CN-121979656-A - UFS chip management method, terminal device and computer readable storage medium

CN121979656ACN 121979656 ACN121979656 ACN 121979656ACN-121979656-A

Abstract

The application discloses a management method of a UFS chip, terminal equipment and a computer readable storage medium, relating to the technical field of storage; the management method of the UFS chip is applied to a storage system and comprises the steps of collecting data of the UFS chip to obtain basic data, extracting characteristics of the basic data to obtain characteristic values, inputting the characteristic values into a hybrid artificial intelligent model to obtain execution decisions, and executing actions according to the execution decisions.

Inventors

GUO HAOBIN
HUANG SHANYONG
QIU JIAYANG

Assignees

深圳市时创意电子股份有限公司

Dates

Publication Date: 20260505
Application Date: 20251210

Claims (10)

1. The management method of the UFS chip is applied to a storage system and is characterized by comprising the following steps: collecting data of the UFS chip to obtain basic data; extracting the characteristics of the basic data to obtain characteristic values; Inputting the characteristic values into the hybrid artificial intelligence model to obtain execution decisions, and And executing the action according to the execution decision.
2. The method of managing UFS chips of claim 1, wherein said step of collecting data of the UFS chips to obtain base data comprises: Collecting data of a UFS chip; Obtaining basic data; making a time stamp corresponding to the basic data; the base data is aligned by the time stamp to form a data snapshot.
3. The method for managing UFS chips as defined in claim 1, wherein said step of extracting features from said basic data to obtain feature values includes: extracting the characteristics of the basic data to obtain characteristic values; Respectively splicing the feature vectors according to the types of the features to obtain feature vectors corresponding to the types; and performing dimension reduction processing on the feature vector to obtain a dimension reduction feature vector.
4. The method for managing UFS chips as defined in claim 3, wherein said step of performing feature vector concatenation according to the type of the feature, respectively, to obtain feature vectors comprises: Dividing the feature value into statistical features, time sequence features and semantic features; And respectively carrying out statistics feature vector splicing, time sequence feature vector splicing and semantic feature vector splicing according to the statistics features, time sequence features and semantic features to obtain statistics feature vectors, time sequence feature vectors and semantic feature vectors.
5. The method for managing UFS chips as defined in claim 4, wherein said step of performing a dimension-reduction process on said feature vector to obtain a dimension-reduced feature vector includes: Normalizing the statistical feature vector, the time sequence feature vector and the semantic feature vector; calculating covariance matrixes of the standardized statistical feature vectors, the timing feature vectors and the semantic feature vectors to obtain corresponding feature vectors; Selecting the number of principal components to generate a dimension-reduction feature vector; And forming a projection matrix according to the number of the selected principal components, and generating a dimension-reducing feature vector by using the projection matrix.
6. A method of managing UFS chips as defined in claim 3, wherein said step of inputting the feature values into a hybrid artificial intelligence model to obtain execution decisions comprises: Calling a hybrid artificial intelligence model; Inputting the dimension-reducing feature vector into a hybrid artificial intelligence model; Outputting an execution decision by the hybrid artificial intelligence model; the output execution decision comprises execution actions with m dimensions, wherein m is greater than or equal to 4.
7. The method of managing UFS chips of claim 1, wherein said step of performing an action based on the execution decision comprises: Analyzing and executing the decision to obtain a decision vector; inquiring a strategy mapping table, and obtaining a corresponding hardware parameter combination according to the value of the decision vector; and sending out instruction commands according to the hardware parameter combination, and executing actions.
8. The method of managing UFS chips of claim 1, wherein the hybrid artificial intelligence model is formed by a hybrid of at least two artificial intelligence models.
9. A terminal device comprising a storage system, a processor and a computer program stored in the storage system and executable on the processor, the processor implementing the steps of the method of managing UFS chips of any one of claims 1 to 8 when the computer program is executed by the processor.
10. A computer-readable storage medium storing a computer program, characterized in that the computer program, when executed by a processor, implements the steps of the management method of a UFS chip as claimed in any one of claims 1 to 8.

Description

UFS chip management method, terminal device and computer readable storage medium Technical Field The present application relates to the field of storage technologies, and in particular, to a management method of a UFS chip, a terminal device, and a computer-readable storage medium. Background In UFS (Universal Flash Storage, universal flash memory storage) chip storage equipment, a static DVFS scheme (Dynamic Voltage and Frequency Scaling) cannot adapt to burst loads, wherein the static DVFS depends on historical loads or triggers frequency modulation by a fixed threshold, but a mobile terminal load has burstiness, for example, a user suddenly opens a camera continuous shooting, a UFS chip needs to switch from a PWM-Gear1 low-power consumption mode to an HS-Gear4 high-performance mode, the static DVFS needs time to detect the load and switch, so that the writing of the previous few photos is delayed, the user experiences the clamping, and if heuristic scheduling based on rules (such as fixed queue depth, priority strategy, cache strategy and the like) is adopted, the user behavior cannot be learned, the difference of using modes of different users is huge, the rules are fixed, high queue depth is needed for game players to process a large number of small file loads in parallel, the low queue depth can meet the requirement of orderly writing large files for document editors, and the video and picture loading needs burst processing for App users; The current resource management scheme of the UFS chip storage device cannot quickly respond to the sudden load demand of the mobile terminal, is difficult to adapt to the personalized use mode of the user, causes the dislocation of storage resource allocation and actual demand, and cannot realize the accurate balance of performance and power consumption. Therefore, how to quickly adjust the hardware resource configuration of the UFS flash memory chip based on different usage modes of the user, so as to realize management of the UFS chip, is a current problem to be solved. Disclosure of Invention The application aims to provide a management method, terminal equipment and a computer readable storage medium of a UFS chip, which realize dynamic adjustment of hardware parameters through a hybrid artificial intelligent model and realize balanced management of power consumption and performance of the UFS chip. The application discloses a management method of a UFS chip, which is applied to a storage system and comprises the following steps: collecting data of the UFS chip to obtain basic data; extracting the characteristics of the basic data to obtain characteristic values; Inputting the characteristic values into the hybrid artificial intelligence model to obtain execution decisions, and And executing the action according to the execution decision. Optionally, the step of acquiring data of the UFS chip to obtain the basic data includes: Collecting data of a UFS chip; Obtaining basic data; making a time stamp corresponding to the basic data; the base data is aligned by the time stamp to form a data snapshot. Optionally, the step of extracting the features from the basic data to obtain the feature value includes: extracting the characteristics of the basic data to obtain characteristic values; Respectively splicing the feature vectors according to the types of the features to obtain feature vectors corresponding to the types; and performing dimension reduction processing on the feature vector to obtain a dimension reduction feature vector. Optionally, the step of respectively performing feature vector splicing according to the type of the feature to obtain the feature vector includes: Dividing the feature value into statistical features, time sequence features and semantic features; And respectively carrying out statistics feature vector splicing, time sequence feature vector splicing and semantic feature vector splicing according to the statistics features, time sequence features and semantic features to obtain statistics feature vectors, time sequence feature vectors and semantic feature vectors. Optionally, the step of performing the dimension reduction processing on the feature vector to obtain a dimension-reduced feature vector includes: Normalizing the statistical feature vector, the time sequence feature vector and the semantic feature vector; calculating covariance matrixes of the standardized statistical feature vectors, the timing feature vectors and the semantic feature vectors to obtain corresponding feature vectors; Selecting the number of principal components to generate a dimension-reduction feature vector; And forming a projection matrix according to the number of the selected principal components, and generating a dimension-reducing feature vector by using the projection matrix. Optionally, the step of inputting the feature value into the hybrid artificial intelligence model to obtain the execution decision includes: Calling a hybrid artificial intelligence