An energy saving optimization data storage system for a paper machine vacuum system

Abstract

The present application relates to the technical field of industrial data processing and storage, and particularly relates to an energy-saving optimization data storage system for a paper machine vacuum system. The system comprises: a data acquisition module that acquires vacuum system operation data in real time; a data processing module that cleans, classifies and selects a block strategy for the data; a block storage module that uses three block methods of fixed byte length, data difference and data characteristics to cut and store the data; a redundant data processing module that extracts features, identifies and deletes redundant data using an optimal classification hyperplane; a storage node allocation module that allocates data to cloud storage nodes using a preset strategy; a load monitoring module that adjusts the allocation strategy according to node performance; and a data retrieval interface module that provides an external access interface. The present application can efficiently store paper machine vacuum system operation data, save storage space, improve data access efficiency and provide reliable support for energy-saving optimization.

Description

Technical Field

[0001] This invention relates to the field of industrial data processing and storage technology, specifically to an energy-saving and optimized data storage system for a papermaking machine vacuum system. Background Technology

[0002] The vacuum system of a papermaking machine is a core energy-consuming component in the papermaking process, and its operating status directly affects paper quality and production efficiency. Modern papermaking machine vacuum systems are typically equipped with numerous sensors and control systems that generate massive amounts of operational data in real time, including vacuum level, pumping volume, motor speed, power consumption, bearing vibration, and critical point temperature. This data is of great value for energy-saving optimization control, fault early warning and diagnosis, and equipment health management, and forms the basis for supporting intelligent algorithms such as deep reinforcement learning and fuzzy neural networks to achieve deep energy-saving optimization.

[0003] However, existing industrial data storage technologies have significant shortcomings in data management for paper machine vacuum systems. First, the data types generated by paper machine vacuum systems are diverse and varied in format, making it difficult for traditional single-block storage systems to balance storage efficiency and retrieval performance for different types of data. Second, there is a large amount of redundant information in the operational data, such as duplicate records under steady-state conditions and invalid data generated by sensor anomalies. This redundant data consumes a large amount of storage resources and affects the efficiency of subsequent data analysis. In addition, existing storage systems mostly use simple hash modulo methods to allocate data without considering the real-time load differences of storage nodes, which can easily lead to some nodes being overloaded while others are idle. Furthermore, large-scale data migration is required when nodes are expanded or fail. More importantly, there is a lack of data storage architecture specifically optimized for energy saving in vacuum systems, making it difficult to provide high-quality, low-latency data support for intelligent algorithms such as deep reinforcement learning and fuzzy neural networks, thus hindering the exploration of the deep energy-saving potential of paper machine vacuum systems. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention provides an energy-saving and optimized data storage system for a paper machine vacuum system. The system includes: a data acquisition module that collects vacuum system operating data in real time; a data processing module that cleans, classifies, and selects a block partitioning strategy for the data; a block storage module that partitions and stores data using three methods: fixed byte length, data difference, and data characteristics; a redundant data processing module that extracts features, identifies and deletes redundant data using an optimal classification hyperplane; a storage node allocation module that allocates data to cloud storage nodes using a preset strategy; a load monitoring module that adjusts the allocation strategy based on node performance; and a data retrieval interface module that provides an external access interface. This invention can efficiently store paper machine vacuum system operating data, save storage space, improve data access efficiency, and provide reliable support for energy-saving optimization.

[0005] The present invention adopts the following technical solution: an energy-saving and optimized data storage system for a paper machine vacuum system, comprising a data acquisition module, a data processing module, a block storage module, a redundant data processing module, a storage node allocation module, a load monitoring module, and a data retrieval interface module; The data acquisition module is used to collect real-time operating data of the paper machine's vacuum system; The data processing module is used to clean and classify the operational data collected by the data acquisition module. The block storage module is used to divide and store the running data according to the morphological distribution characteristics of the running data using multiple block methods; the block methods include: fixed byte length block method, data difference block method, and data feature block method; The redundant data processing module includes a feature extraction unit, a redundancy identification unit, and a redundancy deletion unit; The feature extraction unit is used to extract feature vectors of each category of running data in the block storage module using a local analysis method; The redundancy identification unit is used to classify the extracted feature vectors using the optimal classification hyperplane algorithm, output the classification label corresponding to each running data, and identify all running data whose classification labels are redundant as redundant data. The redundancy removal unit is used to remove all redundant data from the block storage module; The storage node allocation module is used to allocate the running data processed by the redundant data processing module to each physical storage node of the cloud computing platform according to a preset allocation strategy. The load monitoring module is used to collect the performance parameters of each storage node in real time, calculate the load weight of each storage node based on the performance parameters of each storage node, and adjust the preset allocation strategy based on the load weight of each storage node. The data retrieval interface module is used to provide a data access interface for external applications to provide data retrieval and query functions.

[0006] Furthermore, the data acquisition module collects operational data including at least: vacuum level, pumping volume, motor speed, energy consumption, vibration, and temperature parameters.

[0007] Furthermore, in the data processing module, the running data collected by the data acquisition module is cleaned and classified. Specifically, the cleaning operation includes removing outliers and null values; the classification operation uses the K-means clustering method to divide the running data into multiple categories.

[0008] Furthermore, in the block storage module, the fixed byte length block segmentation method is specifically as follows: The size of the data block is determined by a preset byte length based on the type of running data. Based on the preset byte length, the running data is sequentially divided into multiple data blocks starting from the beginning of the running data; Each data block is assigned a sequence number according to the order of partitioning. The sequence number is calculated as follows:

[0009] In the formula, Indicates the sequence number of the data block. This represents the byte offset from the starting position of the running data to the starting position of the current data block. Indicates the preset byte length.

[0010] Furthermore, in the block storage module, the data difference block segmentation method is specifically as follows: Map the running data to a spatial coordinate system to obtain its coordinate position and running data width in the spatial coordinate system; By using the preset width and height of the differential blocks, the running data is divided into grids in a top-to-bottom and left-to-right order to obtain multiple data blocks; After dividing the data into grids, sort the data blocks by differences and obtain the number of each data block, represented as:

[0011] In the formula, Indicates the data block number. This indicates the width of the preset differential blocks. Indicates the height of the preset differential blocks. The width of the running data.

[0012] Furthermore, in the block storage module, the data feature block segmentation method is specifically as follows: Based on object-oriented principles, the runtime data is transformed into multiple types of data objects; these multiple types of data objects include initial task data, product data objects, and vector data. The runtime data is stored in blocks according to the type of each data object.

[0013] Furthermore, in the redundant identification unit, the extracted feature vectors are classified using the optimal classification hyperplane algorithm, including: Using the feature vector extracted by the feature extraction unit as input, an optimal classification hyperplane is constructed; Based on the optimal classification hyperplane, the optimal classification function is derived. Each feature vector is substituted into the optimal classification function for calculation. The classification label of the feature vector is determined according to the sign of the output of the optimal classification function.

[0014] Furthermore, the optimal classification function is expressed as:

[0015] In the formula, Represents the optimal classification function. The coordinates of the eigenvectors are represented. Represents the discriminant function. Indicates the classification threshold, subscript These are the indices of the eigenvectors. For category attributes, This indicates a sign function used to map the value within parentheses to 1, -1, or 0 to output the category label.

[0016] Furthermore, in the storage node allocation module, the preset allocation strategy is specifically as follows: The preset allocation strategy employs a consistent hashing algorithm. The consistent hashing algorithm is used to map each physical storage node of the cloud computing platform to multiple virtual nodes on a hash ring; The landing point on the hash ring is obtained based on the hash value of each running data to determine the virtual node corresponding to each running data; The physical storage node for storing data is determined based on the virtual node corresponding to each piece of running data.

[0017] Furthermore, the load monitoring module calculates the load weight of each storage node based on its performance parameters, and adjusts the preset allocation strategy according to the load weight of each storage node, including: The performance parameters include current performance parameters, maximum performance parameters, and storage node utilization. The node performance ratio is calculated based on the ratio of the current performance parameter of each storage node to its maximum performance parameter. The overall performance value of each storage node is calculated based on the node performance ratio and the utilization rate of the corresponding storage node. The load weight of each storage node is calculated based on the overall performance value and current performance parameters of each storage node. The number of virtual nodes on the hash ring is adjusted according to the load weight of each storage node.

[0018] The beneficial effects of this invention are as follows: This invention uses a data acquisition module and a data processing module to collect key parameters such as vacuum level, pumping volume, motor speed, electrical power, bearing vibration, and temperature in real time for cleaning and classification, providing a comprehensive data foundation for energy-saving optimization; it improves data quality and subsequent processing efficiency; the block storage module adopts three block methods—fixed byte length, data difference, and data characteristics—to select the optimal block strategy based on the characteristics of different types of data, balancing storage efficiency and retrieval performance; the redundant data processing module extracts feature vectors through local analysis methods and uses the optimal classification hyperplane algorithm to identify and delete redundant data, effectively saving storage space while improving data quality; the storage node allocation module allocates data to cloud storage nodes using a preset allocation strategy, ensuring the rationality of data distribution; the load monitoring module collects node performance parameters in real time and dynamically adjusts the allocation strategy to avoid node overload and achieve load balancing; the data retrieval interface module provides efficient data access capabilities for external energy-saving optimization algorithms. Combining the functions of the above modules, it can provide high-quality, low-latency data storage support for in-depth energy-saving optimization of the paper machine vacuum system. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of an energy-saving and optimized data storage system structure for a papermaking machine vacuum system according to an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] A schematic diagram of an energy-saving and optimized data storage system for a paper machine vacuum system according to an embodiment of the present invention is shown below. Figure 1 As shown, it includes: a data acquisition module, a data processing module, a block storage module, a redundant data processing module, a storage node allocation module, a load monitoring module, and a data retrieval interface module. Specifically: The data acquisition module is used to collect real-time operating data of the paper machine's vacuum system; In this embodiment of the invention, the data acquisition module can acquire data by deploying various sensors or control system interfaces at key parts of the paper machine vacuum system. The acquired operating data includes at least vacuum degree, pumping volume, motor speed, energy consumption, vibration, and temperature parameters. The vacuum degree can be obtained from multiple monitoring points based on the location of the sensors. Pumping volume and motor speed refer to the pumping volume and motor speed of the turbine vacuum pump, respectively. Energy consumption is the electrical power consumption of the vacuum system. Vibration refers to the vibration amplitude of the bearing of the turbine vacuum pump. Temperature parameters can also be acquired by deploying multiple acquisition points on the turbine vacuum pump to obtain temperature parameters at multiple measurement points.

[0023] The data processing module is used to clean and classify the operational data collected by the data acquisition module. In this embodiment of the invention, the data processing module performs cleaning operations including removing outliers and null values. Specifically, for operating parameters such as vacuum level, pumping volume, motor speed, energy consumption, vibration, and temperature, corresponding reasonable value ranges are set. These reasonable value ranges can be obtained by statistically distributing historical normal operating data. Data exceeding these ranges are identified as outliers and removed. For null values ​​generated during the acquisition process due to transmission interruption or acquisition failure, linear interpolation or direct deletion are used to remove them.

[0024] In this embodiment of the invention, after the cleaning operation is completed, the operating data is further divided into multiple categories based on the speed and basis weight of the paper machine using the K-means clustering method. Specifically, the data processing module extracts the paper machine speed and paper basis weight parameters corresponding to each operating data, and uses these two parameters as clustering features to divide the operating data into multiple categories using the K-means clustering method. Each category corresponds to a typical paper machine operating condition.

[0025] The block storage module is used to divide and store the running data according to the shape and distribution characteristics of the running data using multiple block methods; In this embodiment of the invention, the segmentation methods include: fixed byte length segmentation, data difference segmentation, and data feature segmentation. Specifically, for operational data that is mainly characterized by time-series changes, such as continuously collected operational data like vacuum level, temperature, and vibration, a fixed byte length segmentation method can be selected. For data such as pumping volume and vacuum level, vibration, and temperature at different monitoring points, since the data collected from different monitoring points generally have spatial distribution characteristics, a data difference segmentation method can be selected. If the collected operational data needs to be encapsulated according to time windows, a data feature segmentation method can be selected to abstract the operational data into units with independent semantics.

[0026] In a specific embodiment of the present invention, when using a fixed-byte-length block method to divide runtime data with temporal continuity characteristics: First, based on the type of running data, the data block size is set to a preset byte length. For example, if the vacuum level data collected per second is divided into multiple data blocks of a fixed 64KB byte length, each data block can contain vacuum level records for a continuous time period. Then, based on the preset byte length, the running data is divided into multiple data blocks starting from the beginning, and the data blocks are of equal length. Finally, each data block is assigned a sequence number according to the order of division. The sequence number is calculated as follows:

[0027] In the formula, Indicates the sequence number of the data block. This represents the byte offset from the starting position of the running data to the starting position of the current data block. This indicates the preset byte length. This block partitioning method is simple to operate and fast to calculate, making it very suitable for raster datasets and object datasets.

[0028] For operational data with spatial distribution characteristics, the process of using data difference partitioning is as follows: Map the running data to a spatial coordinate system to obtain its coordinate position and running data width in the spatial coordinate system; By using the preset width and height of the differential blocks, the running data is divided into grids in a top-to-bottom and left-to-right order to obtain multiple data blocks; After dividing the data into grids, sort the data blocks by differences and obtain the number of each data block, represented as:

[0029] In the formula, Indicates the data block number. This indicates the width of the preset differential blocks. Indicates the height of the preset differential blocks. This defines the data width for operation. This block partitioning method uses a spatial grid to store spatially adjacent data points in the same data block, which is beneficial for subsequent spatial analysis and querying.

[0030] For runtime data that needs to be encapsulated according to time windows, such as encapsulating a set of runtime data collected within a continuous time period into an initial task data object, the data feature segmentation method is as follows: Based on object-oriented principles, runtime data is transformed into multiple types of data objects; these data objects include initial task data, product data objects, and vector data, which can be constructed according to the following model:

[0031] in, Represents a data object. This represents initial task data, such as a time series record of all parameters within one minute. This represents a product data object, such as intermediate data generated after preprocessing. Represents vector data, such as processed feature vectors; Based on the type of each data object, each data object is stored as an independent data block. Data is abstracted and managed in an object-oriented manner, which facilitates subsequent data retrieval and retrieval according to semantic units.

[0032] Based on the block storage module proposed above, the optimal block strategy can be adaptively selected according to the morphological distribution characteristics of the running data, thereby improving data storage efficiency and the convenience of subsequent data access.

[0033] The redundant data processing module includes a feature extraction unit, a redundancy identification unit, and a redundancy deletion unit; The feature extraction unit is used to extract feature vectors of each category of running data in the block storage module using local analysis methods; In this embodiment of the invention, the feature extraction unit extracts the main features of the various types of operational data that have been segmented and stored in the block storage module, such as vacuum degree time series data blocks, pumping volume spatial distribution data blocks, and objectified data blocks, using a local feature analysis method. This can obtain the main features of each type of redundant data in the energy-saving storage of task data under differential cloud computing, and use all features as an important basis for subsequent classification processing.

[0034] In a specific embodiment of the present invention, for time-series data blocks, features such as time-domain rate of change, frequency-domain distribution characteristics, and numerical repetition patterns can be extracted. The time-domain rate of change represents the degree of difference between adjacent data points; the frequency-domain distribution characteristics can be obtained by obtaining the main frequency components through Fast Fourier Transform; and the numerical repetition patterns are obtained by extracting continuously repeating numerical sequences. For spatially distributed data blocks, spatial gradient features and local anomaly characteristics can be extracted. The spatial gradient features represent the variation amplitude of adjacent grid points; and the local anomaly characteristics are obtained by acquiring the deviation between the data block and its surrounding grid points. For objectified data blocks, statistical features such as mean, variance, and extreme values ​​can be extracted. Finally, the extracted features are organized into feature vectors, and each feature vector establishes a corresponding mapping relationship with its corresponding data block. It should be noted that all the above feature extraction methods can be implemented using existing technologies. The specific extraction process and deviation acquisition methods are all achievable using existing technologies and will not be elaborated upon in this embodiment of the present invention.

[0035] The redundancy identification unit is used to classify the extracted feature vectors using the optimal classification hyperplane algorithm, output the classification label corresponding to each running data, and identify all running data whose classification labels are redundant as redundant data. In this embodiment of the invention, the redundancy identification unit uses the feature vector extracted by the feature extraction unit as input to construct the optimal classification hyperplane, thereby transforming the redundant data classification problem into an optimal hyperplane solution problem. Then, the optimal classification function is derived based on the optimal classification hyperplane, as follows:

[0036] In the formula, Represents the optimal classification function. The coordinates of the eigenvectors are represented. Represents the discriminant function. Indicates the classification threshold, subscript These are the indices of the eigenvectors. For category attributes, This indicates a sign function used to map the value within parentheses to 1, -1, or 0 to output the category label.

[0037] Each feature vector is substituted into the optimal classification function for calculation, and the classification label of the feature vector is determined according to the sign of the function output value. For example, a positive value indicates non-redundancy, and a negative value indicates redundancy. This is just an example. The running data corresponding to the feature vector with the classification label of redundancy is identified as redundant data.

[0038] A redundancy removal unit is used to remove all redundant data from the block storage module; In this embodiment of the invention, the redundancy deletion unit obtains the identifier of the corresponding running data based on the redundant data identified by the redundancy identification unit, and then locates the corresponding data block in the block storage module according to the identifier to perform the deletion operation, so that the storage space can be released and the data quality can be improved.

[0039] The storage node allocation module is used to allocate the running data processed by the redundant data processing module to each physical storage node of the cloud computing platform according to a preset allocation strategy. In this embodiment of the invention, the preset allocation strategy adopts the consistent hashing algorithm, specifically: the consistent hashing algorithm is used to map each physical storage node of the cloud computing platform to multiple virtual nodes on a hash ring; the hash value of each running data is used to obtain its landing point on the hash ring to determine the virtual node corresponding to each running data; and the physical storage node for storing data is determined based on the virtual node corresponding to each running data.

[0040] In one specific embodiment of the present invention, when a new physical storage node is added to the cloud computing platform, the storage node allocation module adds several virtual nodes corresponding to the new node to the hash ring. This involves splitting the hash space between the original two nodes and adding a completely new node. This ensures that in the event of a storage failure, there will be no large-scale data migration; only the changed task data needs to be maintained. Specifically, only the data within the affected hash range after the new node is added needs to be redistributed to the new node; data on other nodes does not need to be migrated. When a physical storage node fails, the virtual node corresponding to that physical storage node is removed from the hash ring, and the data originally stored on that node will be redistributed to the physical node corresponding to the next virtual node in a clockwise direction, thus achieving fault tolerance.

[0041] The load monitoring module is used to collect the performance parameters of each storage node in real time, calculate the load weight of each storage node based on the performance parameters of each storage node, and adjust the preset allocation strategy according to the load weight of each storage node. In this embodiment of the invention, an adaptive load weight calculation method is designed based on the hardware performance and space utilization of storage nodes. This method adjusts the number of virtual nodes corresponding to each physical node within the consistent hashing ring to achieve load balancing in the distributed storage system. The specific calculation steps and algorithm characteristics are as follows: The node performance ratio is defined as the ratio of a node's current performance parameters to its maximum performance parameters. It is used to characterize the degree of resource utilization of a node at present. The calculation formula is as follows: In the formula, This represents the performance ratio of the k-th storage node. These are the current performance parameters of this node. This represents the maximum performance parameter that sets the upper limit for the storage node hardware.

[0042] Combining node performance ratio and storage space utilization, a scaling factor is introduced to obtain the overall node carrying capacity score, i.e., the comprehensive performance value, expressed as:

[0043] In the formula, This represents the overall performance value of the k-th storage node. This represents the storage space utilization rate of the node.

[0044] The load weight of each node is obtained by normalizing the overall performance of all nodes. Nodes with stronger performance carrying capacity are assigned higher weights, as shown in the formula:

[0045] In the formula, For the first Load weight of storage nodes The total number of physical storage nodes in the system, all weights satisfy the following: , This represents the overall performance value of the k-th storage node. For the first The one that is different from the first The overall performance value of each storage node.

[0046] The number of virtual nodes on the consistent hash ring is allocated according to the load weight ratio, and the allocation formula is as follows: = In the formula, Preset the total number of virtual nodes for the hash ring. The final number of virtual nodes allocated to the k-th physical node is used to optimize system load balancing through adaptive adjustment of the number of virtual nodes.

[0047] Compared to traditional consistent hashing strategies with a fixed number of virtual nodes, this adaptive load weighting algorithm has several advantages: Firstly, it no longer relies solely on storage space utilization to assess node load, but also incorporates hardware performance ratio parameters, comprehensively considering node computing power and storage occupancy. This results in load assessments that better reflect the actual operating conditions of distributed storage, avoiding the implicit load imbalance problem of "idle storage but saturated computing power." Secondly, it uses a forward normalization method to generate load weights, achieving an adaptive scheduling logic of heavier loads for stronger nodes and lighter loads for weaker nodes, completely resolving the load distribution disorder caused by the inverted weights in the original formula. Thirdly, it incorporates scaling factors in the overall performance value. The system allows for customizable hyperparameters, enabling flexible adjustment of the weighting of performance and utilization metrics based on actual cluster test results. This adapts to distributed storage clusters with different architectures and business types, resulting in strong engineering feasibility. Furthermore, the scheduling mechanism dynamically adjusts only the number of virtual nodes without altering the underlying consistent hashing mapping rules. This results in low computational overhead, good system compatibility, and the ability to optimize the uniformity of data load distribution over the long term without significantly increasing cluster computational pressure.

[0048] The number of virtual nodes on the hash ring is adjusted according to the load weight of each storage node. Specifically, for nodes with a high load weight threshold, the storage node allocation module reduces the number of virtual nodes, thereby reducing the probability of new data being written to that node; for nodes with a low load weight threshold, the storage node allocation module can increase the number of virtual nodes, thereby increasing the probability of new data being written to that node, thus realizing the adjustment of the preset allocation strategy.

[0049] The data retrieval interface module provides a data access interface for external applications to enable data retrieval and query functions.

[0050] In this embodiment of the invention, the data retrieval interface module can provide at least two access methods: one is web access, which allows authorized users or administrators to log in to the system through a browser and query relevant data in real time on the web page; the other is mobile access, which can provide data query functions through a client APP.

[0051] When the data retrieval interface module provides data retrieval and query functions, based on the above-mentioned block storage method for operating data, it can support users to search by time range, data type, device identifier, or any combination of the above three methods. This provides external applications with flexible and efficient data access capabilities, supports multiple application scenarios, and realizes intelligent management and energy-saving optimization of the paper machine vacuum system.

[0052] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An energy-saving and optimized data storage system for a papermaking machine vacuum system, characterized in that, include: The system includes a data acquisition module, a data processing module, a block storage module, a redundant data processing module, a storage node allocation module, a load monitoring module, and a data retrieval interface module. The data acquisition module is used to collect real-time operating data of the paper machine's vacuum system; The data processing module is used to clean and classify the operational data collected by the data acquisition module. The block storage module is used to divide and store the running data using multiple block methods according to the morphological distribution characteristics of the running data; The block segmentation methods include: fixed byte length block segmentation, data difference block segmentation, and data feature block segmentation. The redundant data processing module includes a feature extraction unit, a redundancy identification unit, and a redundancy deletion unit; The feature extraction unit is used to extract feature vectors of each category of running data in the block storage module using a local analysis method; The redundancy identification unit is used to classify the extracted feature vectors using the optimal classification hyperplane algorithm, output the classification label corresponding to each running data, and identify all running data whose classification labels are redundant as redundant data. The redundancy removal unit is used to remove all redundant data from the block storage module; The storage node allocation module is used to allocate the running data processed by the redundant data processing module to each physical storage node of the cloud computing platform according to a preset allocation strategy. The load monitoring module is used to collect the performance parameters of each storage node in real time, calculate the load weight of each storage node based on the performance parameters of each storage node, and adjust the preset allocation strategy based on the load weight of each storage node. The data retrieval interface module is used to provide a data access interface to external applications to provide data retrieval and query functions.

2. The energy-saving and optimized data storage system for a papermaking machine vacuum system according to claim 1, characterized in that: The data acquisition module collects operational data including at least: vacuum level, pumping volume, motor speed, energy consumption, vibration, and temperature parameters.

3. The energy-saving and optimized data storage system for a papermaking machine vacuum system according to claim 1, characterized in that: In the data processing module, the running data collected by the data acquisition module is cleaned and classified. Specifically, the cleaning operation includes removing outliers and null values; the classification operation uses the K-means clustering method to divide the running data into multiple categories.

4. The energy-saving and optimized data storage system for a papermaking machine vacuum system according to claim 1, characterized in that: In the block storage module, the fixed byte length block segmentation method is specifically as follows: The size of the data block is determined by a preset byte length based on the type of running data. Based on the preset byte length, the running data is sequentially divided into multiple data blocks starting from the beginning of the running data; Each data block is assigned a sequence number according to the order of partitioning. The sequence number is calculated as follows: In the formula, Indicates the sequence number of the data block. This represents the byte offset from the starting position of the running data to the starting position of the current data block. Indicates the preset byte length.

5. The energy-saving and optimized data storage system for a papermaking machine vacuum system according to claim 1, characterized in that: In the block storage module, the data difference block segmentation method is specifically as follows: Map the running data to a spatial coordinate system to obtain its coordinate position and running data width in the spatial coordinate system; By using the preset width and height of the differential blocks, the running data is divided into grids in a top-to-bottom and left-to-right order to obtain multiple data blocks; After dividing the data into grids, sort the data blocks by differences and obtain the number of each data block, represented as: In the formula, Indicates the data block number. This indicates the width of the preset differential blocks. Indicates the height of the preset differential blocks. For the running data width, The coordinates of the eigenvectors are represented.

6. The energy-saving and optimized data storage system for a papermaking machine vacuum system according to claim 1, characterized in that: In the block storage module, the data feature block segmentation method is specifically as follows: Based on object-oriented principles, the runtime data is transformed into multiple types of data objects; these multiple types of data objects include initial task data, product data objects, and vector data. The runtime data is stored in blocks according to the type of each data object.

7. The energy-saving and optimized data storage system for a papermaking machine vacuum system according to claim 1, characterized in that: In the redundant identification unit, the extracted feature vectors are classified using the optimal classification hyperplane algorithm, including: Using the feature vector extracted by the feature extraction unit as input, an optimal classification hyperplane is constructed; Based on the optimal classification hyperplane, the optimal classification function is derived. Each feature vector is substituted into the optimal classification function for calculation. The classification label of the feature vector is determined according to the sign of the output of the optimal classification function.

8. The energy-saving and optimized data storage system for a papermaking machine vacuum system according to claim 7, characterized in that: The optimal classification function is expressed as: In the formula, Represents the optimal classification function. The coordinates of the eigenvectors are represented. Represents the discriminant function. Indicates the classification threshold, subscript These are the indices of the eigenvectors. For category attributes, This represents a sign function used to map the numerical value within parentheses to 1, -1, or 0 to output the category label.

9. The energy-saving and optimized data storage system for a papermaking machine vacuum system according to claim 1, characterized in that: The preset allocation strategy in the storage node allocation module is as follows: The preset allocation strategy employs a consistent hashing algorithm. The consistent hashing algorithm is used to map each physical storage node of the cloud computing platform to multiple virtual nodes on a hash ring; The landing point on the hash ring is obtained based on the hash value of each running data to determine the virtual node corresponding to each running data; The physical storage node for storing data is determined based on the virtual node corresponding to each piece of running data.

10. The energy-saving and optimized data storage system for a papermaking machine vacuum system according to claim 1, characterized in that: The load monitoring module calculates the load weight of each storage node based on its performance parameters, and adjusts the preset allocation strategy according to the load weight of each storage node, including: The performance parameters include current performance parameters, maximum performance parameters, and storage node utilization. The node performance ratio is calculated based on the ratio of the current performance parameter of each storage node to its maximum performance parameter. The overall performance value of each storage node is calculated based on the node performance ratio and the utilization rate of the corresponding storage node. The load weight of each storage node is calculated based on the overall performance value and current performance parameters of each storage node. The number of virtual nodes on the hash ring is adjusted according to the load weight of each storage node.

Images