Monitoring implementation and data acquisition system and method based on Internet of Things

Abstract

The invention discloses a monitoring implementation and data acquisition system and method based on the Internet of Things, and relates to the technical field of data acquisition. Edge nodes are arranged, and a cloud end and the edge nodes are combined in an energy consumption product working environment to construct a lightweight distributed acquisition network; dividing the energy-using product into a normal state and an abnormal state by using the characteristic data, and setting a reference sampling rate for different states of the energy-using product; adjusting the reference sampling rate of the energy-using product in a real-time working state by using the characteristic data gradient; deploying a virtual acquisition model in the physical simulation model by using a digital twinning technology; and acquiring a measured value of the energy-using product data by using the adjusted sampling rate, predicting the data of the energy-using product during real-time working by using the virtual acquisition model to obtain a predicted value, and fusing the measured value and the predicted value to obtain a real value of the energy-using product and outputting the real value of the energy-using product.

Description

Technical Field

[0001] The present invention relates to the technical field of data acquisition, and in particular to an implementation monitoring and data acquisition system and method based on the Internet of Things. Background Art

[0002] As industrial production scale and complexity increase, companies need real-time visibility into the operating status of equipment and various parameters in the production process to promptly identify problems, optimize production processes, and improve production efficiency and product quality. For example, in automobile manufacturing plants, real-time monitoring of the operating status of various equipment on the production line is necessary to ensure smooth operation. The continuous advancement of sensor technology, wireless communication technology, computer technology, and network technology has provided technical support for the implementation of real-time monitoring and data acquisition systems in the Internet of Things. Sensors can accurately sense various physical information, such as temperature, humidity, pressure, current, and voltage. Wireless communication technology enables the rapid and reliable transmission of sensor data to the central processing unit. Computer technology and network technology provide a powerful platform for data storage, processing, and analysis.

[0003] However, when sensors collect data from energy-consuming products today, the data sampling rates for energy-consuming products in different working states are mostly statically set. However, the data variation range of energy-consuming products in different working states is not constant, so how to dynamically adjust the sampling rate is crucial. Summary of the Invention

[0004] The purpose of the present invention is to provide a monitoring and data acquisition system and method based on the Internet of Things to solve the problems raised in the prior art.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A monitoring and data collection method based on the Internet of Things, comprising the following steps:

[0007] S100: Construct a plane reference system within the working area of the energy-consuming product, determine the location coordinates of each energy-consuming product, set up edge nodes, and combine the cloud and edge nodes within the working environment of the energy-consuming product to build a lightweight distributed data collection network.

[0008] Furthermore, the specific steps to build a lightweight distributed collection network are:

[0009] S101, establish a plane coordinate system in the working area of the energy-consuming product, mark the position of each energy-consuming product in the coordinate system to obtain coordinates (x i ,y i ), divide the working area of the energy-consuming product into grid-like sub-areas, and set the area of each sub-area as S block= L × W, L represents the length of the sub-region, W represents the width of the sub-region; calculate the number of sub-regions N block =S total / S block ;S total Indicates the total area of the working area of the energy-consuming product, S block represents the area of the sub-region;

[0010] S102: Extract the location coordinates of energy-consuming products in each sub-area and calculate the optimal location of the edge node. The formula is:

[0011]

[0012] In the formula, (xb, yb) represents the optimal position of the edge node, and n represents the number of energy-consuming products in the sub-region. The edge node is set at the optimal position of the edge node in each sub-region;

[0013] S103. Calculate the relative distance d(i, b) between the edge node and the farthest energy-consuming product in the sub-area, extract the communication radius of the edge node as R, and determine that the edge node is successfully set when d(i, b) ≤ R; when d(i, b) > R, determine that the edge node cannot be covered, and add an edge node in the corresponding sub-area;

[0014] When it is determined that an edge node needs to be added, the relative distances between the energy-consuming products and the set edge nodes in all sub-areas are calculated, the relative distances are determined, and all energy-consuming products whose relative distances are greater than the communication radius R are extracted. The extracted energy-consuming product coordinates are input into the formula described in S102 to calculate the optimal location for adding the edge node and set it.

[0015] S104: After setting up edge nodes in each sub-area, a cloud is set up in the working area of the energy-consuming product, and the cloud and edge nodes are combined to build a lightweight distributed data collection network.

[0016] Edge nodes enable preliminary data processing and analysis close to energy-consuming products. For large amounts of real-time data, edge nodes can filter out irrelevant information and transmit only critical data to the cloud. This significantly reduces data transmission volume and alleviates network bandwidth pressure, especially in scenarios with large data volumes, such as the massive sensor data collection in smart buildings, effectively avoiding network congestion.

[0017] By utilizing a lightweight distributed data collection network, even if some nodes fail, other nodes can continue to function, ensuring continuous data collection and system operation. Furthermore, the integration of cloud and edge nodes enables multiple data backups and redundancy, improving the reliability and fault tolerance of the entire system. For example, in a smart grid, even if some edge nodes fail to collect data, the cloud can still obtain grid operation information through other nodes, ensuring stable grid operation.

[0018] S200: For each node, collect historical operating status data of energy-consuming products, extract feature data from the status data, use the feature data to classify the energy-consuming products into normal and abnormal states, and set benchmark sampling rates for different states of the energy-consuming products;

[0019] Furthermore, the specific steps for setting the baseline sampling rate for different states of energy-consuming products are as follows:

[0020] S201. Collect all historical status data of energy-consuming products during operation, extract status data of energy-consuming products during abnormal operation and status data of energy-consuming products during normal operation, plot the status data of energy-consuming products during abnormal operation and normal operation based on time series, and obtain status data curves of abnormal and normal operation. Fit all types of status data curves using a polynomial fitting algorithm to obtain the change function of each status data when the energy-consuming product is abnormal operation as G(s) and the change function of each status data when the energy-consuming product is normal operation as F(s); extract all independent variable coefficients of the abnormal operation and normal operation change functions and calculate the average value, assuming that the average independent variable coefficient of G(s) is a1 and the average independent variable coefficient of F(s) is a2; calculate the change range of the two average independent variable coefficients, using the formula:

[0021] a c =|a1-a2|;

[0022] In the formula, a c Indicates the variation range of the average independent variable coefficient of the variation function when the energy-consuming product is working abnormally and normally; the variation range of the average independent variable coefficient of the variation function of all state data is obtained by the same calculation; let the number of types of state data be m; sort the variation ranges of the average independent variable coefficients of all calculated state data from small to large, calculate the difference between adjacent variation ranges, and then calculate the average of m-1 variation range differences. The characteristic data screening threshold is calculated using the average value. The formula is:

[0023] Cm=max|Ca j -Ca_p|;

[0024] In the formula, Cm represents the maximum difference between the change amplitude difference and the average value, Ca jrepresents the difference between the jth adjacent change amplitudes, Ca_p represents the average value of all change amplitude differences, the formula represents the calculation of the difference between m-1 change amplitude differences and the average value in sequence and the extraction of the maximum value, and the adjacent change amplitude of the change amplitude difference corresponding to the maximum difference value is (a c，j , a c，j+1 ), select a c，j+1 As a feature data screening threshold;

[0025] The change range of all state data is judged by using the screening threshold, and the state data with a change range greater than the screening threshold is used as the feature data;

[0026] Calculate the average value and standard deviation of all characteristic data of the energy-consuming product during historical abnormal and normal operation, and subtract the standard deviation from the average value to obtain the energy-consuming product abnormal threshold In based on each characteristic data;

[0027] S202. Setting a baseline sampling rate for different characteristic data of energy-consuming products when they are operating abnormally and normally. The formula is:

[0028]

[0029] In the formula, f base-nor Indicates the sampling rate when the energy-consuming product is working normally, f base-ab Indicates the sampling rate when the energy-consuming product is working abnormally, a2 min Indicates the minimum value of all independent variable coefficients of the characteristic data that works properly, a2 max Indicates the maximum value of all independent variable coefficients of the characteristic data that works properly, a1 min The minimum value of all independent variable coefficients of characteristic data indicating abnormal working conditions, a1 max The maximum value among all the independent variable coefficients of the characteristic data indicating abnormal working conditions; f min Indicates the minimum sampling rate; the baseline sampling rate of all feature data is calculated according to the formula.

[0030] By collecting historical status data and extracting characteristic data to categorize normal and abnormal states, we can identify the key characteristics of energy-consuming products under different conditions. Setting baseline sampling rates for different conditions allows for more targeted data collection. In normal conditions, a lower sampling rate satisfies the need to monitor basic equipment operating conditions while reducing data collection and storage requirements. In abnormal conditions, a higher sampling rate captures more detailed information about abnormal equipment changes, providing richer data support for fault diagnosis and analysis.

[0031] Setting a reference sampling rate according to the different states of energy-consuming products provides a basis for formulating differentiated maintenance strategies. For equipment in a normal state, regular inspections and maintenance can be carried out according to the conventional maintenance plan; while for equipment in an abnormal state, it is necessary to strengthen the monitoring and maintenance efforts, and arrange professional personnel for repair in a timely manner, achieving a reasonable allocation of maintenance resources and avoiding the situations of over-maintenance or under-maintenance.

[0032] S300. In the distributed acquisition network, detect the characteristic data of the energy-consuming product during operation in real time, judge the real-time working state of the energy-consuming product, calculate the gradient of the characteristic data of the real-time working state of the energy-consuming product, and adjust the reference sampling rate of the energy-consuming product under the real-time working state by using the characteristic data gradient;

[0033] Furthermore, the specific steps for adjusting the reference sampling rate of the energy-consuming product under the real-time working state by using the sliding window gradient are as follows:

[0034] S301. In the distributed acquisition network, detect the characteristic data of the energy-consuming product during operation in real time. Let the real-time characteristic data be Ts, and judge the real-time characteristic data by using the abnormal threshold. When any one of the characteristic data satisfies Ts≥In, it is judged that the energy-consuming product is working abnormally; when all the characteristic data satisfy Ts<In, it is judged that the energy-consuming product is working normally;

[0035] S302. After judging the real-time working state of the energy-consuming product, analyze each real-time characteristic data based on time series, and calculate the gradient of each real-time characteristic data. The formula is:

[0036]

[0037] In the formula, △st represents the characteristic data gradient, s(t) represents the characteristic data value at time series t, s(t + △t) represents the characteristic data value at time series t + △t, and △t represents the time series change value;

[0038] S303. Adjust the reference sampling rate by using the characteristic data gradient. The formula is:

[0039]

[0040] In the formula, fs represents the adjusted sampling rate, k represents the adjustment coefficient, and k is set according to experience and k<1.

[0041] The baseline sampling rate is adjusted based on the characteristic data gradient, achieving adaptive optimization of the sampling strategy. When the characteristic data gradient is large, indicating that the energy-consuming product's status is changing dramatically, possibly indicating an abnormal or critical operating phase, increasing the sampling rate allows for more intensive data collection and captures more detailed information. When the characteristic data gradient is small, indicating a relatively stable state, reducing the sampling rate avoids collecting excessive redundant data and conserves system resources.

[0042] Dynamically adjusting the sampling rate effectively balances data quality and resource consumption. This ensures high-quality, high-resolution data is acquired during critical periods for accurate analysis and diagnosis, while also avoiding the waste of storage, transmission, and processing resources caused by excessive data collection during non-critical periods. This improves the operational efficiency and cost-effectiveness of the entire distributed data collection network.

[0043] S400: Establish a physical simulation model of energy-consuming products in each node of the distributed data collection network, deploy a virtual data collection model in the physical simulation model using digital twin technology, collect historical data of energy-consuming products, and input it into the virtual data collection model for training;

[0044] Furthermore, the specific steps for collecting historical data on energy-consuming products and inputting it into the virtual acquisition model for training are as follows:

[0045] S401. Establish a physical simulation model of the energy-consuming product in each node in the distributed data collection network, deploy a virtual data collection model in the physical simulation model using digital twin technology, collect characteristic data from the historical work of the energy-consuming product and input it into the virtual data collection model for training, generate a prediction curve for each characteristic data, and use the time series as the independent variable and the characteristic data as the variable.

[0046] S500. Collect measured values of energy-consuming product data using the adjusted sampling rate, predict the real-time working data of the energy-consuming product using the virtual acquisition model to obtain a predicted value, and fuse the measured value and the predicted value to obtain a true value output of the energy-consuming product.

[0047] Furthermore, the specific steps for fusing the measured value and the predicted value to obtain the actual value output of the energy consumption product are as follows:

[0048] S501. Collect characteristic data of energy-consuming products using the adjusted sampling rate, use the collected characteristic data as the measured value sc, use the virtual acquisition model to predict each characteristic data as the predicted value sy, and calculate the reliability of the predicted value. The formula is:

[0049]

[0050] In the formula, S tIt represents the confidence level of the predicted value, and α represents a small constant to prevent division by zero; when St < H, it is determined that the corresponding predicted value is not credible and is removed. H represents the confidence threshold, which is set according to experience;

[0051] S502. Fuse the measured value and the predicted value of the feature data to obtain the true value. The formula is:

[0052]

[0053] In the formula, sz represents the true value of the feature data, β represents the fusion coefficient, and δ represents the smoothing parameter. Use the formula to fuse each type of feature data to obtain the true value, and output the fused true value.

[0054] In actual measurement, errors are inevitable. By predicting data through the virtual acquisition model and fusing it with the measured value, the predicted value can be used to correct and supplement the measured value. For example, when collecting power consumption data of power equipment, the measuring instrument may be deviated due to environmental interference. The virtual acquisition model predicts based on the operation rules and historical data of the equipment, and can correct the deviation of the measured value to a certain extent, improving the data accuracy.

[0055] The measured value and the predicted value verify each other. If the difference between the two is large, data anomalies can be detected in a timely manner. For example, when monitoring the energy consumption of an air conditioning system, if the measured value suddenly deviates significantly from the predicted value of the virtual acquisition model, it may mean that the sensor fails or the equipment has abnormal operating conditions, which is convenient for timely troubleshooting of problems and ensuring the reliability of the data.

[0056] An implementation monitoring and data acquisition system based on the Internet of Things. The implementation monitoring and data acquisition system includes a distributed acquisition network module, a reference sampling rate setting module, a reference sampling rate adjustment module, a virtual acquisition module, and a data fusion module;

[0057] The distributed acquisition network module is used to calculate and determine the position coordinates of each energy-consuming product, set edge nodes, and combine the cloud and edge nodes in the working environment of the energy-consuming product to construct a lightweight distributed acquisition network;

[0058] The reference sampling rate setting module is used to extract the feature data of the energy-consuming product and then set the reference sampling rate when the energy-consuming product is working abnormally and normally respectively;

[0059] The reference sampling rate adjustment module is used to calculate the gradient of the feature data when the energy-consuming product is working in real time, and use the gradient to adjust the reference sampling rate of each type of feature data;

[0060] The virtual acquisition module is used to establish a physical simulation model of the energy-consuming product in each node of the distributed acquisition network, and deploy a virtual acquisition model in the physical simulation model using digital twin technology;

[0061] The data fusion module is used to obtain the predicted value of the feature data by using the virtual acquisition model, obtain the measured value by real-time acquisition, and fuse the predicted value and the measured value to obtain the true value.

[0062] The reference sampling rate setting module includes a feature data extraction unit and a reference sampling rate setting unit;

[0063] The characteristic data extraction unit is used to calculate a characteristic data screening threshold value and extract characteristic data of energy-consuming products using the screening threshold value;

[0064] The reference sampling rate setting unit is used to calculate the reference sampling rates of different working states using the independent variable coefficient of each characteristic data curve.

[0065] The reference sampling rate adjustment module includes a characteristic data gradient calculation unit and a reference sampling rate adjustment unit;

[0066] The feature data gradient calculation unit is used to analyze each type of real-time feature data based on the time series and calculate the gradient of each type of real-time feature data;

[0067] The reference sampling rate adjustment unit is used to adjust the reference sampling rate using the characteristic data gradient.

[0068] The data fusion module includes a prediction value screening unit and a fusion unit;

[0069] The predicted value screening unit is used to calculate the credibility of the predicted value, set a credibility threshold, and use the credibility threshold to screen the predicted value;

[0070] The fusion unit is used to obtain a fusion coefficient by using the credibility calculation, and to fuse the measured value and the predicted value by using the fusion coefficient to obtain a true value for output.

[0071] Compared with the prior art, the present invention has the following beneficial effects:

[0072] 1. This invention uses a lightweight distributed data collection network. Even if some nodes fail, other nodes can continue to work, ensuring the continuity of data collection and the normal operation of the system. At the same time, the combination of cloud and edge nodes enables multiple data backups and redundancy, improving the reliability and fault tolerance of the entire system.

[0073] 2. This invention sets a baseline sampling rate based on the different states of energy-consuming products, providing a basis for developing differentiated maintenance strategies. For equipment in normal conditions, regular inspection and maintenance can be carried out according to the conventional maintenance plan; however, for equipment in abnormal conditions, enhanced monitoring and maintenance efforts are required, and professional personnel are promptly dispatched for inspection and repair. This ensures the rational allocation of maintenance resources and avoids over- or under-maintenance.

[0074] 3. This invention verifies measured and predicted values against each other, identifying data anomalies promptly if there are significant discrepancies. For example, when monitoring the energy consumption of an air conditioning system, if the measured value suddenly deviates significantly from the predicted value of the virtual acquisition model, this could indicate a sensor failure or abnormal equipment operation, facilitating prompt troubleshooting and ensuring data reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 This is a module distribution diagram of an implementation monitoring and data acquisition system based on the Internet of Things of the present invention;

[0076] Figure 2 The figure is a schematic diagram of the steps of implementing a monitoring and data collection method based on the Internet of Things of the present invention. DETAILED DESCRIPTION

[0077] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0078] Example: Figure 1-Figure 2 As shown, the present invention provides a technical solution.

[0079] A monitoring and data collection method based on the Internet of Things, comprising the following steps:

[0080] S100: Construct a plane coordinate system within the working area of the energy-consuming product, determine the location coordinates of each energy-consuming product, set up edge nodes, and combine the cloud and edge nodes in the working environment of the energy-consuming product to build a lightweight distributed data collection network;

[0081] The specific steps to build a lightweight distributed acquisition network are:

[0082] S101, establish a plane coordinate system in the working area of the energy-consuming product, mark the position of each energy-consuming product in the coordinate system to obtain coordinates (x i ,y i ), divide the working area of the energy-consuming product into grid-like sub-areas, and set the area of each sub-area as S block = L × W, L represents the length of the sub-region, W represents the width of the sub-region; calculate the number of sub-regions N block =S total / S block ;S total Indicates the total area of the working area of the energy-consuming product, S block represents the area of the sub-region;

[0083] S102: Extract the location coordinates of energy-consuming products in each sub-area and calculate the optimal location of the edge node. The formula is:

[0084]

[0085] In the formula, (x b ,y b ) represents the optimal position of the edge node, n represents the number of energy-consuming products in the sub-area; an edge node is set at the optimal position of the edge node in each sub-area;

[0086] S103. Calculate the relative distance d(i, b) between the edge node and the farthest energy-consuming product in the sub-area, extract the communication radius of the edge node as R, and determine that the edge node is successfully set when d(i, b) ≤ R; when d(i, b) > R, determine that the edge node cannot be covered, and add an edge node in the corresponding sub-area;

[0087] When it is determined that an edge node needs to be added, the relative distances between the energy-consuming products and the set edge nodes in all sub-areas are calculated, the relative distances are determined, and all energy-consuming products whose relative distances are greater than the communication radius R are extracted. The extracted energy-consuming product coordinates are input into the formula described in S102 to calculate the optimal location for adding the edge node and set it.

[0088] S104: After setting up edge nodes in each sub-area, a cloud is set up in the working area of the energy-consuming product, and the cloud and edge nodes are combined to build a lightweight distributed data collection network.

[0089] Edge nodes enable preliminary data processing and analysis close to energy-consuming products. For large amounts of real-time data, edge nodes can filter out irrelevant information and transmit only critical data to the cloud. This significantly reduces data transmission volume and alleviates network bandwidth pressure, especially in scenarios with large data volumes, such as the massive sensor data collection in smart buildings, effectively avoiding network congestion.

[0090] By utilizing a lightweight distributed data collection network, even if some nodes fail, other nodes can continue to function, ensuring continuous data collection and system operation. Furthermore, the integration of cloud and edge nodes enables multiple data backups and redundancy, improving the reliability and fault tolerance of the entire system. For example, in a smart grid, even if some edge nodes fail to collect data, the cloud can still obtain grid operation information through other nodes, ensuring stable grid operation.

[0091] S200: For each node, collect historical operating status data of energy-consuming products, extract feature data from the status data, use the feature data to classify the energy-consuming products into normal and abnormal states, and set benchmark sampling rates for different states of the energy-consuming products;

[0092] The specific steps to set the baseline sampling rate for different states of energy-consuming products are as follows:

[0093] S201. Collect all historical status data of energy-consuming products during operation, extract status data of energy-consuming products during abnormal operation and status data of energy-consuming products during normal operation, plot the status data of energy-consuming products during abnormal operation and normal operation based on time series, and obtain status data curves of abnormal and normal operation. Fit all types of status data curves using a polynomial fitting algorithm to obtain the change function of each status data when the energy-consuming product is abnormal operation as G(s) and the change function of each status data when the energy-consuming product is normal operation as F(s); extract all independent variable coefficients of the abnormal operation and normal operation change functions and calculate the average value, assuming that the average independent variable coefficient of G(s) is a1 and the average independent variable coefficient of F(s) is a2; calculate the change range of the two average independent variable coefficients, using the formula:

[0094] a c =|a1-a2|;

[0095] In the formula, a c Indicates the variation range of the average independent variable coefficient of the variation function when the energy-consuming product is working abnormally and normally; the variation range of the average independent variable coefficient of the variation function of all state data is obtained by the same calculation; let the number of types of state data be m; sort the variation ranges of the average independent variable coefficients of all calculated state data from small to large, calculate the difference between adjacent variation ranges, and then calculate the average of m-1 variation range differences. The characteristic data screening threshold is calculated using the average value. The formula is:

[0096] Cm=max|Ca j -Ca_p|;

[0097] In the formula, Cm represents the maximum difference between the change amplitude difference and the average value, Ca j represents the difference between the jth adjacent change amplitudes, Ca_p represents the average value of all change amplitude differences, the formula represents the calculation of the difference between m-1 change amplitude differences and the average value in sequence and the extraction of the maximum value, and the adjacent change amplitude of the change amplitude difference corresponding to the maximum difference value is (a c，j , a c，j+1 ), select a c，j+1 As a feature data screening threshold;

[0098] The change range of all state data is judged by using the screening threshold, and the state data with a change range greater than the screening threshold is used as the feature data;

[0099] Calculate the average value and standard deviation of all characteristic data of the energy-consuming product during historical abnormal and normal operation, and subtract the standard deviation from the average value to obtain the energy-consuming product abnormal threshold In based on each characteristic data;

[0100] S202. Setting a baseline sampling rate for different characteristic data of energy-consuming products when they are operating abnormally and normally. The formula is:

[0101]

[0102] In the formula, f base-nor Indicates the sampling rate when the energy-consuming product is working normally, f base-ab Indicates the sampling rate when the energy-consuming product is working abnormally, a2 min Indicates the minimum value of all independent variable coefficients of the characteristic data that works properly, a2 max Indicates the maximum value of all independent variable coefficients of the characteristic data that works properly, a1 min The minimum value of all independent variable coefficients of characteristic data indicating abnormal working conditions, a1 max The maximum value among all the independent variable coefficients of the characteristic data indicating abnormal working conditions; f min Indicates the minimum sampling rate; the baseline sampling rate of all feature data is calculated according to the formula.

[0103] By collecting historical status data and extracting characteristic data to categorize normal and abnormal states, we can identify the key characteristics of energy-consuming products under different conditions. Setting baseline sampling rates for different conditions allows for more targeted data collection. In normal conditions, a lower sampling rate satisfies the need to monitor basic equipment operating conditions while reducing data collection and storage requirements. In abnormal conditions, a higher sampling rate captures more detailed information about abnormal equipment changes, providing richer data support for fault diagnosis and analysis.

[0104] Setting baseline sampling rates based on the different states of energy-consuming products provides a basis for developing differentiated maintenance strategies. Equipment in normal condition can be regularly inspected and maintained according to the standard maintenance plan; equipment in abnormal condition requires enhanced monitoring and maintenance, with timely dispatch of specialized personnel for repairs. This ensures the rational allocation of maintenance resources and avoids over- or under-maintenance.

[0105] S300. Detect the characteristic data of the energy-consuming product in real time in the distributed acquisition network, judge the real-time working state of the energy-consuming product, calculate the gradient of the characteristic data of the real-time working state of the energy-consuming product, and adjust the reference sampling rate of the energy-consuming product in the real-time working state by using the gradient of the characteristic data;

[0106] The specific steps for adjusting the reference sampling rate of the energy-consuming product in the real-time working state by using the sliding window gradient are as follows:

[0107] S301. Detect the characteristic data of the energy-consuming product in real time in the distributed acquisition network. Let the real-time characteristic data be Ts, and judge the real-time characteristic data by using the abnormal threshold. When any kind of characteristic data satisfies Ts≥In, it is judged that the energy-consuming product works abnormally; when all characteristic data satisfy Ts<In, it is judged that the energy-consuming product works normally;

[0108] S302. After judging the real-time working state of the energy-consuming product, analyze each real-time characteristic data based on time series, and calculate the gradient of each real-time characteristic data. The formula is:

[0109]

[0110] In the formula, △st represents the gradient of the characteristic data, s(t) represents the value of the characteristic data at time series t, s(t + △t) represents the value of the characteristic data at time series t + △t, and △t represents the change value of the time series;

[0111] S303. Adjust the reference sampling rate by using the gradient of the characteristic data. The formula is:

[0112]

[0113] In the formula, fs represents the adjusted sampling rate, k represents the adjustment coefficient, and k is set according to experience and k<1.

[0114] Adjusting the reference sampling rate according to the gradient of the characteristic data realizes the adaptive optimization of the sampling strategy. When the gradient of the characteristic data is large, it indicates that the state of the energy-consuming product changes violently and may be in an abnormal or critical operation stage. At this time, increasing the sampling rate can collect data more densely and capture more detailed information; while when the gradient of the characteristic data is small, that is, the state is relatively stable, reducing the sampling rate can avoid collecting too much redundant data and save system resources.

[0115] Adopting the method of dynamically adjusting the sampling rate effectively balances the relationship between data quality and resource consumption. It not only ensures that high-quality and high-resolution data can be obtained during critical periods for accurate analysis and diagnosis, but also avoids the waste of storage, transmission and processing resources caused by excessive data collection during non-critical periods, and improves the operation efficiency and economy of the entire distributed acquisition network.

[0116] S400. Establish a physical simulation model of the energy-consuming product in each node of the distributed acquisition network. Deploy a virtual acquisition model in the physical simulation model using digital twin technology, and collect historical data of the energy-consuming product and input it into the virtual acquisition model for training;

[0117] The specific steps of collecting historical data of the energy-consuming product and inputting it into the virtual acquisition model for training are as follows:

[0118] S401. Establish a physical simulation model of the energy-consuming product in each node of the distributed acquisition network. Deploy a virtual acquisition model in the physical simulation model using digital twin technology, collect the characteristic data of the energy-consuming product in its historical operation and input it into the virtual acquisition model for training, generate a prediction curve for each type of characteristic data, with the time series as the independent variable and the characteristic data as the variable.

[0119] S500. Collect the measured values of the energy-consuming product data using the adjusted sampling rate, use the virtual acquisition model to predict the data during the real-time operation of the energy-consuming product to obtain the predicted values, and fuse the measured values and the predicted values to obtain the true value output of the energy-consuming product.

[0120] The specific steps of fusing the measured values and the predicted values to obtain the true value output of the energy-consuming product are as follows:

[0121] S501. Collect the characteristic data of the energy-consuming product using the adjusted sampling rate, take the collected characteristic data as the measured value sc, use the virtual acquisition model to predict each type of characteristic data as the predicted value sy, and calculate the credibility of the predicted value. The formula is:

[0122]

[0123] In the formula, S t represents the credibility of the predicted value, α represents a small constant to prevent division by zero; when St < H, it is determined that the corresponding predicted value is not credible and is removed. H represents the credibility threshold, which is set according to experience;

[0124] S502. Fuse the measured value and the predicted value of the characteristic data to obtain the true value. The formula is:

[0125]

[0126] In the formula, sz represents the true value of the characteristic data, β represents the fusion coefficient, δ represents the smoothing parameter. Use the formula to fuse each type of characteristic data to obtain the true value, and output the fused true value.

[0127] Errors are inevitable in actual measurements. By using a virtual acquisition model to predict data and integrate it with the measured values, the predicted values can be used to correct and supplement the measured values. For example, when collecting power data from power equipment, the measuring instrument may be affected by environmental interference and cause deviations. The virtual acquisition model, based on the equipment's operating patterns and historical data, can make predictions and correct the measured value deviations to a certain extent, improving data accuracy.

[0128] Measured and predicted values are mutually verified, and significant discrepancies between them can be detected promptly. For example, when monitoring the energy consumption of an air conditioning system, if the measured value suddenly deviates significantly from the predicted value of the virtual acquisition model, it may indicate a sensor failure or abnormal equipment operation, facilitating timely troubleshooting and ensuring data reliability.

[0129] An implementation monitoring and data acquisition system based on the Internet of Things, comprising a distributed acquisition network module, a reference sampling rate setting module, a reference sampling rate adjustment module, a virtual acquisition module and a data fusion module;

[0130] The distributed data collection network module is used to calculate and determine the location coordinates of each energy-consuming product, set up edge nodes, and combine the cloud and edge nodes in the working environment of the energy-consuming product to build a lightweight distributed data collection network;

[0131] The reference sampling rate setting module is used to extract characteristic data of the energy-consuming product and set the reference sampling rate when the energy-consuming product is working abnormally and normally respectively;

[0132] The reference sampling rate adjustment module is used to calculate the characteristic data gradient of the energy-consuming product when it is working in real time, and use the gradient to adjust the reference sampling rate of each characteristic data;

[0133] The virtual acquisition module is used to establish a physical simulation model of the energy-consuming product in each node in the distributed acquisition network, and deploy the virtual acquisition model in the physical simulation model using digital twin technology;

[0134] The data fusion module is used to obtain the predicted value of the feature data by using the virtual acquisition model, obtain the measured value by real-time acquisition, and fuse the predicted value and the measured value to obtain the true value.

[0135] The reference sampling rate setting module includes a feature data extraction unit and a reference sampling rate setting unit;

[0136] The characteristic data extraction unit is used to calculate a characteristic data screening threshold value and extract characteristic data of energy-consuming products using the screening threshold value;

[0137] The reference sampling rate setting unit is used to calculate the reference sampling rates of different working states using the independent variable coefficient of each characteristic data curve.

[0138] The reference sampling rate adjustment module includes a characteristic data gradient calculation unit and a reference sampling rate adjustment unit;

[0139] The feature data gradient calculation unit is used to analyze each type of real-time feature data based on the time series and calculate the gradient of each type of real-time feature data;

[0140] The reference sampling rate adjustment unit is used to adjust the reference sampling rate using the characteristic data gradient.

[0141] The data fusion module includes a prediction value screening unit and a fusion unit;

[0142] The predicted value screening unit is used to calculate the credibility of the predicted value, set a credibility threshold, and use the credibility threshold to screen the predicted value;

[0143] The fusion unit is used to obtain a fusion coefficient by using the credibility calculation, and to fuse the measured value and the predicted value by using the fusion coefficient to obtain a true value for output.

[0144] Example: Data collection and status monitoring are now performed on energy-consuming products in a certain test site, and a lightweight distributed collection network is set up; energy-consuming products represent working equipment that generates energy consumption in the test site;

[0145] The characteristic data of energy-consuming products are temperature, vibration and power. Taking vibration data as an example, when the work is normal, the coefficients of all independent variables in the polynomial change function are 1.5, 2.6 and 5 respectively; when the work is abnormal, the coefficients of all independent variables in the polynomial change function are 4, 3.8 and 8 respectively. Based on the historical equipment working records of the test site, the minimum sampling rate is set to 0.1, and the formula is used for calculation.

[0146]

[0147] When working normally, power a2=3, a2 min =1.5, a2 max =5, substitute into the calculation to get the power reference sampling rate f when working normally base-nor =2.4Hz; abnormal working reference sampling rate f base-ab =3.1Hz; let the sampling rate represent the number of samples per unit time;

[0148] Collect real-time characteristic data of energy-consuming products, calculate the vibration data gradient to be 6, and judge that the real-time vibration data is greater than the power abnormality threshold; set the adjustment coefficient to 0.1 based on experience; calculate using the formula,

[0149]

[0150] Fs=3.1+0.1×6=3.7, and the real-time vibration sampling rate is adjusted to 3.7 Hz.

[0151] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A monitoring and data collection method based on the Internet of Things, characterized by: The method comprises the following steps: S100: Construct a plane coordinate system within the working area of the energy-consuming product, determine the location coordinates of each energy-consuming product, set up edge nodes, and combine the cloud and edge nodes in the working environment of the energy-consuming product to build a lightweight distributed data collection network; S200: For each node, collect historical operating status data of energy-consuming products, extract feature data from the status data, use the feature data to classify the energy-consuming products into normal and abnormal states, and set benchmark sampling rates for different states of the energy-consuming products; S300, detecting characteristic data of the energy-consuming product in real time during operation in the distributed data collection network, determining the real-time operating state of the energy-consuming product, calculating the characteristic data gradient of the real-time operating state of the energy-consuming product, and adjusting the reference sampling rate of the energy-consuming product in the real-time operating state using the characteristic data gradient; S400: Establish a physical simulation model of energy-consuming products in each node of the distributed data collection network, deploy a virtual data collection model in the physical simulation model using digital twin technology, collect historical data of energy-consuming products, and input it into the virtual data collection model for training; S500. Collect measured values of energy-consuming product data using the adjusted sampling rate, predict the real-time working data of the energy-consuming product using the virtual acquisition model to obtain a predicted value, and fuse the measured value and the predicted value to obtain a true value output of the energy-consuming product.

2. The method for monitoring and data collection based on the Internet of Things according to claim 1, wherein: The specific steps of constructing a lightweight distributed acquisition network in S100 are: S101, establish a plane coordinate system in the working area of the energy-consuming product, mark the position of each energy-consuming product in the coordinate system to obtain coordinates (x i ,y i ), divide the working area of the energy-consuming product into grid-like sub-areas, and set the area of each sub-area as S block = L × W, L represents the length of the sub-region, W represents the width of the sub-region; calculate the number of sub-regions N block =S total / S block ;S total Indicates the total area of the working area of the energy-consuming product, S block represents the area of the sub-region; S102: Extract the location coordinates of energy-consuming products in each sub-area and calculate the optimal location of the edge node. The formula is: In the formula, (x b ,y b ) represents the optimal position of the edge node, n represents the number of energy-consuming products in the sub-area; an edge node is set at the optimal position of the edge node in each sub-area; S103. Calculate the relative distance d(i, b) between the edge node and the farthest energy-consuming product in the sub-area, extract the communication radius of the edge node as R, and determine that the edge node is successfully set when d(i, b) ≤ R; when d(i, b) > R, determine that the edge node cannot be covered, and add an edge node in the corresponding sub-area; When it is determined that an edge node needs to be added, the relative distances between the energy-consuming products and the set edge nodes in all sub-areas are calculated, the relative distances are determined, and all energy-consuming products whose relative distances are greater than the communication radius R are extracted. The extracted energy-consuming product coordinates are input into the formula described in S102 to calculate the optimal location for adding the edge node and set it. S104: After setting up edge nodes in each sub-area, a cloud is set up in the working area of the energy-consuming product, and the cloud and edge nodes are combined to build a lightweight distributed data collection network.

3. The method for monitoring and data collection based on the Internet of Things according to claim 1, wherein: The specific steps of setting the reference sampling rate for different states of the energy-consuming product in S200 are: S201. Collect all the status data of the energy-consuming product during its historical operation, extract the status data when the energy-consuming product has abnormal historical operation and the status data when it operates normally, plot the status data of the energy-consuming product during abnormal and normal operation respectively based on the time series, obtain the status data curve graphs during abnormal and normal operation, and use the polynomial fitting algorithm to fit all types of status data curve graphs, so as to obtain the change function G(s) of each type of status data during abnormal operation of the energy-consuming product and the change function F(s) during normal operation; extract all the independent variable coefficients of the abnormal and normal change functions respectively and calculate their averages. Let the average independent variable coefficient of G(s) be a1 and the average independent variable coefficient of F(s) be a2; calculate the change amplitude of the two average independent variable coefficients. The formula is: <h2 style=";text-align:left;direction:ltr">a<h2 style=";text-align:left;direction:ltr"> c <h2 style=";text-align:left;direction:ltr"> (a1-a2) In the formula, a c The variation range of the average independent variable coefficient of the variation function when the energy-consuming product is working abnormally and normally is expressed. The variation range of the average independent variable coefficient of the variation function of all state data is obtained by the same calculation. Assume that the number of types of state data is m. The variation ranges of the average independent variable coefficients of all calculated state data are sorted from small to large, the difference between adjacent variation ranges is calculated, and then the average value of m-1 variation range differences is calculated. The characteristic data screening threshold is calculated using the average value. The formula is: Cm=max|Ca j -Ca_p|; In the formula, Cm represents the maximum difference between the change amplitude difference and the average value, Ca j represents the difference between the jth adjacent change amplitudes, Ca_p represents the average value of all change amplitude differences, the formula represents the calculation of the difference between m-1 change amplitude differences and the average value in sequence and the extraction of the maximum value, and the adjacent change amplitude of the change amplitude difference corresponding to the maximum difference value is (a c，j , a c，j+1 ), select a c，j+1 As a feature data screening threshold; Use the screening threshold to judge the change amplitude of all status data, and regard the status data with a change amplitude greater than the screening threshold as characteristic data; Calculate the average value and standard deviation of all characteristic data of the energy-consuming product during abnormal and normal historical operation, and use the average value minus the standard deviation to obtain the abnormal threshold In of the energy-consuming product based on each type of characteristic data; S202. Set the reference sampling rate for different characteristic data of the energy-consuming product during abnormal and normal operation. The formula is: In the formula, f base-nor Indicates the sampling rate when the energy-consuming product is working normally, f base-ab Indicates the sampling rate when the energy-consuming product is working abnormally, a2 min Indicates the minimum value of all independent variable coefficients of the characteristic data that works properly, a2 max Indicates the maximum value of all independent variable coefficients of the characteristic data that works properly, a1 min The minimum value of all independent variable coefficients of characteristic data indicating abnormal working conditions, a1 max The maximum value among all the independent variable coefficients of the characteristic data indicating abnormal working conditions; f min Indicates the minimum sampling rate; the baseline sampling rate of all feature data is calculated according to the formula.

4. The method for monitoring and data collection based on the Internet of Things according to claim 1, wherein: The specific steps for adjusting the reference sampling rate of the energy-consuming product in real-time working state using the sliding window gradient in S300 are as follows: S301. Detect the characteristic data of the energy-consuming product during operation in real-time in the distributed acquisition network. Let the real-time characteristic data be Ts, and use the abnormal threshold to judge the real-time characteristic data. When any type of characteristic data satisfies Ts≥In, it is judged that the energy-consuming product operates abnormally; when all characteristic data satisfy Ts<In, it is judged that the energy-consuming product operates normally; S302. After judging the real-time working state of the energy-consuming product, analyze each real-time characteristic data based on the time series and calculate the gradient of each real-time characteristic data. The formula is: In the formula, △st represents the characteristic data gradient, s(t) represents the value of the characteristic data at time series t, s(t + △t) represents the value of the characteristic data at time series t + △t, and △t represents the change value of the time series; S303. Use the characteristic data gradient to adjust the reference sampling rate. The formula is: In the formula, fs represents the adjusted sampling rate, k represents the adjustment coefficient, and k is set according to experience and k<1.

5. The method for monitoring and data collection based on the Internet of Things according to claim 1, wherein: The specific steps for collecting the historical data of the energy-consuming product in S400 and inputting it into the virtual acquisition model for training are as follows: S401. Establish a physical simulation model of the energy-consuming product in each node of the distributed acquisition network. Deploy the virtual acquisition model in the physical simulation model using digital twin technology, collect the characteristic data of the energy-consuming product during historical operation and input it into the virtual acquisition model for training, generate the prediction curve of each type of characteristic data, with the time series as the independent variable and the characteristic data as the variable.

6. The method for monitoring and data collection based on the Internet of Things according to claim 1, characterized in that: The specific steps for fusing the measured value and the predicted value to obtain the true value output of the energy-consuming product in S500 are as follows: S501. Collect characteristic data of energy-consuming products using the adjusted sampling rate, use the collected characteristic data as the measured value sc, use the virtual acquisition model to predict each characteristic data as the predicted value sy, and calculate the reliability of the predicted value. The formula is: In the formula, S t represents the credibility of the predicted value, α represents a small constant to prevent division by zero; when St < H, it is determined that the corresponding predicted value is not credible and is removed. H represents the credibility threshold, which is set according to experience; S502: Fusing the measured value and predicted value of the feature data to obtain the true value. The formula is: sz=β×sc+(1-β)×sy, In the formula, sz represents the true value of the feature data, β represents the fusion coefficient, and δ represents the smoothing parameter. The formula is used to fuse each feature data to obtain the true value, and the fused true value is output.

7. An implementation monitoring and data acquisition system based on the Internet of Things, characterized by: The monitoring and data acquisition system includes a distributed acquisition network module, a baseline sampling rate setting module, a baseline sampling rate adjustment module, a virtual acquisition module, and a data fusion module; The distributed data collection network module is used to calculate and determine the location coordinates of each energy-consuming product, set up edge nodes, and combine the cloud and edge nodes in the working environment of the energy-consuming product to build a lightweight distributed data collection network; The reference sampling rate setting module is used to extract characteristic data of the energy-consuming product and set the reference sampling rate when the energy-consuming product is working abnormally and normally respectively; The reference sampling rate adjustment module is used to calculate the characteristic data gradient of the energy-consuming product when it is working in real time, and use the gradient to adjust the reference sampling rate of each characteristic data; The virtual acquisition module is used to establish a physical simulation model of the energy-consuming product in each node in the distributed acquisition network, and deploy the virtual acquisition model in the physical simulation model using digital twin technology; The data fusion module is used to obtain the predicted value of the feature data by using the virtual acquisition model, obtain the measured value by real-time acquisition, and fuse the predicted value and the measured value to obtain the true value.

8. The Internet of Things-based monitoring and data acquisition system according to claim 7, characterized in that: The reference sampling rate setting module includes a feature data extraction unit and a reference sampling rate setting unit; The characteristic data extraction unit is used to calculate a characteristic data screening threshold value and extract characteristic data of energy-consuming products using the screening threshold value; The reference sampling rate setting unit is used to calculate the reference sampling rates of different working states using the independent variable coefficient of each characteristic data curve.

9. The Internet of Things-based monitoring and data acquisition system according to claim 7, characterized in that: The reference sampling rate adjustment module includes a feature data gradient calculation unit and a reference sampling rate adjustment unit; The feature data gradient calculation unit is used to analyze each type of real-time feature data based on the time series and calculate the gradient of each type of real-time feature data; The reference sampling rate adjustment unit is used to adjust the reference sampling rate using the characteristic data gradient.

10. The Internet of Things-based monitoring and data acquisition system according to claim 7, characterized in that: The data fusion module includes a prediction value screening unit and a fusion unit; The predicted value screening unit is used to calculate the credibility of the predicted value, set a credibility threshold, and use the credibility threshold to screen the predicted value; The fusion unit is used to obtain a fusion coefficient by using the credibility calculation, and to fuse the measured value and the predicted value by using the fusion coefficient to obtain a true value for output.

Images