of the parameters before and after migration, an adaptation rule system is established. This rule system includes parameter adjustment thresholds, related parameter adaptation priorities, and exception handling mechanisms. The parameter adjustment thresholds are dynamically set based on the processing capacity of the target architecture, and the related parameter adaptation priorities decrease sequentially from the core layer to the middle layer and then to the edge layer. This step effectively solves the problem of parameter incompatibility between different architectures. By establishing scientific migration adaptation rules, it ensures that parameters maintain good functional characteristics after cross-architecture transmission, laying the foundation for subsequent heterogeneous parameter fusion.

[0028] Step S4 utilizes a heterogeneous parameter fusion and inference algorithm to perform multi-dimensional fusion operations on the cross-architecture adapted parameters, generating a unified intermediate parameter set. During implementation, the cross-architecture adapted parameters are first classified into heterogeneous types, clarifying the proportions of numerical, logical, and character parameters, requiring each type to account for approximately 60%, 30%, and 10%, respectively. In the algorithm initialization phase, fusion weight coefficients are set: 0.5 for numerical parameters, 0.3 for logical parameters, and 0.2 for character parameters. Simultaneously, the strength parameter of the feature enhancement function is configured to 0.7, and the adjustment coefficient of the difference balance function is set to 0.3. During the fusion operation, a phased fusion strategy is adopted. In the first phase, parameters of the same type are internally fused. Numerical parameters are fused using the arithmetic mean method, logical parameters are fused using majority voting, and character parameters are fused using string matching degree to select the optimal data. In the second phase, cross-type fusion is performed on parameters of different types. By establishing a parameter association matrix, the state of logical parameters is converted into numerical identifiers, key features of character parameters are extracted and encoded into numerical form, and then weighted and fused with numerical parameters. During the fusion process, the rationality of parameter fusion is monitored in real time. A deviation threshold of 0.1 is set for the fusion result. When the deviation between the fusion result and each input parameter exceeds this threshold, the difference balancing function is activated to adjust the fusion weights, ensuring the accuracy of the fusion result. After the fusion operation is completed, the generated intermediate parameter set undergoes consistency verification. The verification includes parameter format, value range, and correlation relationships. The verification pass rate must reach over 99%. Parameters that fail the verification are marked and re-enter the fusion process until a compliant intermediate parameter set is generated. This step achieves the organic integration of heterogeneous parameters, eliminates the differences between different types of parameters, forms a unified parameter expression form, provides standardized data support for the extraction of common parameters, and improves the uniformity and efficiency of parameter processing.

[0029] Step S5 extracts a subset of universally applicable common parameters based on the intermediate parameter set. By accurately separating common and non-common parameters, the parameter reuse rate is improved. During implementation, a hierarchical distribution analysis is first performed on the intermediate parameter set. Statistical analysis methods are used to calculate the frequency of each parameter in the core, intermediate, and edge layers, setting a frequency threshold of 0.7. Parameters with a frequency higher than this threshold in at least two layers are selected, initially forming a high-frequency parameter subset. Subsequently, further screening is conducted based on the parameters' universality characteristics. Universality characteristics include the number of times a parameter is applicable in different application scenarios, the breadth of its association with other parameters, and its irreplaceable function. The number of applicable scenarios must be no less than 8, the number of associated parameters no less than 20, and the irreplaceable function score is set from 0 to 10, retaining only parameters with a score higher than 7. Through the above screening, a candidate set of common parameters is obtained. Redundancy detection is then performed on the candidate set using a similarity clustering algorithm, setting a clustering threshold of 0.9. Parameters with similarity higher than this threshold are grouped into one class, with only one core parameter retained in each class, and the remaining redundant parameters deleted. Simultaneously, parameters with a correlation strength below 0.3 are removed to ensure the conciseness and effectiveness of the common parameter subset. Finally, the effectiveness of the optimized common parameter subset is verified through a global parameter fusion intelligent analysis engine. The verification scenarios cover more than 80% of the original data collection scenarios, and the verification indicators include parameter functional stability, data accuracy, and adaptation compatibility. The pass rate of each indicator must reach more than 98%. The final output is a global common parameter subset, in which the number of parameters is controlled between 30% and 40% of the total number of intermediate parameter sets, ensuring both the comprehensiveness of the common parameters and avoiding parameter redundancy.

[0030] Step S6 involves localizing and adapting a subset of common parameters to ensure they accurately match the operational requirements of the target application scenario and guarantee stable system operation. During implementation, the architectural characteristics and operational requirements of the target application scenario are first obtained, including hardware configuration, software environment, business processes, and performance metrics. Hardware configuration requires specifying key parameters such as processor model, memory size, and storage capacity. The software environment requires determining the operating system type, database version, and development framework. Performance metrics are set as follows: response time no more than 100 milliseconds, concurrent processing capacity no less than 1000 users / second, and data transmission error rate less than 0.01%. Based on this information, a localized adaptation and adjustment model is established, and the step size parameter for adaptation and adjustment is set according to parameter sensitivity levels. The adjustment step size is 0.01 for high-sensitivity parameters, 0.05 for medium-sensitivity parameters, and 0.1 for low-sensitivity parameters. The sensitivity level is determined based on the degree of impact of the parameter on system performance: a score higher than 8 indicates high sensitivity, 5 to 8 indicates medium sensitivity, and below 5 indicates low sensitivity. During the adaptation and adjustment process, an iterative adjustment strategy is adopted. After each iteration, the effect of parameter configuration is tested. Test indicators include parameter call success rate, system resource utilization, and business processing efficiency. Adjustment stops when the test indicators meet the target requirements; otherwise, the adaptation direction and step size are adjusted based on the test results. Simultaneously, a dynamic feedback mechanism for adaptation parameters is established to monitor the operational status of the target scenario in real time. When the scenario's operating environment changes, the parameter adaptation and adjustment process is automatically triggered to ensure that the parameter configuration always remains consistent with the scenario requirements. Finally, a localized parameter configuration result adapted to the target scenario is generated. This result is stored in the form of a configuration file, which uses an encrypted format with a 256-bit key to ensure the security of the parameter configuration. This step achieves accurate matching between common parameters and the target scenario, fully leveraging the reusability of common parameters and improving the system's operational efficiency and stability.

[0031] Preferably, the expression of the hierarchical nested parameter-aware model includes:

[0032]

[0033] The perceived output value of the j-th parameter at level 1. Let be the perceptual weight coefficient of the i-th level. For hierarchical parameter-aware activation functions, The feature scaling factor for the j-th original parameter. The collected value of the j-th original parameter. Let be the correlation coefficient between the i-th level and the level above it. For the first The perceived output value of the j-th parameter at each level. Let be the perceptual bias correction value for the i-th level. The result is a multi-level parameter fusion sensing result. Let be the fusion weight of the parameters at level i. This is the parameter difference adjustment function between levels, where n is the total number of parameter-aware levels.

[0034] Specifically, the hierarchical nested parameter perception model, based on the hierarchical correlation characteristics and feature transfer rules of parameters, performs single-level parameter perception output and multi-level parameter fusion to form a complete hierarchical perception system. Single-level parameter perception needs to consider the influence of both the original parameter features and the perception results of the previous level, optimizing perception accuracy through weight coefficients and deviation correction values. Multi-level fusion needs to be adjusted based on the differences in importance and parameters at each level to ensure the comprehensiveness of the fusion result. In implementation, the hierarchical perception weight coefficients are set according to the functional priority of the parameter hierarchy: 0.7 for the core layer, 0.5 for the middle layer, and 0.3 for the edge layer. The feature scaling coefficient is adjusted according to the parameter type: 0.9 for numerical parameters, and 0.6 for both logical and character parameters. The hierarchical correlation coefficient is set to 0.4 to ensure a reasonable influence of the previous level on the current level. The deviation correction value is determined based on the parameter acquisition error statistics: 0.02 for the core layer, 0.03 for the middle layer, and 0.05 for the edge layer. The fusion weight is consistent with the hierarchical perception weight, and the adjustment strength of the parameter difference adjustment function is set to 0.8. After the model starts, the original parameters of each level are first scaled and weighted and superimposed with the perception results of the previous level. After processing by an activation function, a bias correction value is added to obtain the single-level perception output. Then, the outputs of each level are weighted by fusion weights, and the differences between levels are balanced by a parameter difference adjustment function to generate the final fused perception result. The establishment of this model fully considers the heterogeneity and correlation of the level parameters. The parameter values ​​have been verified through multiple experiments to ensure a balance between perception accuracy and efficiency. After implementation, it can deeply mine the features of multiple levels of parameters, provide accurate data support for subsequent processing, and improve the comprehensiveness and accuracy of parameter feature extraction.

[0035] Preferably, the expression for the cross-architecture parameter migration adaptation model includes:

[0036]

[0037] This is the initial value for parameter migration across architectures. For architectural migration coefficients, Adaptation functions for the differences between the source and target architectures. For the characteristic parameters of the target application architecture, These are characteristic parameters of the source application architecture. This is the cross-architecture migration deviation compensation value. The value of the parameter after cross-architecture adaptation. For the target architecture attribute adjustment function, These are the attribute parameters of the target architecture. To adapt and optimize the coefficients, The adaptation weights for the k-th transfer parameter, is the initial migration value for the k-th migration parameter, and m is the total number of parameters participating in the migration.

[0038] Specifically, the cross-architecture parameter migration and adaptation model, based on the differences between the source and target architectures, performs initial calculations for cross-architecture parameter migration and completes post-migration adaptation optimization. Cross-architecture migration requires first eliminating the impact of architectural differences on parameters, and then performing targeted optimization based on the target architecture attributes to ensure the effectiveness of parameter functionality. During implementation, the architecture migration coefficient is determined based on the ratio of the processing capabilities of the source and target architectures, ranging from 0.6 to 0.9. The upper limit is used when the target architecture's processing capability is higher than the source architecture, and the lower limit is used otherwise. The sensitivity parameter of the difference adaptation function is set to 0.7 to ensure accurate capture of architectural feature differences, and the migration deviation compensation value is set to 0.04 based on the error analysis results of historical migration data. The adjustment coefficient of the target architecture attribute adjustment function is configured according to the architecture type: 0.8 for industrial control architecture, 0.7 for intelligent terminal architecture, and 0.6 for cloud service architecture. The adaptation optimization coefficient is set to 0.5, and the migration parameter adaptation weight is allocated according to parameter hierarchical priority: 0.6 for the core layer, 0.3 for the middle layer, and 0.1 for the edge layer. During model runtime, initial parameter migration values ​​across architectures are first calculated using architecture migration coefficients, difference adaptation functions, and migration deviation compensation values. Then, these initial migration values ​​are optimized and adjusted using target architecture attribute adjustment functions and adaptation weights to obtain the final adaptation result. By combining architectural differences and parameter characteristics, parameter values ​​are calibrated through multiple architecture migration experiments. After implementation, this effectively solves cross-architecture parameter incompatibility issues, ensures stable parameter operation across different architectures, and enhances the ability to reuse parameters across scenarios.

[0039] Preferably, the expression for the heterogeneous parameter fusion and deduction algorithm is:

[0040] in, The results of heterogeneous parameter fusion simulation are as follows. Let p be the fusion weight coefficient of the p-th heterogeneous parameter. For the p-th heterogeneous parameter value after cross-architecture adaptation, For heterogeneous parameter feature enhancement functions, For the heterogeneous parameter difference balance function, q represents the difference between different types of heterogeneous parameters, and q is the total number of heterogeneous parameter types.

[0041] Specifically, the heterogeneous parameter fusion and inference algorithm is based on the feature complementarity and difference balance of heterogeneous parameters. It integrates the features of different types of parameters through weighted fusion, while introducing feature enhancement and difference balancing mechanisms to eliminate conflicts and redundancies between heterogeneous parameters. In implementation, the heterogeneous parameter fusion weight coefficient is set according to the importance ratio of parameter type: numerical parameters, due to their explicit quantitative characteristics, have a value of 0.6; logical parameters, due to their key status identifiers, have a value of 0.3; and character parameters, due to their weak feature correlation, have a value of 0.1. The enhancement strength parameter of the feature enhancement function is set to 0.8 to improve the fusion effectiveness by strengthening the core features of the parameters. The balance coefficient of the difference balancing function is set to 0.4 to adjust the difference magnitude between different heterogeneous parameters, avoiding a single type of parameter dominating the fusion result. The algorithm implementation process is as follows: first, the heterogeneous parameters after cross-architecture adaptation are identified and their features extracted; then, each type of parameter is substituted into the formula, and weighted calculation is performed through the fusion weight coefficient. Simultaneously, key features are strengthened through the feature enhancement function, and parameter differences are adjusted through the difference balancing function, ultimately obtaining the fusion and inference result. The values ​​of each parameter in the formula were determined through extensive heterogeneous parameter fusion experiments to ensure a reasonable fusion ratio for different types of parameters and a moderate balance between feature enhancement and difference. This algorithm overcomes the limitations of traditional heterogeneous parameter fusion methods, enabling the organic integration of heterogeneous parameters to form a unified and accurate parameter expression. This provides a high-quality data foundation for common parameter extraction and enhances the scientific rigor and effectiveness of parameter fusion.

[0042] Preferably, the global parameter fusion intelligent analysis engine adopts a multi-core heterogeneous processor architecture, integrating an FPGA chip and a GPU acceleration module. The FPGA chip is used for parallel acquisition and rapid filtering of raw parameters, while the GPU acceleration module is responsible for parallel processing of multi-level parameters. The parameter perception layer number is set to a range of 3-8 layers, and the parameter mapping threshold for cross-architecture migration adaptation is set to be dynamically adjustable, adjusted according to the architectural complexity of the target scene through an adaptive algorithm. In the process of extracting common parameters, an extraction strategy based on hierarchical density clustering is adopted. The adjustment step size of localized parameter adaptation is set according to the parameter sensitivity level. The adjustment step size of highly sensitive parameters is controlled in the range of 1 / 3-1 / 2 of that of low-sensitivity parameters. In the adaptation process, real-time feedback adjustment is performed through edge computing nodes to ensure the real-time performance and accuracy of parameter adaptation.

[0043] Preferred, such as Figure 2As shown, S2 includes the following sub-steps: S21, constructing a multi-level perception framework based on the attribute features of the parameters, determining the perception dimension and parameter selection conditions of each level, allocating the parameter cluster to be processed to the corresponding perception level according to the hierarchical division rules, and establishing a transmission channel for parameters between levels; S22, starting a hierarchical nested parameter perception model in each perception level, extracting features from the parameters allocated to that level, and capturing the local features and global correlation features of the parameters by adjusting the size of the model's perception window; S23, based on the transmission channel for parameters between levels, transmitting the parameter perception results of each level to the next level, while receiving the parameter feature information of the next level, and performing cross-level parameter feature fusion; S24, verifying the fused cross-level parameter features, removing abnormal feature data, updating the perception weights and correlation coefficients of each level, and outputting the parameter feature set after multi-level nested perception processing.

[0044] Specifically, step S2 involves multi-level feature mining of a layered nested parameter perception model. S21 first constructs a three-layer perception framework—core, intermediate, and edge—based on parameter attribute features. The perception dimensions for each layer are determined to be 8 dimensions for the core layer, 6 dimensions for the intermediate layer, and 4 dimensions for the edge layer. Parameter selection criteria are set as follows: numerical fluctuation range not exceeding ±10%, and the number of associated parameters not less than 5. The parameter clusters to be processed are allocated to the corresponding layers according to these rules. Simultaneously, a high-speed data transmission channel is established between layers, with a bandwidth of 10Gbps to ensure delay-free parameter transmission. S22, after starting the model at each layer, the perception window size is initialized to 64 and dynamically adjusted every 500 parameter data points processed, with an adjustment range of ±16. The parameter data is traversed through a sliding window, capturing local features while establishing a global correlation mapping. The feature extraction frequency for the core layer parameters is set to 100 times per second, while the intermediate and edge layers are set to 80 and 50 times per second, respectively. S23 transmits the perception results of each level upwards through a preset transmission channel, prioritizing the core layer, while simultaneously receiving feature data from the next level. A weighted summation method is used for cross-level fusion, with the fusion weights consistent with the level priorities: 0.6 for the core layer, 0.3 for the intermediate layer, and 0.1 for the edge layer. S24 performs anomaly detection on the fused feature data, setting an anomaly threshold of 0.05. Feature points exceeding the threshold are removed. Based on the detection results, the perception weights and correlation coefficients of each level are updated once per second. The final output is a set of parameter features including hierarchical correlation information. The entire process, through precise parameter configuration and step-by-step execution, ensures the depth and accuracy of feature mining, providing high-quality data support for subsequent processing.

[0045] Preferred, such as Figure 3As shown, step S3 includes the following sub-steps: S31, collecting feature parameters of the source application architecture and the target application architecture, establishing an architecture feature database, and standardizing the architecture feature parameters based on the requirements of the cross-architecture parameter migration adaptation model; S32, inputting the parameter feature set after hierarchical nested parameter awareness processing, determining the migration mapping relationship of parameters between the source architecture and the target architecture through model calculation, and generating migration path planning results; S33, migrating the parameters from the source architecture to the target architecture according to the migration path planning results, monitoring the transmission status and integrity of the parameters in real time during the migration process, and dynamically correcting deviations that occur during the transmission process; S34, performing adaptability testing on the migrated parameters based on the operating requirements of the target architecture, adjusting the architecture migration coefficient and difference adaptation function parameters in the model according to the test results, and outputting the parameter set after cross-architecture adaptation.

[0046] Specifically, step S3 includes S31: First, collect the core feature parameters of the source and target architectures, including processing speed, interface type, resource quota, etc., and establish an architecture feature database. The database storage capacity is set to 100GB, supporting read and write operations of 2000 data entries per second. Then, the feature parameters are processed according to the requirements of unified data format and standardized value range. The error of the processed parameters is controlled within ±0.02. S32: Input the parameter feature set after hierarchical nested perception into the model, calculate the mapping relationship between the parameters in the source and target architectures through a similarity matching algorithm, set the matching threshold to 0.85, and generate planning results including migration path, priority, and transmission order. The migration priority of the core layer parameters is set to level 1, the middle layer to level 2, and the edge layer to level 3. S33 initiates parameter migration according to the planning results, adopting a fragmented transmission mode with each fragment set to 2048 bytes. A cyclic redundancy check mechanism is used for transmission verification to ensure data integrity. During the migration process, the transmission status is monitored in real time. When the transmission delay exceeds 50 milliseconds, the transmission rate is automatically adjusted, with a maximum adjustment of 50% of the initial rate. A retransmission mechanism is initiated in case of transmission failure, with a maximum of 3 retransmissions. S34 performs compatibility testing on the migrated parameters based on the target architecture's operational requirements. Testing indicators include functional compatibility, data validity, and response timeliness, with a pass threshold of 95% for each indicator. The architecture migration coefficient and difference adaptation function parameters are dynamically adjusted based on the test results, with an adjustment step size of 0.05, until the parameter compatibility rate reaches over 98%. Finally, a cross-architecture adapted parameter set is output. Through step-by-step refined operations, parameter incompatibility issues between different architectures are resolved, improving the cross-scenario usability of parameters.

[0047] Preferred, such as Figure 4As shown, step S4 includes the following sub-steps: S41, classifying the parameter set after cross-architecture adaptation by type, distinguishing parameters of different heterogeneous types, clarifying the characteristic attributes and data formats of each type of parameter, and establishing a heterogeneous parameter classification index; S42, initializing various parameters of the heterogeneous parameter fusion inference algorithm, including fusion weight coefficients, feature enhancement function parameters, and difference balance function parameters, and setting different initial values ​​according to parameter types; S43, inputting various heterogeneous parameters into the algorithm model for fusion operation, monitoring the rationality of parameter fusion in real time during the operation, and adjusting the fusion ratio of different types of parameters through the difference balance function; S44, performing consistency verification on the parameter results after fusion operation, removing abnormal data generated during the fusion process, generating an intermediate parameter set that meets the requirements, and establishing attribute labeling and index information for the intermediate parameters.

[0048] Specifically, step S4 includes: S41 First, classify the parameters after cross-architecture adaptation, clarifying the specific ranges of the three types of parameters: numeric, logical, and character. Establish a classification index library, setting the index query response time to no more than 10 milliseconds. Simultaneously, record the characteristic attributes of each type of parameter, including data format, value range, and correlation strength. The completeness of the characteristic attribute records must reach 100%. S42 Initialize the algorithm parameters. Configure the fusion weight coefficients according to the ratio of 0.5 for numeric, 0.3 for logical, and 0.2 for character. Set the feature enhancement function strength parameter to 0.7 and the difference balance function adjustment coefficient to 0.3. Control the parameter initialization error within ±0.01 to ensure the stability of the algorithm's initial state. S43 Input the classified heterogeneous parameters into the algorithm model and start the phased fusion operation. The first stage completes the fusion of parameters of the same type. The second stage achieves the fusion of cross-type parameters. During the fusion process, monitor the fusion deviation in real time. When the deviation exceeds 0.1, activate the difference balance mechanism to dynamically adjust the fusion weights. The adjustment cycle is once every 100 parameter data points processed. S44 performs consistency checks on the fused parameter results. The checks include data format consistency, reasonableness of value range, and validity of correlation. The pass rate must be above 99%. Parameters that fail the checks are marked and re-enter the fusion process. At the same time, attribute annotations and index information for intermediate parameters are established. The annotation information includes fusion time, weight configuration, and verification results. The index update frequency is set to 5 times per second. Through rigorous execution in steps, the organic integration of heterogeneous parameters is achieved, eliminating interference caused by parameter differences and laying the foundation for the extraction of common parameters.

[0049] Preferred, such as Figure 5As shown, step S5 includes the following sub-steps: S51, performing hierarchical distribution analysis on the intermediate parameter set, statistically analyzing the frequency of occurrence and correlation strength of parameters at each level, establishing a parameter hierarchical distribution map, and identifying parameter subsets with high-frequency correlation characteristics; S52, based on the universality judgment criteria of common parameters, verifying each parameter subset with high-frequency correlation characteristics, selecting parameters that have effective effects in multiple scenarios, and initially determining the candidate set of common parameters; S53, performing redundancy detection on the candidate set of common parameters, deleting duplicate parameters and parameters with extremely low correlation, and optimizing the combination structure of common parameters through a global parameter fusion intelligent analysis engine; S54, validating the optimized set of common parameters to ensure the universality and compatibility of each common parameter in different application scenarios, and finally outputting a universally applicable subset of common parameters.

[0050] Specifically, step S5 includes S51, which involves performing a hierarchical distribution analysis on the intermediate parameter set. A statistical sampling method is used, with the sampling ratio set at 20% of the total parameters. The frequency and correlation strength of each parameter in the core, intermediate, and edge layers are calculated. The correlation strength is calculated using cosine similarity, and the results are rounded to two decimal places. Based on the analysis results, a parameter hierarchical distribution map is constructed, updated twice per second, clearly presenting the hierarchical distribution characteristics of the parameters. S52, based on the universality criteria for common parameters, the subset of high-frequency correlated parameters is verified one by one. The universality criteria include at least 8 applicable scenarios, at least 30 cross-level correlations, and a functional irreplaceability score higher than 7. The verification process uses parallel processing, processing 50 parameters simultaneously, with a verification efficiency controlled at over 30 parameters per second. S53 performs redundancy detection on the candidate set of common parameters. A clustering algorithm is used to group parameters with a similarity higher than 0.9 into one class, retaining only the core parameters in each class. Simultaneously, the correlation strength between parameters is calculated, and parameters with a correlation strength lower than 0.3 are removed. The proportion of redundant parameters removed is controlled between 20% and 30% of the total number of candidate sets, ensuring a concise and efficient parameter subset. S54 uses a global parameter fusion intelligent analysis engine to verify the effectiveness of the optimized common parameter subset. The verification scenarios cover more than 80% of the original data collection scenarios. Verification indicators include functional stability, data accuracy, and adaptation compatibility. Each indicator is tested at least 100 times, with a pass rate of over 98%. Finally, a globally applicable common parameter subset is output. Through step-by-step screening and verification, the universality and effectiveness of the common parameters are ensured, improving parameter reuse rate.

[0051] A method for extracting common parameters and adapting localized parameters based on a layered architecture is disclosed. This method is implemented through a system for extracting common parameters and adapting localized parameters based on a layered architecture, comprising: a multi-level parameter sensing and acquisition unit, a cross-architecture migration and adaptation processing unit, a heterogeneous parameter fusion and deduction unit, a global common parameter extraction unit, a localized parameter adaptation and optimization unit, and an intelligent analysis and control unit. The multi-level parameter sensing and acquisition unit and the cross-architecture migration and adaptation processing unit are connected via a high-speed data transmission bus for transmitting the parameter data after hierarchical sensing and acquisition to the cross-architecture migration and adaptation processing unit. The cross-architecture migration and adaptation processing unit and the heterogeneous parameter fusion and deduction unit... The units communicate bidirectionally via a distributed data interface to transmit migration adaptation parameters and provide feedback on fusion results. The heterogeneous parameter fusion and deduction unit is connected to the global common parameter extraction unit through a parameter feature transmission channel, providing intermediate parameters after fusion for the common parameter extraction. The global common parameter extraction unit is connected to the localized parameter adaptation and optimization unit through a scenario-based adaptation interface, transmitting a subset of common parameters to the localized parameter adaptation and optimization unit for adaptation and adjustment. The intelligent analysis and control unit is connected to each unit through control signal lines, dynamically adjusting various parameter configurations based on the operating status data of each unit, and coordinating the working timing and data transmission rate of each unit.

[0052] Based on a layered architecture, a common parameter extraction and localized parameter adaptation method is constructed, establishing a hierarchical, end-to-end parameter processing system. This system forms a coherent and collaborative technical link from parameter acquisition to localized adaptation. Through the deep integration of core technologies such as layered nested perception, cross-architecture migration, and heterogeneous fusion, efficient connection between each stage of parameter processing is achieved. Hardware and software are co-optimized, employing a multi-core heterogeneous processor architecture paired with dynamically adjustable software parameter configurations, balancing processing speed and adaptation flexibility, significantly improving parameter reuse rate and scenario compatibility. The system is highly intelligent, utilizing a full-domain parameter fusion intelligent analysis engine and a multi-unit collaborative control mechanism to achieve dynamic optimization of parameter feature mining, common parameter extraction, and localized adaptation, adapting to the differentiated needs of different architectures and scenarios.

[0053] This method addresses the insufficient targeting of parameter awareness and transfer adaptation. It deeply mines the correlation features of parameters at different levels through a layered nested parameter awareness model, and combines this with a cross-architecture parameter transfer adaptation model to fully consider the differences between the source and target architectures, establishing precise parameter mapping and adaptation rules to achieve rapid and accurate adaptation in different scenarios. To address the low level of intelligence in common parameter extraction and localization adaptation, it optimizes parameter fusion effects through a heterogeneous parameter fusion and inference algorithm, dynamically adjusts common parameter extraction standards based on parameter hierarchical distribution patterns, and employs flexible localization adaptation strategies and intelligent end-to-end control to effectively avoid adaptation lag and unreasonable configuration issues, comprehensively improving the accuracy and efficiency of parameter processing. In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," "link," and "fix" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication between two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0054] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various equivalent changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for extracting common parameters and adapting localized parameters based on a layered architecture, characterized in that, Includes the following steps: S1. A hierarchical sensing and collection process is performed on the original parameter set in a multi-source heterogeneous scenario using a global parameter fusion intelligent analysis engine. Preliminary classification and screening are conducted based on parameter attribute features to obtain a cluster of parameters to be processed that possess hierarchical correlation characteristics. S2. A multi-level nested feature mining process is used on the parameter cluster to be processed using a hierarchical nested parameter sensing model. By dynamically adjusting the sensing dimension and hierarchical depth, the correlation mapping relationship between parameters at different levels is identified. S3. A cross-architecture parameter migration and adaptation model is used to perform cross-scenario architecture migration processing on the parameters corresponding to the correlation mapping relationship, establishing a parameter adaptation rule system under different application architectures. S4. A heterogeneous parameter fusion inference algorithm is used to perform multi-dimensional fusion operations on the migrated and adapted parameters to generate a fused intermediate parameter set. S5. Based on the intermediate parameter set, a common parameter extraction operation is performed. Combining the general characteristics and hierarchical distribution patterns of the parameters, a subset of universally applicable common parameters is obtained. S6. The common parameter subset is localized for parameter adaptation and adjustment. Based on the architectural characteristics and operational requirements of the target application scenario, the common parameters are optimized for scenario-based adaptation, forming a localized parameter configuration result adapted to the target scenario.

2. The method for extracting common parameters and adapting localized parameters based on a layered architecture according to claim 1, characterized in that, The expression for the hierarchical nested parameter-aware model includes: ； ； The perceived output value of the j-th parameter at level 1. Let be the perceptual weight coefficient of the i-th level. For hierarchical parameter-aware activation functions, The feature scaling factor for the j-th original parameter. The collected value of the j-th original parameter. Let be the correlation coefficient between the i-th level and the level above it. For the first The perceived output value of the j-th parameter at each level. Let be the perceptual bias correction value for the i-th level. The result is a multi-level parameter fusion sensing result. Let be the fusion weight of the parameters at level i. This is the parameter difference adjustment function between levels, where n is the total number of parameter-aware levels.

3. The method for extracting common parameters and adapting localized parameters based on a layered architecture according to claim 1, characterized in that, The expression for the cross-architecture parameter transfer adaptation model includes: ； ； This is the initial value for parameter migration across architectures. For architectural migration coefficients, Adaptation functions for the differences between the source and target architectures. For the characteristic parameters of the target application architecture, These are characteristic parameters of the source application architecture. This is the cross-architecture migration deviation compensation value. The value of the parameter after cross-architecture adaptation. For the target architecture attribute adjustment function, These are the attribute parameters of the target architecture. To adapt and optimize the coefficients, The adaptation weights for the k-th transfer parameter. is the initial migration value for the k-th migration parameter, and m is the total number of parameters participating in the migration.

4. The method for extracting common parameters and adapting localized parameters based on a layered architecture according to claim 1, characterized in that, The expression for the heterogeneous parameter fusion and inference algorithm is: ； in, The results of heterogeneous parameter fusion simulation are as follows. Let p be the fusion weight coefficient of the p-th heterogeneous parameter. For the p-th heterogeneous parameter value after cross-architecture adaptation, For heterogeneous parameter feature enhancement functions, For the heterogeneous parameter difference balance function, q represents the difference between different types of heterogeneous parameters, and q is the total number of heterogeneous parameter types.

5. The method for extracting common parameters and adapting localized parameters based on a layered architecture according to claim 1, characterized in that, The comprehensive parameter fusion intelligent analysis engine adopts a multi-core heterogeneous processor architecture, integrating an FPGA chip and a GPU acceleration module. The FPGA chip is used for parallel acquisition and rapid filtering of raw parameters, while the GPU acceleration module is responsible for parallel processing of multi-level parameters. The parameter perception layer is set to a range of 3-8 layers, and the parameter mapping threshold for cross-architecture migration adaptation is set to be dynamically adjustable, adjusted according to the architectural complexity of the target scene through an adaptive algorithm. In the process of extracting common parameters, an extraction strategy based on hierarchical density clustering is adopted. The adjustment step size of localized parameter adaptation is set according to the parameter sensitivity level. The adjustment step size of highly sensitive parameters is controlled in the range of 1 / 3-1 / 2 of that of low-sensitivity parameters. During the adaptation process, real-time feedback adjustment is performed through edge computing nodes to ensure the real-time performance and accuracy of parameter adaptation.

6. The method for extracting common parameters and adapting localized parameters based on a layered architecture according to claim 1, characterized in that, S2 includes the following steps: S21, constructing a multi-level perception framework based on the attribute features of the parameters, determining the perception dimension and parameter selection conditions of each level, allocating the cluster of parameters to be processed to the corresponding perception level according to the hierarchical division rules, and establishing a transmission channel for parameters between levels; S22, starting a hierarchical nested parameter perception model in each perception level, extracting features from the parameters allocated to that level, and capturing the local features and global correlation features of the parameters by adjusting the size of the model's perception window; S23, based on the transmission channel for parameters between levels, transmitting the parameter perception results of each level to the next level, while receiving parameter feature information from the next level, and performing cross-level parameter feature fusion. S24 verifies the fused cross-level parameter features, removes abnormal feature data, updates the perception weights and correlation coefficients of each level, and outputs the parameter feature set after multi-level nested perception processing.

7. The method for extracting common parameters and adapting localized parameters based on a layered architecture according to claim 1, characterized in that, S3 includes the following steps: S31, collecting feature parameters of the source application architecture and the target application architecture, establishing an architecture feature database, and standardizing the architecture feature parameters based on the requirements of the cross-architecture parameter migration adaptation model; S32, inputting the parameter feature set after hierarchical nested parameter awareness processing, determining the migration mapping relationship of parameters between the source architecture and the target architecture through model calculation, and generating migration path planning results; S33, migrating the parameters from the source architecture to the target architecture according to the migration path planning results, monitoring the transmission status and integrity of the parameters in real time during the migration process, and dynamically correcting any deviations that occur during the transmission process; S34, performing adaptability testing on the migrated parameters based on the operating requirements of the target architecture, adjusting the architecture migration coefficient and difference adaptation function parameters in the model according to the test results, and outputting the parameter set after cross-architecture adaptation.

8. The method for extracting common parameters and adapting localized parameters based on a layered architecture according to claim 1, characterized in that, S4 includes the following sub-steps: S41, classifying the parameter set after cross-architecture adaptation by type, distinguishing different heterogeneous parameter types, clarifying the characteristic attributes and data formats of each type of parameter, and establishing a heterogeneous parameter classification index; S42, initializing various parameters of the heterogeneous parameter fusion inference algorithm, including fusion weight coefficients, feature enhancement function parameters, and difference balance function parameters, setting different initial values ​​according to parameter types; S43, inputting various heterogeneous parameters into the algorithm model for fusion operation, monitoring the rationality of parameter fusion in real time during the operation, and adjusting the fusion ratio of different types of parameters through the difference balance function; S44, performing consistency verification on the parameter results after fusion operation, removing abnormal data generated during the fusion process, generating a qualified intermediate parameter set, and establishing attribute labeling and index information for the intermediate parameters.

9. The method for extracting common parameters and adapting localized parameters based on a layered architecture according to claim 1, characterized in that, S5 includes the following steps: S51, performing hierarchical distribution analysis on the intermediate parameter set, statistically analyzing the frequency of occurrence and correlation strength of parameters at each level, establishing a parameter hierarchical distribution map, and identifying parameter subsets with high-frequency correlation characteristics; S52, based on the universality judgment criteria of common parameters, verifying each parameter subset with high-frequency correlation characteristics, selecting parameters that have effective effects in multiple scenarios, and initially determining the candidate set of common parameters; S53, performing redundancy detection on the candidate set of common parameters, deleting duplicate parameters and parameters with extremely low correlation, and optimizing the combination structure of common parameters through a global parameter fusion intelligent analysis engine; S54, validating the optimized set of common parameters to ensure the universality and compatibility of each common parameter in different application scenarios, and finally outputting a universally applicable subset of common parameters.

10. A method for extracting common parameters and adapting localized parameters based on a hierarchical architecture according to any one of claims 1-9, characterized in that, This method is implemented through a hierarchical architecture-based common parameter extraction and localized parameter adaptation system, comprising: a multi-level parameter sensing and acquisition unit, a cross-architecture migration and adaptation processing unit, a heterogeneous parameter fusion and deduction unit, a global common parameter extraction unit, a localized parameter adaptation and optimization unit, and an intelligent analysis and control unit. The multi-level parameter sensing and acquisition unit and the cross-architecture migration and adaptation processing unit are connected via a high-speed data transmission bus for transmitting the hierarchically sensed and acquired parameter data to the cross-architecture migration and adaptation processing unit. The cross-architecture migration and adaptation processing unit and the heterogeneous parameter fusion and deduction unit communicate bidirectionally via a distributed data interface. Communication is used to transmit migration adaptation parameters and provide feedback on fusion results; the heterogeneous parameter fusion and deduction unit is connected to the global common parameter extraction unit through a parameter feature transmission channel, providing intermediate parameters after fusion for common parameter extraction; the global common parameter extraction unit is connected to the localized parameter adaptation and optimization unit through a scenario-based adaptation interface, transmitting a subset of common parameters to the localized parameter adaptation and optimization unit for adaptation and adjustment; the intelligent analysis and control unit is connected to each unit through control signal lines, dynamically adjusting various parameter configurations based on the operating status data of each unit, and coordinating the working sequence and data transmission rate of each unit.

Images