Granted Patent B2

US 12,646,944 · App. 18/313,324 · Granted Jun 2, 2026

Inventor: Lin Fu (Chendu, CN)

Assignee: CHENGDU JIUGUAN SMART ENERGY TECHNOLOGY CO., LTD.

H02J3/17G06Q10/06H02J2101/10

Abstract

Claims (77)

1 . A non-transitory computer-readable storage medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:

data collection: collecting operation and maintenance data of an LNG smart terminal and personal data of each of different safety inspectors;

data analysis: monitoring an operation and maintenance situation of the LNG smart terminal in real-time for analysis and generating an inspection order reminder and an inspection instruction;

task matching: matching an inspection requirement of the LNG smart terminal with inspection data of each safety inspector and sending the inspection instruction and the inspection order reminder to an optimal safety inspector for safety inspection; wherein the task matching includes:

decrypting the operation and maintenance data uploaded by an LNG smart terminal to be maintained and obtaining location information of the LNG smart terminal to be maintained; analyzing the personal data of each safety inspector, obtaining personal current location information and a working status of each safety inspector, and judging whether each safety inspector is in the working status; and performing an optimal matching calculation on the working status and the personal current location information of each safety inspector with the location information of the LNG smart terminal to be maintained, determining the optimal safety inspector according to a calculation result, and sending the inspection order reminder and the inspection instruction to the optimal safety inspector for the safety inspection; wherein the optimal matching calculation includes:

first calculating a distance between a personal current location of each safety inspector and a location of the LNG smart terminal and selecting safety inspectors with a distance less than a preset distance; and giving priority matching to a safety inspector who is not in the working status, estimating an inspection completion time of the safety inspector arriving at the location of the LNG smart terminal, and if the inspection completion time exceeds a preset time threshold, estimating completion times of all safety inspectors in the working status, and selecting a safety inspector with a shortest completion time as the optimal safety inspector for the safety inspection; wherein the estimating an inspection completion time of the safety inspector arriving at the location of the LNG smart terminal includes:

obtaining personnel features of the safety inspector, terminal features of the LNG smart terminal, and location features of a start point and an end point; predicting the inspection completion time by processing the personnel features, the terminal features, and the location features through an inspection completion time prediction model, wherein the inspection completion time prediction model is a machine learning model; wherein

the inspection completion time prediction model includes a route planning layer, a time prediction layer, and a time integration layer; wherein the route planning layer is a machine learning model for determining candidate inspection routes, an input of the route planning layer include the location features of the start point and the end point, and an output of the route planning layer include one or more candidate inspection routes; the time prediction layer is a machine learning model for predicting a candidate completion time of a candidate inspection route, an input of the time prediction layer include the personnel features of the safety inspector, the terminal features of the LNG smart terminal, and the candidate inspection route, and an output of the time prediction layer include a candidate inspection completion time of the candidate inspection route; the time integration layer is an algorithm model for determining the inspection completion time, the time integration layer processes candidate inspection completion times of a plurality of candidate inspection routes and outputs the inspection completion time; and

the route planning layer is obtained by training a plurality of first training samples with first labels, the plurality of first training samples with the first labels are input into an initial route planning layer, a loss function is constructed through the first labels and results of the initial route planning layer, parameters of the initial route planning layer are iteratively updated based on the loss function, and when the loss function of the initial route planning layer satisfies a first preset iteration condition, training is completed, and the route planning layer is obtained, wherein the first preset iteration condition is that the loss function converges, or a count of iterations reaches a threshold; wherein a construction method of the first label of the first training sample includes: generating a plurality of sample candidate inspection routes based on sample location features of a sample start point and a sample end point through a heuristic pathfinding algorithm or a dynamic programming method; ranking the plurality of sample candidate inspection routes according to lengths of the inspection routes from short to long and a failure probability distribution of the LNG smart terminal; and selecting first N sample candidate inspection routes as the first label of the first training sample; and the failure probability distribution comprises a probability distribution of various failures in the LNG smart terminal, including a probability distribution of gas leakage, a failure to pre-cool, and an excessive pressure in a storage tank of the LNG smart terminal; and

task completion confirmation: after completing the safety inspection, obtaining inspection completion information sent by the LNG smart terminal and the optimal safety inspector, respectively, and confirming completion of the safety inspection based on the inspection completion information.

2 . The non-transitory computer-readable storage medium according to claim 1 , wherein the data collection executed by the one or more processors includes:

encrypting plaintext of the operation and maintenance data of the LNG smart terminal using an Advanced Encryption Standard (AES) algorithm by the LNG smart terminal; and

obtaining the personal data of each of different safety inspectors, the personal data including personal current location information and a working status.

3 . The non-transitory computer-readable storage medium according to claim 2 , wherein the encrypting plaintext of the operation and maintenance data of the LNG smart terminal using an Advanced Encryption Standard (AES) algorithm by the LNG smart terminal via the one or more processors includes:

generating in advance a key for the LNG smart terminal and distributing the key to the LNG smart terminal; and encrypting the operation and maintenance data using the key and obtaining an encrypted operation and maintenance data file uploaded to the LNG smart terminal by the LNG smart terminal.

4 . The non-transitory computer-readable storage medium according to claim 1 , wherein the data analysis executed by the one or more processors includes:

after receiving encrypted operation and maintenance data, decrypting the encrypted operation and maintenance data using a key corresponding to the LNG smart terminal, monitoring the operation and maintenance situation of the LNG smart terminal after obtaining plaintext of the operation and maintenance data, and judging whether the LNG smart terminal needs to be maintained by judging the operation and maintenance situation of the LNG smart terminal according to a preset maintenance condition; and

if the LNG smart terminal needs to be maintained, forming the inspection order reminder and the inspection instruction according to the operation and maintenance data corresponding to the LNG smart terminal.

5 . The non-transitory computer-readable storage medium according to claim 1 , wherein the task completion confirmation executed by the one or more processors includes:

after the optimal safety inspector receives the inspection order reminder and the inspection instruction, instructing the optimal safety inspector to go to a designated location according to a location of the LNG smart terminal to be maintained in the inspection order reminder; and after the optimal safety inspector arrives at the designated location, instructing the optimal safety inspector to photograph the LNG smart terminal to be maintained via a handheld terminal to obtain an inspection image, based on the inspection image and a receiving time of the inspection image, extracting information of the LNG smart terminal to be maintained from the inspection image, and matching the information with information of the LNG smart terminal to be maintained in the inspection order reminder, if the matching is successful, judging whether the receiving time of the inspection image is time-out according to a preset inspection arrival time, and if not, sending a task matching success reminder to the handheld terminal of the optimal safety inspector; and after the optimal safety inspector completes inspection and maintenance, instructing the optimal safety inspector to photograph a maintained LNG smart terminal to obtain an inspection completion image and upload the inspection completion image, obtain an inspection and maintenance operation result through the maintained LNG smart terminal, and send the inspection completion information, and confirming completion of the task after receiving the inspection completion image and the inspection completion information.

6 . The non-transitory computer-readable storage medium according to claim 1 , wherein the one or more processors are further caused to perform operations including:

obtaining a terminal location distribution of LNG smart terminals that currently need inspections; and

determining a preset inspection distance condition based on the terminal location distribution.

7 . The non-transitory computer-readable storage medium according to claim 6 , wherein the one or more processors are further caused to perform operations including:

determining an inspection geometric center of locations of the LNG smart terminals that currently need inspections based on the terminal location distribution, including: regarding n LNG smart terminals that currently need inspections as mass points, whose masses are m 1 , m 2 , m 3 , . . . , and m n , respectively, wherein a distribution of terminal locations is [(x 1 , y 1 ), (x 2 , y 2 ), (x 3 , y 3 ) . . . (x n , v n )], and calculating the inspection geometric center (x a , y a ) by equations:

x

a

=

∑

i

=

1

n

⁢

m

i

⁢

x

i

∑

i

=

1

n

⁢

m

i

,

y

a

=

∑

i

=

1

n

⁢

m

i

⁢

y

i

∑

i

=

1

n

⁢

m

i

,

wherein x i is a horizontal coordinate of a location of an i-th LNG smart terminal (1≤i≤n); y i is a vertical coordinate of the location of the i-th LNG smart terminal; m i is a mass of the i-th LNG smart terminal when it is regarded as a mass point; wherein when the n LNG smart terminals that currently need inspections are regarded as mass points, the masses of the n LNG smart terminals that currently need inspections are positively correlated with failure probability distributions of the LNG smart terminals;

determining an inspection radius based on the inspection geometric center, wherein the inspection radius refers to a distance between a safety inspector and the inspection geometric center; and

determining the preset inspection distance condition based on the inspection radius, including: determining the preset inspection distance condition to be that a distance between the safety inspector and the inspection geometric center is smaller than the inspection radius.

8 . The non-transitory computer-readable storage medium according to claim 1 , wherein the terminal features include a failure probability distribution.

9 . The non-transitory computer-readable storage medium according to claim 1 , wherein the time prediction layer is obtained by training the plurality of first training samples with the first labels, the plurality of first training samples with the first labels are input into an initial time prediction layer, a loss function is constructed through the first labels and results of the initial time prediction layer, and parameters of the initial time prediction layer are iteratively updated based on the loss function, and when the loss function of the initial time prediction layer satisfies a second preset iteration condition, training is completed, and the time prediction layer is obtained, wherein the second preset iteration condition is that the loss function converges, or a count of iterations reaches a threshold.

10 . The non-transitory computer-readable storage medium according to claim 1 , wherein the time integration layer calculates an average of candidate completion times of a plurality of candidate inspection routes, and outputs the average as the inspection completion time.

Assignments (2)

Priority Claims (1)

Continuity (1)

References Cited (13)