CN-121723117-B - Pipeline displacement stress prediction method, device, storage medium and equipment

CN121723117BCN 121723117 BCN121723117 BCN 121723117BCN-121723117-B

Abstract

The invention discloses a pipeline displacement stress prediction method, a device, a storage medium and equipment, relates to the technical field of pipeline state monitoring, and aims to realize high-precision prediction of pipeline displacement stress by taking full-pipe system coverage, low-cost deployment and real-time response into consideration. The method comprises the steps of obtaining historical temperature, historical pressure and actual measurement historical displacement and actual measurement historical stress of an actual measurement preset node on site, calculating simulation historical displacement and simulation historical stress of the preset node of the whole pipe system under the historical temperature and the historical pressure based on a pipe model, training an initial pipe displacement stress prediction model according to the historical temperature, the historical pressure, the simulation historical displacement, the simulation historical stress, the actual measurement historical displacement and the actual measurement historical stress, if training is not completed, executing a preset adjustment strategy until the pipe displacement stress prediction model is obtained, collecting the operating temperature and pressure of the pipe in real time, and predicting the displacement and the stress of the preset node of the whole pipe system according to the pipe displacement stress prediction model.

Inventors

LIU XIN
LIU WEIDONG
WANG TAN
WU SHAOJIE
ZHANG BINGQI
ZHAO JINGQI
SUN XU
SUN YUE
ZHOU HONGYANG
ZHANG YANMING
HOU XINGLONG
XIE JUNJUN

Assignees

大唐东北电力试验研究院有限公司
郑州大学

Dates

Publication Date: 20260512
Application Date: 20260226

Claims (9)

1. A method of predicting pipeline displacement stress, the method comprising: acquiring the historical temperature of pipeline operation, the historical pressure of pipeline operation, the actual measurement historical displacement of a field actual measurement preset node and the actual measurement historical stress of the field actual measurement preset node; Based on the constructed pipeline model, calculating the simulated historical displacement of the all-pipe system preset node and the simulated historical stress of the all-pipe system preset node according to the historical temperature and the historical pressure; Training an initial pipeline displacement stress prediction model by using the historical temperature, the historical pressure, the full-pipe system preset node, the simulated historical displacement and the simulated historical stress until the initial pipeline displacement stress prediction model converges to obtain a pipeline displacement stress prediction model to be verified, judging whether the pipeline displacement stress prediction model to be verified is trained according to the historical temperature, the historical pressure, the full-pipe system preset node, the simulated historical displacement, the simulated historical stress, the actual measurement historical displacement and the actual measurement historical stress, and if not, executing a preset adjustment strategy until the pipeline displacement stress prediction model to be verified is obtained; Acquiring the temperature of pipeline operation and the pressure of pipeline operation in real time, inputting time series data formed by the temperature and the pressure and the all-pipe system preset node into the pipeline displacement stress prediction model, and predicting to obtain the displacement of the all-pipe system preset node and the stress of the all-pipe system preset node; training an initial pipeline displacement stress prediction model by using the historical temperature, the historical pressure, the full pipe system preset node, the simulated historical displacement and the simulated historical stress until the initial pipeline displacement stress prediction model converges to obtain a pipeline displacement stress prediction model to be verified, wherein the training comprises the following steps of: Determining different historical time series data according to different continuous historical moments according to the historical temperature and the historical pressure, dividing training historical time series data from all the historical time series data, inputting the training historical time series data and the all-pipe system preset nodes into an initial pipeline displacement stress prediction model, and obtaining a first prediction historical displacement of the all-pipe system preset nodes and a first prediction historical stress of the all-pipe system preset nodes output by the initial pipeline displacement stress prediction model; Taking the simulated history displacement corresponding to the training history time series data as a training simulated history displacement, and calculating a first loss function value according to the training simulated history displacement and the first prediction history displacement of the same all-pipe system preset node, wherein the simulated history displacement at the last history time of the training history time series data is the simulated history displacement corresponding to the training history time series data; Taking the simulated historical stress corresponding to the training historical time series data as a training simulated historical stress, and calculating a second loss function value according to the training simulated historical stress and the first predicted historical stress of the same all-pipe system preset node, wherein the simulated historical stress at the last historical moment of the training historical time series data is the simulated historical stress corresponding to the training historical time series data; And if at least one first loss function value is greater than or equal to a preset displacement threshold value or at least one second loss function value is greater than or equal to a preset stress threshold value, determining that the initial pipeline displacement stress prediction model is not converged, and adjusting parameters of the initial pipeline displacement stress prediction model until the initial pipeline displacement stress prediction model is converged, so as to obtain the pipeline displacement stress prediction model to be verified.
2. The method according to claim 1, wherein the method further comprises: Acquiring a preset displacement safety interval, a displacement early warning interval and a displacement dangerous interval, if the displacement is in the displacement safety interval, displaying the displacement safety of the all-pipe preset node corresponding to the displacement, if the displacement is in the displacement early warning interval, displaying the displacement early warning of the all-pipe preset node corresponding to the displacement, and if the displacement is in the displacement dangerous interval, displaying the displacement warning of the all-pipe preset node corresponding to the displacement so as to perform displacement on-line monitoring; And acquiring a preset stress safety interval, a stress early warning interval and a stress dangerous interval, if the stress is in the stress safety interval, displaying the stress safety of the all-pipe preset node corresponding to the stress, if the stress is in the stress early warning interval, displaying the stress early warning of the all-pipe preset node corresponding to the stress, and if the stress is in the stress dangerous interval, displaying the stress warning of the all-pipe preset node corresponding to the stress so as to perform stress on-line monitoring.
3. The method of claim 1, wherein the obtaining the historical temperature of the pipe operation, the historical pressure of the pipe operation, the measured historical displacement of the field measured preset node, and the measured historical stress of the field measured preset node comprises: Collecting historical temperature of pipeline operation of a pipeline reservation interface based on a temperature sensor, and collecting historical pressure of pipeline operation of the pipeline reservation interface based on a pressure sensor; Measuring actual measurement history displacement of an actual measurement preset node on site based on a binocular stereoscopic vision displacement measuring device; based on the vibrating wire type surface strain gauge arranged at the on-site actually-measured preset node, collecting the vibration frequency of the pipeline, and carrying out numerical reading on a data acquisition instrument in the vibrating wire type surface strain gauge to obtain the surface temperature of the pipeline at the position of the vibrating wire type surface strain gauge; obtaining preset calculation parameters, and calculating actual measurement historical stress of the on-site actual measurement preset node according to the pipeline vibration frequency, the pipeline surface temperature and the preset calculation parameters, wherein the preset calculation parameters at least comprise a reference frequency, a reference temperature, a material elastic modulus, a linear expansion coefficient, a temperature correction coefficient and a strain gauge sensitivity coefficient.
4. The method of claim 1, wherein the determining whether the pipe displacement stress prediction model to be validated is trained based on the historical temperature, the historical pressure, the full tubing preset node, the simulated historical displacement, the simulated historical stress, the measured historical displacement, and the measured historical stress comprises: Dividing verification historical time series data from all the historical time series data, inputting the verification historical time series data and the all-pipe system preset nodes into the pipeline displacement stress prediction model to be verified, obtaining the pipeline displacement stress prediction model to be verified, and outputting second prediction historical displacement of the all-pipe system preset nodes and second prediction historical stress of the all-pipe system preset nodes by the pipeline displacement stress prediction model to be verified; Taking the simulated historical displacement corresponding to the verification historical time series data as a verification simulated historical displacement, calculating a first displacement deviation rate according to the verification simulated historical displacement and the second prediction historical displacement of the same all-pipe system preset node, taking the simulated historical stress corresponding to the verification historical time series data as a verification simulated historical stress, calculating a first stress deviation rate according to the verification simulated historical stress and the second prediction historical stress of the same all-pipe system preset node, calculating a second displacement deviation rate according to the actual measurement historical displacement and the second prediction historical displacement of the same field actual measurement preset node, and calculating a second stress deviation rate according to the actual measurement historical stress and the second prediction historical stress of the same field actual measurement preset node, wherein the simulated historical displacement at the last historical moment of the verification historical time series data is the simulated historical displacement corresponding to the verification historical time series data, and the simulated historical stress at the last historical moment of the verification historical time series data is the simulated historical stress corresponding to the verification historical time series data; Judging whether the pipeline displacement stress prediction model to be verified is trained according to the first displacement deviation rate, a first preset displacement threshold, the first stress deviation rate, a first preset stress threshold, the second displacement deviation rate, a second preset displacement threshold, the second stress deviation rate and a second preset stress threshold.
5. The method of claim 4, wherein if not, executing a preset adjustment strategy until a trained pipeline displacement stress prediction model is obtained, comprising: If at least one first displacement deviation rate is greater than or equal to the first preset displacement threshold value, or at least one first stress deviation rate is greater than or equal to the first preset stress threshold value, or at least one second displacement deviation rate is greater than or equal to the second preset displacement threshold value, or at least one second stress deviation rate is greater than or equal to the second preset stress threshold value, determining that the pipeline displacement stress prediction model to be verified does not complete training; and determining an added training sample as a preset adjustment strategy, acquiring a new added sample set, and retraining the pipeline displacement stress prediction model to be verified by using the new added sample set until the retrained pipeline displacement stress prediction model to be verified completes training, so as to obtain a pipeline displacement stress prediction model after training is completed.
6. The method according to claim 4, wherein the method further comprises: if all the first displacement deviation rates are smaller than the first preset displacement threshold value, all the first stress deviation rates are smaller than the first preset stress threshold value, all the second displacement deviation rates are smaller than the second preset displacement threshold value, all the second stress deviation rates are smaller than the second preset stress threshold value, the pipeline displacement stress prediction model to be verified is trained, and the trained pipeline displacement stress prediction model is obtained.
7. A pipeline displacement stress prediction apparatus, the apparatus comprising: The acquisition module is used for acquiring the historical temperature of pipeline operation, the historical pressure of pipeline operation, the actual measurement historical displacement of the on-site actual measurement preset node and the actual measurement historical stress of the on-site actual measurement preset node; the calculation module is used for calculating the simulation history displacement of the all-pipe system preset node and the simulation history stress of the all-pipe system preset node according to the history temperature and the history pressure based on the constructed pipeline model; The first training module is configured to train an initial pipeline displacement stress prediction model by using the historical temperature, the historical pressure, the full-pipe system preset node, the simulated historical displacement and the simulated historical stress until the initial pipeline displacement stress prediction model converges to obtain a pipeline displacement stress prediction model to be verified, judge whether the pipeline displacement stress prediction model to be verified is trained according to the historical temperature, the historical pressure, the full-pipe system preset node, the simulated historical displacement, the simulated historical stress, the actual measurement historical displacement and the actual measurement historical stress, and if not, execute a preset adjustment strategy until the pipeline displacement stress prediction model to be verified is obtained; the prediction module is used for collecting the temperature of pipeline operation, the pressure of pipeline operation, time series data formed by the temperature and the pressure, and the all-pipe system preset node in real time, inputting the pipeline displacement stress prediction model, and predicting and obtaining the displacement of the all-pipe system preset node and the stress of the all-pipe system preset node; The first training module is configured to determine different historical time series data according to the historical temperature and the historical pressure and different continuous historical moments, divide training historical time series data from all the historical time series data, input an initial pipeline displacement stress prediction model into the training historical time series data and the all-pipe system preset node, and obtain the initial pipeline displacement stress prediction model to output a first prediction historical displacement of the all-pipe system preset node and a first prediction historical stress of the all-pipe system preset node; the method comprises the steps of taking the simulated historical displacement corresponding to the training historical time series data as training simulated historical displacement, calculating a first loss function value according to the training simulated historical displacement and the first prediction historical displacement of the same all-pipe system preset node, wherein the simulated historical displacement at the last historical moment of the training historical time series data is the simulated historical displacement corresponding to the training historical time series data, taking the simulated historical stress corresponding to the training historical time series data as training simulated historical stress, calculating a second loss function value according to the training simulated historical stress and the first prediction historical stress of the same all-pipe system preset node, wherein the simulated historical stress at the last historical moment of the training historical time series data is the simulated historical stress corresponding to the training historical time series data, determining that the initial pipeline displacement stress prediction model does not converge if at least one of the first loss function value is larger than or equal to a preset displacement threshold or at least one of the second loss function value is larger than or equal to a preset stress threshold, and adjusting parameters of the initial pipeline displacement stress prediction model until the initial pipeline displacement stress prediction model converges to obtain a pipeline displacement stress prediction model to be verified.
8. A storage medium having stored thereon a computer program, wherein the program when executed by a processor implements the pipe displacement stress prediction method of any one of claims 1 to 6.
9. A computer device comprising a memory, a processor and a computer program stored on the storage medium and running on the processor, characterized in that the processor implements the pipe displacement stress prediction method of any one of claims 1 to 6 when executing the program.

Description

Pipeline displacement stress prediction method, device, storage medium and equipment Technical Field The present invention relates to the field of pipeline state monitoring technologies, and in particular, to a method, an apparatus, a storage medium, and a device for predicting pipeline displacement stress. Background In a thermal power generation system, four large pipelines are used as core components for energy transmission and are used in severe environments of high temperature, high pressure and continuous vibration for a long time. With the great improvement of the deep peak shaving capacity requirement of the thermal power generating unit, the thermal load and the mechanical load born by the pipeline frequently fluctuate, and the load condition is more complex. If the displacement of the pipeline is abnormal and the stress is accumulated to exceed the allowable value, the pipeline is easy to crack, break and even impact damage of equipment, and the safe and stable operation of a power plant is seriously threatened. The existing prediction method for the pipeline displacement stress has the obvious defects that 1. The actual measurement method is high in cost and has large limitation that a large number of points of a pipeline are provided with sensors, the cost is high due to mass sensors and wiring cost, a small number of points are provided with sensors which are difficult to cover the whole pipeline (such as complex elbows and header areas) and dead zones exist, 2. The existing prediction model is narrow in coverage range, because sensor data which cover the whole pipeline cannot be obtained, training samples of the existing prediction model are few, and the method is limited to sensor data of sensor arrangement positions such as single pipeline elbows or local points, and is difficult to accurately predict the pipeline displacement stress of multiple nodes of the whole pipeline. Disclosure of Invention In view of the above, the invention provides a method, a device, a storage medium and equipment for predicting pipeline displacement stress, which take full pipe system coverage, low-cost deployment and real-time response into consideration, and realize high-precision prediction of pipeline displacement stress. According to one aspect of the present invention, there is provided a method of pipeline displacement stress prediction, the method comprising: acquiring the historical temperature of pipeline operation, the historical pressure of pipeline operation, the actual measurement historical displacement of a field actual measurement preset node and the actual measurement historical stress of the field actual measurement preset node; Based on the constructed pipeline model, calculating the simulated historical displacement of the all-pipe system preset node and the simulated historical stress of the all-pipe system preset node according to the historical temperature and the historical pressure; Training an initial pipeline displacement stress prediction model by using the historical temperature, the historical pressure, the full-pipe system preset node, the simulated historical displacement and the simulated historical stress until the initial pipeline displacement stress prediction model converges to obtain a pipeline displacement stress prediction model to be verified, judging whether the pipeline displacement stress prediction model to be verified is trained according to the historical temperature, the historical pressure, the full-pipe system preset node, the simulated historical displacement, the simulated historical stress, the actual measurement historical displacement and the actual measurement historical stress, and if not, executing a preset adjustment strategy until the pipeline displacement stress prediction model to be verified is obtained; And acquiring the temperature of pipeline operation and the pressure of pipeline operation in real time, inputting time series data formed by the temperature and the pressure and the all-pipe system preset node into the pipeline displacement stress prediction model, and predicting to obtain the displacement of the all-pipe system preset node and the stress of the all-pipe system preset node. Preferably, the method further comprises: Acquiring a preset displacement safety interval, a displacement early warning interval and a displacement dangerous interval, if the displacement is in the displacement safety interval, displaying the displacement safety of the all-pipe preset node corresponding to the displacement, if the displacement is in the displacement early warning interval, displaying the displacement early warning of the all-pipe preset node corresponding to the displacement, and if the displacement is in the displacement dangerous interval, displaying the displacement warning of the all-pipe preset node corresponding to the displacement so as to perform displacement on-line monitoring; And acquiring a preset stress safety interval, a stress earl