Industrial data acquisition and analysis method based on industrial internet of things

Abstract

The invention discloses an industrial data acquisition and analysis method based on the industrial Internet of Things, and belongs to the field of the industrial Internet of Things. Industrial equipment parameters are acquired by setting edge detection points, an acquisition and evaluation value is calculated according to equipment importance, an operation environment and parameter fluctuation, and an acquisition frequency is matched; then sorting equipment and collection frequency of each node responsible for collection, evaluating a node load value, and distributing a sharing node for the heavy load node; analyzing the collected data, calculating a diagnosis area, judging a fault parameter and performing early warning; for fault equipment, the similarity between the diagnosis curve and the database curve is analyzed through a dynamic time warping algorithm, and the fault type is determined; the system optimizes the data acquisition and analysis process, improves the fault diagnosis precision, reduces the maintenance cost and equipment downtime, is suitable for equipment management and maintenance in an industrial Internet of Things environment, and effectively improves the intelligent level and efficiency of industrial production.

Description

Technical Field

[0001] This invention belongs to the field of industrial Internet of Things (IoT) technology, specifically relating to a method for industrial data acquisition and analysis based on the Industrial Internet of Things. Background Technology

[0002] With the rapid development of industrial automation and informatization, the Industrial Internet of Things (IIoT) has become a key technology driving the transformation and upgrading of the manufacturing industry. The IIoT deeply integrates industrial equipment, sensors, networks, cloud computing, and other technologies to achieve comprehensive perception, data transmission, and intelligent analysis of industrial production processes. However, in actual industrial production, monitoring the operating status and diagnosing faults of industrial equipment still face many challenges. Low data acquisition efficiency: Traditional industrial data acquisition methods rely heavily on centralized acquisition systems, which make it difficult to dynamically adjust the acquisition frequency according to the importance of the equipment and the operating environment, resulting in low data acquisition efficiency and an inability to capture abnormal equipment status in a timely manner. Uneven device load: In large-scale industrial IoT environments, the load distribution of edge computing nodes is unreasonable. Some nodes may become overloaded due to excessive data acquisition tasks, affecting the stability and real-time performance of data acquisition. Insufficient fault diagnosis accuracy: Existing fault diagnosis methods mostly rely on empirical rules or simple data analysis models, which makes it difficult to accurately identify the fault types of industrial equipment, resulting in low fault diagnosis accuracy.

[0003] Insufficient support for fault type identification and maintenance: After equipment failure, existing technologies are unable to quickly identify the fault type, making it difficult for maintenance personnel to develop effective maintenance plans, resulting in longer maintenance cycles and increased equipment downtime. To address this, we propose an industrial data acquisition and analysis method based on the Industrial Internet of Things. Summary of the Invention

[0004] The purpose of this invention is to provide an industrial data acquisition and analysis method based on the Industrial Internet of Things (IIoT) to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an industrial data acquisition and analysis method based on the Industrial Internet of Things, comprising the following steps: Step 1: Set up edge detection points and collect industrial equipment parameters. Based on the equipment importance, operating environment, and parameter standard deviation, calculate the equipment core value, operational quality value, and fluctuation value respectively. Combine the three to obtain the evaluation value. Match the corresponding industrial parameter collection frequency according to the evaluation value of the equipment. Step 2: Identify the equipment responsible for collecting industrial parameters at each node and their corresponding collection frequencies, analyze the node's collection frequency load, and evaluate the node's CPU utilization, memory utilization, and response time at the time of collection to obtain the node load value. Determine whether a node is a heavily loaded node based on the node load value, and assign load-sharing nodes to the heavily loaded nodes. Step 3: Analyze the collected industrial parameter data to obtain data analysis values. Based on the relationship curve of the data analysis values ​​over time, analyze the diagnostic area. Determine whether the industrial parameter is a fault diagnosis parameter based on the diagnostic area. If the industrial parameter is a fault diagnosis parameter, immediately mark the corresponding industrial equipment as faulty equipment and trigger a fault warning prompt. Step 4: For each faulty device, obtain all its fault diagnosis parameters and corresponding data analysis value change curves, and build a diagnostic database. By analyzing the similarity distance between the diagnostic curves of the fault diagnosis parameters and the diagnostic curves of the same parameters in the database, analyze the fault type of the faulty device.

[0006] Preferably, in step one, the specific process of setting edge detection points to collect industrial equipment parameters and analyzing the equipment's core values, operational degradation values, and fluctuation values ​​is as follows: The industrial IoT factory area is divided into several detection zones, and each detection zone is equipped with an edge computing node. The industrial equipment in the detection zone is connected to the edge computing node in the corresponding zone to communicate with it. Through the edge computing node, the industrial parameters of each industrial equipment in the detection zone are collected. The industrial parameters include: electrical parameters, physical parameters, chemical parameters, mechanical parameters, and operating status parameters. For industrial equipment in the Industrial Internet of Things (IIoT) factory area, the industrial equipment is divided into three levels according to its importance: Level 1 important equipment, Level 2 important equipment, and Level 3 important equipment. Each level corresponds to a core value of the equipment. For each piece of industrial equipment, the core value HP is obtained according to its importance level. For each industrial device in the industrial IoT factory area, environmental parameters of the location are obtained based on the operating environment of the industrial device's location. These environmental parameters include: temperature, humidity, electromagnetic interference intensity, dust concentration, and corrosive substance concentration. A harsh environment assessment database was constructed, which includes various harsh environment types: high temperature environment, high humidity environment, strong electromagnetic interference environment, high dust environment, and strong corrosive environment. For each harsh environment category, multiple environmental parameter value ranges were set, and each range corresponds to a harshness value. The environmental parameters of the location of the industrial equipment are compared and matched with the environmental parameter value ranges corresponding to each type of harsh environment in the harsh environment assessment database according to the category. The harshness value corresponding to each environmental parameter is output. Then, the harshness values ​​corresponding to all environmental parameters are added together to obtain the harshness value of the industrial equipment operating environment, which is denoted as the equipment operating condition value YL. A preset testing period is set up. For each industrial device, various industrial parameters are collected in real time during the testing period. A series of parameter values ​​for each industrial parameter are obtained during the testing period. For each industrial parameter, the standard deviation of all parameter values ​​during the testing period is calculated to obtain the parameter standard deviation. A preset weighting coefficient is assigned to the standard deviation of all parameters of the industrial device during the testing period. Each parameter standard deviation is multiplied by the corresponding preset weighting coefficient, and then these products are added together to obtain the device fluctuation value BD.

[0007] Preferably, in step one, the specific process of analyzing the sampling and evaluation values ​​and matching the corresponding industrial parameter sampling frequency according to the sampling and evaluation values ​​of the equipment is as follows: For each piece of industrial equipment, the core equipment value HP, operational performance value YL, and fluctuation value BD are normalized and then processed using the formula: The evaluation value CPZ is obtained, where a1, a2, and a3 are preset weighting coefficients, and C is a preset correction value. Several sampling value ranges are preset, and each sampling value range corresponds to a sampling frequency. For each industrial device, its corresponding sampling value is matched with all sampling value ranges, the corresponding sampling frequency is output, and the industrial parameters of the industrial device are collected through edge computing nodes according to its corresponding sampling frequency.

[0008] Preferably, the specific process for classifying industrial equipment according to its importance is as follows: For each piece of industrial equipment, in terms of the criticality of the production process: if the failure of industrial equipment will directly cause the entire production line to stop, or if the failure of industrial equipment will cause the production line to stop for a longer period of time than the set time, it is judged as high criticality. In terms of maintenance complexity and cost: If the maintenance process of industrial equipment requires a technical team with senior electromechanical engineers or above, and the unit price of the replacement parts exceeds the set amount, and the maintenance cycle exceeds the set time, then it is classified as equipment with high maintenance difficulty and cost. Regarding the degree of safety impact: If the consequences of industrial equipment failure meet one of the following conditions, it is classified as high safety impact equipment: First, it may cause a safety accident; second, it causes the concentration of sulfur dioxide in the environment to exceed the national ambient air quality standards. Regarding the feasibility of equipment substitution: If it is difficult to find substitute products for industrial equipment on the market, or if the purchase cost of the substitute equipment exceeds twice the purchase cost of the original equipment, or if the installation and commissioning cycle exceeds the set time, and the production interruption time exceeds the set time, then it is classified as difficult to substitute equipment. Based on the above criteria, industrial equipment that meets three or more of the high-level evaluation criteria in four dimensions—production process criticality, maintenance complexity and cost, safety impact, and equipment replacement feasibility—is classified as Level 1 important equipment; industrial equipment that meets two of the high-level evaluation criteria in the four dimensions is classified as Level 2 important equipment; and industrial equipment that meets only one of the high-level evaluation criteria or none of the four dimensions is classified as Level 3 important equipment.

[0009] Preferably, in step two, the specific process of identifying the equipment responsible for collecting industrial parameters at each node and their corresponding collection frequencies, analyzing the node's collection frequency load, and obtaining the node's load value is as follows: For each edge computing node, identify all the industrial devices responsible for collecting industrial parameters, and obtain their respective collection frequencies. Summarize the collection frequencies of all industrial devices responsible for the edge computing node, and then calculate the average to obtain the collection frequency of the edge computing node. Based on the load sampling frequency of the edge computing node, obtain the CPU utilization rate CS and memory utilization rate NS of the edge computing node at the sampling point; at the same time, send test packets to the edge computing node according to the load sampling frequency, record the time of data packet transmission and reception, calculate the time difference between the two, and obtain the response time XT; after normalizing the CPU utilization rate CS and memory utilization rate NS and the response time XT at the sampling time, use the formula: FDZ=CS×d1+NS×d2+XT×d3 to obtain the node load value FDZ, where d1, d2, and d3 are preset weight coefficients.

[0010] Preferably, in step two, the node is determined to be an overloaded node based on its load value; the specific process for assigning load-sharing nodes to overloaded nodes is as follows: A preset node burden threshold is set by comparing the node burden value at each acquisition time with the preset node burden threshold. If the node burden value is greater than the corresponding threshold, the acquisition time corresponding to it is marked as a burden time. Starting from the moment when the load occurs, record the duration of consecutive load moments, which is called the continuous overload duration. If the continuous overload duration of the edge computing node is greater than the continuous overload duration threshold at the current moment, then mark the edge computing node as an overload node. For overloaded nodes, if the duration of consecutive overload-free moments exceeds the overload elimination duration threshold, starting from the moment when the node was last free of load, then the node's label will be restored from overloaded node to normal edge computing node. For each heavily loaded node, obtain the core values ​​of all industrial equipment under the responsibility of that node, and sum them to obtain the node core value; The node core values ​​corresponding to the overloaded nodes are sorted from largest to smallest to obtain the node core sequence. Then, in descending order of the node core values ​​in the node core sequence, a sharing node is selected for each overloaded node.

[0011] Preferably, the selection process for the sharing node is as follows: All edge computing nodes not marked as overloaded nodes are organized into a candidate node pool. Each node in the candidate node pool is marked as a sharing node. For each sharing node, its corresponding CPU remaining rate CY, memory remaining amount NY, total memory ZN, network bandwidth remaining WY, and total network bandwidth ZW are obtained, and after normalization, they are used with the formula: The score FPZ is obtained, where f1, f2, and f3 are preset weight coefficients; All load-sharing nodes in the candidate node pool are sorted in descending order of their score values. Then, for the overloaded nodes in the node core sequence, they are processed one by one in descending order of their node core values. For each overloaded node, load-sharing nodes are selected sequentially from the beginning of the sorted list of load-sharing nodes.

[0012] Preferably, in step three, the specific process of analyzing the data analysis values ​​is as follows: For each industrial device, a preset extraction period is set to acquire industrial parameter data collected within the extraction period, resulting in a series of industrial parameter data. For each type of industrial parameter, its mean, maximum, minimum, variance, and median within the extraction period are calculated, and these calculation results are uniformly recorded as statistics. Weight coefficients are assigned to each statistic, and then each statistic is multiplied by its corresponding weight coefficient. Finally, these multiplication results are summed to obtain the data analysis value.

[0013] Preferably, in step three, the diagnostic area is analyzed, and the industrial parameter is determined to be a fault diagnosis parameter based on the diagnostic area. If the industrial parameter is a fault diagnosis parameter, the corresponding industrial equipment is immediately marked as faulty equipment, and a fault warning prompt is triggered. The specific process is as follows: For each type of industrial parameter, a two-dimensional Cartesian coordinate system is established with the data analysis value as the vertical axis and time as the horizontal axis. The data analysis values ​​corresponding to each extraction period are sequentially marked on the Cartesian coordinate system according to their temporal relationship, resulting in a series of data points. Then, adjacent data points are connected by curves to obtain a data analysis value variation curve. In the analysis value variation curve, two threshold lines are preset: a data overshoot threshold line and a data undershoot threshold line. The area enclosed by the portion of the data analysis value variation curve that exceeds the data overshoot threshold line and this threshold line is marked as the overshoot area. All overshoot areas are summed to obtain the total overshoot product P. C. Simultaneously, the area enclosed by the portion of the data analysis value change curve below the low data threshold line and the threshold line is marked as the deviation area. Then, all deviation areas are summed to obtain the total deviation product PJ. Using the formula: ZDS=PC×s1+PJ×s2, the diagnostic area ZDS is obtained, where s1 and s2 are preset weight coefficients and the preset diagnostic area threshold is used. The diagnostic area corresponding to the industrial parameter is compared with the corresponding threshold. If the diagnostic area is greater than the corresponding threshold, the industrial parameter is marked as a fault diagnosis parameter, and the industrial equipment corresponding to the industrial parameter is marked as faulty equipment, and a fault warning is issued.

[0014] Preferably, the specific process in step four is as follows: For each faulty device, obtain all its fault diagnosis parameters and corresponding data analysis value change curves, and mark these curves as diagnostic curves; construct a diagnostic database, which stores the diagnostic curves corresponding to each industrial parameter of the faulty device under different fault conditions; For each fault diagnosis parameter's diagnostic curve, retrieve all diagnostic curves for that parameter from the diagnostic database; use the dynamic time warping algorithm to calculate the similarity distance between the diagnostic curve of the fault diagnosis parameter and the diagnostic curves of the same parameter in the database one by one. If the similarity distance between the diagnostic curve of a certain fault diagnosis parameter and a fault curve of the same parameter in the database is less than the corresponding threshold, then the two are considered to be highly similar. In this case, the matching relationship between the diagnostic curve of the fault diagnosis parameter and the fault curve in the database is marked as a match, and the fault type corresponding to the similar curve is recorded. After performing the above operations of calculating the similarity distance using the dynamic time warping algorithm, determining whether a match is found, and recording the fault type for the diagnostic curves of all fault diagnostic parameters, the number of matches for each fault type is counted. If the proportion of a certain fault type's matching count to the total matching count exceeds a preset proportion threshold, then that fault type is determined to be the current main fault type of the device; This fault type and related diagnostic curve information were then sent to the maintenance personnel.

[0015] Compared with the prior art, the beneficial effects of the present invention are: (1) The industrial data acquisition and analysis method and system based on the Industrial Internet of Things calculates the acquisition evaluation value of the equipment by comprehensively considering the importance of the equipment, the operating environment and parameter fluctuations, and dynamically matches the acquisition frequency of industrial parameters accordingly. This dynamic adjustment mechanism can ensure high-frequency acquisition of important equipment or equipment in harsh operating environments, timely capture potential faults, and avoid unnecessary over-acquisition of ordinary equipment, thus optimizing the resource utilization efficiency of data acquisition.

[0016] (2) The industrial data acquisition and analysis method and system based on the Industrial Internet of Things monitors the workload of each edge computing node in real time, calculates the node load value, and determines whether the node is a heavy-load node. For heavy-load nodes, the nodes with strong sharing capacity are selected from the candidate node pool according to the core value ranking of the equipment they are responsible for, thereby optimizing the node load distribution and avoiding data acquisition delays or loss due to node overload.

[0017] (3) The industrial data acquisition and analysis method and system based on the Industrial Internet of Things calculates the statistical quantities of various parameters, including average, maximum, minimum, variance and median, analyzes the data analysis value, and calculates the diagnostic area based on the relationship curve of the data analysis value with time, so as to accurately determine whether the industrial parameter is a fault diagnosis parameter. This method not only considers the change of a single parameter, but also comprehensively evaluates the operating status of the equipment by comprehensively analyzing the changing trends of multiple parameters, avoiding misjudgment caused by fluctuation of a single parameter. At the same time, once the diagnostic area exceeds the preset threshold, the fault diagnosis parameter is quickly locked, the faulty equipment is located and an all-round early warning is triggered, which makes it easier to focus on the root cause of the problem, quickly clarify the specific parameter range involved in the equipment fault, and carry out targeted testing, greatly shortening the fault investigation time and improving maintenance efficiency.

[0018] (4) The industrial data acquisition and analysis method and system based on the Industrial Internet of Things calculates the similarity distance between the diagnostic curve of the fault diagnosis parameter and the diagnostic curve of the same parameter in the database by using the dynamic time warping algorithm, and presets the similarity distance threshold to quickly determine the similarity between the curves. If the match is successful, the corresponding fault type is recorded. After all diagnostic curve operations are completed, the number of times each fault type is matched is counted according to the preset fault type matching frequency ratio threshold to determine the main fault type. Finally, the key information is pushed to the maintenance personnel in a timely manner. This method can facilitate the quick location of problems, formulate maintenance plans, greatly shorten the maintenance cycle, reduce equipment downtime, and ensure production continuity. Attached Figure Description

[0019] Figure 1 This is a flowchart of the present invention; Figure 2 This is a graph showing the changes in the data analysis values ​​of this invention. Detailed Implementation

[0020] 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. Example

[0021] Please see Figures 1-2 This invention provides an industrial data acquisition and analysis method based on the Industrial Internet of Things, including: a data acquisition module, an acquisition collaboration module, an aggregation analysis module, and a data diagnosis module; The data acquisition module sets edge detection points and collects industrial equipment parameters. Based on equipment importance, operating environment, and parameter standard deviation, it calculates the equipment's core value, operational degradation value, and fluctuation value. These three values ​​are then combined to obtain the assessment value. The industrial parameter acquisition frequency is matched according to the equipment assessment value. The specific process is as follows: The industrial IoT factory area is divided into several monitoring zones, each equipped with an edge computing node. Industrial equipment within each monitoring zone is connected to this edge computing node for communication. The edge computing node collects industrial parameters from each piece of equipment within the monitoring zone. These parameters include electrical, physical, chemical, mechanical, and operational status parameters. Electrical parameters include voltage, current, power, and frequency; physical parameters include temperature, pressure, flow rate, liquid level, and speed; chemical parameters include pH, concentration, and composition; mechanical parameters include vibration, noise, torque, and wear; and operational status parameters include equipment operating time, number of start-ups and shutdowns, and fault codes. For industrial equipment within an Industrial Internet of Things (IIoT) factory, it is divided into three levels based on its importance: Level 1 Important Equipment, Level 2 Important Equipment, and Level 3 Important Equipment. Each level corresponds to a core value. For each piece of industrial equipment, its core value (HP) is obtained based on its importance level. The higher the core value, the greater the importance of the equipment and the higher its level. Level 1 is the most important, and Level 3 is the least important. The process of classifying industrial equipment according to its importance is as follows: For each piece of industrial equipment, in terms of the criticality of the production process: if the failure of industrial equipment will directly cause the entire production line to stop, or if the failure of industrial equipment will cause the production line to stop for a longer period of time than the set time, it is judged as high criticality. In terms of maintenance complexity and cost: If the maintenance process of industrial equipment requires a technical team with senior electromechanical engineers or above, and the unit price of the replacement parts exceeds the set amount, and the maintenance cycle exceeds the set time, then it is classified as equipment with high maintenance difficulty and cost. Regarding the degree of safety impact: If the consequences of industrial equipment failure meet one of the following conditions, it is classified as high safety impact equipment: First, it may cause a safety accident; second, it causes the concentration of sulfur dioxide in the environment to exceed the national ambient air quality standards. Regarding the feasibility of equipment substitution: If it is difficult to find substitute products for industrial equipment on the market, or if the purchase cost of the substitute equipment exceeds twice the purchase cost of the original equipment, or if the installation and commissioning cycle exceeds the set time, and the production interruption time exceeds the set time, then it is classified as difficult to substitute equipment. Based on the above criteria, industrial equipment that meets three or more of the high-level evaluation criteria in four dimensions—production process criticality, maintenance complexity and cost, safety impact, and equipment replacement feasibility—is classified as Level 1 important equipment; industrial equipment that meets two of the high-level evaluation criteria in the four dimensions is classified as Level 2 important equipment; and industrial equipment that meets only one of the high-level evaluation criteria or none of the four dimensions is classified as Level 3 important equipment. For each industrial device in the industrial IoT factory area, environmental parameters of the location are obtained based on the operating environment of the industrial device's location. These environmental parameters include: temperature, humidity, electromagnetic interference intensity, dust concentration, and corrosive substance concentration. A harsh environment assessment database was constructed, which includes various harsh environment types: high temperature environment, high humidity environment, strong electromagnetic interference environment, high dust environment, and strong corrosive environment. For each harsh environment category, multiple environmental parameter value ranges were set, and each range corresponds to a harshness value. The environmental parameters of the location of the industrial equipment are compared and matched with the environmental parameter value ranges corresponding to each type of harsh environment in the harsh environment assessment database according to the category. The harshness value corresponding to each environmental parameter is output. Then, the harshness values ​​corresponding to all environmental parameters are added together to obtain the harshness value of the industrial equipment operating environment, which is denoted as the equipment operating condition value YL. A preset testing period is set up. For each industrial device, various industrial parameters are collected in real time during the testing period. A series of parameter values ​​for each industrial parameter are obtained during the testing period. For each industrial parameter, the standard deviation of all parameter values ​​during the testing period is calculated to obtain the parameter standard deviation. A preset weighting coefficient is assigned to the standard deviation of all parameters of the industrial device during the testing period. Each parameter standard deviation is multiplied by the corresponding preset weighting coefficient, and then these products are added together to obtain the device fluctuation value BD. For each piece of industrial equipment, the core equipment value HP, operational performance value YL, and fluctuation value BD are normalized and then processed using the formula: The evaluation value CPZ is obtained, where a1, a2, and a3 are preset weight coefficients, and C is a preset correction value. The correction value can be set according to the data processing capability of the edge computing node. The larger the evaluation value for industrial equipment, the higher the frequency of collecting its industrial parameters. Several sampling value ranges are preset, and each sampling value range corresponds to a sampling frequency. The larger the lower and upper bounds of the sampling value range, the faster the corresponding sampling frequency. For each industrial device, its corresponding sampling value is matched with all sampling value ranges to output the corresponding sampling frequency. Based on its corresponding sampling frequency, the industrial parameters of the industrial device are collected through edge computing nodes. It should be noted that by subdividing the detection area and using edge computing nodes to collect industrial parameters, preliminary data acquisition is achieved. When assessing the importance of equipment, a comprehensive judgment is made from four dimensions: criticality of the production process, complexity and cost of maintenance, degree of safety impact, and feasibility of equipment replacement, ensuring that the core value of the equipment accurately reflects its importance. When analyzing the operating environment, environmental parameters are quantified into equipment performance values ​​using a harsh environment assessment library. The equipment fluctuation value is obtained by calculating the standard deviation of parameters, comprehensively considering the stability of industrial parameters. The assessment value is obtained by analyzing the core value, performance value, and fluctuation value of the equipment, and the corresponding industrial parameter collection frequency is matched according to the assessment value of the equipment. Determining the sampling frequency based on the operating environment and parameter fluctuations allows for increased sampling frequency when equipment is in complex environments or experiencing significant parameter fluctuations. This enables timely detection of equipment anomalies, early warning of potential faults, and improved timeliness and effectiveness of equipment maintenance. Furthermore, setting the sampling frequency avoids using a single sampling frequency for all equipment, improving the targeting of data collection, optimizing the resource utilization efficiency of edge computing nodes, and reducing data transmission and storage costs. This provides strong support for the efficient and stable operation of industrial IoT plants.

[0022] The data acquisition and coordination module analyzes the data acquisition frequency of each node by sorting out the equipment responsible for acquiring industrial parameters and their corresponding acquisition frequencies. Based on this, it assesses the CPU utilization, memory utilization, and response time of each node during data acquisition, thereby obtaining a node load value. The node load value is then used to determine whether a node is overloaded. For overloaded nodes, a load-sharing node is assigned. The specific process is as follows: For each edge computing node, identify all the industrial devices responsible for collecting industrial parameters, and obtain their respective collection frequencies. Summarize the collection frequencies of all industrial devices responsible for the edge computing node, and then calculate the average to obtain the collection frequency of the edge computing node. Based on the edge computing node's workload sampling frequency, the CPU utilization (CS) and memory utilization (NS) of the edge computing node at the sampling point are obtained. Simultaneously, test packets are sent to the edge computing node according to the workload sampling frequency, and the sending and receiving times of the data packets are recorded. The time difference between the two is calculated to obtain the response time (XT). After normalizing the CPU utilization (CS), memory utilization (NS), and response time (XT) at the sampling time, the node workload value (FDZ) is obtained using the formula: FDZ = CS × d1 + NS × d2 + XT × d3. Here, d1, d2, and d3 are preset weighting coefficients; the larger the node workload value, the heavier the data acquisition task burden on the edge computing node at the sampling time. A preset node burden threshold is set by comparing the node burden value at each acquisition time with the preset node burden threshold. If the node burden value is greater than the corresponding threshold, the acquisition time corresponding to it is marked as a burden time. Starting from the moment when the load occurs, record the duration of consecutive load moments, which is called the continuous overload duration. Two key duration thresholds are preset: one is the continuous overload duration threshold, and the other is the overload elimination duration threshold. If the continuous overload duration of the edge computing node is greater than the continuous overload duration threshold at the current moment, then the edge computing node is marked as an overload node. For overloaded nodes, starting from the last time they are not burdened, if the node burden value corresponding to the collection time is less than or equal to the node burden threshold, then no burdened time will occur. If the duration of consecutive burdenless times is greater than the overload elimination duration threshold at the current time, then the node's label will be restored from overloaded node to normal edge computing node. For each heavy-load node, obtain the core values ​​of all industrial equipment under the responsibility of that node, and sum them to obtain the node core value. The larger the node core value, the more important the heavy-load node is. The nodes are sorted from largest to smallest according to their core values, resulting in a core sequence. Then, in descending order of the core values, a sharing node is selected for each heavy-load node, enabling the heavy-load node and the sharing node to work together and jointly undertake the data collection task. The process of selecting the sharing node is as follows: Based on all edge computing nodes not marked as overloaded nodes, a candidate node pool is formed. Each edge computing node in the candidate node pool is marked as a sharing node. For each sharing node, its corresponding CPU remaining rate CY, memory remaining amount NY, total memory ZN, network bandwidth remaining WY, and total network bandwidth ZW are obtained, and after normalization, they are calculated using the formula: The score value FPZ is obtained, where f1, f2, and f3 are preset weight coefficients. The larger the score value of the sharing node, the stronger the sharing task of the node. All load-sharing nodes in the candidate node pool are sorted in descending order of their score values. Then, for the heavy-load nodes in the node core sequence, they are processed one by one in descending order of their node core values. For each heavy-load node, starting from the beginning of the sorted list of load-sharing nodes, load-sharing nodes are selected in sequence so that the load-sharing nodes and heavy-load nodes work together to undertake the data collection task.

[0023] It should be noted that the data acquisition and collaboration module assesses node load and intelligently allocates resources from multiple dimensions. Specifically, in the load assessment stage, it first sorts out the devices and their acquisition frequencies for each edge computing node, calculates the node's acquisition frequency, and obtains the CPU utilization, memory utilization, and response time of the node at the acquisition time, and comprehensively analyzes them to obtain the node load value. The node load value is compared with a preset threshold, the load time is recorded, and the node is determined to be overloaded based on the continuous overload duration threshold. For overloaded nodes, the sum of the core values ​​of the devices they are responsible for is calculated and sorted to determine their importance. In terms of resource allocation, a candidate node pool is built from the unoverloaded nodes, and the evaluation value of the sharing nodes is calculated using parameters such as CPU remaining rate, memory remaining amount, and network bandwidth remaining. The sharing nodes are sorted in descending order of evaluation value, and collaborative sharing nodes are selected in order for the overloaded nodes with high importance. The above methods enable real-time monitoring of the workload of each edge computing node, timely detection of overloaded nodes, and prevention of data acquisition delays or loss due to node overload, ensuring the stability and reliability of acquisition tasks. Based on the resource allocation strategy of device core values ​​and shared node evaluation values, nodes with strong sharing capabilities are prioritized for important overloaded nodes, optimizing the resource utilization efficiency of the entire acquisition system and improving the overall efficiency of data acquisition. Simultaneously, when the burden on overloaded nodes is reduced, their normal state is restored according to the overload elimination time threshold, allowing the system to dynamically adapt to changes in acquisition tasks, enhancing the flexibility and adaptability of the acquisition collaboration module, ultimately improving the quality and efficiency of industrial data acquisition and assisting in intelligent management and decision-making in industrial production.

[0024] The aggregation analysis module analyzes the collected industrial parameter data to obtain data analysis values. Based on the relationship between the data analysis values ​​and time, it analyzes the diagnostic area and determines whether the industrial parameter is a fault diagnosis parameter. If the industrial parameter is a fault diagnosis parameter, the corresponding industrial equipment is immediately marked as faulty equipment, and a fault warning is triggered. The specific process is as follows: For each industrial device, a preset extraction period is set to acquire industrial parameter data collected within the extraction period, resulting in a series of industrial parameter data. For each type of industrial parameter, the mean, maximum, minimum, variance, and median within the extraction period are calculated, and these calculation results are uniformly recorded as statistics. Weight coefficients are assigned to each statistic, and then each statistic is multiplied by its corresponding weight coefficient. Finally, these multiplication results are summed to obtain the data analysis value. For each type of industrial parameter, a two-dimensional Cartesian coordinate system is established with the data analysis value as the vertical axis and time as the horizontal axis. The data analysis values ​​corresponding to each extracted time period are sequentially marked on the Cartesian coordinate system according to their temporal relationship, resulting in a series of data points. Then, adjacent data points are connected by curves to obtain a curve showing the change in the data analysis value, as shown below. Figure 2 In the analysis value change curve, two threshold lines are preset: a data high threshold line and a data low threshold line. The area enclosed by the portion of the data analysis value change curve that exceeds the data high threshold line and this threshold line is marked as the over-excess area. All over-excess areas are summed to obtain the total over-excess product PC. Simultaneously, the area enclosed by the portion of the data analysis value change curve that is below the data low threshold line and this threshold line is marked as the under-excess area. Then, all under-excess areas are summed to obtain the total under-excess product PJ. Using the formula: ZDS=PC×s1+PJ×s2, the diagnostic area ZDS is obtained, where s1 and s2 are preset weight coefficients. A preset diagnostic area threshold is set. The diagnostic area corresponding to the industrial parameter is compared with the corresponding threshold. If the diagnostic area is greater than the corresponding threshold, the industrial parameter is marked as a fault diagnosis parameter, and the corresponding industrial equipment is marked as faulty equipment. At the same time, a fault warning is issued, and maintenance personnel are arranged to repair the equipment when a fault occurs.

[0025] It should be noted that during the data processing stage, the parameter data collected for each industrial device within the preset extraction period is analyzed in depth. By calculating statistical quantities such as mean, maximum and minimum values, variance, and median, preliminary data aggregation and feature extraction are achieved. This not only reflects the key characteristics of the data but also significantly reduces the data volume, refining representative indicators and laying a solid foundation for subsequent trend analysis, thereby improving analysis efficiency and operability. In the fault diagnosis stage, a coordinate system is constructed with data analysis values ​​on the vertical axis and time on the horizontal axis to plot change curves. Preset high and low threshold lines are used to quantify the areas of over- and under-biased deviations to obtain the diagnostic area. Once the diagnostic area exceeds the preset threshold, the fault diagnosis parameters are quickly locked, the faulty equipment is located, and a comprehensive early warning is triggered. This facilitates focusing on the root cause of the problem, quickly identifying the specific parameter range involved in the equipment fault, and thus conducting targeted testing, greatly shortening the fault diagnosis time and improving maintenance efficiency.

[0026] The data diagnostic module acquires all fault diagnostic parameters and corresponding data analysis value change curves for each faulty device, and constructs a diagnostic database. By analyzing the similarity distance between the diagnostic curves of the fault diagnostic parameters and the diagnostic curves of the same parameters in the database, the fault type of the faulty device is analyzed. The specific process is as follows: For each faulty device, obtain all its fault diagnosis parameters and corresponding data analysis value change curves, and mark these curves as diagnostic curves; construct a diagnostic database, which stores the diagnostic curves corresponding to each industrial parameter of the faulty device under different fault conditions; For each fault diagnosis parameter's diagnostic curve, retrieve all diagnostic curves for that parameter from the diagnostic database; use the dynamic time warping algorithm to calculate the similarity distance between the diagnostic curve of the fault diagnosis parameter and the diagnostic curves of the same parameter in the database one by one. A preset similarity distance threshold is set. If the similarity distance between the diagnostic curve of a certain fault diagnosis parameter and a fault curve of the same parameter in the database is less than the corresponding threshold, the two are considered to be highly similar. At this time, the matching relationship between the diagnostic curve of the fault diagnosis parameter and the fault curve in the database is marked as a match, and the fault type corresponding to the similar curve is recorded. After setting a threshold for the percentage of times a fault type matches, and performing the above-mentioned operations of calculating the similarity distance using the dynamic time warping algorithm on the diagnostic curves of all fault diagnosis parameters, determining whether a match is found, and recording the fault type, the number of times each fault type matches is counted. If the proportion of a certain fault type's matching count to the total matching count exceeds a preset proportion threshold, then that fault type is determined to be the current main fault type of the device; The fault type and related diagnostic curve information are then pushed to the maintenance personnel, and fault records are kept, including the time of occurrence, type, and parameters involved in the fault. It should be noted that for each fault diagnosis parameter's diagnostic curve, all curves with the same parameter are retrieved from the database. The dynamic time warping algorithm is used to meticulously calculate the similarity distance between curves. By setting a preset similarity distance threshold, the degree of similarity between curves is quickly determined. If a match is successful, the corresponding fault type is recorded. After completing the operation of all diagnostic curves, the number of matches for each fault type is counted based on the preset fault type matching frequency percentage threshold to determine the main fault types. Finally, key information is promptly pushed to maintenance personnel, and fault records covering multi-dimensional information are retained. This method facilitates rapid problem location, the development of repair plans, significantly shortens the repair cycle, reduces equipment downtime, and ensures production continuity.

[0027] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An industrial data acquisition and analysis method based on an industrial Internet of Things, characterized in that: The method comprises the following steps: Step one: set edge detection points and collect industrial equipment parameters, calculate equipment core value, operation value and fluctuation value according to equipment importance, operation environment and parameter standard deviation, obtain evaluation value by comprehensively considering the three values, and match corresponding industrial parameter collection frequency according to the corresponding evaluation value of the equipment; Step two: sort the equipment responsible for collecting industrial parameters and the corresponding collection frequency of each node, analyze the node burden collection frequency, and evaluate the CPU usage, memory usage and response time of the node at the collection time according to the node burden value, and then obtain the node burden value, and determine whether the node is a heavy load node according to the node burden value; And allocate a sharing node for the heavy load node; Step three: analyze the collected industrial parameter data to obtain data analysis value, analyze the diagnosis area according to the change relationship curve of the data analysis value with time, judge whether the industrial parameter is a fault diagnosis parameter according to the diagnosis area, and if the industrial parameter is a fault diagnosis parameter, mark the corresponding industrial equipment as a fault equipment, and trigger a fault warning prompt; Step four: for each fault equipment, obtain all fault diagnosis parameters and corresponding data analysis value change curves, and construct a diagnosis database, analyze the fault type of the fault equipment by analyzing the similarity distance between the diagnosis curve of the fault diagnosis parameter and the diagnosis curve of the same parameter in the database. 2.The industrial data acquisition and analysis method based on industrial Internet of Things according to claim 1, characterized in that: In step one, the specific process of setting edge detection points to collect industrial equipment parameters and analyzing equipment core value, operation value and fluctuation value is as follows: Divide the industrial internet factory into several detection areas, set edge computing nodes in each detection area, communicate the industrial equipment in the detection area with the edge computing nodes in the corresponding area, collect the industrial parameters of each industrial equipment in the detection area through the edge computing nodes; the industrial parameters include: electrical parameters, physical parameters, chemical parameters, mechanical parameters and running state parameters; For the industrial equipment in the industrial internet factory, according to the importance of the industrial equipment, it is divided into three levels, namely first important equipment, second important equipment and third important equipment, each level corresponds to a device core value, for each industrial equipment, according to its importance level, obtain its device core value HP; For each industrial equipment in the industrial internet factory, according to the running environment of the position where the industrial equipment is located, obtain the environmental parameters of the position, the environmental parameters include: temperature, humidity, electromagnetic interference intensity, dust concentration, corrosive substance concentration; Construct a harsh environment evaluation library, which has multiple harsh environment types, including: high temperature environment, high humidity environment, strong electromagnetic interference environment, high dust environment and strong corrosive environment; for each harsh environment category, multiple environmental parameter value intervals are set, and each interval corresponds to a harsh value; Compare and match the environmental parameters of the position where the industrial equipment is located with the environmental parameter value intervals corresponding to each harsh environment type in the harsh environment evaluation library according to the category, output the harsh value corresponding to each environmental parameter, and then add all the harsh values corresponding to the environmental parameters to obtain the harsh value of the running environment of the industrial equipment, which is recorded as the equipment operation value YL; A preset detection period is set, and for each industrial equipment, various types of industrial parameters are collected in real time within the detection period to obtain a series of parameter values of each industrial parameter within the detection period. For each industrial parameter, the standard deviation of all parameter values within the detection period is calculated to obtain a parameter standard deviation. A preset weight coefficient is assigned to all parameter standard deviations of the industrial equipment within the detection period, and each parameter standard deviation is multiplied by the corresponding preset weight coefficient. Then, the products are added to obtain the equipment fluctuation value BD. 3.The industrial data acquisition and analysis method based on industrial Internet of Things according to claim 2, characterized in that: In step one, the evaluation value is analyzed, and the specific process of matching the corresponding industrial parameter collection frequency according to the corresponding evaluation value of the equipment is as follows: For each industrial equipment, the equipment core value HP, the equipment performance value YL and the equipment fluctuation value BD corresponding to the industrial equipment are normalized, and the following formula is used to obtain the evaluation value CPZ: , wherein a1, a2 and a3 are preset weight coefficients, and C is a preset correction value. A plurality of evaluation value intervals are preset, each of which corresponds to a collection frequency. For each industrial equipment, the corresponding evaluation value is matched with all evaluation value intervals, and the corresponding collection frequency is output. The industrial parameters of the industrial equipment are collected by the edge computing node according to the corresponding collection frequency. 4.The industrial data acquisition and analysis method based on industrial Internet of Things according to claim 2, characterized in that: According to the importance of the industrial equipment, the specific process of classifying the industrial equipment is as follows: For each industrial equipment, in terms of production process criticality: if the industrial equipment failure will directly cause the entire production line to stall, or the industrial equipment failure will cause the production line to stall for more than a set time, it is determined to be high criticality. In terms of maintenance complexity and cost: if the industrial equipment maintenance process requires a technical team with a senior mechanical and electrical engineer or above qualification, and the replacement parts required have a unit price exceeding a set amount, and the maintenance period exceeds a set time, it is classified as a high maintenance difficulty and cost equipment. In terms of safety impact: if the consequences that may be caused by the failure of the industrial equipment meet one of the following situations, it is classified as a high safety impact equipment: one is that it may cause a safety accident; two is that it causes the concentration of sulfur dioxide in the environment to exceed the national environmental air quality standard. In terms of equipment replacement feasibility: if the industrial equipment is difficult to obtain a replacement product in the market, or the procurement cost of the replacement equipment exceeds twice the procurement cost of the original equipment, the installation and debugging period exceeds a set time, and the production interruption time exceeds a set time, it is classified as a difficult-to-replace equipment. Based on the above standards, among the four dimensions of production process criticality, maintenance complexity and cost, safety impact, and equipment replacement feasibility, the industrial equipment that meets three or more high-level evaluation standards is classified as a first-level important equipment. Among the four dimensions, the industrial equipment that meets two high-level evaluation standards is classified as a second-level important equipment. Among the four dimensions, the industrial equipment that meets only one high-level evaluation standard or none is classified as a third-level important equipment. 5.The industrial data acquisition and analysis method based on industrial Internet of Things according to claim 3, characterized in that: In step two, the devices responsible for collecting industrial parameters and the corresponding collection frequencies of each node are sorted out, and the specific process of analyzing the node burden collection frequency and obtaining the node burden value is as follows: For each edge computing node, sort out all industrial equipment responsible for collecting industrial parameters by the node. For these industrial equipment, obtain their respective collection frequencies. The collection frequencies of all industrial equipment responsible by the edge computing node are summarized and then averaged to obtain the edge computing node burden collection frequency. According to the burden collection frequency of the edge computing node, the CPU usage rate CS and the memory usage rate NS of the edge computing node corresponding to the collection point are obtained; at the same time, test packets are sent to the edge computing node according to the burden collection frequency, the time difference between the sending time and the receiving time of the data packet is calculated, and the response time XT is obtained; After the CPU usage rate CS and the memory usage rate NS corresponding to the collection time and the response time XT are normalized, the node burden value FDZ is obtained by using the formula: FDZ=CS×d1+NS×d2+XT×d3, wherein d1, d2 and d3 are preset weight coefficients. 6.The industrial data acquisition and analysis method based on industrial Internet of Things according to claim 5, characterized in that: In step two, whether the node is a heavy load node is determined according to the node burden value; the specific process of allocating a sharing node for the heavy load node is: A preset node burden threshold is set, and the node burden value corresponding to each collection time is compared with the preset node burden threshold; if the node burden value is greater than the corresponding threshold, the collection time corresponding to the node burden value is marked as a burden time; Starting from the burden time, the duration of the continuous burden time is recorded, which is recorded as the continuous heavy load duration; if the current time, the continuous heavy load duration of the edge computing node is greater than the continuous heavy load duration threshold, the edge computing node is marked as a heavy load node; For the heavy load node, starting from the last time when the burden time does not appear, if the current time, the duration of the continuous burden time is greater than the heavy load elimination duration threshold, the mark of the node is recovered from the heavy load node to the normal edge computing node; For each heavy load node, the device core value corresponding to all industrial devices responsible for the node is obtained, and the sum is calculated to obtain the node core value; According to the node core value corresponding to the heavy load node from large to small, the node core sequence is obtained, and the heavy load node is selected in turn according to the order from high to low in the node core sequence. 7.The industrial data acquisition and analysis method based on industrial internet of things according to claim 6, characterized in that: The selection process of the sharing node is: All edge computing nodes that are not marked as heavy load nodes are sorted to form a candidate node pool. Each node in the candidate node pool is marked as a sharing node. For each sharing node, its corresponding CPU remaining rate CY, memory remaining amount NY, total memory ZN, network bandwidth remaining WY and total network bandwidth ZW are obtained, and after normalization, a partial evaluation value FPZ is obtained by using the formula: , wherein f1, f2 and f3 are preset weight coefficients. According to the order from large to small, all the sharing nodes in the candidate node pool are arranged in descending order, then the heavy load nodes in the node core sequence are processed one by one according to the order from high to low of their node core values; for each heavy load node, the sharing nodes are selected from the beginning of the sorted sharing node list. 8.The industrial data acquisition and analysis method based on industrial internet of things according to claim 7, characterized in that: In step three, the specific process of analyzing the data analysis value is: For each industrial device, a preset extraction period is set, the industrial parameter data collected in the extraction period is obtained, a series of industrial parameter data is obtained, for each type of industrial parameter, the average value, the maximum value, the minimum value, the variance and the median value in the extraction period are calculated, and these calculation results are uniformly recorded as statistical quantities; each statistical quantity is assigned a weight coefficient, then each statistical quantity is multiplied by the corresponding weight coefficient, and the product results are accumulated to obtain the data analysis value. 9.The industrial data acquisition and analysis method based on industrial internet of things according to claim 8, characterized in that: In step three, the analysis diagnosis area is used to determine whether the industrial parameter is a fault diagnosis parameter; if the industrial parameter is a fault diagnosis parameter, the corresponding industrial device is marked as a fault device, and the specific process of triggering the fault warning prompt is: For each type of industrial parameter, a two-dimensional rectangular coordinate system is established with data analysis values as the ordinate and time as the abscissa. The data analysis values corresponding to each extraction period are sequentially marked in the rectangular coordinate system according to the time relationship to obtain a series of data points. Then, the adjacent data points are connected to obtain a data analysis value change curve graph. In the analysis value change curve graph, two threshold lines are preset, namely a data high threshold line and a data low threshold line. The area surrounded by the part of the data analysis value change curve that exceeds the data high threshold line and the threshold line is marked as a super-area. The sum of all super-areas is obtained to obtain a super-total PC. At the same time, the area surrounded by the part of the data analysis value change curve that is lower than the data low threshold line and the threshold line is marked as a drop-area. Then, the sum of all drop-areas is obtained to obtain a drop-total PJ. Using the formula: ZDS = PC × s1 + PJ × s2, the diagnostic area ZDS is obtained, where s1 and s2 are preset weight coefficients. A preset diagnostic area threshold is compared with the diagnostic area corresponding to the industrial parameter. If the diagnostic area is greater than the corresponding threshold, the industrial parameter is marked as a fault diagnosis parameter, and the industrial equipment corresponding to the industrial parameter is marked as a fault equipment. At the same time, a fault warning prompt is given. 10.The industrial data acquisition and analysis method based on industrial Internet of Things according to claim 9, characterized in that: The specific process in step four is as follows: For each fault equipment, all fault diagnosis parameters and corresponding data analysis value change curves are obtained. These curves are marked as diagnosis curves. A diagnosis database is constructed, which stores the diagnosis curves of each industrial parameter of the fault equipment under different fault conditions. For each fault diagnosis parameter diagnosis curve, all diagnosis curves of the parameter are retrieved from the diagnosis database. The dynamic time warping algorithm is used to calculate the similarity distance between the fault diagnosis parameter diagnosis curve and the diagnosis curves of the same parameter in the database one by one. If the similarity distance between the diagnosis curve of a fault diagnosis parameter and a fault curve of the same parameter in the database is less than the corresponding threshold, it is determined that they are highly similar. At this time, the matching relationship between the diagnosis curve of the fault diagnosis parameter and the fault curve in the database is marked as matching, and the fault type corresponding to the similar curve is recorded. After the above operations of calculating the similarity distance by the dynamic time warping algorithm, determining whether they are matched, and recording the fault type are completed for the diagnosis curves of all fault diagnosis parameters, the matching times of each fault type are counted. If the proportion of the matching times of a fault type in the total matching times exceeds a preset proportion threshold, the fault type is determined as the current main fault type of the equipment. Then, the fault type and related diagnosis curve information are pushed to the maintenance personnel.

Images