Search

CN-121996416-A - Model calculation force optimization method and device, electronic equipment, chip and medium

CN121996416ACN 121996416 ACN121996416 ACN 121996416ACN-121996416-A

Abstract

The application discloses a model calculation force optimization method and device, electronic equipment, a chip and a medium, relates to the technical field of graph neural networks, and aims to solve the problems of single calculation force evaluation dimension, inaccurate analysis and untimely early warning. The method comprises the steps of obtaining multi-source data, wherein the multi-source data comprises performance index data and equipment connection data, determining a data processing coefficient corresponding to the performance index data and an equipment connection coefficient corresponding to the equipment connection data based on the multi-source data, determining a calculation force optimization rationality index based on the data processing coefficient and the equipment connection coefficient, outputting a normal signal if the calculation force optimization rationality index is larger than a preset index value, and outputting an early warning signal if the calculation force optimization rationality index is not larger than the preset index value.

Inventors

  • XIE PAN
  • ZHOU SHIQI

Assignees

  • 中国移动通信集团广东有限公司
  • 中国移动通信集团有限公司

Dates

Publication Date
20260508
Application Date
20260106

Claims (14)

  1. 1. A model calculation force optimization method, comprising: acquiring multi-source data, wherein the multi-source data comprises performance index data and equipment connection data; Determining a data processing coefficient corresponding to the performance index data and a device connection coefficient corresponding to the device connection data based on the multi-source data; Determining a computing power optimization rationality index based on the data processing coefficients and the device connection coefficients; if the calculated force optimization rationality index is larger than a preset index value, outputting a normal signal; Otherwise, outputting an early warning signal.
  2. 2. The method of claim 1, wherein the performance index data comprises one or more of data processing delay data and data processing accuracy data; The data processing delay data comprises one or more of edge node processing delay, cloud processing delay and end-to-end processing delay, and the data processing accuracy data comprises one or more of data error rate, data integrity check failure rate and abnormal data filtering proportion.
  3. 3. The method of claim 2, wherein the device connection data includes one or more of device access characteristic data and device connection stability data; the device access characteristic data comprises one or more of heterogeneous network access proportion, device authentication average duration and low-power consumption device wake-up period, and the device connection stability data comprises one or more of connection interruption frequency, average reconnection time and signal intensity standard deviation.
  4. 4. A method according to claim 3, wherein the data processing coefficients comprise a delay pressure coefficient and a data quality coefficient, and the device connection coefficients comprise an access energy efficiency coefficient and a connection risk coefficient.
  5. 5. The method of claim 4, wherein the delay pressure coefficient is positively correlated with the edge node processing delay and the cloud processing delay, and wherein the delay pressure coefficient is negatively correlated with the end-to-end processing delay.
  6. 6. The method of claim 4, wherein the data quality coefficient is positively correlated with the data error rate and the data integrity check failure rate, and wherein the data quality coefficient is negatively correlated with the abnormal data filtering ratio.
  7. 7. The method of claim 4, wherein the access energy efficiency coefficient is positively correlated with the heterogeneous network access ratio, and wherein the access energy efficiency coefficient is negatively correlated with the device authentication average duration and the low power consumption device wake-up period.
  8. 8. The method of claim 4, wherein the connection risk factor is positively correlated with the connection break frequency and the signal strength standard deviation, and wherein the connection risk factor is negatively correlated with the average reconnection time.
  9. 9. The method of claim 4, wherein the determining a computational effort optimization rational index based on the data processing coefficients and the device connection coefficients comprises: Determining the computational effort optimization rationality index based on the delay pressure coefficient, the data quality coefficient, the access energy efficiency coefficient, and the connection risk coefficient; The computational effort optimization rationality index is positively correlated with the product of the delay pressure coefficient, the data quality coefficient, the access energy efficiency coefficient, and the connection risk coefficient, and negatively correlated with the square of the sum of the delay pressure coefficient and the data quality coefficient.
  10. 10. The method according to claim 1, wherein the method further comprises: Updating the multi-source data in response to the pre-warning signal; determining a target computing power optimization rationality index based on the updated multi-source data; If the target calculation force optimization rationality index is larger than the preset index value, stopping outputting the early warning signal and outputting the normal signal; Otherwise, generating an optimization log.
  11. 11. A model calculation force optimizing apparatus, characterized by comprising: The device comprises an acquisition module, a processing module and a processing module, wherein the acquisition module is used for acquiring multi-source data, and the multi-source data comprises performance index data and equipment connection data; The determining module is used for determining a data processing coefficient corresponding to the performance index data and a device connection coefficient corresponding to the device connection data based on the multi-source data; The determining module is further used for determining a computing power optimization rationality index based on the data processing coefficient and the equipment connection coefficient; the processing module is used for outputting a normal signal if the calculated force optimization rationality index is larger than a preset index value; otherwise, the processing module is also used for outputting an early warning signal.
  12. 12. An electronic device comprising a processor and a memory, the memory storing a computer program that, when executed by the processor, implements the method of any of claims 1-10.
  13. 13. A computer storage medium having instructions stored therein which, when executed, implement the method of any one of claims 1-10.
  14. 14. A chip comprising a processor and a communication interface coupled to the processor for running a computer program or instructions to implement the method of any one of claims 1-10.

Description

Model calculation force optimization method and device, electronic equipment, chip and medium Technical Field The application relates to the technical field of graphic neural networks, in particular to a model calculation force optimization method and device, electronic equipment, a chip and a medium. Background With the development of the internet of things technology, a large model based on a graph neural network is widely applied to the field of the internet of things, but the calculation power requirement is large and becomes a bottleneck. The existing calculation force optimization method comprises the steps of data acquisition and preprocessing, graph structure construction, calculation resource allocation and scheduling, performance evaluation and scheduling and the like. However, the method has the problems of incomplete data acquisition, missing data acquisition of partial areas or equipment, and influence on model training and prediction accuracy, and the analysis method has low processing efficiency and inaccurate result in the face of high-dimensional, large-scale and strong noise data, reduces the reliability and practicality of the model, and influences on effective management and optimization of the system. Disclosure of Invention The application aims to provide a model calculation force optimization method and device, electronic equipment, a chip and a medium, so as to solve the problems of single calculation force evaluation dimension, inaccurate analysis and untimely early warning in the prior art. In order to achieve the above object, the present application provides the following technical solutions: a model calculation force optimization method, comprising: acquiring multi-source data, wherein the multi-source data comprises performance index data and equipment connection data; Determining a data processing coefficient corresponding to the performance index data and a device connection coefficient corresponding to the device connection data based on the multi-source data; Determining a computing power optimization rationality index based on the data processing coefficients and the device connection coefficients; if the calculated force optimization rationality index is larger than a preset index value, outputting a normal signal; Otherwise, outputting an early warning signal. Compared with the prior art, in the model calculation power optimization method, the multi-source data is collected, the multi-source data comprises the performance index data and the equipment connection data, so that the performance index data reflecting the health condition of the data processing chain and the equipment connection data reflecting the stability of the network access chain are obtained, and the multi-angle data collection provides solid and rich data support for the subsequent optimization strategy. By analyzing and calculating the acquired data, the data processing coefficient corresponding to the performance index data and the device connection coefficient corresponding to the device connection data can be determined, so that key factors which can influence calculation force can be accurately identified. And finally, by carrying out system evaluation on the data processing result, the feasibility and effectiveness of the optimization strategy are improved. By means of a real-time monitoring and dynamic tracking technology, once a calculation force bottleneck or performance abnormality is found, an intelligent adjustment mechanism is started immediately, so that operation and maintenance personnel can quickly master the real-time state and potential problems of calculation force of a large model, and timely adjust and optimize a resource allocation scheme or system configuration, so that solid guarantee is provided for realizing efficient and stable calculation force support of the large model and scientific and reasonable technical deployment, and efficient operation and wide application of the large model in complex tasks are forcefully promoted. The application also provides a model calculation force optimizing device, which comprises: The device comprises an acquisition module, a processing module and a processing module, wherein the acquisition module is used for acquiring multi-source data, and the multi-source data comprises performance index data and equipment connection data; The determining module is used for determining a data processing coefficient corresponding to the performance index data and a device connection coefficient corresponding to the device connection data based on the multi-source data; The determining module is further used for determining a computing power optimization rationality index based on the data processing coefficient and the equipment connection coefficient; the processing module is used for outputting a normal signal if the calculated force optimization rationality index is larger than a preset index value; otherwise, the processing module is also used for