CN117057239B - Optical fiber temperature measuring point positioning system based on laser - Google Patents
Optical fiber temperature measuring point positioning system based on laser Download PDFInfo
- Publication number
- CN117057239B CN117057239B CN202311023694.6A CN202311023694A CN117057239B CN 117057239 B CN117057239 B CN 117057239B CN 202311023694 A CN202311023694 A CN 202311023694A CN 117057239 B CN117057239 B CN 117057239B
- Authority
- CN
- China
- Prior art keywords
- sequence
- temperature
- temperature measurement
- trained
- temperature distribution
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 239000013307 optical fiber Substances 0.000 title claims abstract description 38
- 238000009826 distribution Methods 0.000 claims abstract description 144
- 238000009529 body temperature measurement Methods 0.000 claims abstract description 141
- 238000012549 training Methods 0.000 claims abstract description 41
- 230000005540 biological transmission Effects 0.000 claims abstract description 16
- 238000004590 computer program Methods 0.000 claims abstract description 10
- 230000006870 function Effects 0.000 claims description 52
- 238000005070 sampling Methods 0.000 claims description 27
- 238000000034 method Methods 0.000 claims description 20
- 230000004927 fusion Effects 0.000 claims description 17
- 238000000605 extraction Methods 0.000 claims description 12
- 230000014759 maintenance of location Effects 0.000 claims description 7
- 238000013507 mapping Methods 0.000 claims description 5
- 239000000835 fiber Substances 0.000 claims description 2
- 238000012544 monitoring process Methods 0.000 description 4
- 238000003062 neural network model Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 3
- 238000012163 sequencing technique Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 238000002372 labelling Methods 0.000 description 2
- 230000000717 retained effect Effects 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 230000003416 augmentation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000010287 polarization Effects 0.000 description 1
- 230000008054 signal transmission Effects 0.000 description 1
- 238000004861 thermometry Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01K—MEASURING TEMPERATURE; MEASURING QUANTITY OF HEAT; THERMALLY-SENSITIVE ELEMENTS NOT OTHERWISE PROVIDED FOR
- G01K11/00—Measuring temperature based upon physical or chemical changes not covered by groups G01K3/00, G01K5/00, G01K7/00 or G01K9/00
- G01K11/32—Measuring temperature based upon physical or chemical changes not covered by groups G01K3/00, G01K5/00, G01K7/00 or G01K9/00 using changes in transmittance, scattering or luminescence in optical fibres
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
- G06N3/0455—Auto-encoder networks; Encoder-decoder networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/0895—Weakly supervised learning, e.g. semi-supervised or self-supervised learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/08—Thermal analysis or thermal optimisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2123/00—Data types
- G06F2123/02—Data types in the time domain, e.g. time-series data
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Measuring Temperature Or Quantity Of Heat (AREA)
Abstract
The invention provides a laser-based optical fiber temperature measuring point position recommending system, which comprises: the system comprises a laser, a transmission optical fiber, a database, a processor and a memory storing a computer program, wherein the laser generates laser, the laser is transmitted through the transmission optical fiber, the transmission optical fiber is deployed in a temperature measuring scene, and when the computer program is executed by the processor, the following steps are realized: obtaining target temperature distribution according to N scene temperature distribution, obtaining an adjusted temperature measurement point position sequence according to the initial temperature measurement point position sequence and the target temperature distribution, using a position reconstruction model, training the position reconstruction model according to the adjusted temperature measurement point position sequence, the initial temperature measurement point position sequence and the target temperature distribution, and obtaining a recommended temperature measurement point position sequence according to the trained position reconstruction model, the initial temperature measurement point position sequence and the target temperature distribution.
Description
Technical Field
The invention relates to the field of optical fiber sensing, in particular to an optical fiber temperature measuring point positioning system based on a laser.
Background
When a conventional broadband light source is applied to a transmission optical fiber, the signal-to-noise ratio of transmission and the resolution of an optical fiber sensor are low, so that the conventional method adopts an optical fiber laser as a light source, thereby effectively improving the signal-to-noise ratio of optical signal transmission and the resolution of the optical fiber sensor, and for example, the optical fiber laser can adopt an optical fiber DFB laser, a DBR laser, a ring-shaped resonant cavity laser and the like.
The optical fiber temperature measurement technology can be used for various application scenes such as automatic fire alarm, monitoring point positioning, line self-detection and positioning, and the like, and can be generally divided into point type temperature measurement, quasi-distributed temperature measurement and complete distributed temperature measurement, wherein the point type temperature measurement is widely applied to the optical fiber temperature measurement technology due to lower cost.
In order to realize the intellectualization of the deployment of the temperature measuring points, a deep neural network model can be adopted to predict the deployment position of the temperature measuring points of the optical fibers so as to recommend the temperature measuring points to operation staff as references during the deployment of the temperature measuring points of the optical fibers.
However, the deep neural network model usually needs to use the manually marked temperature measurement points for the temperature measurement scene as the label data during training, so that the deployment information of the manually deployed temperature measurement scene is usually directly applied to the augmentation model data set, so that further temperature measurement point position recommendation is difficult to be performed on the temperature measurement scene where the temperature measurement points are deployed, and the model is difficult to adapt to the temperature measurement scene with continuous change because the manually marked temperature measurement scene is usually processed in a shorter period of time, so that how to improve the accuracy of optical fiber temperature measurement point recommendation through the deep neural network model on the premise of not increasing the marking cost becomes an urgent problem to be solved.
Disclosure of Invention
Aiming at the technical problems, the invention adopts the following technical scheme: a laser-based fiber optic temperature measurement point positioning system, comprising: the system comprises a laser, a transmission optical fiber, a database, a processor and a memory storing a computer program, wherein the laser generates laser light, the laser light is transmitted through the transmission optical fiber, and when the computer program is executed by the processor, the following steps are realized:
s1, obtaining target temperature distribution according to scene temperature distribution corresponding to N sequential sampling frames;
s2, according to a preset initial temperature measurement point position sequence and target temperature distribution, using a position reconstruction model to obtain an adjusted temperature measurement point position sequence;
s3, training the position reconstruction model according to the temperature measurement point position sequence, the initial temperature measurement point position sequence and the target temperature distribution to obtain a trained position reconstruction model;
s4, obtaining a recommended temperature measurement point position sequence according to the trained position reconstruction model, the initial temperature measurement point position sequence and the target temperature distribution.
The invention has at least the following beneficial effects:
the invention provides an optical fiber temperature measuring point positioning system based on a laser, which comprises the following components: the system comprises a laser, a transmission optical fiber, a database, a processor and a memory storing a computer program, wherein the laser generates laser light, the laser light is transmitted through the transmission optical fiber, and when the computer program is executed by the processor, the following steps are realized: according to the scene temperature distribution corresponding to each of N sequential continuous sampling frames, target temperature distribution is obtained, according to a preset initial temperature measurement point position sequence and target temperature distribution, a position reconstruction model is used to obtain an adjustment temperature measurement point position sequence, according to the adjustment temperature measurement point position sequence, the initial temperature measurement point position sequence and the target temperature distribution, the position reconstruction model is trained to obtain a trained position reconstruction model, according to the trained position reconstruction model, the initial temperature measurement point position sequence and the target temperature distribution, a recommended temperature measurement point position sequence is obtained, scene temperature distribution obtained through a plurality of sequential continuous sampling frames is combined to obtain target temperature distribution, so that the target temperature distribution can represent temperature information in a long period, temperature measurement point position reconstruction is carried out based on the target temperature distribution, the adjustment temperature measurement point position sequence which is more suitable for a temperature measurement scene can be obtained, and therefore the accuracy of temperature measurement point position recommendation is improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flowchart of a computer program executed by a laser-based optical fiber temperature measurement point positioning system according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to fall within the scope of the invention.
The embodiment of the invention provides an optical fiber temperature measuring point positioning system based on a laser, which comprises the following steps: the laser device generates laser light, the laser light is transmitted through the transmission optical fiber, in the embodiment, the laser device selects an optical fiber DFB laser device to generate linear polarization to obtain single-frequency laser light by using distributed feedback under the whole optical fiber structure, stable and efficient single-frequency laser output is achieved, the single-frequency laser light is transmitted through the transmission optical fiber, and in order to ensure stability in the laser transmission process, a temperature measuring point needs to be determined in a laser transmission scene, namely a real temperature measuring scene, so that the temperature condition during laser transmission is accurately monitored under the condition of lower temperature measuring cost.
When the computer program is executed by a processor, the following steps are implemented:
s1, obtaining target temperature distribution according to scene temperature distribution corresponding to N time sequence continuous sampling frames.
Wherein N is an integer greater than one, the sequential succession may refer to each sampling frame being sampled according to a time sequence, the sampling frames may refer to acquisition time points for performing scene temperature distribution, the scene temperature distribution may be used for characterizing a temperature distribution condition in a temperature measurement scene at a time corresponding to the sampling frames, and the target temperature distribution may be used for characterizing a temperature distribution condition in the temperature measurement scene at a time period composed of N sampling frames.
Specifically, the temperature measurement scene comprises a plurality of scene space points, the scene temperature distribution can be represented by a three-dimensional tensor, the dimension corresponding to a single element in the three-dimensional tensor can be used for representing three-dimensional coordinate information of the corresponding scene space point, the element value corresponding to the single element in the three-dimensional tensor can be used for representing the scene temperature value of the corresponding scene space point, and the target temperature distribution can be obtained by carrying out superposition processing on the scene temperature distribution corresponding to each of N sampling frames.
The time interval between each sampling frame may be a preset fixed value, which is set to 1 hour in this embodiment, and the practitioner may adjust the fixed value according to the actual situation, for example, to 2 hours, 4 hours, 1 day, or the like.
In this embodiment, the scene temperature distribution may be acquired through a preset temperature measurement point, that is, a temperature measurement point at the initial temperature measurement point position, and after the preset temperature measurement point acquires the temperature data, the scene temperature distribution may be further acquired through interpolation processing.
In one embodiment, the practitioner may also obtain the scene temperature distribution by modeling the temperature distribution of each heat source in the thermometry scene and then performing the superposition process.
In a specific embodiment, in step S1, the method further includes the following steps:
s11, initializing iteration times i to be 1 in N sampling frames with continuous time sequences, and taking scene temperature distribution corresponding to the ith sampling frame in the time sequences as a history frame;
s12, taking the scene temperature distribution corresponding to the (i+1) th sampling frame in the time sequence as a current frame;
s13, multiplying the history frame by a preset first superposition coefficient to obtain history retention information;
s14, multiplying the current frame by a preset second superposition coefficient to obtain current reserved information;
s15, adding the historical reserved information and the current reserved information to obtain comprehensive reserved information;
s16, taking the comprehensive reservation information as a history frame, adding 1 to i, and returning to execute the steps S12 to S16 until the preset condition is met, stopping iteration, and taking the comprehensive reservation information obtained when the iteration is stopped as target temperature distribution.
The iteration number can be used for representing the execution number of the iteration process, an initial value of the iteration number is set to be 1, N sampling frames are sequenced according to the acquired time sequence, each sampling frame is provided with a corresponding sequencing position after sequencing, the sampling frame with the sequencing position at the ith position is initially used as a historical frame, the historical frame can be used for representing temperature distribution superposition information obtained by previous superposition in the iteration superposition process, the current frame can refer to temperature distribution information needing to be superposed under the current iteration number, the first superposition coefficient can refer to weight of the historical frame, and the second superposition coefficient can refer to weight of the current frame.
The history retaining information may refer to temperature distribution stacking information obtained by previous stacking in the iterative stacking process, the information retained after being processed by the first stacking coefficient, the current retaining information may refer to temperature distribution information required to be stacked under the current iteration number, the information retained after being processed by the second stacking coefficient, and the comprehensive retaining information may refer to a result after the current frame is stacked to the history frame.
Steps S12 to S16 may refer to the above iterative process, and a preset condition may be used to determine whether to stop the iteration.
Specifically, in this embodiment, the sum of the first stacking coefficient and the second stacking coefficient is 1, where the first stacking coefficient may be set to 0.05, and correspondingly, the second stacking coefficient is set to 0.95, so that the information of the history frame is kept in a smaller portion, and the information of the current frame is kept in a larger portion, so that by stacking the integrated history frame information and the current frame information, it can be known that, in the stacking manner, if a position is basically maintained at a higher temperature value in a plurality of sampling frames, the temperature value in the position is also higher in the integrated retention information, and if a position is only at the higher temperature value in a small number of sampling frames, the temperature value in the position is relatively lower in the integrated retention information along with the increase of the number of iterations, so that the continuous high-temperature position information under the temperature measurement scene is effectively extracted, and it is required to be explained that an practitioner can adjust the values of the first stacking coefficient and the second stacking coefficient according to the actual situation, so that the value range of the first stacking coefficient is recommended to be within (0,0.1 ].
In this embodiment, the preset condition may be stopping iteration when i is equal to N, and when iteration is stopped, all scene distributions corresponding to N sampling frames are subjected to superposition processing.
In an embodiment, the preset condition may be adjusted to stop iteration when i is equal to N, which is an integer greater than zero and less than N.
In this embodiment, the first superposition coefficient and the second superposition coefficient are combined to perform iterative superposition of the scene temperature distribution, so that the obtained target temperature distribution can effectively represent the information of the continuous high-temperature position, thereby effectively avoiding noise interference, further providing more accurate temperature distribution information for the subsequent position reconstruction model, and improving the accuracy of the position reconstruction process.
According to the steps of obtaining the target temperature distribution according to the scene temperature distribution corresponding to the N sampling frames with continuous time sequences, the scene temperature distribution of the sampling frames is combined to obtain the target temperature distribution, so that the position information which basically maintains high temperature under the sampling frames is determined, namely, more reliable positions which need to be subjected to temperature monitoring can be provided for position recommendation, and the accuracy of position recommendation is improved.
S2, according to a preset initial temperature measurement point position sequence and target temperature distribution, a position reconstruction model is used to obtain an adjusted temperature measurement point position sequence.
The initial temperature measurement point position sequence may include position coordinates corresponding to a plurality of initial temperature measurement points, and the position reconstruction model may be used to obtain an adjusted temperature measurement point position sequence after adjustment corresponding to the initial temperature measurement point sequence according to the initial temperature measurement point position sequence and the target temperature distribution.
Specifically, the initial temperature measurement point position sequence may be actually deployed in a temperature measurement scene, or may be manually marked, which is not limited herein.
In a specific embodiment, the position reconstruction model includes a position sequence encoder, a temperature distribution encoder, and a sequence reconstruction decoder;
the step S2 further includes the following steps:
s21, inputting the initial temperature measuring point position sequence into a position sequence encoder for feature extraction to obtain position sequence features;
s22, inputting the target temperature distribution into a temperature distribution encoder for feature extraction to obtain temperature distribution features;
s23, fusing the position sequence features and the temperature distribution features to obtain fusion features;
s24, inputting the fusion characteristics into a sequence reconstruction decoder to perform characteristic reconstruction, and obtaining an adjustment temperature measurement point position sequence.
The position sequence encoder can be used for extracting characteristic information of an initial temperature measurement point position sequence, the temperature distribution encoder can be used for extracting characteristic information of target temperature distribution, the fusion characteristic can comprise temperature distribution characteristics and position sequence characteristics, and the sequence reconstruction decoder can be used for reconstructing and obtaining an adjustment temperature measurement point position sequence.
Specifically, in this embodiment, the output sizes of the temperature distribution encoder and the position sequence encoder are consistent, so that the position sequence feature and the temperature distribution feature are fused, the feature fusion can be implemented in a feature splicing manner, the sizes of the position sequences of the temperature measurement points are adjusted to be consistent with the sizes of the position sequences of the initial temperature measurement points, the target sizes are all target sizes, the target sizes can be determined through the preset maximum number of the acceptable temperature measurement points, and if the number of the initial temperature measurement points is smaller than the maximum number of the acceptable temperature measurement points, the coordinates of the initial temperature measurement points form the initial temperature measurement point sequence meeting the target sizes in a zero filling manner.
In a specific embodiment, the position reconstruction model is pre-trained;
the pre-training process of the position reconstruction model comprises the following steps:
inputting a sample temperature measuring point position sequence into a position sequence encoder for feature extraction to obtain sample sequence features, and inputting zero temperature distribution into a temperature distribution encoder for feature extraction to obtain zero distribution features;
fusing the sample sequence features and the zero distribution features to obtain sample fusion features, and inputting the sample fusion features into a sequence reconstruction decoder to reconstruct the features to obtain a sample adjustment position sequence;
pre-training a position sequence encoder, a temperature distribution encoder and a sequence reconstruction decoder according to the position sequence of the sample temperature measuring point and the sample adjustment position sequence to obtain a pre-trained position sequence encoder, a pre-trained temperature distribution encoder and a pre-trained sequence reconstruction decoder;
a position reconstruction model is formed from the pre-trained position sequence encoder, the pre-trained temperature distribution encoder, and the pre-trained sequence reconstruction decoder.
The sample sequence features can be used for representing feature information of a sample temperature measurement point sequence, the zero temperature distribution can refer to all zero scene temperature distribution, the zero distribution features can be used for representing feature information of the zero temperature distribution, and it can be expected that element values in the zero distribution features are all zero, that is, a temperature distribution encoder does not substantially participate in a pre-training process, and only the input size and the output size of a position sequence encoder, a temperature distribution encoder and a sequence reconstruction decoder in a position reconstruction model are ensured to be in accordance with the sizes in a normal reasoning process.
The sample fusion feature may include a sample sequence feature and a zero distribution feature, and the sample adjustment position sequence may refer to a result of adjusting the sample temperature measurement point sequence according to the sample fusion feature.
Specifically, since the temperature distribution encoder does not substantially participate in the pre-training process, the pre-training process is essentially a supervision of training by the reconstruction task, so that the position sequence encoder can extract the feature information that accurately characterizes the sample temperature measurement point sequence, and the sequence reconstruction decoder can accurately reconstruct according to the feature information that characterizes the sample temperature measurement point sequence.
In this embodiment, the calculation of the mean square error loss function is performed according to the sample temperature measurement point position sequence and the sample adjustment position sequence to obtain a pre-training loss, and according to the pre-training loss, the position sequence encoder, the temperature distribution encoder and the sequence reconstruction decoder are pre-trained by using a gradient descent method, and only the position sequence encoder and the sequence reconstruction decoder are actually pre-trained until the pre-training loss converges, so as to obtain a pre-trained position sequence encoder, a pre-trained temperature distribution encoder and a pre-trained sequence reconstruction decoder, and a position reconstruction model is formed by the pre-trained position sequence encoder, the pre-trained temperature distribution encoder and the pre-trained sequence reconstruction decoder.
In this embodiment, the pre-training of the position reconstruction model is performed through the reconstruction task, so that the position sequence encoder in the position reconstruction model can extract effective features, and the sequence reconstruction decoder can accurately reconstruct the position sequence according to the extracted features, thereby providing a basis for subsequent retraining and avoiding the situation that the subsequent retraining process is difficult to converge due to gradient loss.
And the step of obtaining the position sequence of the temperature measuring point by using the position reconstruction model according to the preset initial temperature measuring point position sequence and target temperature distribution, and providing basic information for training of a subsequent position reconstruction model, thereby improving the position reconstruction accuracy of the trained position reconstruction model.
And S3, training the position reconstruction model according to the temperature measurement point position sequence, the initial temperature measurement point position sequence and the target temperature distribution to obtain a trained position reconstruction model.
The trained position reconstruction model can adjust the initial temperature measurement point position sequence, and not reconstruct the initial temperature measurement point position sequence only.
In a specific embodiment, in step S3, the method further includes the following steps:
s31, acquiring the number of the temperature-measuring point adjustment in the temperature-measuring point position adjustment sequence, and obtaining a first loss function according to the number of the temperature-measuring point adjustment;
s32, obtaining a second loss function according to the temperature measuring point position sequence and the initial temperature measuring point position sequence;
s33, according to the distribution information of the temperature measuring points in the temperature measuring point position adjusting sequence, a third loss function is obtained;
s34, performing weighted addition on the first loss function, the second loss function and the third loss function to obtain a position loss function;
and S37, training the position reconstruction model according to the position loss function to obtain a trained position reconstruction model.
The adjusting the number of the temperature measuring points may refer to adjusting the number of the temperature measuring points in the temperature measuring point position sequence, and the first loss function may be used to monitor and adjust the number of the temperature measuring points so as to ensure that the deployment cost of the temperature measuring points is not too high.
The second loss function can be used for supervising and adjusting the difference between the temperature measuring point position sequence and the initial temperature measuring point position sequence so as to supervise and adjust the temperature measuring point position sequence to be similar to the initial temperature measuring point position sequence as much as possible, reduce the cost of temperature measuring point adjustment, and simultaneously can keep the deployment information of the initial temperature measuring point position sequence.
The third loss function can be used for monitoring the distribution situation of the adjustment temperature measurement points in the adjustment temperature measurement point position sequence so as to ensure that the distribution of the adjustment temperature measurement points is dispersed as much as possible, and avoid the adjustment temperature measurement points being too concentrated, namely a plurality of adjustment temperature measurement points monitor the same position at the same time, so that the adjustment temperature measurement points can be distributed in the whole temperature measurement scene as much as possible, and the position loss function can be used for monitoring the training of the position reconstruction model.
Specifically, the first loss function may be an exponential mapping, in this embodiment, to adjust the number of temperature measurement pointsThe first loss function is determined by multiplying the square of (a) by a preset coefficient, which can be determined by the number of expected temperature points set by the practitioner, the number of expected temperature points being denoted as a, the number of adjusted temperature points being denoted as b, the first loss function can be expressed as (1/b) 2 )*a 2 。
The second loss function may be obtained by calculating a mean square error loss between the sequence of adjusted temperature measurement points and the sequence of initial temperature measurement points.
The third loss function may be obtained by calculating a mean square error loss between the volume of the temperature measurement scene and the volume of the smallest circumscribed moment body of the adjusted temperature measurement point in the sequence of the adjusted temperature measurement points.
In this embodiment, weights corresponding to the weighted addition of the first loss function, the second loss function, and the third loss function may be set to 1, so as to obtain the position loss function.
In this embodiment, the training of the position reconstruction model is jointly supervised through a plurality of loss functions, and meanwhile, the quantity and distribution conditions of the adjustment temperature measurement points are supervised, so that the situation that the quantity of the adjustment temperature measurement points in the adjustment temperature measurement point position sequence output by the trained position reconstruction model is too small or the distribution of the adjustment temperature measurement points is too concentrated is avoided, and meanwhile, the deployment information of the initial temperature measurement points in the initial temperature measurement point position sequence can be kept, so that the training effect of the position reconstruction model is improved from a plurality of dimensions, the initial temperature measurement point position sequence can be adjusted by the trained position reconstruction model, and the adjustment temperature measurement point position sequence more suitable for a temperature measurement scene is obtained.
In a specific embodiment, in step S34, the following steps are further included:
s341, mapping to obtain a first weight according to the number of the temperature measuring points;
s342, mapping to obtain a second weight according to training rounds of training the position reconstruction model;
s343, obtaining a third weight according to the preset total weight, the first weight and the second weight;
s344, multiplying the first loss function by the first weight to obtain a first multiplication result, multiplying the second loss function by the second weight to obtain a second multiplication result, multiplying the third loss function by the third weight to obtain a third multiplication result, and taking the sum of the first multiplication result, the second multiplication result and the third multiplication result as a position loss function.
The first weight may refer to a weight when the first loss function is weighted, the second weight may refer to a weight when the second loss function is weighted, and the third weight may refer to a weight when the third loss function is weighted.
Specifically, the number of the expected temperature measurement points is denoted as a, the number of the adjusted temperature measurement points is denoted as b, the first weight may be denoted as 1/[ e (a-b) +1], if the number of the adjusted temperature measurement points is greater than the number of the expected temperature measurement points, the first weight is greater than 0.5, if the number of the adjusted temperature measurement points is equal to the number of the expected temperature measurement points, the first weight is equal to 0.5, if the number of the adjusted temperature measurement points is less than the number of the expected temperature measurement points, the first weight is less than 0.5, and it is required to be noted that the first weight is greater than 0 and less than 1.
The training round is denoted as c, and the second weight may be denoted as e (-c), that is, as the training round increases, the reconstructed model may be considered to mainly refer to the deployment information of the initial temperature measurement points when the training round is smaller, and to be mainly supervised according to the number and distribution of the adjustment temperature measurement points when the training round is larger, so that the second weight may be gradually reduced as the training round increases.
The preset total weight may be set to 2, and the third weight may be obtained by subtracting the first weight and the second weight from the preset total weight.
In this embodiment, by setting different weights for different loss functions, attention information in the training process is effectively controlled, so as to ensure rapid convergence of the training process, and improve training efficiency of the training process of the position reconstruction model.
In a specific embodiment, in step S3, the method further includes the following steps:
s35, acquiring a plurality of attention points in target temperature distribution;
s36, obtaining a distribution loss function according to all the attention points and the temperature measurement point sequence;
accordingly, in step S37, it includes:
and training the position reconstruction model according to the position loss function and the distribution loss function to obtain a trained position reconstruction model.
The target temperature distribution is a target temperature distribution, and the target temperature distribution is a target temperature distribution, wherein the target temperature distribution is a target temperature distribution, and the target temperature distribution is a target temperature distribution.
Specifically, for any one point of interest, determining an adjusted temperature measuring point position closest to the point of interest from all the adjusted temperature measuring point positions in the adjusted temperature measuring point position sequence, and calculating a distance difference value between the point of interest and the whole temperature measuring point position closest to the point of interest to obtain a distance difference value of the corresponding point of interest;
and adding the distance difference values of all the attention points, taking the added result as a distribution loss function, and training the position reconstruction model by combining the position loss function and the distribution loss function, so that the position reconstruction model ensures that the distribution of the temperature measurement points is discrete as far as possible on one hand and the temperature measurement points and the attention points are close as far as possible on the other hand during training, thereby avoiding the condition that the attention points of the internal area of the temperature measurement scene are not correspondingly adjusted due to the fact that the temperature measurement points are distributed on the periphery of the temperature measurement scene.
According to the method, the position reconstruction model is trained according to the temperature measurement point position sequence, the initial temperature measurement point position sequence and the target temperature distribution, and the trained position reconstruction model is obtained, so that the position reconstruction model is trained in a self-supervision mode, and the accuracy of optical fiber temperature measurement point position recommendation through the position reconstruction model is improved on the premise of not increasing the labeling cost.
S4, obtaining a recommended temperature measurement point position sequence according to the trained position reconstruction model, the initial temperature measurement point position sequence and the target temperature distribution.
The recommended temperature measurement point sequence can be sent to an operation and maintenance person for optical fiber deployment to recommend the deployment position of the optical fiber temperature measurement point.
In a specific embodiment, the trained position reconstruction model comprises: a trained position sequence encoder, a trained temperature profile encoder, and a trained sequence reconstruction decoder;
the step S4 also comprises the following steps:
s41, inputting an initial temperature measuring point position sequence into a trained position sequence encoder for feature extraction to obtain initial sequence features;
s42, inputting the target temperature distribution into a trained temperature distribution encoder for feature extraction to obtain target distribution features;
s43, fusing the initial sequence features and the target distribution features to obtain target fusion features;
s44, inputting the target fusion characteristics into a trained sequence reconstruction decoder for characteristic reconstruction, and obtaining a recommended temperature measurement point sequence.
The initial sequence feature may be used to characterize feature information of the initial temperature measurement point sequence, the target distribution feature may be used to characterize feature information of the target temperature distribution, and the target fusion feature may include the initial sequence feature and the target distribution feature.
According to the training position reconstruction model, the initial temperature measurement point position sequence and the target temperature distribution, the recommended temperature measurement point position sequence is obtained, the temperature measurement point position reconstruction is carried out based on the target temperature distribution, and the temperature measurement point position sequence which is more suitable for the temperature measurement scene can be obtained, so that the accuracy rate of temperature measurement point position recommendation is improved.
In the embodiment of the invention, the target temperature distribution is obtained through the combination of scene temperature distribution obtained by a plurality of time sequence continuous sampling frames, so that the target temperature distribution can represent temperature information under a long period, the temperature measuring point position reconstruction is carried out based on the target temperature distribution, an adjustment temperature measuring point position sequence which is more suitable for the temperature measuring scene can be obtained, the accuracy of temperature measuring point position recommendation is improved, and the position reconstruction model can be trained in a self-supervision mode based on a preset initial temperature measuring point position sequence, so that the accuracy of optical fiber temperature measuring point recommendation through a deep neural network model is improved on the premise of not increasing the labeling cost.
While certain specific embodiments of the invention have been described in detail by way of example, it will be appreciated by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the invention. Those skilled in the art will also appreciate that many modifications may be made to the embodiments without departing from the scope and spirit of the invention. The scope of the invention is defined by the appended claims.
Claims (3)
1. A laser-based optical fiber temperature measurement point recommendation system, the system comprising: the system comprises a laser, a transmission optical fiber, a database, a processor and a memory storing a computer program, wherein the laser generates laser, the laser is transmitted through the transmission optical fiber, and when the computer program is executed by the processor, the following steps are realized:
s1, obtaining target temperature distribution according to scene temperature distribution corresponding to N sampling frames with continuous time sequences, wherein the step S1 further comprises the following steps:
s11, initializing iteration times i to be 1 in the N sampling frames with continuous time sequences, and taking scene temperature distribution corresponding to the ith sampling frame in the time sequences as a history frame;
s12, taking the scene temperature distribution corresponding to the (i+1) th sampling frame in the time sequence as a current frame;
s13, multiplying the history frame by a preset first superposition coefficient to obtain history retention information;
s14, multiplying the current frame by a preset second superposition coefficient to obtain current reserved information;
s15, adding the history retention information and the current retention information to obtain comprehensive retention information;
s16, taking the comprehensive reservation information as the history frame, adding 1 to i, and returning to the execution steps S12 to S16 until a preset condition is met, stopping iteration, and taking the comprehensive reservation information obtained when the iteration is stopped as the target temperature distribution;
s2, obtaining an adjusted temperature measurement point position sequence by using a position reconstruction model according to a preset initial temperature measurement point position sequence and the target temperature distribution, wherein the position reconstruction model comprises a position sequence encoder, a temperature distribution encoder and a sequence reconstruction decoder;
the step S2 further includes the following steps:
s21, inputting the initial temperature measuring point position sequence into the position sequence encoder for feature extraction to obtain position sequence features;
s22, inputting the target temperature distribution into the temperature distribution encoder for feature extraction to obtain temperature distribution features;
s23, fusing the position sequence features and the temperature distribution features to obtain fusion features;
s24, inputting the fusion characteristics into the sequence reconstruction decoder for characteristic reconstruction to obtain the temperature measurement point adjustment sequence;
wherein the position reconstruction model is pre-trained;
the pre-training process of the position reconstruction model comprises the following steps:
inputting a sample temperature measuring point position sequence into the position sequence encoder for feature extraction to obtain sample sequence features, and inputting zero temperature distribution into the temperature distribution encoder for feature extraction to obtain zero distribution features;
fusing the sample sequence features and the zero distribution features to obtain sample fusion features, and inputting the sample fusion features into the sequence reconstruction decoder to reconstruct the features to obtain a sample adjustment position sequence;
pre-training the position sequence encoder, the temperature distribution encoder and the sequence reconstruction decoder according to the sample temperature measurement point position sequence and the sample adjustment position sequence to obtain a pre-trained position sequence encoder, a pre-trained temperature distribution encoder and a pre-trained sequence reconstruction decoder;
forming the position reconstruction model from the pre-trained position sequence encoder, the pre-trained temperature distribution encoder, and the pre-trained sequence reconstruction decoder;
s3, training the position reconstruction model according to the temperature measurement point position adjustment sequence, the initial temperature measurement point position sequence and the target temperature distribution to obtain a trained position reconstruction model, wherein the step S3 further comprises the following steps:
s31, acquiring the number of the temperature-adjusting points in the temperature-adjusting point temperature sequence, and obtaining a first loss function according to the number of the temperature-adjusting points;
s32, obtaining a second loss function according to the temperature measuring point position adjusting sequence and the initial temperature measuring point position sequence;
s33, according to the distribution information of the temperature measuring points in the temperature measuring point position adjusting sequence, a third loss function is obtained;
s34, carrying out weighted addition on the first loss function, the second loss function and the third loss function to obtain a position loss function;
s37, training the position reconstruction model according to the position loss function to obtain the trained position reconstruction model;
wherein, in the step S3, the method further comprises the following steps:
s35, acquiring a plurality of attention points in the target temperature distribution;
s36, obtaining a distribution loss function according to all the attention points and the temperature measurement point adjustment sequence;
accordingly, in step S37, it includes:
training the position reconstruction model according to the position loss function and the distribution loss function to obtain the trained position reconstruction model;
s4, obtaining a recommended temperature measurement point sequence according to the trained position reconstruction model, the initial temperature measurement point sequence and the target temperature distribution.
2. The laser-based optical fiber temperature sensing point recommendation system as claimed in claim 1, further comprising the steps of, in step S34:
s341, mapping to obtain a first weight according to the number of the temperature measuring points;
s342, mapping to obtain a second weight according to training rounds of training the position reconstruction model;
s343, obtaining a third weight according to the preset total weight, the first weight and the second weight;
s344, multiplying the first loss function and the first weight to obtain a first multiplication result, multiplying the second loss function and the second weight to obtain a second multiplication result, multiplying the third loss function and the third weight to obtain a third multiplication result, and taking the sum of the first multiplication result, the second multiplication result and the third multiplication result as the position loss function.
3. The laser based fiber optic temperature measurement point recommendation system of claim 1, wherein the trained position reconstruction model comprises: a trained position sequence encoder, a trained temperature profile encoder, and a trained sequence reconstruction decoder;
the step S4 also comprises the following steps:
s41, inputting the initial temperature measurement point position sequence into the trained position sequence encoder for feature extraction to obtain initial sequence features;
s42, inputting the target temperature distribution into the trained temperature distribution encoder for feature extraction to obtain target distribution features;
s43, fusing the initial sequence features and the target distribution features to obtain target fusion features;
s44, inputting the target fusion characteristics into the trained sequence reconstruction decoder for characteristic reconstruction, and obtaining the recommended temperature measurement point sequence.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311023694.6A CN117057239B (en) | 2023-08-14 | 2023-08-14 | Optical fiber temperature measuring point positioning system based on laser |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311023694.6A CN117057239B (en) | 2023-08-14 | 2023-08-14 | Optical fiber temperature measuring point positioning system based on laser |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117057239A CN117057239A (en) | 2023-11-14 |
| CN117057239B true CN117057239B (en) | 2024-04-12 |
Family
ID=88652994
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311023694.6A Active CN117057239B (en) | 2023-08-14 | 2023-08-14 | Optical fiber temperature measuring point positioning system based on laser |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117057239B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109974892A (en) * | 2019-03-25 | 2019-07-05 | 清华大学 | Based on optical fiber temperature-measurement the temperature field forming method and system complementary with Digital Measurement of Temperature |
| CN114526821A (en) * | 2022-03-07 | 2022-05-24 | 优链加(重庆)科技有限公司 | Temperature measuring device capable of automatically positioning temperature measuring point for industrial field equipment |
| CN116499607A (en) * | 2023-06-27 | 2023-07-28 | 之江实验室 | Optical fiber sensing signal noise reduction method and device and storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11176330B2 (en) * | 2019-07-22 | 2021-11-16 | Advanced New Technologies Co., Ltd. | Generating recommendation information |
-
2023
- 2023-08-14 CN CN202311023694.6A patent/CN117057239B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109974892A (en) * | 2019-03-25 | 2019-07-05 | 清华大学 | Based on optical fiber temperature-measurement the temperature field forming method and system complementary with Digital Measurement of Temperature |
| CN114526821A (en) * | 2022-03-07 | 2022-05-24 | 优链加(重庆)科技有限公司 | Temperature measuring device capable of automatically positioning temperature measuring point for industrial field equipment |
| CN116499607A (en) * | 2023-06-27 | 2023-07-28 | 之江实验室 | Optical fiber sensing signal noise reduction method and device and storage medium |
Non-Patent Citations (2)
| Title |
|---|
| 基于主动学习的半监督领域自适应方法研究;姚明海;黄展聪;;高技术通讯;20200815(08);全文 * |
| 遗传优化神经网络的激光散斑数据建模与分析;庞亮;孟雪井;;激光杂志;20180925(09);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117057239A (en) | 2023-11-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111191841B (en) | Power load prediction method and device, computer equipment and storage medium | |
| CN110456026B (en) | Soil moisture content monitoring method and device | |
| CN116150897A (en) | Machine tool spindle performance evaluation method and system based on digital twin | |
| WO2024104401A1 (en) | Cloud network abnormality detection model training method based on reinforcement learning, and storage medium | |
| CN114004430B (en) | Wind speed forecasting method and system | |
| CN114330647A (en) | Model training method and device and silicon rod weight prediction method | |
| CN113037577A (en) | Network traffic prediction method, device and computer readable storage medium | |
| CN112651519A (en) | Secondary equipment fault positioning method and system based on deep learning theory | |
| CN116050627A (en) | Photovoltaic power prediction method and photovoltaic power prediction model training method | |
| CN114528190B (en) | Single index abnormality detection method and device, electronic equipment and readable storage medium | |
| CN117057234B (en) | Optical fiber temperature measuring point positioning system based on laser | |
| CN117057239B (en) | Optical fiber temperature measuring point positioning system based on laser | |
| CN116258991A (en) | Abnormality detection apparatus, abnormality detection system, abnormality detection method, and computer program product | |
| CN113988210A (en) | Method and device for restoring distorted data of structure monitoring sensor network and storage medium | |
| FI20195790A1 (en) | Machine-learning-based quality prediction of manufactured fiber optic cable | |
| KR102489473B1 (en) | Method and system for predicting climate indices using artificial intelligence model capable of providing uncertainty and relevance for climate index prediction result | |
| CN115344621A (en) | Multi-variable time sequence anomaly detection method and system based on unwrapping network | |
| CN115713044B (en) | Method and device for analyzing residual life of electromechanical equipment under multi-condition switching | |
| CN116362138A (en) | Artificial intelligence park carbon monitoring method based on big data | |
| CN114705177A (en) | Fiber-optic gyroscope attitude measurement data processing method based on error analysis | |
| CN115795991A (en) | Landslide prediction method and device, electronic equipment and readable storage medium | |
| CN117787109B (en) | Wind speed conversion reliability assessment method and system | |
| US11256244B2 (en) | Adaptive alarm and dispatch system using incremental regressive model development | |
| CN117057235B (en) | Optical fiber temperature measuring point deployment system based on laser | |
| CN113379455B (en) | Order quantity prediction method and equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |