Construction of digital twin model of ALOT equipment in smart park and data collaborative processing method

Through the classification and historical data analysis of ALOT equipment data, combined with acquisition time and device status, the problem of low data accuracy is solved, the accurate data processing and accurate modeling of digital twin models are realized, and the reliability of data applications is improved. sex.

CN118734235BActive Publication Date: 2024-11-19HUNAN TELECOMM CONSTR CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411224424.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-03
Publication Date
2024-11-19
Estimated Expiration
2044-09-03

AI Technical Summary

Technical Problem

The existing data collaborative processing methods fail to effectively combine the data's own characteristics and past data characteristics, resulting in low data accuracy and affecting subsequent applications.

Method used

By classifying the static and dynamic data of ALOT devices, combining the acquisition time and device status, the data is classified normally and abnormally, and using historical data to match and correct data to improve data accuracy; at the same time, a digital twin model is built to improve construction Accurate model.

Benefits of technology

It realizes reasonable data management and accuracy analysis, improves the accuracy of data processing, and further improves modeling accuracy through digital twin models.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118734235B_ABST
    Figure CN118734235B_ABST
Patent Text Reader

Abstract

The present invention discloses a method for constructing a digital twin model of ALOT equipment in a smart park and for collaborative data processing. The present invention relates to the technical field of model construction and collaborative data processing, and solves the technical problem that the data accuracy is not high due to the lack of comprehensive processing combining the characteristics of the data itself with the characteristics of past data. The present invention classifies the collected data to achieve rational management of the data, and then performs accuracy analysis on the classified data. The acquired data is analyzed in combination with historical data to improve the accuracy of the data. At the same time, further analysis is performed based on the characteristics of the data, and erroneous data is corrected. Secondly, a secondary accuracy judgment is performed on the corrected data, and correct data is used for digital twin modeling processing to further improve the accuracy of modeling.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of model construction and data collaborative processing, and specifically to a method for constructing a digital twin model of ALOT equipment in a smart park and collaborative data processing. Background Art

[0002] ALOT devices refer to smart devices that combine artificial intelligence (AI) and Internet of Things (LOT) technologies. They can realize intelligent data processing and network communication. The core of ALOT devices is to realize the autonomous perception, learning and decision-making capabilities of the devices through advanced technical means, so that the devices can interact with the environment and other devices more intelligently.

[0003] According to the patent with application number CN202110407787.3, the patent specifically includes: the mobile device obtains the data to be processed, and sends the data to be processed to multiple fog nodes whose coverage includes the current location of the mobile device; after receiving the data to be processed, the multiple fog nodes select the target fog node closest to the mobile device from the multiple fog nodes, and establish a processing task for the data to be processed in the target fog node; if the current processing resources of the target fog node cannot process the processing task, according to the fault type of the current processing resources, a processing plan for the processing task is determined by a preset deep reinforcement learning network DQN algorithm; the target fog node collaborates with the collaborative fog node to process the processing task according to the processing method in the processing plan.

[0004] The above patent selects the fog node closest to the mobile device as the target fog node to establish a processing task for the data to be processed after the fog node receives the data to be processed from the mobile device. When the target fog node cannot process the processing task due to load saturation or the type of service it can provide, a processing solution for the processing task is obtained through analysis of a pre-trained DQN (Deep QNetwork, deep reinforcement learning network). The target fog node collaborates with the collaborative fog node to process the processing task according to the processing solution.

[0005] However, some existing data collaborative processing methods improve the efficiency of data processing through multi-party cooperation when processing data, but there are still problems with inadequate data processing. For example, they fail to combine the characteristics of the data itself with the characteristics of past data for comprehensive analysis, which will affect the accuracy of the processed data and further cause trouble for the subsequent data application. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention provides a digital twin model construction and data collaborative processing method for ALOT equipment in a smart park, which solves the problem of low data accuracy caused by not combining the characteristics of the data itself with the features of past data for comprehensive processing.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: a method for collaboratively processing data of ALOT devices in a smart park, the method specifically comprising the following steps:

[0008] Step 1: Collect static data and dynamic data corresponding to the ALOT device, and classify the collected data into the same type to obtain classified data;

[0009] Step 2: Classify the device data in the classified data into device-origin data and non-device-origin data according to the data source;

[0010] Step 3: Classify the original data of the device into normal data and abnormal data in combination with the acquisition time and the device status, and classify the normal data into overdue data and non-overdue data according to the acquisition time;

[0011] Step 4: Match the non-device source data with historical data to obtain matching results, and classify the matching results into normal matching results and abnormal matching results according to the device status. At the same time, compare the deviation value between the normal matching result and the non-device source data with the threshold value, and classify the non-device source data into normal data and abnormal data;

[0012] Step 5: Correct the abnormal data and outdated data to obtain corrected data, and compare and judge the corrected data to generate corrected correct data or corrected erroneous data.

[0013] As a further solution of the present invention: the specific method of classifying the device data in step 2 is:

[0014] Take any set of classified data as the analysis target, then judge the source of the device data in the analysis data, and classify the device data from the ALOT device as device source data, otherwise classify it as non-device source data.

[0015] As a further solution of the present invention: the specific method of analyzing the device source data in step 3 is:

[0016] Obtain the acquisition time and the corresponding device status of the device source data. When the device status is normal, classify the corresponding device source data as normal data, otherwise classify it as abnormal data;

[0017] The method for obtaining normal data and performing secondary analysis is: comparing the acquisition time of the normal data with the preset time value. When the acquisition time is greater than the preset value, the corresponding normal data is marked as overdue data. Conversely, when the acquisition time is less than the preset value, the corresponding normal data is marked as not overdue data.

[0018] The data that has not exceeded the validity period is labeled as i, where i=1, 2, ..., j, and the number of uses corresponding to the data i that has not exceeded the validity period within the acquisition time Ti is obtained, which is recorded as Ci. At the same time, the usage duration corresponding to the usage frequency Ci is obtained, which is recorded as Cit. The average usage duration corresponding to the usage frequency Ci within the acquisition time Ti is calculated, which is recorded as Cip, and the ratio of the average usage duration to the usage frequency is calculated.

[0019] As a further solution of the present invention: the analysis method of the non-device source data in step 4 is as follows:

[0020] All non-device source data and their corresponding historical data are obtained, and the two are matched to obtain matching results. At the same time, the matching results are classified as normal matching results or abnormal matching results according to the device status corresponding to the matching results.

[0021] The deviation value between the normal matching result and the non-device source data is calculated, and the deviation value is compared with the threshold. The specific value of the threshold is set by the operator. When the deviation value is greater than the threshold, the non-device source data is classified as abnormal data. Conversely, when the deviation value is less than the threshold, the non-device source data is classified as normal data.

[0022] As a further solution of the present invention: the specific method of analyzing abnormal data and overdue data in step 5 is:

[0023] The generation conditions of abnormal data and outdated data are obtained, and the generation conditions are adjusted. The data obtained after the generation conditions are adjusted are recorded as corrected data. Then, the historical data under the same conditions as the corrected data is obtained, and the corrected data is compared with the historical data to generate corrected correct data or corrected erroneous data.

[0024] As a further solution of the present invention: a method for constructing a digital twin model of ALOT equipment in a smart park, the method for constructing a digital twin model specifically includes the following steps:

[0025] S1: Obtain the collaboratively processed device data and determine the application areas and objectives of the digital twin model based on the device data;

[0026] S2: Collect physical data and data features based on the acquired device data, and clean the acquired device data;

[0027] S3: Obtain the structure and establish the parameter model, and select the appropriate modeling algorithm according to the data characteristics;

[0028] S4: Perform model training, verification and optimization;

[0029] S5: Connect and verify the digital twin model with the actual system, and make adjustments and updates.

[0030] The present invention provides a method for constructing a digital twin model of ALOT equipment in a smart park and for collaborative data processing. Compared with the prior art, it has the following beneficial effects:

[0031] The present invention can achieve rational management of data by classifying the collected data, and secondly perform accuracy analysis on the classified data. The acquired data is analyzed in combination with historical data to improve the accuracy of the data. At the same time, further analysis is performed according to the characteristics of the data, and erroneous data is corrected. Secondly, a secondary accuracy judgment is made on the corrected data, and the correct data is used for digital twin modeling to further improve the accuracy of modeling. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 is a flow chart of the collaborative processing method of the present invention;

[0033] Figure 2 This is a diagram of the model construction method of the present invention. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0035] See also Figure 1 , the present application provides a method for collaborative processing of ALOT device data in a smart park, the method specifically comprising the following steps:

[0036] Step 1: Collect the device data corresponding to the ATOL device. The data generated by the ATOL device includes static data and dynamic data. Static data is usually the inherent attribute of the device and does not change frequently, such as the location of the device, GIS information, device name and number, asset attribute-related tags, device specifications, etc. Dynamic data is generated by the device during operation and can reflect the real-time status of the device or environmental changes, such as temperature, humidity, pressure, audio and other sensor data, and is classified according to the device data. The specific classification method is: classify the same type of device data according to the type of device data, such as uniformly classifying the acquired temperature data, uniformly classifying the pressure data, and so on, to further acquire the classified data.

[0037] Step 2: Analyze the accuracy of the obtained classification data, and the specific analysis method is as follows:

[0038] All classified data are obtained, and any group of classified data is obtained as the analysis target. Then, the device data in the analysis target is classified. The device source data and non-device source data are obtained by classification according to the source of the device data. Specifically, the classified data obtained here is represented by judging whether it is data directly obtained from the device. If so, it is classified as device source data, otherwise it is classified as non-device source data. Non-device source data may be obtained by calculation and other methods, and then the two are analyzed separately.

[0039] Step 3: Obtain all the device source data, and at the same time obtain the acquisition time and the corresponding device status of the device source data. The specific device status is represented by whether the corresponding device is in a normal working state or an abnormal working state when the device source data is obtained. Then, the device source data is classified into normal data and abnormal data according to the device status, and the normal data is analyzed again. The specific analysis method is as follows:

[0040] The acquisition time corresponding to the normal data is obtained, and then the acquisition time is compared with the preset time value. The preset time value here represents the effective time retained for normal data collection, and also represents the timeliness of the data. When the acquisition time is greater than the preset value, the corresponding normal data is marked as over-time data. Conversely, when the acquisition time is less than the preset value, the corresponding normal data is marked as not over-time data.

[0041] Next, the data that has not exceeded the validity period is analyzed by analyzing the number of times the data that has not exceeded the validity period is used within the acquisition time period. The specific analysis method is as follows:

[0042] All the data that have not exceeded the validity period are obtained and labeled as i, where i=1, 2, ..., j. The number of uses corresponding to the data i that has not exceeded the validity period within the acquisition time Ti is obtained and recorded as Ci. At the same time, the usage time corresponding to the usage time Ci is obtained and recorded as Cit. The usage time here is represented by the overall operating time of the equipment in the process of using the overdue data for analysis. The average usage time corresponding to the usage time Ci at the acquisition time Ti is calculated and recorded as Cip. At the same time, the ratio of the average usage time to the usage times is calculated, and it is converted into a percentage and used as the usage accuracy.

[0043] Taking one set of overdue data as an example, the number of uses within the corresponding acquisition time is 4 times, and the usage time corresponding to a single use is 2h, 5h, 1.5h and 3h. Then the overall average of the usage time is calculated to be 2.8, and then the average 2.8 is calculated by the ratio of the number of times 4 to get the final percentage of 70%, which is therefore used as the usage accuracy.

[0044] Marked outdated data will not be processed.

[0045] Step 4: Obtain all non-device source data and their corresponding historical data. The specific historical data includes all data generated by the device, such as temperature, pressure, etc., and match the historical data with the non-device source data as the standard to obtain a matching result. The matching result here is represented by the same type of data belonging to the non-device source data in the historical data. Then, the matching result is judged according to the corresponding device status. When the device status corresponding to the matching result is normal, the matching result is classified as a normal matching result. On the contrary, when the device status corresponding to the matching result is abnormal, the matching result is classified as an abnormal matching result. Then, a secondary analysis is performed on the obtained normal matching result. The specific secondary analysis method is as follows:

[0046] The deviation value between the normal matching result and the non-device source data is obtained, and the deviation value is compared with the threshold. The threshold here represents the deviation value within the allowable range, and the specific setting range of the threshold is determined according to the actual situation. When the deviation value is greater than the threshold, the non-device source data is classified as abnormal data. Conversely, when the deviation value is less than the threshold, the non-device source data is classified as normal data.

[0047] For example, the non-device source data here is temperature. The specific normal matching result indicates the temperature data obtained when the device is working normally. Then the deviation value between the two is calculated, that is, the temperature difference is, for example, 15°C. Then, the temperature difference of 5°C is compared with the threshold, and the threshold here is 10°C. Therefore, the deviation value between the two exceeds the normal threshold range, which further indicates that the non-device source data obtained here is abnormal.

[0048] Step 5: Analyze the data classified as abnormal data and overdue data, adjust and correct the corresponding generation conditions of the two, and obtain the data after the conditions are corrected as corrected data. Then judge the correctness of the corrected data, obtain the historical data under the same conditions as the corrected data, and match the two. When the corrected data is the same as the historical data, it means that the corrected data is correct. Conversely, when the corrected data is different from the historical data, it means that the corrected data is wrong.

[0049] Combined with actual analysis, both abnormal data and overdue data are assumed to be erroneous data by default, so the corresponding generation conditions of the two are adjusted. For example, the abnormal data is pressure, and the obtained pressure is obtained under the equipment temperature condition of 60 degrees Celsius. Under normal circumstances, the optimal temperature for obtaining pressure is 45 degrees Celsius. Therefore, the temperature needs to be adjusted and the pressure data needs to be obtained at 45 degrees Celsius.

[0050] The acquired pressure data is compared again to determine whether there is an abnormality, and then the pressure data is judged to be correct or incorrect.

[0051] The digital twin model of ALOT equipment in the smart park is constructed, and the construction method of the digital twin model is as follows:

[0052] S1: Obtain the equipment data after collaborative processing, and determine the application field and target of the digital twin model based on the equipment data. Specifically, the equipment data obtained here are all normal data, and the non-timeout data with an accuracy rate of more than 90% can also be used as subsequent modeling and analysis data.

[0053] Before starting to build a digital twin model, it is necessary to clarify the application field and objectives of the model, such as industrial manufacturing, energy, transportation, etc., as well as the goals of the model, such as optimizing the design of the physical object, improving efficiency, or predicting the performance of the physical object.

[0054] S2: Collect physical data and data features based on the acquired device data, and clean the acquired device data. Data cleaning mainly includes: missing value processing, outlier detection and correction, etc. The specific data features are mainly reflected in real-time and diversity. First, ALOT devices deploy a large number of sensors of various types. Each sensor is an information source that can collect and update environmental information in real time. This real-time nature makes the data highly timely and valuable.

[0055] Secondly, due to the use of multiple types of sensors, the data collected by the ALOT device is diverse. The information content and information format captured by different categories of sensors may be different, thus providing a rich data source and diverse data dimensions.

[0056] S3: Obtain the structure and establish the parameter model, and select the appropriate modeling algorithm according to the data characteristics. Various algorithms collected based on mathematical models, feature physics, model building, digital statistical model, twin model neural network, etc.

[0057] S4: Perform model training, verification and optimization. Specific model optimization includes hyperparameter tuning, model simplification and ensemble learning. Hyperparameter tuning uses grid search, random search or Bayesian optimization to adjust the model's hyperparameters to find the optimal solution.

[0058] Model simplification means that if the model is too complex, consider using model simplification techniques such as principal component analysis (PCA) or feature selection methods to reduce the complexity of the model.

[0059] Ensemble learning uses ensemble learning methods, such as Bagging, Boosting, or Stacking, which can usually improve the performance of the model.

[0060] Combined with actual analysis, in the modeling process, model training, verification and optimization are required to ensure the accuracy and reliability of the model. The accuracy of the model can be verified by comparing and verifying with actual data. If there are differences or deviations between the model and the actual data, the model needs to be adjusted and optimized until the simulation results are consistent with the actual situation.

[0061] S5: Connect and verify the digital twin model with the actual system, and make adjustments and updates. Specifically, the established theoretical model is implemented in the computer system and assembled. Then the model is verified with actual data to ensure the accuracy and reliability of the model.

[0062] Some of the data in the above formulas are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0063] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. Smart Park ALOT Equipment Data Collaborative Processing Method, It is characterized in that The data collaborative processing method specifically includes the following steps: Step 1: Collect static data and dynamic data corresponding to the ALOT device, and classify the collected data into the same type to obtain classified data. Static data represents the inherent attributes of the device and does not change frequently. Dynamic data represents the data generated during the operation of the device and can reflect the real-time status of the device or environmental changes. Step 2: Classify the device data in the classified data into device-origin data and non-device-origin data according to the data source, and the specific classification method is: Get any set of classified data as the analysis target, then judge the source of the device data in the analysis data, and classify the device data whose source is ALOT device as device source data, otherwise classify it as non-device source data; Step 3: Classify the original data of the device into normal data and abnormal data in combination with the acquisition time and the device status, and classify the normal data into overdue data and non-overdue data according to the acquisition time; Step 4: Match the non-device source data with historical data to obtain matching results, and classify the matching results into normal matching results and abnormal matching results according to the device status. At the same time, compare the deviation value between the normal matching result and the non-device source data with the threshold value, and classify the non-device source data into normal data and abnormal data; Step 5: Correct the abnormal data and outdated data to obtain corrected data, and compare and judge the corrected data to generate corrected correct data or corrected erroneous data.

2. The method for collaboratively processing ALOT device data in a smart park according to claim 1, It is characterized in that The specific method of analyzing the device source data in step 3 is: Obtain the acquisition time and the corresponding device status of the device source data. When the device status is normal, classify the corresponding device source data as normal data, otherwise classify it as abnormal data; The method for obtaining normal data and performing secondary analysis is: comparing the acquisition time of the normal data with the preset time value. When the acquisition time is greater than the preset value, the corresponding normal data is marked as overdue data. Conversely, when the acquisition time is less than the preset value, the corresponding normal data is marked as not overdue data.

3. The method for collaboratively processing ALOT device data in a smart park according to claim 2, It is characterized in that The analysis method for the non-time-out data is as follows: The data that has not exceeded the validity period is labeled as i, where i=1, 2, ..., j, and j represents the total number of data that has not exceeded the validity period. The number of uses corresponding to the data i that has not exceeded the validity period within the acquisition time Ti is obtained, which is recorded as Ci. At the same time, the usage duration corresponding to the usage times Ci is obtained, which is recorded as Cit. The average usage duration corresponding to the usage times Ci within the acquisition time Ti is calculated, which is recorded as Cip. At the same time, the ratio of the average usage duration to the number of uses is calculated.

4. The method for collaboratively processing ALOT device data in a smart park according to claim 1, It is characterized in that The analysis method for non-device source data in step 4 is as follows: All non-device source data and their corresponding historical data are obtained, and the two are matched to obtain matching results. At the same time, the matching results are classified as normal matching results or abnormal matching results according to the device status corresponding to the matching results.

5. The method for collaboratively processing ALOT device data in a smart park according to claim 1, It is characterized in that The specific method of analyzing the normal matching results in step 4 is: The deviation value between the normal matching result and the non-device source data is calculated, and the deviation value is compared with the threshold. The specific value of the threshold is set by the operator. When the deviation value is greater than the threshold, the non-device source data is classified as abnormal data. Conversely, when the deviation value is less than the threshold, the non-device source data is classified as normal data.

6. The method for collaboratively processing data of ALOT devices in a smart park according to claim 1, It is characterized in that The specific method of analyzing abnormal data and overdue data in step 5 is as follows: The generation conditions of abnormal data and outdated data are obtained, and the generation conditions are adjusted. The data obtained after the generation conditions are adjusted are recorded as corrected data. Then, the historical data under the same conditions as the corrected data is obtained, and the corrected data is compared with the historical data to generate corrected correct data or corrected erroneous data.

7. A method for constructing a digital twin model of an ALOT device in a smart park, the method being applied to the method for collaborative data processing of ALOT devices in a smart park as claimed in any one of claims 1 to 6. It is characterized in that The method for constructing the digital twin model specifically includes the following steps: S1: Obtain the collaboratively processed device data and determine the application areas and objectives of the digital twin model based on the device data; S2: Collect physical data and data features based on the acquired device data, and clean the acquired device data; S3: Obtain the structure and establish the parameter model, and select the appropriate modeling algorithm according to the data characteristics; S4: Perform model training, validation, and optimization; S5: Connect and verify the digital twin model with the actual system, and make adjustments and updates.