CN111798089B - Urban natural gas high-pressure pipe network running state risk evaluation method - Google Patents

Urban natural gas high-pressure pipe network running state risk evaluation method Download PDF

Info

Publication number
CN111798089B
CN111798089B CN202010446748.XA CN202010446748A CN111798089B CN 111798089 B CN111798089 B CN 111798089B CN 202010446748 A CN202010446748 A CN 202010446748A CN 111798089 B CN111798089 B CN 111798089B
Authority
CN
China
Prior art keywords
data
output
gate
natural gas
value
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
Application number
CN202010446748.XA
Other languages
Chinese (zh)
Other versions
CN111798089A (en
Inventor
朱丹
何胜冬
余祖强
张远望
曾梽芮
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chengdu China Resources Gas Design Co ltd
Original Assignee
Chengdu China Resources Gas Design Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Chengdu China Resources Gas Design Co ltd filed Critical Chengdu China Resources Gas Design Co ltd
Priority to CN202010446748.XA priority Critical patent/CN111798089B/en
Publication of CN111798089A publication Critical patent/CN111798089A/en
Application granted granted Critical
Publication of CN111798089B publication Critical patent/CN111798089B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0635—Risk analysis of enterprise or organisation activities
    • 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
    • G06N3/044—Recurrent networks, e.g. Hopfield 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/048—Activation functions
    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Health & Medical Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Human Resources & Organizations (AREA)
  • Software Systems (AREA)
  • Economics (AREA)
  • Data Mining & Analysis (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Computational Linguistics (AREA)
  • Mathematical Physics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Strategic Management (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Tourism & Hospitality (AREA)
  • Game Theory and Decision Science (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Development Economics (AREA)
  • Educational Administration (AREA)
  • Public Health (AREA)
  • Water Supply & Treatment (AREA)
  • Primary Health Care (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention relates to a risk evaluation method for the running state of a city natural gas high-pressure pipe network, in particular to the field of natural gas supply. The method comprises the following steps: s1: acquiring original data; s2: preprocessing the original data to obtain normalized data; s3: a long-term memory network LSTM is built, and comprises three gates, namely a forgetting gate, an input gate and an output gate, wherein each gate is provided with a storage unit, and each gate uses sigmod as an activation function to output a value between 0 and 1. When the sigmod output is 1, all information can pass through; when the sigmod output is 0, any information cannot pass; the method solves the technical problem of risk analysis and evaluation on the pressure value of the urban natural gas high-pressure pipe network under the premise of considering the influence of the sequence factors, and is suitable for prediction of the running state of the urban natural gas high-pressure pipe network.

Description

Urban natural gas high-pressure pipe network running state risk evaluation method
Technical Field
The invention relates to the field of natural gas supply, in particular to a risk evaluation method for the running state of a city natural gas high-pressure pipe network.
Background
Natural gas reserves in China are quite rich, and natural gas is increasingly favored by people along with continuous healthy development of economy, gradual shortage of energy sources such as petroleum, coal and the like and continuous enhancement of environmental protection consciousness of people. The natural gas can burn and release heat, is used by urban residents and industrial enterprises, is a clean, high-energy and high-quality green energy source, is easy to transmit, has the advantages that other fuels are incomparable, and in most cities in the south, the gas conveyed by a gas pipeline network is mainly natural gas (the specific gravity of the gas pipeline network accounts for about 90 percent). With the rapid development of the economy of China and the implementation of the Western gas east transport engineering of China, most of the large and medium cities of China successively establish natural gas pipeline facilities and use a large amount of natural gas. However, natural gas using methane as a main component has the dangers of inflammability, explosiveness, easy diffusion, toxicity (monoscopic gas), thermal expansibility, compressibility and the like, and accidents such as poisoning, fire disaster, explosion and the like can occur in each link of the production, delivery and use process, so that the life and property safety and social stability of people are seriously endangered. Therefore, the natural gas transmission and distribution pipe network system is taken as an important component of urban infrastructure and urban public utilities, and whether the natural gas transmission and distribution pipe network system can safely run has great significance for daily production and life.
Research shows that many factors in actual production and life can influence the operation of a natural gas pipe network, and the main factors are as follows:
(1) Pipeline leakage:
the cause of the leakage of the pipeline is mainly as follows:
(1) corrosion of equipment: factors that corrode plumbing include electrochemical corrosion, chemical corrosion, and parasitic current corrosion. The buried natural gas pipeline generates electrochemical reaction among the rainwater in the pipeline due to the factors of the rainwater in the soil, the non-uniformity of the structure of the pipeline steel pipe, different surface roughness, higher carbon element content and the like, so that the pipeline is oxidized rapidly; the chemical reaction is more intense and rapid due to the interaction between chemical substances in the pipeline and surrounding substances and the electrochemical reaction, and the superposition of the factors accelerates the corrosion of the natural gas pipeline, so that the natural gas pipeline is locally and rapidly thinned, and the natural gas is leaked.
(2) The natural reasons are as follows: the shearing stress at the pipeline joint is changed due to expansion caused by heat and contraction caused by weather change and other factors, so that natural gas is leaked.
(3) External damage: the urban disordered construction accident occurs, and some areas are not provided with safety marks and natural gas pipeline marks, so that partial residents are randomly constructed on the natural gas pipeline, and the safety operation of urban natural gas pipeline networks and facilities is seriously influenced; in the process of accessing roads, buildings, construction, communication cables and power grids into the ground, natural gas pipelines can be damaged due to the negligence of constructors or low technical level and other factors, so that the pipelines are leaked; when a large truck passes through a middle-low pressure pipeline at an intersection, the natural gas pipeline can receive overload acting force from a road, serious deformation can occur, and even the buried natural gas pipeline is broken when the road is stressed unevenly, so that natural gas leakage is caused.
(2) Gas consumption non-uniformity in natural gas pipe network operation
The non-uniformity of urban natural gas usage is a significant feature of urban gas usage, and is mainly represented by the following two aspects:
(1) regional inhomogeneities: the natural gas pipe network in some areas has the phenomenon that the gas consumption in urban hot spot areas is large and the gas consumption in rural non-hot spot areas is small, which directly causes the problems that the natural gas pipe network runs under-pressure in urban hot spot areas and runs over-pressure in rural non-hot spot areas.
(2) Imbalance in time: the natural gas consumption of different periods of each month and each day in the same area is also unevenly influenced by factors such as regional climate, gas supply scale, resident living standard and living habit, holidays and the like, and the natural gas consumption is particularly uneven in month, day and hour.
The natural gas pipeline leakage and damage caused by the above reasons and the non-uniformity of the natural gas consumption lead to the natural gas transmission and distribution pipeline network to be in an unsafe operation state frequently, serious safety accidents are easy to occur, and the life and property of people and countries are greatly lost, thus being a key problem to be solved in the safe operation of the natural gas pipeline network. Therefore, the research of predicting and predicting factors such as the running state of the urban natural gas high-pressure pipe network by using the artificial intelligence algorithm development has very important practical significance and economic significance in national economy and social development.
At present, the risk of the running state of the high-pressure pipe network at home and abroad is mainly evaluated by fuzzy comprehensive evaluation, bayesian network and other methods, although the Bayesian method has simple prediction process and high speed, naive Bayes has the assumption premise of independent distribution, and the predictions are difficult to be completely independent in the actual engineering running. The regression analysis method is simple and convenient when analyzing the multi-factor model, can accurately measure the correlation degree and regression fitting degree between each factor, and improves the effect of a prediction equation, but what factor is selected in the regression analysis and what expression is adopted by the factor are only one type of speculation, which influences the diversity of the factors and the invisibility of certain factors, so that the regression analysis is limited in certain situations. The pressure data, the flow data, the potential data tested by the cathode protection intelligent test system and the like generated during the operation of the actual pipe network are all sequence data, and the domestic and foreign research methods do not consider the influence of sequence factors.
Disclosure of Invention
The invention aims to solve the technical problem of predicting the running state of the urban natural gas high-pressure pipe network under the premise of considering the influence of sequence factors.
The technical scheme for solving the technical problems is as follows: a risk evaluation method for the running state of a city natural gas high-pressure pipe network comprises the following steps:
s1: acquiring original data;
s2: preprocessing the original data to obtain normalized data;
s3: a long-term memory network LSTM is built, and comprises three gates, namely a forgetting gate, an input gate and an output gate, wherein each gate is provided with a storage unit, and each gate uses sigmod as an activation function to output a value between 0 and 1. When the sigmod output is 1, all information can pass through; when the sigmod output is 0, any information cannot pass;
s4: definition { x } 1 ,x 2 ,......,x t Is an input sequence, whichX in the middle t ∈R k Representing a K-dimensional real number vector, representing the input at time t, the forgetting gate reads the output h of the last cell t-1 And current input x t Adding a bias value b f The forgetting gate output f is obtained by the following formula and sigmod activation function sigma t ，f t Representing the probability of forgetting the state of the upper layer of hidden cells;
f t ＝σ(W xf ·x t +W hf ·h t-1 +b f )；
s5: the input gate obtains input gate output i according to the following formula t ，i t Representing the value that decides to update:
i t ＝σ(W xi ·x t +W hi ·h t-1 +b i )；
obtaining the input gate output g according to the tanh activation function phi and the following formula t ，g t A value representing a state that can be added to:
g t ＝φ(W xg ·x t +W hg ·h t-1 +b g )；
output i according to input gate t Input gate output g t Forgetting gate output f t State s of last cell t And obtaining the memory s of the current cell according to the following formula t ，s t A memory value indicating that the current cell will continue to pass on to the next cell:
s t ＝f t *s t-1 +i t *g t ；
s6: the output gate obtains the hidden state parameter o according to the activation function sigmoid and the following formula t :
o t ＝σ(W xo ·x t +W ho ·h t-1 +b o )；
The output gate also calculates the hidden state parameter o by the following formula t ：
h t ＝o t *φ(s t )；
Predicting the pressure value of the node at the next time point according to all the pressure data of the 4h before the node and the following formula;
wherein W is xf ,W hf ,W xi ,W hi ,W xg ,W hg ,W xo ,W ho Weight matrix representing corresponding nodes of the network, b f ,b i ,b g ,b o Representing the bias value of the corresponding node of the network, representing the element multiplication, σ representing the sigmod activation functionPhi represents the tanh activation function +.>
S7: and training the long-term memory network LSTM by taking the normalized data as the output sequence, obtaining an information retention reference value of each normalized data according to the long-term memory network LSTM, and screening the normalized data according to the information retention reference value to obtain a prediction result.
The beneficial effects of the invention are as follows: at present, the field of deep learning research often needs a large amount of data to train a model, and a data source is thousands of in real life, so that the data entering a database without any processing is likely to violate the requirements of three elements of data quality uniqueness, continuity and consistency. After the data in the scheme are processed, the reliability of the data can be ensured when the subsequent data mining is carried out, and because the used normalized data keeps the time sequence of the original data, the running state of the urban natural gas high-pressure pipe network is predicted under the premise of considering the influence of the sequence factors.
On the basis of the technical scheme, the invention can be improved as follows.
Further, the step S2 specifically includes:
s21: performing data cleaning on the original data to obtain corrected data;
s22: marking the correction data to obtain marking data;
s23: converting the tag data into supervised learning data according to the following formula:
X i ＝[data t-12 ,data t-11 ,......,data t-1 ]
Y i ＝[data t ]
wherein t represents the current time, the marking data corresponding to the current time is Y, the marking data of the first 12 time nodes is X, and i is the time node serial number;
s24: and carrying out normalization processing on the supervised learning data to obtain normalized data.
Further, the step 21 specifically includes:
s211: carrying out missing data processing on the original data, comparing the data at the same time of the same day in a week, if the data at one day is missing, solving the arithmetic average of the data at the same time point of the other 6 days in the same week for supplementation, if the difference between the maximum or minimum value and the second maximum or second minimum value after being sequenced according to the values is larger than 20MPa, solving the arithmetic average of the data at the same time point of the other 6 days in the same week for replacing unreasonable data, and if the data at 10 adjacent time points in the certain day are missing, and all the data in the same day are deleted for meeting the continuity and consistency;
s212: carrying out continuity and consistency test on the original data;
s213: and performing accurate bit number auditing on the original data.
Further, step S22 specifically includes:
s221: numbering the nodes in the correction data according to Huffman coding to obtain node numbering data;
s222: adding a step-down number to the node number data according to the node type and the node to obtain step-down number data;
s223: and obtaining a category interval, and marking the step-down number data as a normal area, a yellow area or a red area according to the category interval to obtain marked data.
The beneficial effect of adopting the above-mentioned further scheme is that in this embodiment, regard a website inside the natural gas high pressure pipe network as a node, the website is pressure regulating station or valve chamber, and the category interval is the numerical value interval of natural gas high pressure pipe network pressure data, divides step-down serial number data into 3 regions according to the use numerical value according to the category interval.
Further, in step S1, the raw data is gas usage pressure data with a collection frequency of 20mi n/time.
Additional aspects of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
FIG. 1 is a flow chart of an embodiment of a method for evaluating risk of an operating state of a municipal natural gas high-pressure pipe network according to the invention;
fig. 2 is a topological structure diagram of the coding completion of other embodiments of the method for evaluating the risk of the running state of the urban natural gas high-pressure pipe network.
Detailed Description
The principles and features of the present invention are described below with reference to the drawings, the examples are illustrated for the purpose of illustrating the invention and are not to be construed as limiting the scope of the invention.
An example is substantially as shown in figure 1:
the risk evaluation method for the running state of the urban natural gas high-pressure pipe network in the embodiment comprises the following steps:
s1: acquiring original data;
s2: preprocessing the original data to obtain normalization data, wherein the normalization data in the embodiment can be shown in table 1;
table 1
S3: the method comprises the steps of building a long-short-term memory network LSTM, wherein the long-short-term memory network LSTM comprises three gates, namely a forgetting gate, an input gate and an output gate, each gate is provided with a storage unit, and each gate uses sigmod as an activation function and outputs a value between 0 and 1. When the sigmod output is 1, all information can pass through; when the sigmod output is 0, any information cannot pass;
s4: definition { x } 1 ,x 2 ,......,x t Is an input sequence, where x t ∈R k Representing a K-dimensional real number vector, representing the input at time t, and the forgetting gate reads the output h at the last time t-1 And current input x t Output h of last time t-1 To add the bias value b to the output of the last cell in the multilayer LSTM f Where the bias value is a constant added to the function for data fitting of the reinforcement model in the neural network, e.g. b in y=kx+b, gives the forgetting gate output f by the following formula and sigmod activation function σ t ，f t Representing a probability of forgetting the state of the upper layer of hidden cells;
f t ＝σ(W xf ·x t +W hf ·h t-1 +b f )；
s5: the input gate obtains the input gate output i according to the following formula t ，i t Indicating the value to be updated, the forget gate in this embodiment decides which information of the last cell state should be discarded or retained. The input gate can be used to update the cell state, and its output, in combination with the output of the forget gate and the last cell state, can update the cell state:
i t ＝σ(W xi ·x t +W hi ·h t-1 +b i )；
obtaining the input gate output g according to the tanh activation function phi and the following formula t ，g t A value representing a state that can be added to:
g t ＝φ(W xg ·x t +W hg ·h t-1 +b g )；
output i according to input gate t Input gate output g t Forgetting gate output f t State s of last cell t-1 And obtaining the memory s of the current cell according to the following formula t ，s t A memory value indicating that the current cell will continue to pass on to the next cell:
s t ＝f t *s t-1 +i t *g t ；
s6: the output gate obtains the hidden state parameter o according to the activation function sigmoid and the following formula t In the present embodiment, the output o of the output gate t From the output h of the last time t-1 And the input of the current moment is substituted into a sigmoid function to obtain o t The value of (2) is in the range of [0,1 ]]In between, the specific output value of each cell output gate is changed continuously during the model training process to obtain the optimized value:
o t ＝σ(W xo ·x t +W ho ·h t-1 +b o )；
the output gate also passes the following formula and the hidden state parameter o t Obtaining an information retention reference value h t In this embodiment, a reference value h is reserved for information t Final output of cell:
h t ＝o t *φ(s t )；
predicting the pressure value of the node at the next time point according to all the pressure data of the 4h before the node and the following formula;
wherein W is xf ,W hf ,W xi ,W hi ,W xg ,W hg ,W xo ,W ho Weight matrix representing corresponding nodes of the network, b f ,b i ,b g ,b o Representing the bias value of the corresponding node of the network, representing the element multiplication, σ representing the sigmod activation functionPhi represents the tanh activation function +.>
S7: training the normalized data as an output sequence for an LSTM (long-short-period memory) in an entry, obtaining an information retention reference value of each normalized data according to the LSTM, and screening the normalized data according to the information retention reference value to obtain a predicted result; when the predicted value is more than 16MPa and less than 20MPa, the next moment is indicated to be in a yellow state, and corresponding observation is needed; and when the predicted value is more than 20MPa and less than 35MPa, indicating that the next moment is in a normal state.
The beneficial effects of the invention are as follows: at present, the field of deep learning research often needs a large amount of data to train a model, and a data source is thousands of in real life, so that the data entering a database without any processing is likely to violate the requirements of three elements of data quality uniqueness, continuity and consistency. After the data in the scheme are processed, the reliability of the data can be ensured when the subsequent data mining is carried out, and because the used normalized data keeps the time sequence of the original data, the running state of the urban natural gas high-pressure pipe network is predicted under the premise of considering the influence of the sequence factors.
On the basis of the technical scheme, the invention can be improved as follows.
Optionally, in some other embodiments, step S2 is specifically:
s21: carrying out data cleaning on the original data to obtain corrected data;
s22: marking the correction data to obtain marking data;
s23: the tag data is converted into supervised learning data according to the following formula:
X i ＝[data t-12 ,data t-11 ,......,data t-1 ]
Y i ＝[data t ]
wherein t represents the current time, the marking data corresponding to the current time is Y, the marking data of the first 12 time nodes is X, and i is the time node serial number;
in this embodiment, after the tag data is converted into the supervised learning data, the table 2 may be as follows:
var(t-12) var(t-11) var(t-10) … var(t-1) var(t)
12 33.93 19.20 20.21 … 31.39 16.36
13 19.20 20.21 23.32 … 16.36 28.61
… … … … … … …
49999 26.55 27.53 27.01 … 22.53 55.4
table 2
S24: and carrying out normalization processing on the supervised learning data to obtain normalized data.
Optionally, in some other embodiments, step 21 is specifically:
s211: carrying out missing data processing on the original data, comparing the data at the same time of the same day in a week, if the data at a certain day is missing, calculating the arithmetic average of the data at the same time of the other 6 days in the same week to supplement, if the difference between the maximum or minimum value and the second maximum or second minimum value after being sequenced according to the values is larger than 20MPa, and if the data at 10 adjacent time points in a certain day are missing, all the data in the same day are deleted in order to meet the continuity and consistency;
s212: carrying out continuity and consistency test on the original data;
the model training stage of the scheme needs to train the data of the same time point of eight voltage regulating stations (valve chambers), so that strict time consistency and continuity checking are needed. If the missing occurs, proper filling is carried out according to the project missing data filling method. If the data at the corresponding time point of a certain pressure regulating station (valve chamber) is more missing and cannot be filled for the reasons described in the above (1), deleting the data at the corresponding time point of all other pressure regulating stations (valve chambers);
s213: performing accurate bit number auditing on the original data;
the real-time operation pressure data unit of the natural gas high-pressure pipe network is MPa, and the number of accurate digits is generally 2 digits after decimal places, so that in the data cleaning stage, data which do not meet the requirement of the number of accurate digits need to be correspondingly processed according to the method in the step S211.
Optionally, in some other embodiments, step S22 is specifically:
s221: numbering nodes in the correction data according to Huffman coding to obtain node numbering data; in computer data processing, huffman coding uses a variable length coding table to code source symbols (such as a letter in a file), wherein the variable length coding table is obtained by a method for evaluating the occurrence probability of the source symbols, letters with high occurrence probability use shorter codes, otherwise, letters with low occurrence probability use longer codes, which reduces the average length and expected value of the character strings after coding, thereby achieving the purpose of lossless compression of data. In the encoding process, the left branch of the Huffman encoding tree is specified to represent 0, the right branch represents 1, and then the sequence of 0 and 1 formed by the path passing from the root node to each leaf node becomes the encoding of the corresponding character of the leaf node. When the project is used for drawing a topological structure diagram of a natural gas high-pressure pipe network, coding each voltage regulating station (valve chamber) abstracted into nodes by referring to a Huffman coding principle, wherein the coded topological structure diagram is shown in a figure 2, the drawing method is that the node at the most upstream is coded into P1, and the downstream nodes are numbered as 1, 2 and 3 in sequence according to branches from left to right;
s222: adding a depressurization number to the node number data according to the node type and the node to obtain depressurization number data;
s223: the class section is acquired, the step-down number data is marked as a normal area, a yellow area or a red area according to the class section, and the marked data is obtained, and the class section in this embodiment may be as shown in table 3:
numerical distribution (Unit: MPa) Regional distribution
P<16 Red area
16<P<20 Yellow region
20<P<35 Normal region
P>35 Red area
TABLE 3
When the numerical value is in the normal distribution interval, the natural gas high-pressure pipe network can normally operate, when the numerical value is in the yellow region, the numerical value can be recovered to be normal in the next time period, and the numerical value can be further reduced to the yellow region, so that the situation is relatively complex, the dynamic state of data change needs to be closely tracked, when the numerical value is in the red region, the pipeline is likely to have some leakage, and related technicians need to be immediately arranged for rush repair.
In this embodiment, a site inside the natural gas high-pressure pipe network is regarded as a node, the site is a pressure regulating station or a valve chamber, the class section is a numerical section of pressure data of the natural gas high-pressure pipe network, and the pressure drop number data is divided into 3 areas according to the use numerical value in the class section.
Optionally, in some other embodiments, in step S1, the raw data is acquired at a frequency of 1/20 min, and the pressure data is used.
Additional aspects of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
It should be noted that, the foregoing embodiments are product embodiments corresponding to the foregoing method embodiments, and description of each structural device and an optional implementation manner in this embodiment may refer to corresponding description in the foregoing method embodiments, which is not repeated herein.
The reader will appreciate that in the description of this specification, a description of terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, the different embodiments or examples described in this specification and the features of the different embodiments or examples may be combined and combined by those skilled in the art without contradiction.
It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the apparatus and units described above may refer to corresponding procedures in the foregoing method embodiments, which are not described herein again.
The present invention is not limited to the above embodiments, and various equivalent modifications and substitutions can be easily made by those skilled in the art within the technical scope of the present invention, and these modifications and substitutions are intended to be included in the scope of the present invention. Therefore, the protection scope of the invention is subject to the protection scope of the claims.

Claims (1)

1. The risk evaluation method for the running state of the urban natural gas high-pressure pipe network is characterized by comprising the following steps of:
s1: acquiring original data;
s2: preprocessing the original data to obtain normalized data;
s3: building a long-term memory network LSTM, wherein the long-term memory network LSTM comprises three gates, namely a forgetting gate, an input gate and an output gate, each gate is provided with a storage unit, each gate uses sigmod as an activation function, a value between 0 and 1 is output, and when the sigmod output is 1, all information can pass through; when the sigmod output is 0, any information cannot pass;
s4: definition { x } 1 ,x 2 ,......,x t Is an input sequence, where x t ∈R k Representing a K-dimensional real number vector, representing an input at time t, said forgetting gate reading the output h at the previous time t-1 And current input x t Adding a bias value b f The forgetting gate output f is obtained by the following formula and sigmod activation function sigma t ，f t Representing the probability of forgetting the state of the upper layer of hidden cells;
f t ＝σ(W xf ·x t +W hf ·h t-1 +b f )；
s5: the input gate obtains input gate output i according to the following formula t ，i t Representing the value that decides to update:
i t ＝σ(W xi ·x t +W hi ·h t-1 +b i )；
obtaining the input gate output g according to the tanh activation function phi and the following formula t ，g t A value representing a state that can be added to:
g t ＝φ(W xg ·x t +W hg ·h t-1 +b g )；
output i according to input gate t Input gate output g t Forgetting gate output f t State s of last cell t-1 And obtaining the memory s of the current cell according to the following formula t ，s t A memory value indicating that the current cell will continue to pass on to the next cell:
s t ＝f t *s t-1 +i t *g t ；
s6: the output gate obtains the hidden state parameter o according to the activation function sigmoid and the following formula t :
o t ＝σ(W xo ·x t +W ho ·h t-1 +b o )；
The output gate also calculates the hidden state parameter o by the following formula t ：
h t ＝o t *φ(s t )；
Predicting the pressure value of the node at the next time point according to all the pressure data of 4h before the node and all the formulas;
wherein W is xf ,W hf ,W xi ,W hi ,W xg ,W hg ,W xo ,W ho Weight matrix representing corresponding nodes of the network, b f ,b i ,b g ,b o Representing the bias value of the corresponding node of the network, representing the element multiplication, σ representing the sigmod activation functionPhi represents the tanh activation function +.>
S7: training the long-term memory network LSTM by taking the normalized data as the output sequence, obtaining an information retention reference value of each normalized data according to the long-term memory network LSTM, and screening the normalized data according to the information retention reference value, namely the final output of a cell to obtain a prediction result;
the step S2 specifically comprises the following steps:
s21: performing data cleaning on the original data to obtain corrected data;
s22: marking the correction data to obtain marking data;
s23: converting the tag data into supervised learning data according to the following formula:
X i ＝[data t-12 ,data t-11 ,......,data t-1 ]
Y i ＝[data t ]
wherein t represents the current time, the marking data corresponding to the current time is Y, the marking data of the first 12 time nodes is X, and i is the time node serial number;
s24: normalizing the supervised learning data to obtain normalized data;
the step S21 specifically includes:
s211: carrying out missing data processing on the original data, comparing the data at the same time of the same day in a week, if the data at one day is missing, solving the arithmetic average of the data at the same time point of the other 6 days in the same week for supplementation, if the difference between the maximum or minimum value and the second maximum or second minimum value after being sequenced according to the values is larger than 20MPa, solving the arithmetic average of the data at the same time point of the other 6 days in the same week for replacing unreasonable data, and if the data at 10 adjacent time points in the certain day are missing, and all the data in the same day are deleted for meeting the continuity and consistency;
s212: carrying out continuity and consistency test on the original data;
s213: performing accurate bit number auditing on the original data;
the step S22 specifically includes:
s221: numbering the nodes in the correction data according to Huffman coding to obtain node numbering data;
s222: adding a step-down number to the node number data according to the node type and the node to obtain step-down number data;
s223: acquiring a category interval, and marking the depressurization serial number data as a normal area, a yellow area or a red area according to the category interval to obtain marked data;
in step S1, the original data are gas use pressure data with the acquisition frequency of 20 min/time.
CN202010446748.XA 2020-05-25 2020-05-25 Urban natural gas high-pressure pipe network running state risk evaluation method Active CN111798089B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010446748.XA CN111798089B (en) 2020-05-25 2020-05-25 Urban natural gas high-pressure pipe network running state risk evaluation method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010446748.XA CN111798089B (en) 2020-05-25 2020-05-25 Urban natural gas high-pressure pipe network running state risk evaluation method

Publications (2)

Publication Number Publication Date
CN111798089A CN111798089A (en) 2020-10-20
CN111798089B true CN111798089B (en) 2023-10-27

Family

ID=72806561

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010446748.XA Active CN111798089B (en) 2020-05-25 2020-05-25 Urban natural gas high-pressure pipe network running state risk evaluation method

Country Status (1)

Country Link
CN (1) CN111798089B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114399113B (en) * 2022-01-14 2024-07-09 成都秦川物联网科技股份有限公司 Natural gas pipe management method and system
CN115631066B (en) 2022-12-22 2023-03-07 成都秦川物联网科技股份有限公司 Intelligent gas pipeline frost heaving safety management method and Internet of things system
CN116913057B (en) * 2023-09-12 2023-12-01 西安中创博远网络科技有限公司 Livestock-raising abnormal early warning system based on thing networking

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108280551A (en) * 2018-02-02 2018-07-13 华北电力大学 A kind of photovoltaic power generation power prediction method using shot and long term memory network
CN109131452A (en) * 2018-08-14 2019-01-04 西安理工大学 A kind of train status on-line prediction method based on long memory network in short-term
CN110533163A (en) * 2019-08-02 2019-12-03 中国石油集团长城钻探工程有限公司 Sand plugging after fracturing method for early warning based on long Memory Neural Networks in short-term
CN111022941A (en) * 2019-12-27 2020-04-17 重庆科技学院 Natural gas pipeline leakage detection method based on LSTM recurrent neural network

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108280551A (en) * 2018-02-02 2018-07-13 华北电力大学 A kind of photovoltaic power generation power prediction method using shot and long term memory network
CN109131452A (en) * 2018-08-14 2019-01-04 西安理工大学 A kind of train status on-line prediction method based on long memory network in short-term
CN110533163A (en) * 2019-08-02 2019-12-03 中国石油集团长城钻探工程有限公司 Sand plugging after fracturing method for early warning based on long Memory Neural Networks in short-term
CN111022941A (en) * 2019-12-27 2020-04-17 重庆科技学院 Natural gas pipeline leakage detection method based on LSTM recurrent neural network

Also Published As

Publication number Publication date
CN111798089A (en) 2020-10-20

Similar Documents

Publication Publication Date Title
CN111798089B (en) Urban natural gas high-pressure pipe network running state risk evaluation method
Huan et al. Assessment and validation of groundwater vulnerability to nitrate based on a modified DRASTIC model: a case study in Jilin City of northeast China
Zhu et al. Tunnel condition assessment via cloud model‐based random forests and self‐training approach
Behnia et al. Predicting crest settlement in concrete face rockfill dams using adaptive neuro-fuzzy inference system and gene expression programming intelligent methods
CN106022596A (en) Urban gas pipeline system danger forecast and evaluation method
Karamoozian et al. A hybrid risk prioritization approach in construction projects using failure mode and effective analysis
Abdelkader et al. A comprehensive comparative analysis of machine learning models for predicting heating and cooling loads
Chen et al. Probabilistic performance prediction of shield tunnels in operation through data mining
Pankratova et al. Evaluating perspectives of urban underground construction using modified morphological analysis method
Zhu et al. Evaluation of deep coal and gas outburst based on RS-GA-BP
Hu et al. Ecological technology evaluation model and its application based on Logistic Regression
Ji et al. Safety Risk Evaluation of Large and Complex Bridges during Construction Based on the Delphi‐Improved FAHP‐Factor Analysis Method
Lee et al. MCDM approach for identifying urban flood vulnerability under social environment and climate change
Xu et al. Spatiotemporal variations of eco-environmental vulnerability in Shiyang River Basin, China
CN108615098A (en) Water supply network pipeline burst Risk Forecast Method based on Bayesian survival analysis
He et al. Integration of interpretive structural modeling with fuzzy Bayesian network for risk assessment of tunnel collapse
Liu et al. Evaluation of the influencing factors of using underground space of abandoned coal mines to store hydrogen based on the improved ANP method
López-Kleine et al. Physical characteristics of pipes as indicators of structural state for decision-making considerations in sewer asset management
Wu et al. Risk assessment method of major unsafe hydroelectric project
Cheng et al. Safety grade evaluation of aqueduct structure based on fuzzy cloud theory analysis
Meng et al. Risk identification and assessment during the excavation of the deep foundation pit
Guo et al. Stability Evaluation of the Goaf Based on Combination Weighting and Cloud Model
Long et al. Forecasting law enforcement frequency of internet+ coal mine safety supervision
Hsueh EVALUATION OF COMMUNITY ENERGY-SAVING EFFECTS USING FUZZY LOGIC MODEL.
Yuan et al. Problems related to single gully debris flow assessment models

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