METHOD FOR MONITORING OPERATION OF LIQUEFIED NATURAL GAS (LNG) STORAGE AND INTERNET OF THINGS SYSTEM (IOT) THEREOF

20230419425

18/464316

September 11, 2023

The present disclosure discloses a method for monitoring operation of liquefied natural gas (LNG) storage, comprising: acquiring operating data of an LNG storage device, physical and chemical parameters of LNG and historical pressure change data in the LNG storage device; determining pseudo data information based on the historical pressure change data; determining at least one set of pressure change data at at least one future time point through a pressure model based on the operating data, the physical and chemical parameters, the historical pressure change data, and the pseudo data information, wherein the pressure model is a machine learning model, the pressure model includes a feature extracting layer and a pressure layer; and determining a pressure adjusting time point and preparing for a pressure adjustment based on the at least one set of pressure change data at the at least one future time point.

CHENGDU PUHUIDAO SMART ENERGY TECHNOLOGY CO., LTD. (Chengdu, CN)

1. A method for monitoring operation of liquefied natural gas (LNG) storage, executed by a distributed energy management platform of an Internet of Things (IoT) system for monitoring operation of LNG storage, comprising: acquiring operating data of an LNG storage device, physical and chemical parameters of LNG and historical pressure change data in the LNG storage device; determining pseudo data information based on the historical pressure change data; determining at least one set of pressure change data at at least one future time point through a pressure model based on the operating data, the physical and chemical parameters, the historical pressure change data, and the pseudo data information, wherein the pressure model is a machine learning model, the pressure model includes a feature extracting layer and a pressure layer, wherein an input of the feature extracting layer includes a forward prediction step size, the historical pressure change data, and the pseudo data information, an output of the feature extracting layer includes a pressure change feature, and the forward prediction step size is a time interval between future time points for sampling; an input of the pressure layer includes the pressure change feature, the operating data, and the physical and chemical parameters, and an output of the pressure layer includes the at least one set of pressure change data at the at least one future time point; and determining a pressure adjusting time point and preparing for a pressure adjustment based on the at least one set of pressure change data at the at least one future time point.

2. The method according to claim 1, wherein the operation data at least include thermal conductivity, an environment temperature, and a storage temperature, the method further comprising: in response to a change in a quality of LNG, determining a changed storage temperature based on the operating data and the physical and chemical parameters.

3. The method according to claim 1, wherein the physical and chemical parameters at least include a type, a pressure, and the quality of the LNG at a plurality of consecutive time points.

4. The method according to claim 1, wherein the at least one future time point is determining by a process including: determining a pseudo data feature based on the pseudo data information; and determining a count of the at least one future time point and the forward prediction step size based on the pseudo data feature.

5. The method according to claim 4, wherein the forward prediction step size is determined by a process including: determining a post-update forward prediction step size of the at least one future time point through processing a pre-update forward prediction step size of the at least one future time point and a pseudo data sequence feature vector by an adjusting model, wherein the pseudo data sequence feature vector is determined based on the pseudo data feature, the forward prediction step size is not exceeded a preset lower limit, and the preset lower limit is determined based on an anomaly processing response time.

6. The method according to claim 1, wherein the determining a pressure adjusting time point based on the at least one set of pressure change data at the at least one future time point includes: determining a candidate time point when the at least one set of pressure change data reaches a preset pressure threshold based on the at least one set of pressure change data at the at least one future time point; and determining the pressure adjusting time point based on the candidate time point and a pseudo data feature.

7. The method according to claim 6, wherein the determining the pressure adjusting time point based on the candidate time point and a pseudo data feature includes: determining the pressure adjusting time point through revising the candidate time point based on a pseudo data feature coefficient, wherein the pseudo data feature coefficient is determined based on the pseudo data feature.

8. The method according to claim 1, further comprising: obtaining actual pressure change data; and performing a failure analysis on the LNG storage device based on the actual pressure change data and the at least one set of pressure change data at the at least one future time point.

9. The method according to claim 8, wherein the performing a failure analysis on the LNG storage device based on the actual pressure change data and the at least one set of pressure change data at the at least one future time point includes: obtaining a difference distribution time point based on the actual pressure change data and the at least one set of pressure change data at the at least one future time point; obtaining a pseudo data distribution time point based on a pseudo data feature; determining a similarity based on the difference distribution time point and the pseudo data distribution time point; and determining a failure type based on the actual pressure change data, the at least one set of pressure change data at the at least one future time point, and the similarity.

10. The method according to claim 1, further comprising: monitoring, by utilizing a data acquiring unit, the LNG storage device, perceiving and acquiring pressure, temperature, and position data on the LNG storage device, obtaining encrypted perception information through performing an analog-to-digital conversion on perception information by the data acquiring unit and symmetrically encrypting the perception information by adopting a microsoft point-to-point encryption (MPPE) and Internet Protocol Security (IPSec) mechanism in a binary mode, and managing key by a public-private key verification; actively sending, by the data acquiring unit, authentication information to an LNG distributed energy management platform at a designated address through an LNG distributed energy storage sensor network platform, after passing a two-way symmetric authentication, establishing a unique communication channel between the data-acquiring unit and the LNG distributed energy management platform to transmit the encrypted perception information; decrypting, by the LNG distributed energy management platform, the encrypted perception information, performing an anomaly judgment on decrypted perception information according to a preset anomaly judgment condition, and screening out abnormal perception information; performing a pseudo data verification on the abnormal perception information utilizing a pseudo data verification manner, identifying and labeling a type of pseudo data caused by an external environmental interference; performing an anomaly prediction analysis on the operating data of the LNG storage device according to an early warning mechanism; sending, by an LNG distributed energy storage maintenance personnel sensor network platform, an alarm prompt to field maintenance personnel for an inspection and processing according to a tank number of the LNG storage device corresponding to the abnormal data obtained by the anomaly prediction analysis and the anomaly judgment; and sending, by the field maintenance personnel, processing information to the LNG distributed energy management platform through the LNG distributed energy storage maintenance personnel sensor network after completing the inspection and processing, and confirming, by the LNG distributed energy management platform, whether the processing is completed; obtaining, by the LNG distributed energy management platform, processed tank perception information through the LNG distributed energy storage sensor network platform and confirming that the field maintenance personnel completes the processing if a status of the processed tank perception information is normal, and feeding back to the field maintenance personnel.

11. The method according to claim 10, wherein the performing an anomaly prediction analysis on the operating data of the LNG storage device according to an early warning mechanism includes: data preprocessing: adopting a Holt double-parameter linear exponential smoothing manner to smooth the decrypted perception information to obtain a monitoring time series xt; model initialization: an initialization model order p=1, a forward predicted step size np=np0; model establishment: establishing an initial auto-regression moving average (ARMA) model based on the monitoring time series xt; determining a length of a modeling sample: determining an integer multiple of an inverse of an interval between two adjacent frequencies in a temporal frequency domain of the perception information as the length of the modeling sample through time series analysis; estimating model parameter: estimating the model parameter by utilizing a least square manner; inspecting the model and determining an order: determining a machine order p of a parameter change trend predicting model to obtain a final parameter trend predicting model ARMA (2p, 2p−1) by adopting an Akaike information criterion (AIC); predicting parameter: obtaining a prediction interval by calculating a continuous forward predicted step size np; and analyzing the abnormal data: obtaining an operating prediction result of the LNG storage device through calculating a best prediction result and a corresponding prediction interval corresponding to the best prediction result by adopting a dynamically correcting ARMA prediction manner, and determining whether the operating prediction result is the abnormal data according to the preset anomaly judgment condition.

12. The method according to claim 10, wherein the performing a pseudo data verification on the abnormal perception information by utilizing a pseudo data verification manner and identifying and labeling a type of pseudo data caused by an external environmental interference includes: establishing the pseudo data verification manner, setting an error code in a sensor program of the data acquiring unit to simulate a sensor value during a real electromagnetic interference for pseudo data generated by an electromagnetic interference in a field maintenance process in advance, and setting an anomaly analysis result of the LNG distributed energy management platform as pseudo data of the electromagnetic interference; for pseudo data generated by a transmission or device failure, randomly creating a sensor or transmission line failure, and setting the anomaly analysis result labeled by the LNG distributed energy management platform as pseudo data of the sensor or transmission line; and performing a pseudo data analysis on the abnormal perception information, and labeling a corresponding type of pseudo data by utilizing the pseudo data verification manner.

13. The method according to claim 11, wherein the data preprocessing includes: processing the abnormal perception data, forming the monitoring time series {xt, t=1, 2, . . . , N} for the perceived and acquired operating data of the LNG storage device, and for abnormal monitoring data being zero or with a low probability sensor value, calculating a one-step smoothing value F t of first Nx numbers by monitoring the first Nx numbers in the monitoring time series to replace the abnormal monitoring data, and selecting actual monitoring operating data to obtain a length Nx of the monitoring time series used for a smoothing calculation; processing missing data, for a missing sequence {xt, t=1, 2 . . . } formed by original monitoring data, firstly obtaining the length Nx of the monitoring time series of the original data required for the smoothing calculation according to an actual monitoring operating data analysis; and setting a count of smoothing steps m, and for gas concentration monitoring values {xt, t=1, 2, . . . , Nx} of the first Nx points of missing data points, continuously performing the smoothing calculation of m steps to obtain a final smoothed value Ft+m, and finally inserting the final smoothed value Ft+m into the missing sequence to form a complete monitoring data time series.

14. The method according to claim 11, wherein the dynamically correcting ARMA prediction manner includes: evaluating a predicted error, for previous j−1 predictions, calculating an average value of prediction errors of previous n predictions, and obtaining an error minimum value and an error subminimum value; determining an effective model order, determining model orders p1 and p2 when the error minimum value and the error subminimum value are obtained as effective model orders of the previous j−1 predictions; modeling with current data, for an analysis sequence formed by operating monitoring data of the current LNG storage device, obtaining an optimal order p0 through the ARMA model for parameter estimation and validity inspection; predicting model, taking p=p0, p1, p2 as an order respectively to perform an operating data parameter prediction, and obtaining prediction results X=[xj1, xj2, xj3]; and calculating the best prediction result, calculating an average value of each element of X=[xj1, xj2, xj3] to obtain a final prediction result as the best prediction result.

15. An Internet of Things (IoT) system for monitoring operation of liquefied natural gas (LNG) storage, which is realized by using a method for monitoring operation of LNG storage, wherein the IoT system includes an object platform, a sensor network platform, a management platform, a service platform, and a user platform; the object platform includes an LNG distributed energy storage object platform and an LNG distributed energy storage maintenance personnel object platform; the LNG distributed energy storage object platform is configured to monitor and perceive operating data of an LNG storage device, physical and chemical parameters of LNG and historical pressure change data in the LNG storage device, and transmit the perception information, physical and chemical parameters and the historical pressure change data after symmetric encryption to the LNG distributed energy management platform through the sensor network platform; and the LNG distributed energy storage maintenance personnel object platform is configured for field maintenance personnel to receive an alarm prompt and feedback on maintenance processing; the sensor network platform includes an LNG distributed energy storage sensor network platform and an LNG distributed energy storage maintenance personnel sensor network platform, which are configured to realize a communication connection for perception and control between the management platform and the object platform; the management platform is configured to acquire the operating data, the physical and chemical parameters and the historical pressure change data, determine pseudo data information based on the historical pressure change data, determining at least one set of pressure change data at at least one future time point through a pressure model based on the operating data, the physical and chemical parameters, the historical pressure change data, and the pseudo data information, wherein the pressure model is a machine learning model, the pressure model includes a feature extracting layer and a pressure layer, wherein an input of the feature extracting layer includes a forward prediction step size, the historical pressure change data, and the pseudo data information, an output of the feature extracting layer includes a pressure change feature, and the forward prediction step size is a time interval between future time points for sampling; an input of the pressure layer includes the pressure change feature, the operating data, and the physical and chemical parameters, and an output of the pressure layer includes the at least one set of pressure change data at the at least one future time point, and determining a pressure adjusting time point and preparing for a pressure adjustment based on the at least one set of pressure change data at the at least one future time point, send an alarm prompt to field maintenance personnel for an inspection and processing according to a tank number of the LNG storage device corresponding to the abnormal data obtained by an anomaly prediction analysis and an anomaly judgment through the sensor network platform; the service platform is configured to obtain perception information demanded by a user from the management platform for analysis and storage, and receive control information sent by the user for processing and send processed control information to the management platform; the user platform is configured to obtain the operating data of the LNG storage device from the service platform for various users and send the control information to the service platform; the service platform is further configured to acquire the operating data, the physical and chemical parameters, the historical pressure change data, the at least one set of pressure change data, the pressure adjusting time point demanded by a user from the management platform for analysis and storage, and receive control information sent by the user for processing and send processed control information to the management platform; and the user platform is further configured to obtain the operating data from the service platform, the physical and chemical parameters, the at least one set of pressure change data, the pressure adjusting time point for the user and send the control information to the service platform.

16. The Internet of Things (IoT) system according to claim 15, wherein the management platform is further configured to: determine a pseudo data feature based on the pseudo data information; and determine a count of the at least one future time point and the forward prediction step size based on the pseudo data feature.

17. The Internet of Things (IoT) system according to claim 16, wherein the management platform is further configured to: determine a post-update forward prediction step size of the at least one future time point through processing a pre-update forward prediction step size of the at least one future time point and a pseudo data sequence feature vector by an adjusting model, wherein the pseudo data sequence feature vector is determined based on the pseudo data feature, the forward prediction step size is not exceeded a preset lower limit, and the preset lower limit is determined based on an anomaly processing response time.

18. The Internet of Things (IoT) system according to claim 15, wherein the management platform is further configured to: determine a candidate time point when the at least one set of pressure change data reaches a preset pressure threshold based on the at least one set of pressure change data at the at least one future time point; and determine the pressure adjusting time point based on the candidate time point and a pseudo data feature.

19. The Internet of Things (IoT) system according to claim 15, wherein the management platform is further configured to: obtain actual pressure change data; and perform a failure analysis on the LNG storage device based on the actual pressure change data and the at least one set of pressure change data at the at least one future time point.

20. A non-transitory computer readable storage medium storing a set of instructions, when executed by at least one processor, causing at least one processor to perform a method for monitoring operation of liquefied natural gas (LNG) storage of claim 1.

06; 04; 06

edspap.20230419425