A cloud computing-based dynamic configuration data acquisition method and system

Abstract

The application provides a cloud computing-based dynamic configuration data acquisition method and system, relates to the field of industrial internet, and solves the technical problem of low data acquisition quality in the prior art. The method comprises the following steps: acquiring a basic task demand parameter configured by a user; acquiring network condition parameters, equipment load parameters and data value evaluation values according to a preset period, and combining the basic task demand parameter to calculate an actual sampling frequency, a resource allocation weight and determine a data processing node; acquiring data according to the actual sampling frequency, and allocating computing resources to different tasks according to the resource allocation weight. The application is used in industrial equipment operation state monitoring and industrial data acquisition processes.

Description

Technical Field

[0001] It belongs to the field of industrial internet, specifically a data acquisition technology based on cloud computing. Background Technology

[0002] In the field of industrial internet data acquisition, existing methods often employ fixed parameter configurations, pre-setting parameters such as basic sampling frequency and accuracy requirements, and periodically executing data acquisition according to a preset strategy. However, when faced with real-time fluctuations in network conditions and equipment load in industrial environments, fixed parameter configurations are difficult to adapt adaptively, often leading to unbalanced resource allocation. For example, high-value data may lose critical information due to insufficient sampling frequency, while low-priority tasks may consume excessive resources, hindering important tasks. Furthermore, the value of data from different devices and types is difficult to measure horizontally, making it impossible to achieve precise resource allocation and resulting in low data acquisition quality, failing to meet the actual needs of the industrial internet for flexible data acquisition and optimized resource allocation. Summary of the Invention

[0003] This application provides a cloud computing-based dynamic configuration data acquisition method and system, which solves the technical problems of low data acquisition quality and unreasonable resource allocation in the prior art.

[0004] To achieve the above objectives, this application adopts the following technical solution:

[0005] Firstly, a cloud computing-based dynamic configuration data acquisition method is provided, including:

[0006] Obtain the user-configured basic task requirement parameters, which include the basic sampling frequency, maximum tolerable delay, and minimum accuracy value;

[0007] Based on a preset cycle, network status parameters, device load parameters, and data value assessment values ​​are collected, and combined with the basic task requirement parameters, the actual sampling frequency, resource allocation weights, and data processing nodes are calculated; wherein,

[0008] The data processing nodes include edge nodes and cloud nodes. The device load parameters are used to characterize the CPU and memory usage of the device and are calculated based on CPU utilization and memory utilization. The data value assessment value is used to characterize the data value of the data source and is calculated based on the data change rate.

[0009] Data is collected based on the actual sampling frequency, and computing resources are allocated to different tasks according to the resource allocation weight.

[0010] Based on the above technical solutions, the cloud-based dynamic configuration data acquisition method provided in this application effectively solves the problems of insufficient flexibility and resource waste in traditional acquisition methods through multi-dimensional parameter dynamic calculation and intelligent decision-making. Specifically, this application dynamically adjusts the sampling frequency according to the equipment load, which can avoid the exhaustion of system resources under high load; it accurately allocates computing resources through data value assessment, ensuring the priority of critical data processing; and it intelligently selects edge or cloud processing nodes based on network conditions and data characteristics, balancing real-time performance and computational depth requirements. This application can improve the adaptive capability, resource utilization efficiency, and processing timeliness of industrial internet data acquisition.

[0011] Furthermore, the network condition parameter N includes: network bandwidth Bw, network latency L, and packet loss rate PI, which are obtained in real time through network probes or 5G base stations deployed in industrial sites.

[0012] Furthermore, the formula for calculating the equipment load parameter D is: Among them, CPU used Indicates CPU utilization, CPU max Indicates maximum CPU capacity, RAM used Indicates memory usage, RAM max α1 and α2 represent the maximum memory capacity and the weighting coefficients, respectively.

[0013] Furthermore, the formula for calculating the data value assessment value V is: Where σ represents the change sensitivity weight, λ represents the data state weight, and dX / dt represents the data change rate. This indicates an indicator function that responds when data X exceeds the normal range [X]. min X max When ], the value is 1, and X min X represents the minimum value of data determined through historical data collection and recording. max This indicates the maximum value of the data determined through historical data collection and recording.

[0014] Furthermore, the formula for calculating the actual sampling frequency is: ; where f base D represents the fundamental sampling frequency. max This indicates the maximum load threshold.

[0015] Furthermore, the formula for calculating the resource allocation weight is: Among them, W i V represents the resource allocation weight for the i-th task. i ρ represents the data value assessment value of the i-th task, β represents the accuracy decay coefficient, and ρ represents the data value assessment value of the i-th task. i ρ represents the precision value of the data source for the i-th task. min∑V represents the minimum precision value of the data source for the i-th task. j This represents the sum of the data value assessment values ​​for all tasks.

[0016] Furthermore, the rules for determining the data processing nodes include:

[0017] When network latency > 0.5 × maximum tolerable latency, or data size / network bandwidth > 0.3 × maximum tolerable latency, the edge node is determined as a data processing node; otherwise, the data processing node is determined as a cloud node.

[0018] Furthermore, when the data processing node is an edge node, the collected data is processed according to the processing intensity level. Edge preprocessing is performed, and the edge preprocessing method corresponds one-to-one with the processing intensity level. Different processing intensity levels have different edge preprocessing methods, and the higher the processing intensity level, the higher the computing resources required for the corresponding edge preprocessing method.

[0019] Furthermore, the edge preprocessing method corresponds one-to-one with the processing intensity level, including:

[0020] If the processing intensity level ProcessLevel=1, then no edge preprocessing will be performed;

[0021] If the processing intensity level is ProcessLevel=2, then the collected data will be packaged.

[0022] If the processing intensity level is ProcessLevel=3, then the collected data will be cleaned.

[0023] If the processing intensity level is ProcessLevel=4, then the user-specified data preprocessing method will be used to perform edge preprocessing on the collected data.

[0024] Secondly, this application provides a cloud computing-based dynamic configuration data acquisition system, comprising: a task initialization module, a dynamic decision-making module, and a strategy execution module; wherein,

[0025] The task initialization module is used to obtain the basic task requirement parameters configured by the user, including the basic sampling frequency, maximum tolerable delay, and minimum accuracy value.

[0026] The dynamic decision-making module is used to collect network status parameters, device load parameters, and data value assessment values ​​according to a preset period, and combine them with the basic task requirement parameters to calculate the actual sampling frequency, resource allocation weight, and determine the data processing nodes. The data processing nodes include edge nodes and cloud nodes. The device load parameters are used to characterize the CPU and memory usage of the device and are calculated based on CPU utilization and memory utilization. The data value assessment value is used to characterize the data value of the data source and is calculated based on the data change rate.

[0027] The strategy execution module is used to collect data according to the actual sampling frequency and allocate computing resources to different tasks according to the resource allocation weight.

[0028] Compared with the prior art, the beneficial effects of the present invention are:

[0029] The cloud-based dynamic configuration data acquisition method and system provided by this invention effectively solves the technical problems of insufficient flexibility and resource waste in traditional data acquisition methods through dynamic calculation of multi-dimensional parameters and intelligent decision-making mechanisms. This solution dynamically adjusts the sampling frequency based on the real-time load of the equipment, avoiding system resource exhaustion under high load; simultaneously, it accurately allocates computing resources through a data value assessment system, ensuring the processing priority of key data, fundamentally improving the adaptive capability and resource utilization efficiency of data acquisition in the industrial internet environment.

[0030] In terms of data processing strategy, this invention can intelligently select edge nodes or the cloud as processing nodes based on network conditions and data characteristics: when network latency or data transmission demand exceeds a threshold, edge nodes are prioritized for real-time preprocessing to reduce transmission latency; conversely, the powerful computing resources of the cloud are utilized for in-depth analysis, achieving a balance between real-time performance and computational depth. Furthermore, through the collaborative work of modules such as task initialization, parameter acquisition, and dynamic decision-making, the system forms a closed-loop feedback optimization mechanism, further improving the timeliness and accuracy of data acquisition. This makes it widely applicable to industrial internet scenarios with high requirements for resource efficiency and real-time performance. Attached Figure Description

[0031] 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.

[0032] Figure 1 A system architecture diagram of a cloud computing-based dynamic configuration data acquisition system provided for embodiments of this application;

[0033] Figure 2 A flowchart illustrating a cloud-based dynamic configuration data acquisition method provided in this application embodiment;

[0034] Figure 3 This is a flowchart illustrating another cloud-based dynamic configuration data acquisition method provided in an embodiment of this application. Detailed Implementation

[0035] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0036] It should be noted that, in this application, the terms "exemplary" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0037] The cloud computing-based dynamic configuration data acquisition method provided in this application can be applied to, for example... Figure 1 In a cloud-based dynamic configuration data acquisition system, as shown, Figure 1 As shown, the communication system includes: a task initialization module, a dynamic decision-making module, and a strategy execution module; wherein,

[0038] The task initialization module is used to obtain the basic task requirement parameters configured by the user, including the basic sampling frequency, maximum tolerable delay, and minimum accuracy value.

[0039] The dynamic decision-making module is used to collect network status parameters, device load parameters, and data value assessment values ​​according to a preset period, and calculate the actual sampling frequency, resource allocation weight, and determine data processing nodes in combination with basic task requirement parameters. Among them, data processing nodes include edge nodes and cloud nodes. Device load parameters are used to characterize the CPU and memory usage of the device, and are calculated based on CPU utilization and memory utilization. Data value assessment values ​​are used to characterize the data value of the data source, and are calculated based on the data change rate.

[0040] The strategy execution module is used to collect data according to the actual sampling frequency and allocate computing resources to different tasks according to the resource allocation weight.

[0041] To address the resource waste inherent in traditional data acquisition methods, this application provides a cloud computing-based dynamically configured data acquisition method, which includes:

[0042] Obtain the basic task requirement parameters configured by the user, including the basic sampling frequency, maximum tolerable delay, and minimum accuracy value;

[0043] Based on the preset period, network status parameters, device load parameters, and data value assessment values ​​are collected, and combined with the basic task requirement parameters, the actual sampling frequency, resource allocation weight, and data processing nodes are calculated.

[0044] Data is collected based on the actual sampling frequency, and computing resources are allocated to different tasks according to the resource allocation weight.

[0045] Based on this, this application dynamically adjusts the sampling frequency according to the equipment load, which can avoid the exhaustion of system resources under high load. At the same time, it can accurately allocate computing resources through data value assessment to ensure the priority of critical data processing. It can also intelligently select edge or cloud processing nodes based on network conditions and data characteristics to balance the needs of real-time performance and computational depth, thereby improving the adaptive capability, resource utilization efficiency and processing timeliness of industrial Internet data acquisition.

[0046] like Figure 2 As shown in the figure, an embodiment of this application provides a dynamic configuration data acquisition method based on cloud computing, including:

[0047] S1. Obtain the basic task requirement parameters configured by the user.

[0048] The basic task requirements parameters include the basic sampling frequency, maximum tolerable delay, and minimum accuracy value, which are used to set initial standards for the data acquisition task and ensure that the acquisition process meets the user's basic needs.

[0049] In some implementations, the default values ​​for the base sampling frequency, maximum tolerable delay, and minimum accuracy can be 1Hz, 100ms, and 0.9, respectively, to suit common industrial data acquisition scenarios.

[0050] It should be noted that in the field of industrial internet, the data that needs to be collected is often in real-time streaming format. Therefore, there is a balance between sampling frequency and sampling accuracy. High-frequency sampling may lead to data redundancy, while low-frequency sampling may lose key information.

[0051] S2. Collect network status parameters, device load parameters, and data value assessment values ​​according to the preset period, and calculate the actual sampling frequency, resource allocation weight, and determine the data processing nodes in combination with the basic task requirement parameters.

[0052] Among them, the preset period represents the dynamic configuration period of the data collection strategy, which is usually 2 hours; the device load parameter is used to characterize the CPU and memory usage of the device; and the data value assessment value is used to measure the importance of the data source.

[0053] In some implementations, device load parameters are usually related to CPU utilization and memory utilization; data value assessment values ​​are usually related to data change rate and abnormal status, or they can be fixed values ​​set by the user, such as setting the value coefficient of a certain key device data to 1.5.

[0054] It should be noted that data processing nodes include edge nodes and cloud nodes, meaning that the system can dynamically select processing nodes based on network conditions and data characteristics to ensure the real-time performance and efficiency of data processing. If there are specific needs, user-defined values ​​can also be accepted to perform data processing at specific endpoints. However, when computing resources are insufficient, data preprocessing will not be performed, or only simple packaging and compression operations will be performed.

[0055] For example, when network latency is high, the system can automatically select edge nodes for data preprocessing to reduce transmission latency.

[0056] S3. Collect data according to the actual sampling frequency and allocate computing resources to different tasks according to the resource allocation weight.

[0057] The actual sampling frequency is determined by the basic task requirements parameters, equipment load parameters, and data value assessment. In industrial internet scenarios, due to the diversity and complexity of data acquisition tasks, data of different priorities often need to be processed. Therefore, it is necessary to dynamically adjust the sampling frequency of different tasks according to their importance. This maximizes the utilization of communication channels while ensuring that user needs are met, and reduces the possibility of other more important data acquisition tasks being unable to proceed due to a low-priority task.

[0058] In some implementations, the sampling frequency for high-value data is automatically increased to ensure the integrity of critical data.

[0059] It should be noted that the calculation of resource allocation weights takes into account both data value and equipment load, ensuring that high-value data receives more computing resources while avoiding equipment overload.

[0060] For example, for data tasks that monitor abnormal equipment status, the system will allocate higher resource weights to ensure the accuracy of real-time early warnings.

[0061] Based on the above technical solutions, the dynamic configuration data acquisition method provided in this application realizes adaptive adjustment of sampling frequency, precise allocation of computing resources, and intelligent selection of data processing nodes through dynamic calculation and intelligent decision-making of multi-dimensional parameters. It effectively solves the problems of insufficient flexibility and resource waste in traditional data acquisition methods, and improves the adaptive capability, resource utilization efficiency, and processing timeliness of industrial Internet data acquisition.

[0062] In one possible implementation of this application embodiment, the above-mentioned S1 can be specifically implemented by the following S101, S102 and S103, which are described in detail below:

[0063] S101 provides a user parameter configuration interaction interface.

[0064] This interface allows industrial users to input basic task requirements parameters through a web interface, local client, or API interface. The interface design follows industrial protocol standards (such as OPCUA) to ensure the stability and security of parameter transmission.

[0065] In some implementations, the interface presets a default parameter value: the basic sampling frequency f. base The default frequency is 1Hz, and the maximum tolerable latency τ is... max The default value is 100ms, and the minimum precision requirement is ρ. min The default value is 0.9 to adapt to general industrial scenarios.

[0066] It should be noted that the interface supports hierarchical permission management of parameters. Different user roles (such as administrators and operators) can configure different ranges of parameters to avoid unauthorized modifications.

[0067] For example, a user of a smart manufacturing production line can use a web interface to... base Set to 5Hz to meet the real-time monitoring needs of high-speed equipment.

[0068] S102. Parse and validate user input parameters.

[0069] The system will perform basic format validation on the input parameters (such as the validity of the numeric type and range), for example, requiring f base >0、τ max >0、0<ρ min ≤1. If the parameters do not meet the requirements, the interface will return an error message and mark the exception.

[0070] In some implementations, the system supports semantic verification of parameters, such as when τ max When set to 5ms, the system will automatically prompt that this value is below the lower limit of physical latency for most industrial networks and suggest that the user confirm. If the user enters ρ... minIf the value is 1.2, the system will prompt "precision value exceeds the range of 0-1" and refuse to save.

[0071] It should be noted that the verification rules can be dynamically adjusted according to industrial scenarios, such as the verification of p in high-risk equipment scenarios. min The verification threshold can be automatically increased.

[0072] S103. Generate and store the basic task requirement parameter set Q0.

[0073] The system encapsulates the validated parameters into a structured data set Q0={f base , τ max , ρ min The configuration is stored in a distributed configuration center, supporting real-time synchronous access between edge nodes and the cloud. For example, when a production line switches production processes, the operator updates f in Q0. base The refresh rate is 10Hz, and the new parameters are pushed to all edge acquisition nodes in real time.

[0074] In some implementations, Q0 supports version management, generating a history record for each parameter update, making it easy to trace the impact of configuration changes on data collection results.

[0075] It should be noted that the stored Q0 parameters can be accessed by the real-time monitoring module. When an abnormal parameter is detected (such as fbase suddenly becoming 0), an early warning mechanism is automatically triggered.

[0076] Based on the above technical solution, S1 ensures the accuracy and traceability of basic task requirements parameters through a standardized parameter configuration-verification-storage process, laying the foundation for the execution of subsequent dynamic data acquisition strategies, while also supporting industrial users to flexibly adjust acquisition standards according to actual production needs.

[0077] In one possible implementation of this application embodiment, the above-mentioned S2 can be specifically implemented by the following S201, S202 and S203, which are described in detail below:

[0078] S201, Periodically collect network status parameters N.

[0079] Network status parameters, including network bandwidth Bw, latency L, and packet loss rate PI, are acquired in real time through network probes or 5G base stations deployed in industrial sites. The update frequency is no less than 1Hz to ensure real-time monitoring of network status.

[0080] In some implementations, the SNMP protocol or the dedicated API interface of 5G base stations can be used for parameter acquisition, supporting multi-protocol adaptation to be compatible with different industrial network environments.

[0081] It should be noted that network latency L is a key indicator affecting the selection of data processing nodes. Therefore, when L exceeds the threshold, edge nodes should be prioritized for data processing.

[0082] S202, Real-time calculation of equipment load parameters D.

[0083] CPU and memory usage are collected via edge device monitoring agents, according to the formula D= Calculate the load value, where α1 and α2 are weighting coefficients, with default values ​​of 0.6 and 0.4 respectively.

[0084] It should be noted that the value of D is in the range of 0-1. When D>0.9, the system will automatically trigger a load alarm and start the resource migration mechanism.

[0085] For example, if the CPU utilization of an edge server reaches 85% and the memory utilization reaches 70%, the calculated value D = 0.6 × 0.85 + 0.4 × 0.7 = 0.79, indicating that the device is close to full load.

[0086] S203, Dynamic evaluation data value parameter V.

[0087] The value assessment of data is calculated through a real-time analysis model at edge nodes. Where σ is the change sensitivity weight, λ is the abnormal state weight, and I is the indicator function, i.e., the data exceeds the normal range [X]. min X max The value of V is 1. The value of V directly affects the sampling frequency and resource allocation weight; data with a high V value will receive a higher processing priority.

[0088] In some implementations, σ and λ are obtained through training with historical data, for example, σ=0.7 and λ=0.3 for vibration data of key equipment.

[0089] It should be noted that, to ensure the comparability of data value assessments for different devices, the data change rate dX / dt of different devices is normalized before being substituted into the calculation formula. Specifically, the normalization process maps the original change rate to the [0, 1] interval by dividing it by the maximum value (or standard deviation) of the historical data change rate for that type of device, thus eliminating numerical differences caused by different physical dimensions (such as temperature / pressure / vibration) or device characteristics. For example, if the original value of dX / dt for a temperature sensor is 5℃ / s, and its historical maximum change rate is 20℃ / s, then the normalized value is 0.25, which is horizontally comparable to the normalized result of a vibration sensor. Furthermore, the normalization factor supports dynamic updates. When the operating status of the equipment changes significantly (such as production line upgrades or process adjustments), the system automatically recalculates the normalization benchmark based on the latest historical data, ensuring the timeliness and accuracy of the assessment mechanism. For example, the maximum rate of change of the temperature sensor of a chemical reactor was 10℃ / s in the early stage of production. After six months of operation, it was adjusted to 15℃ / s due to process optimization. After the system detected this change, it automatically updated the normalization factor from 10 to 15 to avoid the deviation in data value assessment due to outdated benchmark.

[0090] S204. Calculate the actual sampling frequency f based on the basic sampling frequency, equipment load parameters, and data value assessment. actual .

[0091] The formula for calculating the actual sampling frequency is as follows: Among them, D max This indicates the maximum load threshold. When the device load is too high, (D) max -D) / D max The value of (0.4 + 0.6V) will decrease, thus lowering the sampling frequency; when the data value is high, the value of (0.4 + 0.6V) will increase, thereby increasing the sampling frequency.

[0092] In some implementations, the maximum load threshold D max The default setting is 0.9, which means that the sampling frequency is automatically reduced when the device load exceeds 90% to avoid system crash.

[0093] It should be noted that this computing mechanism can adaptively balance the importance of equipment resources and data. For example, high-value data from critical equipment can be allocated according to f when the equipment is under light load. base 1.6 times the high-frequency sampling (when V=1).

[0094] For example, a sensor has a base sampling frequency of 10Hz, and the current device load D=0.7, D max =0.9, data value V=0.8, then f actual=10×min(1,(0.9-0.7) / 0.9)×(0.4+0.6×0.8)=10×0.222×0.88≈1.95Hz, the system automatically adjusts the sampling frequency to 2Hz.

[0095] S205. Calculate resource allocation weights based on data value assessment, accuracy requirements, and equipment load.

[0096] The formula for calculating the resource allocation weight is as follows: V i ρ represents the data value assessment value of the i-th task, β represents the accuracy decay coefficient, and ρ represents the data value assessment value of the i-th task. i ρ represents the precision value of the data source for the i-th task. min ∑V represents the minimum precision value of the data source for the i-th task. j This represents the sum of the data value assessment values ​​of all tasks that currently require data collection, with each task representing a data collection instruction from a different device.

[0097] It should be noted that in the formula for calculating resource allocation weights, V i The data value of a task directly determines the basis for resource allocation. As a precision adjustment term, an exponential function is used to weaken the impact when precision meets the target and to penalize and reduce the weight when precision is insufficient; β is used to control the severity of the penalty. In the denominator, D reflects the equipment load; a high load lowers the overall weight; ∑V j Normalization is performed to ensure that the weights of multiple tasks are comparable and fair.

[0098] In some implementations, the precision attenuation coefficient β is set to 0.5 by default. When the precision ρ of a certain task... i Approaching minimum accuracy ρ min hour, The value approaches 1, and the weight is mainly determined by the value of the data.

[0099] This weighting mechanism ensures that high-value, high-precision data receives priority access to computing resources, such as V for fault early warning data. i =0.9、ρ i =0.95, ρ min When the value is 0.9, the weight will be significantly higher than that of ordinary monitoring data.

[0100] For example, a certain task V i =0.8, ρ i =0.9, ρ min =0.8, β=0.5, equipment load D=0.6, total value of all tasks ΣV j =5, then W i =(0.8×e (-0.5×(1-0.9 / 0.8))) / (0.6×5)≈(0.8×e^(-0.0625)) / 3≈(0.8×0.9394) / 3≈0.251, meaning that the task receives a 25.1% weight in the computing resource allocation.

[0101] S206. Determine the data processing node based on network latency, data size, and network bandwidth.

[0102] The rules for determining data processing nodes include:

[0103] When network latency > 0.5 × maximum tolerable latency, or data transmission time (DataSize / Bw) > 0.3 × maximum tolerable latency, select an edge node;

[0104] Otherwise, a cloud node is selected. This rule balances transmission latency and computing resource utilization through quantitative metrics.

[0105] In some implementations, when the remaining computing resources of an edge node are less than 20%, the task will be automatically migrated to the cloud even if the edge processing conditions are met, thus avoiding overload of the edge node.

[0106] It should be noted that the selection of data processing nodes directly affects the real-time performance of data processing. For example, if the data transmission time due to a sudden failure exceeds 0.3 times τ, the real-time performance will be significantly affected. max The system can force preprocessing to be completed at the edge node before transmission.

[0107] For example, if a batch of data is 10MB in size, the network bandwidth is Bw=100Mbps, the maximum tolerable latency is τmax=200ms, the calculated data transmission time is 10×8 / 100=0.8s=800ms. Since 800ms>0.3×200ms=60ms, the system automatically selects an edge node to perform data compression preprocessing to reduce the amount of data transmitted before sending it to the cloud.

[0108] Based on the above technical solution, through real-time calculation and intelligent decision-making of multi-dimensional parameters, the system achieves dynamic adaptation of sampling frequency, precise allocation of computing resources, and intelligent selection of processing nodes. This enables the system to adaptively adjust data acquisition strategies in complex industrial Internet scenarios, ensuring both the quality and real-time performance of key data acquisition and avoiding waste of equipment resources and network congestion.

[0109] In one possible implementation of the embodiments of this application, combined with Figure 2 ,like Figure 3 As shown, the above S3 can be implemented through the following S301, S302 and S303, which are explained in detail below:

[0110] S301. Collect data according to the actual sampling frequency.

[0111] Among them, the edge device calculates the actual sampling frequency f based on step S2. actual Raw data is acquired from sensors or actuators via industrial protocols such as Modbus and OPCUA. The sampling process follows an "equal-interval triggering" mechanism to ensure the uniformity of data timestamps.

[0112] Simultaneously, computing resources are allocated to different tasks according to resource allocation weights: the system is based on the resource allocation weights W determined in step S2. i The computing resources of edge nodes, such as CPU and memory, are divided into dynamic resource pools. When multiple data collection tasks are performed concurrently, the weight W... i Directly mapped to resource allocation ratios, such as W i Tasks with a value of 0.3 will receive 30% of the available computing resources on the edge nodes, ensuring that the total resource allocation of all tasks does not exceed 100%.

[0113] It should be noted that there are hardware constraints on the sampling frequency adjustment. For example, the minimum sampling interval for some industrial sensors is 10ms, in which case the system will automatically adjust the sampling frequency. actual The maximum value limited by hardware.

[0114] S302. Determine the edge processing intensity level.

[0115] When the data processing node is an edge node, the system will allocate weights W based on resources. i The equipment load D and network bandwidth Bw are calculated according to the formula. The computational processing intensity level is categorized into levels 1-4, where Bw0 is the baseline bandwidth (default 100Mbps). Higher levels involve more complex edge preprocessing.

[0116] In some implementations, when the device load D>0.8, the processing intensity level is automatically reduced to level 2 or below to prioritize data transmission; when the network bandwidth Bw<20Mbps, "edge depth processing" is triggered, and the level is increased to level 4 to reduce the amount of data transmitted.

[0117] It should be noted that the processing intensity level is directly related to the complexity of the preprocessing operation. For example, level 3 processing may include data cleaning operations such as removing noise points and filling in missing values, while level 4 processing uses user-specified data preprocessing methods to preprocess the collected data, such as feature extraction and model inference.

[0118] For example, if W i =0.3, D=0.6, Bw=50Mbps (Bw0=100Mbps), then the processing intensity level =min(4, ⌈5×0.3×(1-0.6) / (50 / 100)⌉)=min(5, ⌈(1.2)⌉)=level 2, and the system performs basic data compression operations.

[0119] S303, Perform edge preprocessing and transmit to the target node.

[0120] The edge devices perform corresponding operations based on the processing intensity level:

[0121] Level 1 processing: No edge preprocessing is performed;

[0122] Level 2 processing: Pack the data and add timestamps and device identifiers;

[0123] Level 3 processing: Perform data cleaning (such as removing noise points and filling in missing values);

[0124] Level 4 processing: Users specify preprocessing methods, such as adding feature extraction, like calculating the mean, variance, and rate of change, or adding anomaly detection, such as based on thresholds or machine learning algorithms.

[0125] The processed data is transmitted through an edge-cloud collaborative channel, and the transmission strategy is dynamically adjusted according to the processing nodes determined in step S2.

[0126] It should be noted that edge preprocessing can significantly reduce the amount of data transmitted, but it will introduce additional computational latency. The system selects the optimal level by balancing bandwidth and latency in real time.

[0127] For example, after the vibration sensor data is processed in four levels, early bearing fault characteristics are identified. The edge node can immediately send the abnormal warning to the cloud, while compressing the original data and transmitting it asynchronously, which not only ensures the timeliness of the warning, but also reduces the network pressure.

[0128] Based on the above technical solutions, S3 achieves an optimized balance between data acquisition efficiency and resource consumption through dynamic sampling frequency control, tiered processing intensity, and edge preprocessing strategies. The quantification and calculation mechanism for processing intensity levels enables the system to intelligently adjust the preprocessing depth according to real-time resource conditions, meeting the need for rapid response to critical data while avoiding resource waste. This makes it particularly suitable for environments with many heterogeneous devices and complex network conditions in industrial internet scenarios.

[0129] Some of the data in the above formula are calculated by removing dimensions and taking their numerical values. The formula is the closest to the real situation obtained by software simulation of a large amount of collected data. The preset parameters and preset thresholds in the formula are set by those skilled in the art according to the actual situation or obtained through simulation of a large amount of data.

[0130] Working principle of the invention:

[0131] This invention first acquires the user-configured basic task requirement parameters to set initial standards for data acquisition. Then, it collects parameters such as network, device load, and data value at preset intervals. Combining these with the basic task requirement parameters, it dynamically calculates the actual sampling frequency, resource allocation weights, and determines data processing nodes to adapt to the complex needs of industrial scenarios. Next, it collects data based on the actual sampling frequency, allocates computing resources to different tasks according to resource allocation weights, and determines the edge processing intensity level based on device load and network bandwidth, performing corresponding edge preprocessing before transmitting the data. Through dynamic adaptation and intelligent decision-making at each stage, it achieves efficient collaboration in industrial internet data acquisition, processing, and transmission, balancing resource utilization and task requirements.

[0132] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for dynamically configuring data acquisition based on cloud computing, characterized in that, include: Obtain the user-configured basic task requirement parameters, which include the basic sampling frequency, maximum tolerable delay, and minimum accuracy value; Based on a preset cycle, network status parameters, device load parameters, and data value assessment values ​​are collected, and combined with the basic task requirement parameters, the actual sampling frequency, resource allocation weights, and data processing nodes are calculated; wherein, The data processing nodes include edge nodes and cloud nodes. The device load parameters are used to characterize the CPU and memory usage of the device and are calculated based on CPU utilization and memory utilization. The data value assessment value is used to characterize the data value of the data source and is calculated based on the data change rate. Data is collected based on the actual sampling frequency, and computing resources are allocated to different tasks according to the resource allocation weight; The formula for calculating the data value assessment value V is: Where σ represents the change sensitivity weight, λ represents the data state weight, and dX / dt represents the data change rate. This indicates an indicator function that responds when data X exceeds the normal range [X]. min ,X max When ], the value is 1, and X min X represents the minimum value of data determined through historical data collection and recording. max This represents the maximum value of the data determined through historical data collection and recording. The formula for calculating the actual sampling frequency is: ; where f base D represents the fundamental sampling frequency. max D represents the maximum load threshold, and D represents the device load parameter. The formula for calculating the resource allocation weight is: Among them, W i V represents the resource allocation weight for the i-th task. i ρ represents the data value assessment value of the i-th task, β represents the accuracy decay coefficient, and ρ represents the data value assessment value of the i-th task. i ρ represents the precision value of the data source for the i-th task. min ∑V represents the minimum precision value of the data source for the i-th task. j This represents the sum of the data value assessments for all tasks.

2. The method for dynamic configuration data acquisition based on cloud computing according to claim 1, characterized in that, The network status parameters include network bandwidth (Bw), network latency, and packet loss rate, which are obtained in real time through network probes or 5G base stations deployed in industrial sites.

3. The method for dynamic configuration data acquisition based on cloud computing according to claim 1, characterized in that, The formula for calculating the equipment load parameter D is: Among them, CPU used Indicates CPU utilization, CPU max Indicates maximum CPU capacity, RAM used Indicates memory usage, RAM max α1 and α2 represent the maximum memory capacity and the weighting coefficients, respectively.

4. The method for dynamic configuration data acquisition based on cloud computing according to claim 1, characterized in that, The rules for determining the data processing nodes include: When network latency > 0.5 × maximum tolerable latency, or data size / network bandwidth > 0.3 × maximum tolerable latency, the edge node is determined as a data processing node; otherwise, the data processing node is determined as a cloud node.

5. The method for dynamic configuration data acquisition based on cloud computing according to claim 1, characterized in that, Also includes: When the data processing node is an edge node, the collected data is processed according to the processing intensity level. Edge preprocessing is performed, with Bw0 as the baseline bandwidth. The edge preprocessing method corresponds one-to-one with the processing intensity level. Different processing intensity levels have different edge preprocessing methods, and the higher the processing intensity level, the higher the computing resources required for the corresponding edge preprocessing method.

6. The method for dynamic configuration data acquisition based on cloud computing according to claim 5, characterized in that, The edge preprocessing methods correspond one-to-one with the processing intensity levels, including: If the processing intensity level ProcessLevel=1, then no edge preprocessing will be performed; If the processing intensity level is ProcessLevel=2, then the collected data will be packaged. If the processing intensity level is ProcessLevel=3, then the collected data will be cleaned. If the processing intensity level is ProcessLevel=4, then the user-specified data preprocessing method will be used to perform edge preprocessing on the collected data.

7. A cloud computing-based dynamic configuration data acquisition system, characterized in that, The method for acquiring dynamic configuration data based on cloud computing as described in any one of claims 1-6 includes: The module consists of a task initialization module, a dynamic decision-making module, and a strategy execution module; among them, The task initialization module is used to obtain the basic task requirement parameters configured by the user, including the basic sampling frequency, maximum tolerable delay, and minimum accuracy value. The dynamic decision-making module is used to collect network status parameters, device load parameters, and data value assessment values ​​according to a preset period, and combine them with the basic task requirement parameters to calculate the actual sampling frequency, resource allocation weight, and determine the data processing nodes. The data processing nodes include edge nodes and cloud nodes. The device load parameters are used to characterize the CPU and memory usage of the device and are calculated based on CPU utilization and memory utilization. The data value assessment value is used to characterize the data value of the data source and is calculated based on the data change rate. The strategy execution module is used to collect data according to the actual sampling frequency and allocate computing resources to different tasks according to the resource allocation weight.

Images