A distributed storage method, medium and system for industrial data assets

Abstract

The present invention provides a distributed storage method, medium and system for industrial data assets, belonging to the field of electronic digital data processing technology. The method first collects manufacturing equipment operating time data, maintenance cycle data, task quantity data and data generation frequency data as industrial original data streams, establishes a resource consumption calculation model, extracts data features and constructs a feature mapping matrix, realizes resource scheduling optimization through a storage optimization neural network, allocates storage units based on deep optimization convolution kernels, dynamically evaluates and optimizes the storage location of data blocks, and finally realizes distributed storage of industrial original data streams. Through real-time calculation of storage performance indicators and load factors, a complete performance evaluation and optimization feedback mechanism is established, ensuring the balanced allocation and efficient utilization of storage resources, and solving the technical problem in the existing technology that industrial data asset distributed storage systems are difficult to realize intelligent scheduling of storage resources.

Description

Technical Field

[0001] The present invention belongs to the technical field of electronic digital data processing, and in particular, relates to a distributed storage method, medium and system for industrial data assets. Background Art

[0002] Industrial data asset management is a crucial component of industrial digital transformation. With the rapid development of the Industrial Internet of Things (IIoT) and smart manufacturing, the scale and complexity of industrial data are growing exponentially. Traditional industrial data storage solutions primarily utilize centralized storage architectures, responding to data growth by expanding the hardware configuration of storage servers. As data volumes continue to expand, the limitations of centralized storage architectures in terms of data access efficiency, system reliability, and resource utilization are becoming increasingly apparent. Distributed storage technology is gradually becoming the mainstream solution for industrial data asset management.

[0003] Current mainstream distributed storage systems typically employ static data sharding strategies and fixed resource allocation schemes, lacking real-time awareness and dynamic adjustment capabilities for storage resource utilization during operation. This static resource scheduling approach can lead to storage hotspots, resource waste, and performance bottlenecks when faced with dynamically changing data access patterns and load distribution. Furthermore, existing distributed storage systems often employ empirical rule-based design for data layout optimization, making them difficult to adapt to the high temporal nature, strong correlation, and burstiness of industrial data, impacting the overall storage efficiency of the system.

[0004] While some load-balancing-based resource scheduling methods have been proposed in the prior art, these methods primarily focus on certain local performance indicators of the system and lack global optimization of the overall operating status of the storage system. Furthermore, existing methods fail to fully leverage the advantages of deep learning technology when processing feature extraction and storage optimization for industrial data assets, making it difficult to achieve intelligent scheduling and adaptive optimization of storage resources. This results in low resource utilization efficiency during actual operation. In other words, the prior art faces the technical problem of difficulty in achieving intelligent scheduling of storage resources in distributed storage systems for industrial data assets. Summary of the Invention

[0005] In view of this, the present invention provides an industrial data asset distributed storage method, medium and system, which can solve the technical problem in the prior art that the industrial data asset distributed storage system is difficult to achieve intelligent scheduling of storage resources.

[0006] The present invention is implemented as follows: In a first aspect, the present invention provides a method for distributed storage of industrial data assets, comprising the following steps: collecting manufacturing equipment operating time data, manufacturing equipment maintenance cycle data, production task quantity data, and data generation frequency data as industrial original data streams; establishing a resource consumption calculation model based on the industrial original data streams; using the resource consumption calculation model to extract data time series eigenvalues, data space eigenvalues, and data association eigenvalues ​​of the industrial original data streams, and constructing a data feature mapping matrix; constructing a resource scheduling matrix based on the data feature mapping matrix, and inputting the resource scheduling matrix into a storage optimization neural network; according to The resource scheduling convolution kernel of the storage optimization neural network obtains an optimized scheduling convolution kernel, wherein the resource weight coefficient of the optimized scheduling convolution kernel is obtained based on the ratio of the corresponding weight of the resource scheduling convolution kernel to the benchmark weight coefficient; the optimized scheduling convolution kernel is applied to the resource scheduling matrix to obtain a resource feature distribution map; a deep optimized convolution kernel having a mapping relationship with the resource scheduling convolution kernel in the deep network of the storage optimization neural network is obtained; when the ratio of the resource weight coefficient of the resource scheduling convolution kernel to the benchmark weight coefficient is greater than the storage capacity threshold, the corresponding weight of the resource scheduling convolution kernel is proportionally reduced.

[0007] Among them, the resource consumption calculation model normalizes the manufacturing equipment operating time data, the manufacturing equipment maintenance cycle data, the production task quantity data and the data generation frequency data, and then calculates the data storage demand index through a polynomial fitting equation. The data storage demand index is used to characterize the storage resource consumption level of the industrial raw data stream.

[0008] Among them, the storage optimized neural network consists of an input layer, a first convolutional layer, a second convolutional layer, a third convolutional layer, a first fully connected layer, a second fully connected layer and an output layer. The training data set of the storage optimized neural network is obtained by collecting the data block storage location, storage space utilization, data access delay parameters, storage node load parameters and data redundancy parameters of the industrial site for 365 days. The storage optimized neural network is trained using a back propagation algorithm, and the network parameters are optimized and adjusted using the training data set.

[0009] Among them, a feature extraction operation is performed on the resource feature distribution map to obtain the space division parameters of the storage basic unit and obtain the data access delay parameters of the storage basic unit, wherein the feature extraction operation includes a weighted accumulation operation on the resource weight coefficient and the feature value of the resource feature distribution map.

[0010] Among them, the distribution parameters of the storage optimization unit are determined based on the deep optimization convolution kernel, and the distribution parameters include storage node load parameters and data redundancy parameters; distributed nodes are allocated to the storage optimization unit, and a storage address mapping table is established based on the storage space utilization and the data access delay parameters.

[0011] Wherein, a data sharding operation is performed according to the storage address mapping table, and a data block storage location is determined based on the storage node load parameter and the data redundancy parameter.

[0012] Among them, the deep network of the storage optimization neural network is used to dynamically evaluate the data block storage location, and a storage performance indicator is generated based on the storage space utilization, the data access delay parameter, the storage node load parameter and the data redundancy parameter.

[0013] Among them, the storage load factor is calculated based on the storage performance indicator. When the storage load factor is less than the storage load threshold, the capacity of the data block storage location is expanded; the data block storage location is optimized and adjusted according to the storage load factor to complete the distributed storage of the industrial original data stream.

[0014] A second aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores program instructions. When the program instructions are run in a computer, the program instructions are used to execute the above-mentioned industrial data asset distributed storage method.

[0015] A third aspect of the present invention provides an industrial data asset distributed storage system, comprising the above-mentioned computer-readable storage medium, wherein the system is any one of a computer, a server, and a single-chip microcomputer, and the computer-readable storage medium is arranged in the system, and the system is provided with a microprocessor for executing program instructions stored in the computer-readable storage medium.

[0016] Compared with the prior art, the present invention provides a distributed storage method, medium, and system for industrial data assets. This method, by constructing a resource consumption calculation model and a storage optimization neural network, achieves intelligent extraction of industrial data asset storage characteristics and dynamic optimization of storage resources. This method first establishes a data feature mapping matrix based on the industrial raw data stream, using a multi-dimensional feature expression method to comprehensively characterize the temporal, spatial, and correlation characteristics of the data. Then, through the collaborative optimization of deep optimization convolution kernels and resource scheduling matrices, intelligent scheduling and adaptive allocation of storage resources are achieved.

[0017] The method of this invention organically combines deep learning technology with distributed storage systems by introducing a storage optimization neural network, overcoming the limitations of static rule design in traditional methods. Based on the feature mapping mechanism of the deep network, the system accurately grasps the dynamic changes in industrial data assets and optimizes storage resources in real time. Furthermore, by establishing an evaluation system for storage performance indicators and load factors, it enables global monitoring and optimization of the storage system's operating status, effectively addressing issues such as storage hotspots and load imbalances.

[0018] This invention successfully solves the technical problems of intelligent storage resource scheduling and load balancing in distributed storage systems for industrial data assets. Through a deep learning-driven resource optimization mechanism, it achieves efficient storage resource utilization and continuous optimization of system performance. This method not only improves the system's overall storage efficiency but also enhances its adaptability to changes in data access patterns, resolving the existing technical problem of the difficulty in intelligently scheduling storage resources in distributed storage systems for industrial data assets. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a flow chart of the method of the present invention.

[0020] Figure 2 This is a graph showing the changing trend of the equipment operation data within 31 days in Example 2.

[0021] Figure 3 This is a distribution diagram of the storage demand index of 50 devices in Example 2.

[0022] Figure 4 This is a graph showing the contribution rate analysis results of each eigenvalue in Example 2.

[0023] Figure 5 This is a comparison chart of various performance indicators before and after system optimization in Example 2. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0025] like Figure 1 FIG. 1 is a flow chart of a method for distributed storage of industrial data assets provided by the first aspect of the present invention. The method includes the following steps:

[0026] S01. Collect manufacturing equipment operation time data, manufacturing equipment maintenance cycle data, production task quantity data, and data generation frequency data as industrial raw data streams;

[0027] S02. Establishing a resource consumption calculation model based on the industrial raw data stream, wherein the resource consumption calculation model generates a data storage demand index based on the manufacturing equipment operation time data, the manufacturing equipment maintenance cycle data, the production task quantity data, and the data generation frequency data;

[0028] S03. Extracting data time series eigenvalues, data space eigenvalues, and data association eigenvalues ​​of the industrial raw data stream using the resource consumption calculation model, and constructing a data feature mapping matrix based on the data time series eigenvalues, the data space eigenvalues, and the data association eigenvalues;

[0029] S04. Constructing a resource scheduling matrix based on the data feature mapping matrix, and inputting the resource scheduling matrix into a storage optimization neural network;

[0030] S05. Obtain an optimized scheduling convolution kernel according to the resource scheduling convolution kernel of the storage optimized neural network, wherein the resource weight coefficient of the optimized scheduling convolution kernel is obtained based on the ratio of the corresponding weight of the resource scheduling convolution kernel to the reference weight coefficient in the resource scheduling convolution kernel;

[0031] S06. Applying the optimized scheduling convolution kernel to the resource scheduling matrix to obtain a resource feature distribution map, and determining the storage space utilization rate of the data storage node based on the resource feature distribution map;

[0032] S07. Perform a feature extraction operation on the resource feature distribution map to obtain a space partition parameter of a storage basic unit and acquire a data access delay parameter of the storage basic unit, wherein the feature extraction operation includes performing a weighted accumulation operation on the resource weight coefficient and a feature value of the resource feature distribution map;

[0033] S08. Obtaining a deep optimized convolution kernel in a deep network of the storage optimized neural network that has a mapping relationship with the resource scheduling convolution kernel;

[0034] S09. When the ratio of the resource weight coefficient of the resource scheduling convolution kernel to the reference weight coefficient is greater than a storage capacity threshold, performing proportional reduction processing on the corresponding weight of the resource scheduling convolution kernel;

[0035] S10. Determine distribution parameters of a storage optimization unit based on the deep optimization convolution kernel, where the distribution parameters include a storage node load parameter and a data redundancy parameter;

[0036] S11, performing distributed node allocation on the storage optimization unit, and establishing a storage address mapping table based on the storage space utilization and the data access delay parameter;

[0037] S12, performing a data sharding operation according to the storage address mapping table, and determining a data block storage location based on the storage node load parameter and the data redundancy parameter;

[0038] S13, using the deep network of the storage optimization neural network to dynamically evaluate the data block storage location, and generating a storage performance indicator based on the storage space utilization, the data access delay parameter, the storage node load parameter, and the data redundancy parameter;

[0039] S14, calculating a storage load factor based on the storage performance indicator, and when the storage load factor is less than a storage load threshold, performing capacity expansion processing on the data block storage location;

[0040] S15. Optimize and adjust the data block storage location according to the storage load factor to complete the distributed storage of the industrial original data stream.

[0041] The resource consumption calculation model normalizes the manufacturing equipment operating time data, the manufacturing equipment maintenance cycle data, the production task quantity data and the data generation frequency data, and then calculates the data storage demand index through a polynomial fitting equation. The data storage demand index is used to characterize the storage resource consumption level of the industrial raw data stream.

[0042] The storage-optimized neural network consists of an input layer, a first convolutional layer, a second convolutional layer, a third convolutional layer, a first fully connected layer, a second fully connected layer, and an output layer. The training data set of the storage-optimized neural network is obtained by collecting data block storage locations, storage space utilization, data access delay parameters, storage node load parameters, and data redundancy parameters from an industrial site for 365 days. The storage-optimized neural network is trained using a backpropagation algorithm, and the network parameters are optimized and adjusted using the training data set.

[0043] The storage performance index is calculated based on the storage space utilization, the data access delay parameter, the storage node load parameter and the data redundancy parameter through a weighted average equation. The storage performance index is used to evaluate the overall operating efficiency of the distributed storage system.

[0044] The storage load factor is calculated based on the storage space utilization of the storage block storage location, the data access delay parameter, the storage node load parameter and the data redundancy parameter through the standard deviation equation. The storage load factor is used to characterize the degree of load balancing between the nodes of the distributed storage system.

[0045] The specific implementation of the above steps is described in detail below. The specific implementation of step S01 is to collect equipment operation data through a data acquisition module deployed on the manufacturing equipment. The module contains a time series database for storing equipment operation time data, and records the equipment startup time, downtime and cumulative operation time in real time through a preset data acquisition frequency; at the same time, a maintenance management system is configured to record the time intervals and completion status of maintenance activities such as regular equipment maintenance and repairs; production task data such as the number of production task dispatch orders and task completion status are obtained in the production execution system; the generation frequency of various types of data is obtained based on the data sampling period of the time series database, and this data is input into the subsequent processing link as the original data stream. The purpose of this step is to establish the basic data source of industrial data assets.

[0046] The specific implementation method of step S02 is to first normalize all types of data in the industrial raw data stream, and uniformly map the numerical range to between 0 and 1. Specifically, the maximum and minimum value normalization method is adopted, that is, for any data x, the normalized value is the data minus the minimum value of the data sequence divided by the difference between the maximum and minimum values ​​of the data sequence; then, the normalized equipment operation time data, maintenance cycle data, task quantity data and data generation frequency data are substituted into the polynomial fitting equation. The equation adopts the form of a cubic polynomial, and the polynomial coefficients are determined by the least squares method to calculate the data storage demand index. The larger the index, the higher the degree of storage resource consumption. This step aims to quantitatively evaluate the degree of storage resource demand of industrial data assets.

[0047] The specific implementation method of step S03 is to first use the resource consumption calculation model to perform time series analysis on the industrial raw data stream, and use the sliding window method to calculate the time correlation of the data. The window size is set to 24 hours and the step size is 1 hour to obtain the time series eigenvalues ​​reflecting the time distribution characteristics of the data; secondly, spatial feature extraction is performed, and the principal component analysis method is used to reduce the dimension of the data, and the principal component with a cumulative contribution rate of 85% is retained as the spatial eigenvalue; again, the correlation characteristics between the data are calculated, and the Pearson correlation coefficient method is used to analyze the correlation between data items to obtain the correlation eigenvalue; finally, these three types of eigenvalues ​​are organized into a three-dimensional feature mapping matrix. The purpose of this step is to characterize the characteristic attributes of industrial data assets from multiple dimensions.

[0048] The specific implementation method of step S04 is to construct a resource scheduling matrix based on the data feature mapping matrix. Specifically, the feature mapping matrix is ​​normalized by using the z-score normalization method, that is, the original data is subtracted from the mean and divided by the standard deviation; then the singular value decomposition algorithm is applied to the standardized matrix to extract the main eigenvectors to construct the resource scheduling matrix; finally, the matrix is ​​input into a pre-trained storage optimization neural network. This step is intended to provide a basic data structure for subsequent storage optimization.

[0049] The specific implementation method of step S05 is to first obtain the resource scheduling convolution kernel in the storage optimization neural network, and the initial weight of the convolution kernel is obtained by random initialization of the Gaussian distribution; then compare the convolution kernel with the preset benchmark weight coefficient, and the benchmark weight coefficient is determined by statistical analysis of historical data and set to 0.5; calculate the ratio of each weight in the resource scheduling convolution kernel to the benchmark weight coefficient, and use the ratio as the resource weight coefficient of the optimized scheduling convolution kernel. The purpose of this step is to construct an optimized convolution kernel that adapts to current storage needs.

[0050] The specific implementation method of step S06 is to apply the optimized scheduling convolution kernel to the resource scheduling matrix, perform convolution operation using a sliding window with a window size of 3×3 and a step size of 1 to obtain a feature map; then perform a pooling operation on the feature map, and use the maximum pooling method to extract local features to obtain a resource feature distribution map; based on the distribution map, calculate the storage space usage of each storage node to obtain the storage space utilization rate. This step aims to obtain the distribution status of storage resources.

[0051] The specific implementation method of step S07 is to extract features from the resource feature distribution map. First, the weighted sum of the resource weight coefficient and the feature value is calculated, and a linear weighting method is used. The weight is set according to the importance of the feature, and features with high importance are given larger weights; then, based on the weighted results, the spatial division parameters of the storage basic unit are determined, including unit size, distribution density, etc.; at the same time, the data access delay of the storage basic unit is obtained through the network detection method, and the time interval between the data request being sent and the response being received is recorded. The purpose of this step is to determine the basic configuration parameters of the storage system.

[0052] The specific implementation method of step S08 is to search for a deep convolution kernel that has a feature mapping relationship with the resource scheduling convolution kernel in the deep network of the storage optimization neural network. First, the feature similarity of the two convolution kernels is calculated using the cosine similarity method; then the deep convolution kernel with the highest similarity is selected as the deep optimization convolution kernel. This step aims to establish a correspondence between shallow features and deep features.

[0053] The specific implementation method of step S09 is to compare the ratio of the resource weight coefficient of the resource scheduling convolution kernel to the benchmark weight coefficient. When the ratio exceeds the storage capacity threshold, the convolution kernel weight needs to be adjusted, and the storage capacity threshold is set to 1.5; specifically, a linear proportional reduction method is used to proportionally reduce the weight that exceeds the threshold to meet the storage capacity requirements. The purpose of this step is to prevent over-allocation of storage resources.

[0054] The specific implementation method of step S10 is to determine the distribution parameters of the storage optimization unit based on the deep optimization convolution kernel. First, the load parameters of the storage node are calculated, including CPU utilization, memory utilization, network bandwidth utilization, etc., and the weighted average method is used to integrate various indicators; secondly, the data redundancy parameter is calculated, that is, the ratio of the number of data copies to the number of storage nodes. This step aims to provide a basis for storage optimization configuration.

[0055] The specific implementation method of step S11 is to distribute nodes to the storage optimization unit, and use the consistent hashing algorithm to map the storage space to the hash ring to ensure that only adjacent nodes are affected when nodes are added or deleted; then a storage address mapping table is established based on the storage space utilization and access delay parameters, and the correspondence between data identifiers and physical storage locations is recorded in the form of key-value pairs. The purpose of this step is to achieve balanced distribution of storage resources.

[0056] The specific implementation method of step S12 is to shard the data according to the storage address mapping table, and use the range sharding method to divide the continuous data into data blocks of equal size; then select the appropriate storage node based on the storage node load parameter and data redundancy parameter, and distribute the data blocks to the nodes with lower load and meeting the redundancy requirements. This step aims to realize distributed storage of data.

[0057] The specific implementation method of step S13 is to use the deep network of the storage optimization neural network to evaluate the storage location of the data block. First, the real-time operating status data of each node is collected; then the storage space utilization, access delay parameters, node load parameters and data redundancy parameters are input into the deep network, and the forward propagation algorithm is used to calculate the storage performance index. This index comprehensively considers various parameters through the weighted average method, and the weight is set according to the importance of the parameters. The purpose of this step is to evaluate the operating effect of the storage system.

[0058] The specific implementation method of step S14 is to calculate the storage load factor based on the storage performance index, and use the standard deviation method to calculate the discrete degree of each node performance index. The smaller the standard deviation, the more balanced the load. When the storage load factor is less than the storage load threshold, capacity expansion is required, and the storage load threshold is set to 0.3. This is specifically achieved by adding storage nodes or expanding the capacity of existing nodes. This step is intended to ensure load balance of the storage system.

[0059] The specific implementation method of step S15 is to optimize the storage location of data blocks according to the storage load factor. When the load factor of a node is too high, some data blocks are migrated to nodes with lower load. The migration process adopts an incremental migration strategy to ensure that system services are not interrupted. Through cyclic optimization and adjustment, balanced storage of industrial raw data streams is finally achieved. The purpose of this step is to maintain the stable operation of the storage system.

[0060] The data storage demand index in the resource consumption calculation model is specifically expressed as follows:

[0061] DSI=α1T n +α2M n +α3P n +α4F n +β(T n M n ) 2 +γ(P n F n ) 2 +δ;

[0062] Where DSI is the data storage demand index; T n is the normalized device operation time data; M n is the normalized equipment maintenance cycle data; P n is the normalized production task quantity data; F n Generate frequency data for the normalized data; α1, α2, α3, α4 are linear term coefficients; β, γ are quadratic term coefficients; δ is the error correction term.

[0063] The data feature mapping matrix is ​​specifically expressed as follows:

[0064]

[0065] Where, t ij is the jth time series characteristic value of the i-th data item; s ij is the jth spatial eigenvalue of the i-th data item; r ij is the jth associated eigenvalue of the i-th data item; m is the number of data items; and n is the feature dimension.

[0066] The resource scheduling matrix is ​​specifically expressed as follows:

[0067]

[0068] Where μ is the mean of the feature mapping matrix; σ is the standard deviation of the feature mapping matrix.

[0069] The resource weight coefficient of the optimized scheduling convolution kernel is specifically expressed as follows:

[0070]

[0071] Where w opt is the optimized weight coefficient; w sch The weight of the convolution kernel for resource scheduling; w base is the baseline weight coefficient; λ is the attenuation coefficient, which is used to control the smoothness of weight adjustment.

[0072] The resource characteristic distribution map is specifically shown as follows:

[0073] F dist =Pool(Conv(M sch , K opt ));

[0074] Where Conv represents the convolution operation; K opt To optimize the scheduling of convolution kernels; Pool represents the maximum pooling operation.

[0075] The storage space utilization is specifically expressed as follows:

[0076]

[0077] Where U is the storage space utilization; S used S is the used storage space; total is the total storage space; T access is the data access time; η is the time decay factor.

[0078] The storage performance indicators are specifically expressed as follows:

[0079]

[0080] Where PI is the storage performance index; L node is the node load parameter; R data is the data redundancy parameter; ω1, ω2, ω3, ω4 are weight coefficients.

[0081] The storage load factor is specifically expressed as follows:

[0082]

[0083] Where LF is the storage load factor; N is the number of storage nodes; PI i is the storage performance index of the i-th node; It is the average storage performance index of all nodes.

[0084] Parameter acquisition method description:

[0085] 1. Normalized data is obtained by the maximum and minimum normalization method:

[0086] 2. The linear term coefficients α1, α2, α3, α4 and the quadratic term coefficients β, γ are obtained by least squares fitting;

[0087] 3. The value range of the error correction term δ is 0.01 to 0.1;

[0088] 4. The value range of the attenuation coefficient λ is 0.1 to 1.0;

[0089] 5. The value range of the time decay factor η is 0.001 to 0.01;

[0090] 6. The weight coefficients ω1, ω2, ω3, and ω4 are determined by the analytic hierarchy process and satisfy

[0091] Explanation of the principle of equation construction:

[0092] 1. The data storage demand index equation is in polynomial form and contains linear and quadratic terms. The linear terms represent the independent effects of each factor, the quadratic terms represent the interactions between factors, and the error correction term is used to compensate for model errors.

[0093] 2. The data feature mapping matrix adopts a three-dimensional structure to uniformly represent temporal, spatial and correlation features, facilitating subsequent processing;

[0094] 3. The resource scheduling matrix eliminates dimensionality effects through standardization and improves the comparability of features;

[0095] 4. An exponential decay term is introduced into the weight coefficient equation of the optimized scheduling convolution kernel to ensure the smoothness and stability of weight adjustment;

[0096] 5. The storage space utilization equation takes into account the impact of access time and introduces an exponential term to represent the nonlinear effect of time on utilization;

[0097] 6. The storage performance index equation uses a weighted sum to comprehensively consider multiple performance parameters. The access time and node load are calculated in the inverse form, indicating that the smaller these parameters, the better.

[0098] 7. The storage load factor equation uses the standard deviation form to measure the degree of balance of the system load.

[0099] The following is a detailed description of the derivation and establishment of some calculation processes.

[0100] The derivation process of the data storage demand index DSI equation is as follows:

[0101] First build the basic linear model: DSI base =α1T n +α2M n +α3P n +α4F n ;

[0102] Considering the mutual influence between equipment operation time and maintenance cycle, the interaction term is introduced: (T n M n ) 2 ;

[0103] Considering the correlation between production tasks and data generation frequency, the interaction term is introduced: (P n F n ) 2 ;

[0104] Finally, the error correction term δ is added to obtain the complete equation.

[0105] The coefficients α1, α2, α3, and α4 are obtained through historical data training. The specific steps are:

[0106] 1. Collect 6 months of historical data and build a training set;

[0107] 2. Use gradient descent to minimize the mean square error;

[0108] 3. Determine the optimal coefficient value through cross-validation.

[0109] Data feature mapping matrix M map The derivation of:

[0110] Time series characteristic value t ij The calculation formula is:

[0111]

[0112] Where x k is the data value of the kth hour; w k is the time weight, satisfying

[0113] Spatial eigenvalue s ij Obtained through principal component analysis:

[0114] S = XV;

[0115] Where X is the original data matrix; V is the eigenvector matrix;

[0116] s ij Take the first few principal components with the largest contribution rate in the S matrix.

[0117] Correlation eigenvalue r ij The calculation formula is:

[0118]

[0119] Where x k ,y k is the data pair for which correlation is to be calculated; is the corresponding mean.

[0120] Resource scheduling matrix M sch Standardized processing instructions:

[0121] Calculation of mean μ:

[0122] Calculation of standard deviation σ:

[0123] Optimize the weight coefficient w of the convolution kernel opt Derivation process:

[0124] 1. Initialize weight coefficients:

[0125] 2. Set the benchmark weight: w base =0.5;

[0126] 3. Introduce distance penalty term: ||w sch -w base ||;

[0127] 4. Add exponential decay adjustment:

[0128] 5. Combine to get the final weight coefficient expression.

[0129] Resource characteristics distribution map F dist Supplementary explanation of calculation:

[0130] Convolution operation:

[0131] Max pooling: Pool(x)=max{x 11 , x 12 , x 21 , x 22}, the pooling window size is 2×2.

[0132] Optimization process of storage space utilization U equation:

[0133] 1. Basic utilization rate:

[0134] 2. Consider the impact of access time:

[0135] 3. Combine to get the final expression.

[0136] Access time T access Obtained through network probing, specifically by sending probe packets and recording the round-trip time.

[0137] The process of constructing the storage performance indicator PI equation is as follows:

[0138] 1. Identify key performance parameters: storage space utilization, access time, node load, and data redundancy;

[0139] 2. Use the inverse form of access time and node load:

[0140] 3. Introduce weight coefficients: ω1, ω2, ω3, ω4;

[0141] 4. Determine the weight value through hierarchical analysis method.

[0142] Supplementary explanation on the derivation of the storage load factor LF equation:

[0143] 1. Calculate the average performance index:

[0144] 2. Calculate the variance of performance indicators;

[0145] 3. Take the square root to get the standard deviation, which is used as the load factor.

[0146] A second aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores program instructions. When the program instructions are run in a computer, the program instructions are used to execute the above-mentioned industrial data asset distributed storage method.

[0147] A third aspect of the present invention provides an industrial data asset distributed storage system, comprising the above-mentioned computer-readable storage medium, wherein the system is any one of a computer, a server, and a single-chip microcomputer, and the computer-readable storage medium is arranged in the system, and the system is provided with a microprocessor for executing program instructions stored in the computer-readable storage medium.

[0148] Specifically, the present invention is based on the collaborative optimization concept of deep learning and distributed systems. By constructing a multi-level feature extraction and resource optimization mechanism, it achieves intelligent scheduling of storage resources. First, a resource consumption calculation model is used to extract features from industrial raw data streams. This model uses a polynomial fitting equation to comprehensively consider the impact of multiple dimensions, such as equipment operation, maintenance cycle, task volume, and data frequency, and accurately reflects the data's demand for storage resources.

[0149] Based on feature extraction, storage optimization is performed using a deep neural network. The network structure comprises multiple convolutional and fully connected layers, enabling automatic learning of deep feature representations of data. By optimizing the design of the scheduling convolution kernels, precise control of storage resource allocation is achieved. A weight adjustment mechanism is also introduced to ensure smooth and stable resource scheduling. During system operation, a comprehensive performance evaluation and optimization feedback mechanism is established through real-time calculation of storage performance indicators and load factors, ensuring balanced allocation and efficient utilization of storage resources.

[0150] The various technical links in this invention are closely logically linked, forming a complete technical closed loop from data feature extraction to resource optimization and scheduling, and then to performance evaluation and optimization feedback. This solution fully leverages the advantages of deep learning in feature learning and optimization decision-making, while also integrating the practical needs of distributed systems. It has a strong theoretical foundation and practical feasibility.

[0151] A specific embodiment 1 of the present invention is provided below. The specific implementation of each step in this embodiment 1 is described in detail as follows.

[0152] The specific implementation of step S01 is to collect equipment operation data through a data acquisition module deployed on the manufacturing equipment. This module contains a time series database for storing equipment operation time data. The module records the equipment startup time, downtime, and cumulative operation time in real time at a preset data acquisition frequency, with the acquisition frequency set to once per minute. At the same time, a maintenance management system is configured to record the time intervals and completion status of equipment maintenance activities such as regular maintenance and repairs. The maintenance data acquisition frequency is set to record immediately after each maintenance activity is completed. Production task data such as the number of production task dispatch orders and task completion status is obtained from the production execution system, with the task data acquisition frequency set to once per hour. The generation frequency of various types of data is obtained based on the data sampling period of the time series database. Data preprocessing techniques are used in the data collection process, including data cleaning, outlier detection, and data completion. Data cleaning uses the median filter method to remove noise, outlier detection uses the 3σ criterion to identify abnormal data points, and data completion uses the linear interpolation method to handle missing values. The purpose of this step is to establish the basic data source of industrial data assets.

[0153] The specific implementation of step S02 is to first normalize the various types of data in the industrial raw data stream and uniformly map the numerical range to between 0 and 1. This is specifically achieved through the maximum and minimum value normalization method. The normalization process uses the following formula: Where, X n is the normalized data value, X is the original data value, and X min is the minimum value of the data sequence, X max is the maximum value of the data sequence; then the data storage demand index calculation model is constructed: DSI = α1T n +α2M n +α3P n +α4F n +β(T n Mn ) 2 +γ(P n F n ) 2 +δ, where DSI is the data storage demand index, T nis the normalized device operation time data, M n is the normalized equipment maintenance cycle data, P n is the normalized production task quantity data, F n Frequency data is generated for the normalized data. α1, α2, α3, and α4 are linear term coefficients determined using the least squares method. β and γ are quadratic term coefficients obtained through gradient descent optimization. δ is an error correction term with a value range of 0.01 to 0.1. The model uses a polynomial fit equation containing linear and quadratic terms. The linear terms represent the independent influence of each factor, the quadratic terms represent the interaction between factors, and the error correction term is used to compensate for model errors. During model training, six months of historical data were used as the training set, and 10-fold cross-validation was used to evaluate model performance. The resulting model can accurately assess the storage resource requirements of industrial data assets.

[0154] The specific implementation of step S03 is to first use the resource consumption calculation model to perform time series analysis on the industrial raw data stream, and use the sliding window method to calculate the time series feature value. The specific calculation formula is: Where x k is the data value of the kth hour, w k is the time weight, satisfying Secondly, spatial feature extraction is performed, and the spatial eigenvalues ​​are calculated using the principal component analysis method. This is achieved through eigenvalue decomposition: S = XV, where X is the original data matrix and V is the eigenvector matrix. The correlation eigenvalues ​​between the data are calculated again using the Pearson correlation coefficient method. The calculation formula is: Where x k ,y k is the data pair to be correlated, is the corresponding mean; finally, these three types of eigenvalues ​​are organized into a three-dimensional feature mapping matrix: Where, t ij is the jth time series characteristic value of the i-th data item, s ij is the jth spatial eigenvalue of the i-th data item, r ij is the jth associated eigenvalue of the i-th data item, m is the number of data items, and n is the feature dimension. During feature extraction, a 24-hour sliding window is used for temporal features, and principal components with a cumulative contribution rate of 85% are retained for spatial features. Data is standardized and preprocessed when calculating associated features.

[0155] The specific implementation of step S04 is to construct a resource scheduling matrix based on the data feature mapping matrix, and eliminate the dimension effect through normalization processing. The normalization formula is: Where μ is the mean of the feature mapping matrix, and the calculation formula is: σ is the standard deviation of the feature map matrix, and the calculation formula is: The standardized matrix improves feature comparability and eliminates the impact of different dimensions. The resource scheduling matrix is ​​then input into a pre-trained storage optimization neural network. This network uses a convolutional neural network structure consisting of an input layer, a first convolutional layer, a second convolutional layer, a third convolutional layer, a first fully connected layer, a second fully connected layer, and an output layer. The network is trained using a backpropagation algorithm, and the training dataset is obtained by collecting 365 days of data from industrial sites. This step provides the foundational data structure for subsequent storage optimization.

[0156] The specific implementation of step S05 is to first obtain the resource scheduling convolution kernel in the storage optimization neural network, and the convolution kernel weight coefficient is randomly initialized through Gaussian distribution: The convolution kernel is then compared with the benchmark weight coefficient, which is determined through statistical analysis of historical data and set to 0.5. The resource weight coefficient of the optimized scheduling convolution kernel is calculated using the following formula: Where w opt is the optimized weight coefficient, w sch The weight of the convolution kernel for resource scheduling, w base is the baseline weight coefficient, λ is the attenuation coefficient, ranging from 0.1 to 1.0, which is used to control the smoothness of the weight adjustment. The calculation process of the weight coefficient introduces the distance penalty term ||w sch -w base || and exponential decay regulation Ensure the smoothness and stability of weight adjustment. The purpose of this step is to construct an optimized convolution kernel that adapts to the current storage requirements.

[0157] The specific implementation of step S06 is to apply the optimized scheduling convolution kernel to the resource scheduling matrix and perform convolution operation in a sliding window manner. The convolution calculation formula is: The window size is 3×3 and the step size is 1; then the convolution result is subjected to the maximum pooling operation: Pool(x)=max{x 11 , x 12 , x 21 , x 22}, the pooling window size is 2×2, and the resource feature distribution map F is obtained dit =Pool(Conv(M sch , K opt )); Calculate the storage space utilization based on the distribution graph using the following formula: Where U is the storage space utilization, S used is the used storage space, S total is the total storage space, T accessis the data access time, and η is the time decay factor, ranging from 0.001 to 0.01. The calculation of storage space utilization takes into account the impact of access time, introducing an exponential term to account for the nonlinear effect of time on utilization. This step aims to capture the distribution of storage resources.

[0158] The specific implementation method of step S07 is to perform feature extraction operations on the resource feature distribution map, calculate the weighted cumulative value of the resource weight coefficient and the feature value, and use a linear weighting method. The weight is set according to the importance of the feature, and features with high importance are given a larger weight; based on the weighted results, the spatial division parameters of the storage basic unit are determined, including unit size and distribution density; at the same time, a network detection method is used to obtain the data access delay parameters of the storage basic unit. Specifically, this is achieved by sending a detection packet and recording the round-trip time. The access delay test of each storage unit is repeated 10 times and the average value is taken. In the feature extraction process, different weight coefficients are used for different types of features. The calculation formula is: Where FE is the feature extraction result, w i is the weight coefficient of the i-th feature, f i is the ith eigenvalue, and n is the number of features. The purpose of this step is to determine the basic configuration parameters of the storage system.

[0159] The specific implementation of step S08 is to search for a deep convolution kernel that has a feature mapping relationship with the resource scheduling convolution kernel in the deep network of the storage optimization neural network, and first calculate the cosine similarity between the convolution kernels: Where K1, K2 are the convolution kernels to be compared, k 1i , k 2i The deep convolution kernel with the highest similarity is then selected as the deep optimization kernel. The similarity threshold is set to 0.8, meaning only kernels with a similarity greater than 0.8 are considered to have a mapping relationship. The selection of deep convolution kernels also takes the feature abstraction level into consideration; deeper convolution kernels are able to extract higher-level feature representations. This step aims to establish a correspondence between shallow and deep features.

[0160] The specific implementation of step S09 is that when the ratio of the resource weight coefficient of the resource scheduling convolution kernel to the reference weight coefficient exceeds the storage capacity threshold, the convolution kernel weight needs to be adjusted, and the storage capacity threshold is set to 1.5. The weight adjustment adopts a linear proportional reduction method, and the reduction calculation formula is: Where w new is the adjusted weight value, w old is the original weight value, T threshold is the storage capacity threshold, w ratiois the original weight ratio. A smoothing factor is introduced in the weight adjustment process to avoid weight mutation: w smooth =αw new +(1-α)w old , where α is the smoothing coefficient, ranging from 0.1 to 0.3, w smooth is the weight value after smoothing. This step ensures that storage resources are not over-allocated by reducing the weight, while smoothing ensures the stability of weight adjustment.

[0161] The specific implementation of step S10 is to determine the distribution parameters of the storage optimization unit based on the deep optimization convolution kernel. First, the load parameters of the storage node are calculated, and the weighted average method is used: L node =ω1C usage +ω2M usage +ω3B usage , where L node is the node load parameter, C usage is the CPU usage, M usage is the memory usage, B usage is the network bandwidth utilization rate, ω1, ω2, ω3 are weight coefficients, determined by the hierarchical analysis method; then calculate the data redundancy parameter: Where N replica is the number of data copies, N node is the number of storage nodes. Considering the load balancing between nodes, the variance term is introduced: Where V load is the load variance, L i is the load value of the i-th node, is the average load value. This step is intended to provide a basis for optimizing storage configuration.

[0162] The specific implementation of step S11 is to distribute nodes to the storage optimization unit and use the consistent hashing algorithm to map the storage space to the hash ring. The hash calculation uses virtual node technology. Each physical node corresponds to multiple virtual nodes. The hash value calculation formula of the virtual node is: v =hash(h p +i), where h v is the hash value of the virtual node, h p is the physical node hash value, and i is the virtual node index. On this basis, a storage address mapping table is established by combining storage space utilization and access delay parameters. The scoring function of the mapping relationship is: In the formula, β1, β2 are weight coefficients, U is the storage space utilization, T access The purpose of this step is to achieve a balanced distribution of storage resources.

[0163] The specific implementation of step S12 is to perform data sharding according to the storage address mapping table, adopt the range sharding method, and dynamically adjust the shard size. The adjustment formula is: S chunk =S base (1+γ·L ratio ), where S chunk is the shard size, S base is the basic shard size, set to 64MB, γ is the adjustment coefficient, ranging from 0 to 0.5, L ratio is the load ratio. The selection of data block storage location is based on a comprehensive scoring mechanism, and the scoring calculation formula is: In the formula, λ1, λ2 are weight coefficients, L node is the node load parameter, R data is the data redundancy parameter. This step aims to achieve distributed storage of data and ensure storage efficiency through dynamic sharding and scoring mechanisms.

[0164] The specific implementation of step S13 is to use the deep network of the storage optimization neural network to dynamically evaluate the storage location of the data block. First, the real-time operating status data of the storage node is collected, including storage space utilization, access latency parameters, node load parameters, and data redundancy parameters. The evaluation process uses the forward propagation algorithm to calculate the storage performance index. The index calculation formula is: In the formula, PI is the storage performance index, U is the storage space utilization, T access For access delay parameters, L node is the node load parameter, R data is the data redundancy parameter, ω1, ω2, ω3, ω4 are weight coefficients, and satisfy The weight coefficient is determined by the hierarchical analysis method to construct the judgment matrix: The weight value is calculated using the eigenvalue method. The purpose of this step is to evaluate the performance of the storage system.

[0165] The specific implementation of step S14 is to calculate the storage load factor based on the storage performance index, and use the standard deviation method to calculate the dispersion degree of each node performance index. The calculation formula is: Where LF is the storage load factor, N is the number of storage nodes, and PI is i is the storage performance index of the i-th node, is the average storage performance index of all nodes. When the storage load factor is less than the storage load threshold, capacity expansion is required. The storage load threshold is set to 0.3. Capacity expansion adopts a dynamic threshold adjustment strategy. The threshold adjustment formula is: Where, T new is the new threshold, T oldis the original threshold, δ is the adjustment coefficient, the value range is 0.1 to 0.3, LF base The baseline load factor is set to 0.5. This step is to ensure load balancing of the storage system.

[0166] The specific implementation of step S15 is to optimize the data block storage location according to the storage load factor, and start data migration when the node load factor is too high. The migration priority calculation formula is: Where, P migrate Migration priority, LF i is the current node load factor, LF avg is the average load factor, S free is the remaining storage space of the target node, S total is the total storage space, μ1 and μ2 are weight coefficients. The migration process adopts an incremental migration strategy, and the calculation formula for the amount of data migrated each time is: Where S migrate is the amount of data to be migrated, S block is the data block size, LF target is the target load factor, P migrate is the migration time, and κ is the time decay coefficient. This step optimizes the storage system load through dynamic data migration, ultimately achieving balanced storage of industrial raw data streams.

[0167] To better understand and implement the present invention, Example 2 of a specific application scenario is provided below: A manufacturing enterprise has 50 CNC machine tools. Before implementing the present invention's method, it used a traditional centralized storage system to manage industrial data. As data volume increased, system performance gradually declined, storage resource utilization became uneven, and data access latency increased significantly. The research team decided to use the present invention's method to transform the system, and the specific implementation process is as follows.

[0168] First, the research team collected operating data from manufacturing equipment over a one-month period. The statistical results of equipment operating time, maintenance cycle data, production task quantity data, and data generation frequency data are shown in Table 1:

[0169] Table 1 Equipment operation data statistics

[0170] Data Type Minimum Maximum average value Standard deviation Running time (hours / day) 16.5 23.8 20.4 1.8 Maintenance cycle (days) 25 45 35 5.2 Production task quantity (pieces / day) 85 168 126 22.5 Data generation frequency (items / minute) 120 360 240 60

[0171] Figure 2The data shows the changing trends of equipment operation data over a 31-day period, including the dynamic changes in three indicators: daily operating time, maintenance cycle, and production task volume. The collected data was normalized to obtain normalized data, and then the data storage demand index was calculated based on the resource consumption calculation model. The model parameters were determined using the least squares fitting method: α1 = 0.35, α2 = 0.25, α3 = 0.20, α4 = 0.20, β = 0.15, γ = 0.12, and δ = 0.05. The distribution of the data storage demand index for each device is shown in Table 2:

[0172] Table 2 Data storage demand index distribution table

[0173]

[0174]

[0175] Figure 3 The distribution of storage demand indices for 50 devices is displayed, with color depth visually indicating the storage demand intensity of different devices. Based on the data feature analysis results, the research team constructed a data feature mapping matrix and used a 24-hour sliding window for time series analysis, with a time weight decay coefficient set to 0.042. Principal component analysis was used to extract spatial features, selecting the principal component with a cumulative contribution rate of 85%. The main eigenvalues ​​obtained are shown in Table 3:

[0176] Table 3 Data characteristic value analysis table

[0177] Feature Type Eigenvalue Contribution rate Timing characteristics 1 0.384 0.35 Timing characteristics 2 0.256 0.23 Spatial Features 1 0.215 0.18 Spatial Features 2 0.168 0.15 Related feature 1 0.125 0.09

[0178] Figure 4 The contribution analysis results of each eigenvalue are presented. A bar chart shows the magnitude of each eigenvalue, and a line chart shows the cumulative contribution rate. The feature map matrix was input into a storage-optimized neural network. The network structure consisted of three convolutional layers and two fully connected layers. The training dataset contained 365 days of historical data, with a batch size of 64, a learning rate of 0.001, and 200 training epochs. The performance indicators of the trained network on the test set are shown in Table 4:

[0179] Table 4 Neural network performance evaluation table

[0180] Evaluation Metrics training set Test set Accuracy 0.945 0.923 Recall 0.932 0.908 F1 score 0.938 0.915 Loss value 0.058 0.075

[0181] Based on the trained neural network, the resource weight coefficient of the optimized scheduling convolution kernel is calculated, the attenuation coefficient λ is set to 0.5, and the baseline weight coefficient is 0.5. Applying the optimized scheduling convolution kernel to the resource scheduling matrix, the storage resource distribution obtained is shown in Table 5:

[0182] Table 5 Storage resource distribution table

[0183] Storage Node Space utilization Access delay (ms) Load parameters Redundancy Nodes 1-5 0.75 25 0.68 2.0 Nodes 6-10 0.68 28 0.62 2.0 Nodes 11-15 0.62 30 0.55 1.5

[0184] The weight coefficients of the storage performance indicators were determined using the analytic hierarchy process: ω1 = 0.3, ω2 = 0.25, ω3 = 0.25, and ω4 = 0.2. The performance evaluation results after one month of system operation are shown in Table 6.

[0185] Table 6 System performance evaluation table

[0186] Performance indicators Before optimization After optimization Average response time (ms) 85 28 Storage space utilization 0.45 0.68 Load balancing 0.58 0.85 System throughput (MB / s) 256 512

[0187] During system operation, when the storage load factor is detected to be less than the threshold of 0.3, the capacity expansion mechanism is automatically triggered. Figure 5 The comparison of various performance indicators before and after system optimization is shown, including average response time, storage space utilization, load balancing, and system throughput. Within one month, the system underwent three capacity expansions and 15 data block migration operations. The relevant parameters of the migration process are shown in Table 7:

[0188] Table 7 Data migration parameters

[0189] Migration batches Migrated data volume (GB) Migration time (min) Impact on business time (s) Batch 1 256 45 0 Batch 2 384 62 0 Batch 3 512 85 0

[0190] Compared with traditional distributed storage solutions, the solution of the present invention has significant advantages: traditional solutions use fixed data sharding strategies and static resource allocation methods, which often lead to storage hotspots and resource waste when faced with dynamically changing industrial data. The present invention uses deep learning technology to achieve intelligent scheduling of storage resources, and the system can adaptively adjust resource allocation strategies based on data characteristics. Experimental results show that the solution of the present invention has achieved significant improvements in storage space utilization, system response time, and load balancing. In addition, the solution of the present invention also achieves zero business interruption in the data migration process, ensuring the continuous availability of the system. Through the resource optimization mechanism driven by deep learning, the present invention successfully solves the problems of intelligent scheduling and load balancing in the distributed storage system of industrial data assets.

[0191] It should be noted that the variables involved in the present invention are explained in detail as shown in Table 8 below.

[0192] Table 8 Variable explanation table

[0193]

[0194]

[0195] The above description is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed by the present invention, which should be covered by the scope of protection of the present invention.

Claims

1. A distributed storage method for industrial data assets, characterized in that: The following steps are involved: Collect manufacturing equipment operation time data, manufacturing equipment maintenance cycle data, production task quantity data and data generation frequency data as industrial raw data streams; Establishing a resource consumption calculation model based on the industrial raw data stream; extracting data time series eigenvalues, data space eigenvalues, and data association eigenvalues ​​of the industrial raw data stream using the resource consumption calculation model, and constructing a data feature mapping matrix; Constructing a resource scheduling matrix based on the data feature mapping matrix, and inputting the resource scheduling matrix into a storage optimization neural network; obtaining an optimized scheduling convolution kernel according to the resource scheduling convolution kernel of the storage optimization neural network, wherein a resource weight coefficient of the optimized scheduling convolution kernel is obtained based on a ratio of a corresponding weight of the resource scheduling convolution kernel to a reference weight coefficient; Applying the optimized scheduling convolution kernel to the resource scheduling matrix to obtain a resource feature distribution map; Obtaining a deep optimized convolution kernel in a deep network of the storage optimized neural network that has a mapping relationship with the resource scheduling convolution kernel; When the ratio of the resource weight coefficient of the resource scheduling convolution kernel to the reference weight coefficient is greater than a storage capacity threshold, performing proportional reduction processing on the corresponding weight of the resource scheduling convolution kernel; Determining distribution parameters of a storage optimization unit based on the deep optimization convolution kernel, wherein the distribution parameters include a storage node load parameter and a data redundancy parameter; Distributed node allocation is performed on the storage optimization unit, and a storage address mapping table is established based on storage space utilization and data access delay parameters.

2. The distributed storage method for industrial data assets according to claim 1, characterized in that: The resource consumption calculation model normalizes the manufacturing equipment operating time data, the manufacturing equipment maintenance cycle data, the production task quantity data and the data generation frequency data, and then calculates the data storage demand index through a polynomial fitting equation. The data storage demand index is used to characterize the storage resource consumption level of the industrial raw data stream.

3. The distributed storage method for industrial data assets according to claim 1, characterized in that: The storage-optimized neural network consists of an input layer, a first convolutional layer, a second convolutional layer, a third convolutional layer, a first fully connected layer, a second fully connected layer, and an output layer. The training data set of the storage-optimized neural network is obtained by collecting data block storage locations, storage space utilization, data access delay parameters, storage node load parameters, and data redundancy parameters from an industrial site for 365 days. The storage-optimized neural network is trained using a backpropagation algorithm, and the network parameters are optimized and adjusted using the training data set.

4. The distributed storage method for industrial data assets according to claim 1, characterized in that: A feature extraction operation is performed on the resource feature distribution map to obtain a space division parameter of the storage basic unit and a data access delay parameter of the storage basic unit, wherein the feature extraction operation includes a weighted accumulation operation on the resource weight coefficient and the feature value of the resource feature distribution map.

5. The distributed storage method for industrial data assets according to claim 4, characterized in that: A data sharding operation is performed according to the storage address mapping table, and a data block storage location is determined based on the storage node load parameter and the data redundancy parameter.

6. The distributed storage method for industrial data assets according to claim 5, characterized in that: The deep network of the storage optimization neural network is used to dynamically evaluate the data block storage location, and a storage performance indicator is generated based on the storage space utilization, the data access delay parameter, the storage node load parameter and the data redundancy parameter.

7. The distributed storage method for industrial data assets according to claim 6, characterized in that: A storage load factor is calculated based on the storage performance indicator. When the storage load factor is less than a storage load threshold, the capacity of the data block storage location is expanded. The data block storage location is optimized and adjusted according to the storage load factor to complete the distributed storage of the industrial raw data stream.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium stores program instructions, and when the program instructions are run in a computer, they are used to execute the industrial data asset distributed storage method according to any one of claims 1 to 7.

9. An industrial data asset distributed storage system, characterized in that: The system comprises the computer-readable storage medium according to claim 8, wherein the system is any one of a computer, a server, and a single-chip microcomputer, the computer-readable storage medium is arranged in the system, and the system is provided with a microprocessor for executing program instructions stored in the computer-readable storage medium.

Images