CN-122019164-A - Data processing method and system for AI practical training power workstation

CN122019164ACN 122019164 ACN122019164 ACN 122019164ACN-122019164-A

Abstract

The invention provides a data processing method and a system for an AI practical training power workstation, which relate to the technical field of data processing, and are characterized in that a four-shot matrix sensor group is built, practical training area information is subjected to practical training four-shot data acquisition to obtain practical training four-shot acquisition data, functional analysis optimization and data characteristic processing are carried out according to the practical training four-shot acquisition data to obtain practical training characteristic processing data, the practical training four-shot acquisition data and the practical training characteristic processing data are subjected to parallel mapping power processing state analysis to obtain power processing state judgment information, power resource allocation analysis is carried out according to the power processing state judgment information, and power resource allocation data is obtained according to the power resource allocation analysis information.

Inventors

SHU HUAJUN
LIN LINGZHEN

Assignees

广州市锐星信息科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260129

Claims (10)

1. A data processing method of an AI practical-training power workstation, the method comprising: S1, building a four-shot matrix sensor group, carrying out real training four-shot data acquisition on real training area information to obtain real training four-shot acquisition data, carrying out functional analysis optimization and data characteristic processing according to the real training four-shot acquisition data to obtain real training characteristic processing data; S2, carrying out parallel mapping calculation force processing state analysis on the training four-shot acquisition data and the training characteristic processing data to obtain calculation force processing state judgment information; and S3, carrying out calculation power resource allocation analysis according to the calculation power processing state judgment information, and carrying out calculation power resource allocation according to calculation power resource allocation analysis information to obtain calculation power resource allocation data.
2. The AI-training power workstation data processing method of claim 1, wherein S1 comprises: acquiring training area information, and building a four-shot matrix sensor group according to the training area information; Acquiring real training preset acquisition target data, and acquiring real training four-shot acquisition data by carrying out four-shot data acquisition on the real training preset acquisition target data of real training area information according to the four-shot matrix sensor group; analyzing the characteristic acquisition data of the training four-shot acquisition data to obtain characteristic acquisition analysis data; Performing four-shot acquisition function adjustment according to the characteristic acquisition analysis data to obtain four-shot acquisition function adjustment data; performing four-shot data acquisition and adjustment according to the four-shot acquisition function adjustment data to obtain training four-shot update data; and carrying out characteristic analysis processing on the training four-shot updated data to obtain training characteristic processing data.
3. The data processing method of AI practical training power workstation according to claim 2, wherein the performing feature collection data analysis on the practical training four-shot collection data to obtain feature collection analysis data comprises: Performing target four-shot function acquisition and division on the training four-shot acquisition data to obtain main shot function acquisition data, ultra-wide angle function acquisition data, long focus function acquisition data and auxiliary function acquisition data; Performing function comparison analysis on the main shooting function acquisition data, the ultra-wide angle function acquisition data, the tele function acquisition data and the auxiliary function acquisition data with training preset acquisition target data respectively to obtain four-shooting function comparison data; determining unqualified function data according to the four-shot function comparison data; And the unqualified functional data is characteristic acquisition analysis data.
4. The data processing method of AI practical training power workstation according to claim 2, wherein the performing practical training data feature analysis processing on the practical training four-shot update data to obtain practical training feature processing data comprises: Extracting training characteristic data from the training four-shot updated data to obtain training characteristic extraction data; Carrying out training feature data classification on the training feature extraction data to obtain training feature class data; Carrying out safety feature analysis on the training feature class data to obtain safety feature analysis data; Performing feature safety extraction of the training feature class data according to the safety feature analysis data to obtain training feature class safety extraction data; The training characteristic category safety extraction data is the training characteristic processing data.
5. The data processing method of AI practical training power workstation according to claim 4, wherein said performing security feature analysis and processing on practical training feature class data to obtain security feature analysis data comprises: performing feature extraction of preset encryption key information on the training feature class data to obtain training encryption feature extraction data; acquiring the duty ratio of the training encryption characteristic extraction data in preset encryption key information, and acquiring a training characteristic sensitivity coefficient; comparing the training characteristic sensitivity coefficient with a preset sensitivity threshold and a preset secret-related threshold respectively to obtain a sensitivity level comparison result and a secret-related level comparison result; the sensitivity level comparison result and the secret related level comparison result are the security feature analysis data.
6. The AI-training power workstation data processing method of claim 1, wherein S2 comprises: parallel mapping is carried out on the training four-shot acquisition data and the training characteristic processing data to obtain parallel mapping data; acquiring calculation force processing information of parallel mapping data; Performing calculation force processing state judgment on the parallel mapping data according to the calculation force processing information to obtain calculation force processing state judgment information; and triggering a calculation force processing allocation instruction according to the calculation force processing state judgment information.
7. The AI-training power workstation data processing method of claim 6, wherein said computing power processing state determination of parallel mapping data based on said computing power processing information to obtain computing power processing state determination information comprises: Grouping the parallel mapping data to obtain a plurality of parallel mapping combinations; CPU core occupancy rate data of parallel mapping data of each camping mapping combination is obtained; Acquiring the ratio of the CPU core occupation rate data to preset CPU core occupation rate data, and acquiring a calculation force utilization coefficient; Comparing the calculated force utilization coefficient with a preset calculated force utilization threshold value to obtain a calculated force utilization comparison result; when the calculation force utilization comparison result is that the calculation force utilization coefficient is larger than a preset calculation force utilization threshold value, the calculation force processing state shortage judgment is carried out; and when the calculated force utilization comparison result is that the calculated force utilization coefficient is smaller than or equal to the pre-imputation force utilization threshold value, performing calculated force processing state redundancy judgment.
8. The AI-training power workstation data processing method of claim 1, wherein S3 comprises: When the calculation force processing allocation instruction is triggered, acquiring the ratio of the calculation force utilization coefficient to a preset calculation force utilization threshold value according to calculation force processing state shortage judgment information, and acquiring the calculation force shortage coefficient; Acquiring the ratio of the calculation force utilization coefficient to a preset calculation force utilization threshold value according to the calculation force processing state redundancy judgment information to acquire the calculation force redundancy coefficient; performing calculation power allocation analysis on the calculation power shortage coefficient according to the calculation power redundancy coefficient to obtain calculation power allocation analysis data; And performing calculation power resource allocation according to the calculation power allocation analysis data to obtain calculation power resource allocation data.
9. The data processing method of AI practical training power workstation according to claim 8, wherein said performing a power allocation analysis on said power shortage coefficient according to said power redundancy coefficient to obtain power allocation analysis data includes: sequencing a plurality of parallel mapping combined calculation redundancy coefficients from large to small to obtain a calculation redundancy sequence; Obtaining a difference value between each calculation force redundancy coefficient and a calculation force shortage coefficient in the calculation force redundancy sequence, and obtaining a calculation force allocation difference value; acquiring parallel mapping combinations corresponding to the computational power redundancy coefficients with the smallest computational power allocation difference value, and determining a computational power allocation source; and performing calculation power resource allocation on the calculation power shortage coefficient through the calculation power redundancy coefficient of the calculation power allocation source to obtain calculation power resource allocation data.
10. A data processing system for an AI practical training power workstation, the system comprising: The four-shot data processing module is used for constructing a four-shot matrix sensor group, carrying out real training four-shot data acquisition on real training area information to obtain real training four-shot acquisition data, carrying out functional analysis optimization and data characteristic processing according to the real training four-shot acquisition data to obtain real training characteristic processing data; The power calculation state analysis module is used for carrying out power calculation processing state analysis of parallel mapping on the training four-shot acquisition data and the training characteristic processing data to obtain power calculation processing state judgment information; And the computing power allocation module is used for carrying out computing power resource allocation analysis according to the computing power processing state judgment information and carrying out computing power resource allocation according to the computing power resource allocation analysis information to obtain computing power resource allocation data.

Description

Data processing method and system for AI practical training power workstation Technical Field The invention provides a data processing method and a data processing system for an AI practical training working station, relates to the technical field of data processing, and particularly relates to the technical field of data processing of the AI practical training working station. Background The current AI practical training relies on single-shot or double-shot equipment to collect data, has the problems of insufficient visual angle coverage, insufficient detail capture and the like, and lacks the targeted function optimization and the safety grading treatment of the collected data, so that the risks of poor data quality or privacy leakage are easily caused. Meanwhile, when multitasking is performed in parallel in the practical training process, the computing power resource allocation is mostly in a static mode, the computing power load states of different tasks cannot be perceived in real time, redundant computing power is idle, and the resource waste that the computing power cannot be timely supplemented is caused, so that the practical training task processing efficiency is affected. The existing calculation force allocation is not differentially controlled by combining the data security level, and potential safety hazards that calculation force of sensitive or confidential data is stolen can be caused. Disclosure of Invention The invention provides a data processing method and a data processing system for an AI practical training calculation workstation, which are used for solving the problems: the invention provides a data processing method and a system for an AI practical training power workstation, wherein the method comprises the following steps: S1, building a four-shot matrix sensor group, carrying out real training four-shot data acquisition on real training area information to obtain real training four-shot acquisition data, carrying out functional analysis optimization and data characteristic processing according to the real training four-shot acquisition data to obtain real training characteristic processing data; S2, carrying out parallel mapping calculation force processing state analysis on the training four-shot acquisition data and the training characteristic processing data to obtain calculation force processing state judgment information; and S3, carrying out calculation power resource allocation analysis according to the calculation power processing state judgment information, and carrying out calculation power resource allocation according to calculation power resource allocation analysis information to obtain calculation power resource allocation data. Further, the step S1 includes: acquiring training area information, and building a four-shot matrix sensor group according to the training area information; Acquiring real training preset acquisition target data, and acquiring real training four-shot acquisition data by carrying out four-shot data acquisition on the real training preset acquisition target data of real training area information according to the four-shot matrix sensor group; analyzing the characteristic acquisition data of the training four-shot acquisition data to obtain characteristic acquisition analysis data; Performing four-shot acquisition function adjustment according to the characteristic acquisition analysis data to obtain four-shot acquisition function adjustment data; performing four-shot data acquisition and adjustment according to the four-shot acquisition function adjustment data to obtain training four-shot update data; and carrying out characteristic analysis processing on the training four-shot updated data to obtain training characteristic processing data. Further, the feature collection data analysis is performed on the training four-shot collection data to obtain feature collection analysis data, including: Performing target four-shot function acquisition and division on the training four-shot acquisition data to obtain main shot function acquisition data, ultra-wide angle function acquisition data, long focus function acquisition data and auxiliary function acquisition data; Performing function comparison analysis on the main shooting function acquisition data, the ultra-wide angle function acquisition data, the tele function acquisition data and the auxiliary function acquisition data with training preset acquisition target data respectively to obtain four-shooting function comparison data; determining unqualified function data according to the four-shot function comparison data; And the unqualified functional data is characteristic acquisition analysis data. Further, the performing the characteristic analysis processing on the training four-shot updated data to obtain training characteristic processing data includes: Extracting training characteristic data from the training four-shot updated data to obtain training characteristic extraction data; Car