CN117350146A - GA-BP neural network-based drainage pipe network health evaluation method - Google Patents

GA-BP neural network-based drainage pipe network health evaluation method Download PDF

Info

Publication number
CN117350146A
CN117350146A CN202311162236.0A CN202311162236A CN117350146A CN 117350146 A CN117350146 A CN 117350146A CN 202311162236 A CN202311162236 A CN 202311162236A CN 117350146 A CN117350146 A CN 117350146A
Authority
CN
China
Prior art keywords
neural network
pipe network
health
network
drainage pipe
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.)
Pending
Application number
CN202311162236.0A
Other languages
Chinese (zh)
Inventor
郭婉茜
郭亮
陶喆
孙慧格
孟昭辉
张宝
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.)
Harbin Institute of Technology
Original Assignee
Harbin Institute of Technology
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 Harbin Institute of Technology filed Critical Harbin Institute of Technology
Priority to CN202311162236.0A priority Critical patent/CN117350146A/en
Publication of CN117350146A publication Critical patent/CN117350146A/en
Pending legal-status Critical Current

Links

Classifications

    • 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
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/18—Network design, e.g. design based on topological or interconnect aspects of utility systems, piping, heating ventilation air conditioning [HVAC] or cabling
    • 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
    • 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
    • G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/10—Numerical modelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2113/00—Details relating to the application field
    • G06F2113/14—Pipes

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)
  • Life Sciences & Earth Sciences (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Mathematical Physics (AREA)
  • Geometry (AREA)
  • Computing Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Computer Hardware Design (AREA)
  • Computational Mathematics (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Analysis (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Physiology (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a drainage pipe network health evaluation method based on a GA-BP neural network, and belongs to the technical field of drainage pipe network health evaluation. Determining a drainage pipe network health evaluation index system by analyzing various factors affecting the drainage pipe network health; establishing a BP neural network hierarchical structure, and determining the input and output of the BP neural network; optimizing and screening the weight and the threshold value of the BP neural network structure by adopting a genetic algorithm, and evaluating the effect of the GA-BP neural network model; the finally determined GA-BP model is used for evaluating the health of the drainage pipe network. The GA-BP model used by the invention has strong interpretability, solves the problem of high evaluation subjectivity in the previous evaluation process of the health of the drainage pipe network, is convenient for determining the pipe network detection priority, is beneficial to the safety transportation and disaster risk active management of pipe network facilities, and simultaneously provides decision support for the optimization and transformation of the pipe network.

Description

GA-BP neural network-based drainage pipe network health evaluation method
Technical Field
The invention relates to the technical field of drainage pipe network health evaluation, in particular to a drainage pipe network health evaluation method based on a GA-BP neural network.
Background
The drainage pipe network of China has huge assets, is an indispensable component part of municipal facilities, and relates to urban industrial construction and normal operation of resident life. The reconstruction, planning, design, construction and engineering acceptance process management of the drainage pipe network facilities is one of the important business contents of the urban drainage pipe management department. Part of the pipelines are long in pipe age, the pipe network ages and the like, the safe operation of the drainage pipe network faces serious challenges, and the risk evaluation of the drainage pipe network is more and more concerned.
At present, the repair and maintenance of the drainage pipe network are more random, the evaluation process is simpler, and even the drainage pipe network is not evaluated and is directly replaced according to the age of the pipe, so that the resource waste is caused. To change this phenomenon, a systematic risk assessment of the network is performed in advance before repair or maintenance of the network. The risk evaluation of the existing drainage pipe network is carried out by adopting an analytic hierarchy process to carry out weight analysis, and then a fuzzy comprehensive evaluation method is utilized to determine the risk grade, so that the evaluation is too subjective, and a more accurate and reliable evaluation method is needed.
Disclosure of Invention
The invention aims to provide a drainage pipe network health evaluation method based on a GA-BP neural network, wherein the weight and the threshold of the BP neural network are optimized through a genetic algorithm, the defects that the BP neural network algorithm is easy to fall into a local minimum and the generalization capability is unstable in the calculation process are overcome, the evaluation model and the parameter have strong interpretability, the pipe network detection priority is determined, the active management of the safety transportation and disaster risk of pipe network facilities is facilitated, and the evaluation of the health risk of the drainage pipe network is more accurate and reliable.
In order to achieve the above purpose, the invention provides a drainage pipe network health evaluation method based on a GA-BP neural network, which comprises the following steps:
s1, analyzing various factors influencing the health of a drainage pipe network, and determining a drainage pipe network health evaluation index system according to the actual drainage conditions of the region and by combining a literature analysis method and national standard requirements and defining an index system construction principle;
s2, determining input and output variables of the BP neural network structure, preprocessing data of the number of nodes of an input layer, the number of nodes of an output layer and the number of nodes of an hidden layer, establishing a BP neural network hierarchical structure, and training the BP neural network structure;
s3, setting genetic algorithm improved BP neural network parameters, including coding setting, determining population size, selecting fitness function and genetic operation setting, and optimizing and screening the determined BP neural network structure weight and threshold according to the genetic algorithm to obtain a GA-BP neural network model;
s4, evaluating the health of the drainage pipe network according to the finally determined GA-BP neural network structure, dividing the health of the drainage pipe network into four grades of higher risk r1, higher risk r2, middle risk r3 and lower risk r4, determining the operation and maintenance priority of the drainage pipe network, providing decision support for pipe network optimization transformation, and scientifically making a pipe transformation optimization scheme.
Preferably, the evaluation index system in step S1 is divided into three layers:
1) A target layer, a drainage pipe network healthiness U;
2) The influence of three aspects of a body factor U1, an external factor U2 and an environmental factor U3 of the pipe network;
3) The sub-influence factors of the three influence factors of the ontology factor U1, the external factor U2 and the environment factor U3 are respectively.
Preferably, step S2 specifically includes:
s21, determining the number of hidden layers and the number of hidden layer neurons, wherein the transfer function of the hidden layers is a sigmoid function, and the transfer function of the output layers is a linear function;
s22, obtaining a predicted value and an actual measured value through the function in the step S21, calculating a root mean square error and a regression R value, and comparing and selecting the hidden layer node number when the RMSE is minimum and the correlation coefficient R value is maximum as the optimal parameter of the model;
s23, designing an input layer and an output layer: determining 10 evaluation index assigned values of an input layer as input parameters, taking 10 evaluation indexes in pipe network health evaluation indexes as the input layer and taking a drainage pipeline health grade as a target output value;
s24, data preprocessing: the independent variables comprise continuous variables and classified variables, wherein the two classified variables are coded according to 0-1, 0 represents a pipeline which is not damaged, 1 represents a pipeline which is damaged, and the multiple classified variables are coded by adopting dummy variables.
Preferably, in step S3, before performing BP neural network training, the population individuals are first used as BP neural network parameters, the error of the result is finally given through the BP neural network, the fitness of the population individuals is calculated, genetic operation setting is performed on the individuals with good fitness, the optimal individuals are used as the initial weight and threshold of the BP neural network structure according to the genetic algorithm, and then the weight and threshold are rapidly adjusted according to the negative gradient direction by the BP algorithm;
and evaluating the effect of the GA-BP neural network model by adopting a confusion matrix method, and sequencing samples by adopting an ROC curve through a prediction result.
Preferably, in step S3,
genetic algorithm coding: the coding mode of the coding length L is real number coding
L＝n×h+h×m+h+m
Wherein n is the number of neurons of an input layer, h is the number of nodes of an hidden layer, and m is the number of neurons of an output layer;
determining population size: is arranged between 10 and 20;
determining a fitness function: the calculation formula of each individual fitness value F in the population is
F＝A∑|P K -t K |
Wherein F is a fitness value, A is a coefficient, t k Corresponding to the actual output value.
Preferably, the genetic manipulation setting includes a selection manipulation, a crossover manipulation, and a mutation manipulation;
selection operation: selection probability P of each individual i based on fitness proportion selection strategy by roulette i Is that
Wherein f i The fitness value of the individual i is obtained, and n is the number of population individuals;
crossover operation: chromosome b of the kth k And chromosome b of the first l Crossing at j bits, the specific formula of crossing is
b Kj ＝b Kj (1-a)+b Ij a
b Ij ＝b Ij (1-a)+b Kj a
Mutation operation: variation of the jth gene of the ith individual, the specific variation being operated as
Wherein b min Is the gene b ij Lower bound of b max Is the gene b ij The upper limit of r 1 Is a random number, G is the iteration number, G max Is the maximum number of evolutions.
Therefore, the drainage pipe network health evaluation method based on the GA-BP neural network has the following beneficial effects:
according to the invention, three angle index types of pipe network equipment body factors, external factors and environment are selected, relative influence factors are selected as sub-indexes, and a genetic algorithm is adopted to optimize the weight and the threshold value of the BP neural network, so that the defects that the BP neural network algorithm is easy to fall into a local minimum value and unstable in generalization capability in the calculation process are overcome, the performance of the model is better, and the health risk evaluation of the drainage pipe network is more accurate and reliable.
The technical scheme of the invention is further described in detail through the drawings and the embodiments.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a diagram of a health index system of a drainage network according to the present invention;
FIG. 3 is a data type and single bitmap of each indicator of the health of the drainage network according to the present invention;
FIG. 4 is a block diagram of a three-layer BP neural network of the present invention;
FIG. 5 is a flow chart of the GA-BP neural network of the present invention.
Detailed Description
The technical scheme of the invention is further described below through the attached drawings and the embodiments.
Unless defined otherwise, technical or scientific terms used herein should be given the ordinary meaning as understood by one of ordinary skill in the art to which this invention belongs. The terms "first," "second," and the like, as used herein, do not denote any order, quantity, or importance, but rather are used to distinguish one element from another. The word "comprising" or "comprises", and the like, means that elements or items preceding the word are included in the element or item listed after the word and equivalents thereof, but does not exclude other elements or items. The terms "disposed," "mounted," "connected," and "connected" are to be construed broadly, and may be, for example, fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. "upper", "lower", "left", "right", etc. are used merely to indicate relative positional relationships, which may also be changed when the absolute position of the object to be described is changed.
Examples
As shown in fig. 1-5, the invention provides a drainage pipe network health evaluation method based on a GA-BP neural network, which comprises the following steps:
1. and (S1) analyzing various factors influencing the health of the drainage pipe network, and determining a drainage pipe network health evaluation index system according to the actual drainage conditions of the region and by combining a literature analysis method and the national standard requirements and defining an index system construction principle (completeness principle, representative principle and operational quantification principle).
As shown in fig. 2-3, the evaluation index system is divided into three layers:
1) A target layer, a drainage pipe network healthiness U;
2) The influence of three aspects of a body factor U1, an external factor U2 and an environmental factor U3 of the pipe network;
3) The sub-influence factors of the three influence factors of the ontology factor U1, the external factor U2 and the environment factor U3 are respectively.
The indexes related to the pipe network health evaluation mainly comprise: the structural characteristics of the pipeline, pipeline load and function, boundary conditions and durability. The invention combines the previous research results, and refers to the standard specifications of the urban drainage pipeline structure grade assessment (DB 11/T1492-2017), the urban drainage pipeline detection and evaluation technical specification (CJJ 181-2012) and the like, and determines the drainage pipe network health risk evaluation system by combining the regional temperature, geology and drainage conditions.
And selecting pipe age, pipe material, pipe diameter, pipe length, pipe burial depth, pipe flow rate, operating pressure, road category, weather temperature and pipe peripheral construction degree from three aspects of pipe structural factors, external factors and environmental factors as evaluation indexes.
BP (Back Propagation) neural network model, namely feedback neural network model, is one of the most mature and wide neural network models at present, and a nonlinear dynamic system can be formed by means of complex connection of a large number of neurons; it generally consists of an input layer, an hidden layer and an output layer.
When the network operates, input data is transmitted from front to back between layers, and the output of each layer only affects the next layer; furthermore, neurons in the same layer are neither directly connected nor have any interference; the errors between neurons must propagate from back to front until the errors reach a set accuracy, and the network will not stop adjusting weights and thresholds based on the errors between the actual and expected outputs.
2. And (S2) determining input and output variables of the BP neural network structure, preprocessing data of the number of nodes of the input layer, the number of nodes of the output layer and the number of nodes of the hidden layer, establishing a BP neural network hierarchical structure, and training the BP neural network structure.
As shown in fig. 4, the specific steps are:
s21, determining the number of hidden layers: the BP neural network mainly plays the learning ability by the hidden layer, and the selection of the hidden layer number is determined according to the characteristics of the problem.
S22, determining the number of hidden layer neurons: if the hidden layer neurons are too many, the network calculated amount is increased, and the fitting problem is easy to occur; too few invisible neurons affect network performance, and the expected result cannot be achieved. The number of hidden layer neurons in the network is directly related to the complexity of the actual problem, the number of input and output layer neurons and the setting of the expected error;
the transfer function of the hidden layer is a sigmoid function, also called hyperbolic tangent S-shaped function, and the formula is
The output layer adopts a linear function as a transfer function, generally adopts a tan sig function, and has the formula of
Obtaining a predicted value, actually measured and calculated Root Mean Square Error (RMSE) and regression R value through the functions, comparing and selecting the hidden layer node number with the minimum RMSE and the maximum correlation coefficient R value, namely the optimal parameter of the model, wherein the formula is
S23, designing an input layer and an output layer: determining 10 evaluation index assigned values of an input layer as input parameters, and taking 10 evaluation indexes { U11, U12, U13, U14, U21, U22, U23, U31, U32 and U33} in the pipe network health evaluation indexes as input layers;
taking the health grade of the drainage pipeline as a target output value; the drainage pipeline health is classified into four classes, so the number of neurons of the output layer is 4.
S24, data preprocessing: the independent variables comprise continuous variables and classified variables; coding the two classification variables according to 0-1, wherein 0 represents a pipeline which is not damaged, and 1 represents a pipeline which is damaged; the attribute values of the multi-classification variables are parallel relations, the meaning of the multi-classification variables cannot be defined by the numerical values of the multi-classification variables, and dummy variable coding is needed. In the index system, the pipe, the road category, the weather temperature and the pericycle construction degree are multi-parallel classification variables, and dummy variable coding is adopted; the data are standardized by maximum and minimum methods of pipe age, pipe diameter, pipe length, pipe burial depth, pipe flow rate and operating pressure.
The dimension and magnitude of other evaluation index data have larger difference, the data are standardized by adopting a maximum and minimum method, the dimension of input data and output data are unified and are classified in a range of 0 to 1, the influence caused by the difference between the magnitude and the magnitude of data samples is avoided, the error is reduced, the convergence speed of the neural network is accelerated, and the formula is that
Wherein y is new sample data, x is original data, x max Is the maximum value of the original data sequence; x is x min Is the minimum of the original data sequence.
80% of the sample set is used for a training set, the former is used for training the drainage pipe network health prediction model, and 20% is used for a testing set, and the latter is used for verifying the drainage pipe network health prediction model.
The genetic algorithm (Genetic Algorithm) is formed by natural selection and natural genetic evolution in the Darwin theory of evolution, and can simulate a genetic mechanism and a species evolution rule so as to form a parallel random search optimization method; the method has the advantages of global searching capability, simplicity in operation, robustness and stronger autonomous learning capability;
in order to solve the problems that the BP neural network is easy to be trapped in local optimum, the hidden layer structure is difficult to determine, the real-time performance is poor and the like, the BP neural network is combined with a genetic algorithm to form a GA-BP neural network optimization algorithm; the genetic algorithm is utilized to optimize the global searching capability, so that the optimized weight and the threshold meet the constructed adaptability requirement, and the optimized weight and the threshold are endowed to the BP network.
3. (step S3) as shown in FIG. 5, setting genetic algorithm to improve BP neural network parameters, including coding setting, determining population size, selecting fitness function, genetic operation setting, and optimizing and screening the determined BP neural network structure weight and threshold according to genetic algorithm to obtain GA-BP neural network model.
Before BP neural network training is carried out, population individuals are taken as BP neural network parameters, errors of results are given through a BP neural network, fitness of the population individuals is calculated, genetic operation setting is carried out on individuals with good fitness, optimal individuals are taken as initial weights and threshold values of the BP neural network structure according to a genetic algorithm, and then the weights and the threshold values are rapidly adjusted according to a negative gradient direction by the BP algorithm;
and evaluating the effect of the GA-BP neural network model by adopting a confusion matrix method, and sequencing samples by adopting an ROC curve through a prediction result.
1) Genetic algorithm coding: the coding types mainly comprise binary coding and real coding, the binary coding has the defect of overlong composition length although being simpler, and therefore, the calculation formula of the coding length L by adopting the real coding as a coding mode is as follows
L＝n×h+h×m+h+m
Wherein n is the number of neurons of the input layer, h is the number of nodes of the hidden layer, and m is the number of neurons of the output layer.
2) Determining population size: population sizes are not uniform and are typically set between 10-20.
3) Determining a fitness function: the fitness function is an evaluation function for evaluating the degree of the individuals in the population, and is used as the basis of optimized search, and the calculation formula of each individual fitness value F in the population is as follows
F＝A∑|P K -t K |
Wherein F is a fitness value, A is a coefficient, t k Corresponding to the actual output value.
Genetic manipulation settings include selection manipulation, crossover manipulation, mutation manipulation:
a) Selection operation: selection probability P of each individual i based on fitness proportion selection strategy by roulette i Is that
Wherein f i The fitness value of the individual i is obtained, and n is the number of population individuals;
b) Crossover operation: the larger the crossover probability is, the faster the convergence speed to the optimal solution range is, but the too large crossover probability makes the model converged to a specific solution, and then the optimizing operation cannot be performed;
chromosome b of the kth k And chromosome b of the first l Crossing at j bits, the specific formula of crossing is
b Kj ＝b Kj (1-a)+b Ij a
b Ij ＝b Ij (1-a)+b Kj a
c) Mutation operation: mutation refers to changing a gene segment of an individual selected in advance according to a set probability, and the mutation probability is generally between 0.001 and 0.1. If the variation probability is too large, the whole process oscillates; the variation probability is too small, so that the optimal solution is difficult to search;
variation of the jth gene of the ith individual, the specific variation being operated as
Wherein b min Is the gene b ij Lower bound of b max Is the gene b ij The upper limit of r 1 Is a random number, G is the iteration number, G max Is the maximum number of evolutions.
Terminating the evolution algebra is typically selected between 50-100;
and setting initial training parameters by using the initial weight and the threshold value of the BP neural network after genetic algorithm optimization, and training the BP neural network structure by using training samples.
And evaluating the model effect by adopting a confusion matrix method: let n be ij Represents the number of i-class samples classified into j-class, K is the sample class, and the recall ratio R of the i-th class samples i Is as follows
And sequencing the samples by adopting the ROC curve through the prediction result, wherein the curve drawn by taking the FPR as the horizontal axis and taking the TPR as the vertical axis is the ROC curve. Calculating a real case rate (TPR) and a false case rate (FPR), wherein the real case rate represents the proportion of the actual case to all cases in the current divided case samples; the false positive rate represents the ratio of the current false positive to the total number of negative samples, and the formula is
TP is a real example, which shows that a sample is actually a positive example, and a model prediction result is a sample of the positive example; FP is a false positive example, representing that the sample is actually a negative example but the model prediction result is a positive example; TN is true and negative, and represents that the sample is actually a negative example and the model prediction result is a negative example; FP is a false negative example, representing that the sample is actually a positive example but the model prediction result is a negative example.
4. And (S4) evaluating the health of the drainage pipe network according to the finally determined GA-BP neural network structure, dividing the health of the drainage pipe network into four grades R= (R1, R2, R3 and R4) with higher risk, middle risk and lower risk of R2 and R3, determining the operation and maintenance priority of the drainage pipe network, providing decision support for pipe network optimization transformation, and scientifically making a pipe transformation optimization scheme.
r1 higher risk: the pipe network cannot meet the expected function, needs to be updated immediately, has obvious worsening signs, has extremely high damage possibility and needs to be maintained immediately;
r2 high risk: the pipe network is in an unreliable state and is lower than the design standard, the function and the safety of the pipe network can be directly influenced, and maintenance planning customization is needed for the pipe network;
risk in r 3: the pipe network is in a normal state, and has general degradation signs, and the pipe network needs to be monitored regularly;
r4 low risk: refers to meeting the expected functional requirements, the pipeline is in a good state and basically does not need to be managed.
Therefore, the invention adopts a drainage pipe network health evaluation method based on the GA-BP neural network, selects three angle index types of pipe network equipment body factors, external factors and environment, selects relative influence factors as sub-indexes, adopts a genetic algorithm to optimize the weight and the threshold value of the BP neural network, overcomes the defects that the BP neural network algorithm is easy to fall into a local minimum value and unstable generalization capability in the calculation process, ensures that the performance of a model is better, and evaluates the health risk of the drainage pipe network more accurately and reliably.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention and not for limiting it, and although the present invention has been described in detail with reference to the preferred embodiments, it will be understood by those skilled in the art that: the technical scheme of the invention can be modified or replaced by the same, and the modified technical scheme cannot deviate from the spirit and scope of the technical scheme of the invention.

Claims (6)

1. A drainage pipe network health evaluation method based on a GA-BP neural network is characterized by comprising the following steps:
s1, analyzing various factors influencing the health of a drainage pipe network, and determining a drainage pipe network health evaluation index system according to the actual drainage conditions of the region and by combining a literature analysis method and national standard requirements and defining an index system construction principle;
s2, determining input and output variables of the BP neural network structure, preprocessing data of the number of nodes of an input layer, the number of nodes of an output layer and the number of nodes of an hidden layer, establishing a BP neural network hierarchical structure, and training the BP neural network structure;
s3, setting genetic algorithm improved BP neural network parameters, including coding setting, determining population size, selecting fitness function and genetic operation setting, and optimizing and screening the determined BP neural network structure weight and threshold according to the genetic algorithm to obtain a GA-BP neural network model;
s4, evaluating the health of the drainage pipe network according to the finally determined GA-BP neural network structure, dividing the health of the drainage pipe network into four grades of higher risk r1, higher risk r2, middle risk r3 and lower risk r4, and determining the operation and maintenance priority of the drainage pipe network.
2. The drainage network health evaluation method based on the GA-BP neural network, which is characterized in that: the evaluation index system in the step S1 is divided into three layers:
1) A target layer, a drainage pipe network healthiness U;
2) The influence factors of the body factor U1, the external factor U2 and the environment factor U3 of the pipe network;
3) The sub-influence factors of the three influence factors of the ontology factor U1, the external factor U2 and the environment factor U3 are respectively.
3. The drainage network health evaluation method based on the GA-BP neural network, which is characterized in that: the step S2 specifically comprises the following steps:
s21, determining the number of hidden layers and the number of hidden layer neurons, wherein the transfer function of the hidden layers is a sigmoid function, and the transfer function of the output layers is a linear function;
s22, obtaining a predicted value and an actual measured value through the function in the step S21, calculating a root mean square error and a regression R value, and comparing and selecting the hidden layer node number when the RMSE is minimum and the correlation coefficient R value is maximum as the optimal parameter of the model;
s23, designing an input layer and an output layer: determining 10 evaluation index assigned values of an input layer as input parameters, taking 10 evaluation indexes in pipe network health evaluation indexes as the input layer and taking a drainage pipeline health grade as a target output value;
s24, data preprocessing: the independent variables comprise continuous variables and classified variables, wherein the two classified variables are coded according to 0-1, 0 represents a pipeline which is not damaged, 1 represents a pipeline which is damaged, and the multiple classified variables are coded by adopting dummy variables.
4. The drainage network health evaluation method based on the GA-BP neural network, which is characterized in that: in the step S3, before BP neural network training is carried out, population individuals are used as BP neural network parameters, errors of results are given through BP neural network terminals, fitness of the population individuals is calculated, genetic operation setting is carried out on individuals with good fitness, optimal individuals are used as initial weights and threshold values of BP neural network structures according to a genetic algorithm, and then the weights and the threshold values are quickly adjusted according to negative gradient directions by the BP algorithm;
and evaluating the effect of the GA-BP neural network model by adopting a confusion matrix method, and sequencing samples by adopting an ROC curve through a prediction result.
5. The method for evaluating the health of the drainage network based on the GA-BP neural network, which is characterized in that: in the step S3 of the process,
genetic algorithm coding: the coding mode of the coding length L is real number coding
L＝n×h+h×m+h+m
Wherein n is the number of neurons of an input layer, h is the number of nodes of an hidden layer, and m is the number of neurons of an output layer;
determining population size: is arranged between 10 and 20;
determining a fitness function: the calculation formula of each individual fitness value F in the population is
F＝A∑|P K -t K |
Wherein F is a fitness value, A is a coefficient, t k Corresponding to the actual output value.
6. The method for evaluating the health of the drainage network based on the GA-BP neural network, which is characterized by comprising the following steps of: genetic manipulation settings include selection manipulation, crossover manipulation, mutation manipulation;
selection operation: selection probability P of each individual i based on fitness proportion selection strategy by roulette i Is that
Wherein f i The fitness value of the individual i is obtained, and n is the number of population individuals;
crossover operation: chromosome b of the kth k And chromosome b of the first l Crossing at j bits, the specific formula of crossing is
b Kj ＝b Kj (1-a)+b Ij a
b Ij ＝b Ij (1-a)+b Kj a
Mutation operation: variation of the jth gene of the ith individual, the specific variation being operated as
Wherein b min Is the gene b ij Lower bound of b max Is the gene b ij The upper limit of r 1 Is a random number, G is the iteration number, G max Is the maximum number of evolutions.
CN202311162236.0A 2023-09-08 2023-09-08 GA-BP neural network-based drainage pipe network health evaluation method Pending CN117350146A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311162236.0A CN117350146A (en) 2023-09-08 2023-09-08 GA-BP neural network-based drainage pipe network health evaluation method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311162236.0A CN117350146A (en) 2023-09-08 2023-09-08 GA-BP neural network-based drainage pipe network health evaluation method

Publications (1)

Publication Number Publication Date
CN117350146A true CN117350146A (en) 2024-01-05

Family

ID=89370069

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311162236.0A Pending CN117350146A (en) 2023-09-08 2023-09-08 GA-BP neural network-based drainage pipe network health evaluation method

Country Status (1)

Country Link
CN (1) CN117350146A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117709204A (en) * 2024-02-05 2024-03-15 国网湖北省电力有限公司经济技术研究院 Method, system and equipment for evaluating stability of pole tower in extreme environment

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2014101636A1 (en) * 2012-12-31 2014-07-03 北京邮电大学 Method for evaluating risk in electric power communications network
CN108009639A (en) * 2017-12-13 2018-05-08 重庆邮电大学 A kind of city ecology construction evaluation method based on GA-BP neural network algorithms
CN108520345A (en) * 2018-03-29 2018-09-11 华南农业大学 Evaluation for cultivated-land method and system based on GA-BP neural network models

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2014101636A1 (en) * 2012-12-31 2014-07-03 北京邮电大学 Method for evaluating risk in electric power communications network
CN108009639A (en) * 2017-12-13 2018-05-08 重庆邮电大学 A kind of city ecology construction evaluation method based on GA-BP neural network algorithms
CN108520345A (en) * 2018-03-29 2018-09-11 华南农业大学 Evaluation for cultivated-land method and system based on GA-BP neural network models

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
邓伟锋 等: "基于GA优化BP神经网络的微电网蓄电池健康状态评估", 电测与仪表, 10 November 2018 (2018-11-10), pages 63 - 67 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117709204A (en) * 2024-02-05 2024-03-15 国网湖北省电力有限公司经济技术研究院 Method, system and equipment for evaluating stability of pole tower in extreme environment
CN117709204B (en) * 2024-02-05 2024-05-07 国网湖北省电力有限公司经济技术研究院 Method, system and equipment for evaluating stability of pole tower in extreme environment

Similar Documents

Publication Publication Date Title
CN108280553B (en) Mountain torrent disaster risk zoning and prediction method based on GIS-neural network integration
CN109142171B (en) Urban PM10 concentration prediction method based on feature expansion and fusing with neural network
CN106650767B (en) Flood forecasting method based on cluster analysis and real-time correction
CN111105332B (en) Highway intelligent pre-maintenance method and system based on artificial neural network
CN110046743B (en) Public building energy consumption prediction method and system based on GA-ANN
CN108009674A (en) Air PM2.5 concentration prediction methods based on CNN and LSTM fused neural networks
Wang et al. Research on prediction of environmental aerosol and PM2. 5 based on artificial neural network
CN114781538B (en) Air quality prediction method and system for GA-BP neural network coupling decision tree
CN103105246A (en) Greenhouse environment forecasting feedback method of back propagation (BP) neural network based on improvement of genetic algorithm
CN116721537A (en) Urban short-time traffic flow prediction method based on GCN-IPSO-LSTM combination model
Ning et al. GA-BP air quality evaluation method based on fuzzy theory.
CN106127242A (en) Year of based on integrated study Extreme Precipitation prognoses system and Forecasting Methodology thereof
CN110598902A (en) Water quality prediction method based on combination of support vector machine and KNN
CN117350146A (en) GA-BP neural network-based drainage pipe network health evaluation method
CN116542382A (en) Sewage treatment dissolved oxygen concentration prediction method based on mixed optimization algorithm
CN114372631A (en) Data-lacking area runoff prediction method based on small sample learning and LSTM
CN116432123A (en) Electric energy meter fault early warning method based on CART decision tree algorithm
CN113435124A (en) Water quality space-time correlation prediction method based on long-time and short-time memory and radial basis function neural network
CN115654381A (en) Water supply pipeline leakage detection method based on graph neural network
CN112765902A (en) RBF neural network soft measurement modeling method based on TentFWA-GD and application thereof
CN115948964A (en) Road flatness prediction method based on GA-BP neural network
Shang et al. Research on intelligent pest prediction of based on improved artificial neural network
CN113269384B (en) Method for early warning health state of river system
CN112434887B (en) Water supply network risk prediction method combining network kernel density estimation and SVM
CN116913098B (en) Short-time traffic flow prediction method integrating air quality and vehicle flow data

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