CN110852505B - Smart city traffic flow prediction method based on quantum genetic optimization LVQ neural network - Google Patents
Smart city traffic flow prediction method based on quantum genetic optimization LVQ neural network Download PDFInfo
- Publication number
- CN110852505B CN110852505B CN201911085728.8A CN201911085728A CN110852505B CN 110852505 B CN110852505 B CN 110852505B CN 201911085728 A CN201911085728 A CN 201911085728A CN 110852505 B CN110852505 B CN 110852505B
- Authority
- CN
- China
- Prior art keywords
- traffic flow
- neural network
- lvq neural
- lvq
- input
- 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
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 79
- 230000002068 genetic effect Effects 0.000 title claims abstract description 25
- 238000000034 method Methods 0.000 title claims abstract description 19
- 238000005457 optimization Methods 0.000 title claims abstract description 15
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 34
- 238000012360 testing method Methods 0.000 claims abstract description 10
- 239000013598 vector Substances 0.000 claims description 29
- 210000002569 neuron Anatomy 0.000 claims description 21
- 238000012549 training Methods 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 6
- 230000002860 competitive effect Effects 0.000 claims description 4
- 230000006870 function Effects 0.000 claims description 4
- 230000009467 reduction Effects 0.000 claims description 4
- 238000004088 simulation Methods 0.000 abstract description 4
- 230000000052 comparative effect Effects 0.000 abstract description 3
- 238000004458 analytical method Methods 0.000 description 7
- 210000000349 chromosome Anatomy 0.000 description 5
- 239000002096 quantum dot Substances 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000013473 artificial intelligence Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000035772 mutation Effects 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001427 coherent effect Effects 0.000 description 1
- 238000010835 comparative analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 230000009191 jumping Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000008707 rearrangement Effects 0.000 description 1
- 238000005728 strengthening Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Images
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
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- G06N3/08—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
- G06Q50/10—Services
- G06Q50/26—Government or public services
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Tourism & Hospitality (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Development Economics (AREA)
- Biophysics (AREA)
- General Business, Economics & Management (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Educational Administration (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a smart city traffic flow prediction method based on a quantum genetic optimization LVQ neural network. By applying the quantum genetic algorithm with a better global solution, the problems that the LVQ neural network is sensitive to the initial weight and is easy to fall into a local minimum value are solved, and the convergence speed is improved. On 5 general traffic flow data sets, comparative simulation experiments were performed with traffic flow predictions based on GA-BP neural networks and wavelet neural networks. The result shows that the short-term traffic flow prediction based on the QGA-LVQ neural network has better accuracy and real-time performance, and the reliability of the short-term traffic flow prediction is verified by the test results on a plurality of data sets.
Description
Technical Field
The invention relates to a smart city traffic flow prediction method based on a quantum genetic optimization LVQ neural network.
Background
Along with the rapid development of economy in China, the travel modes of people are changed greatly, the reserved quantity of private automobiles in China is increased greatly and is promoted year by year, the traffic jam problem in cities is highlighted day by day, and the people and relevant government departments pay more attention to the traffic jam problem. Through long-term investigation and analysis of urban traffic conditions, the traffic congestion is proved to be closely related to the flow. The traditional solution mainly adopts means of increasing road infrastructure, strengthening traffic management and the like, but needs a large amount of manpower and capital investment, has an unobvious effect, and cannot fundamentally solve the problem. The rise and rapid development of artificial intelligence scientific technology provides a new way for smart cities to solve the increasingly serious traffic congestion problem in the future.
An intelligent traffic system based on an artificial neural network becomes a key point and a difficult point of the current research in the field of smart cities. The artificial intelligence management technology can reasonably regulate and control the traffic flow of the whole city, optimize the utilization rate of a road network, effectively relieve traffic jam, reduce energy consumption and reduce pollution. Fig. 1 shows dynamic prediction and selection of a route by using an artificial neural network technology, and table 1 dynamically shows congestion conditions among nodes of a travel route network based on fig. 1, and the congestion conditions of current road sections are dynamically analyzed according to time periods by using the maps and tables, so that sufficient preparation is made for people to predict and select a proper travel route, the travel efficiency of people is improved, and precious time of people is saved; in addition, as can be seen from the figures and tables, the artificial intelligence technology supports the whole road network integrally to perform reasonable autonomous flow regulation and control, and the use efficiency of the road network is improved.
Table 1 congestion situation table based on fig. 1 between nodes of travel network
However, traffic flow data has special attributes such as strong periodicity, nonlinearity and uncertainty, so that real-time accurate traffic flow prediction is difficult. In recent years, various machine learning and artificial neural network models are applied to this field, such as a BP neural network, a wavelet neural network, and the like, but accuracy and real-time performance still cannot meet practical requirements. As an extension of the SOM self-organizing network, the LVQ neural network has the advantages of simple structure, good clustering effect, simple calculation, strong nonlinear fitting capability and great application potential in traffic flow prediction. Therefore, the invention provides a quantum genetic optimization LVQ neural network-based urban traffic network flow prediction method. Because the quantum genetic algorithm has a better global solution, the method can be used for effectively overcoming the defects that the LVQ neural network is sensitive to the initial weight and is easy to fall into a local minimum value, and the convergence speed is improved.
Disclosure of Invention
The invention aims to provide an intelligent urban traffic flow prediction method based on a quantum genetic optimization LVQ neural network, which effectively improves the real-time performance and precision of urban traffic flow prediction.
In order to achieve the purpose, the technical scheme of the invention is as follows: a smart urban traffic flow prediction method based on a quantum genetic optimization LVQ neural network comprises the following steps:
s1, carrying out wavelet noise reduction on the traffic flow data acquired by the ground detector, and removing interference noise signals;
step S2, dividing the denoised data into a training sample part and a test sample part;
step S3, normalizing the two divided samples;
step S4, initializing LVQ neural network algorithm parameters and QGA algorithm parameters;
step S5, inputting a training sample, and carrying out optimization processing on the weight of the LVQ neural network algorithm parameter through a QGA algorithm;
and step S6, inputting the test sample to the optimized LVQ neural network for traffic flow short-term prediction.
In an embodiment of the present invention, the LVQ neural network includes an input layer, a competition layer and an output layer, wherein the competition layer is responsible for completing classification of input vectors in neurons of the input layer.
In an embodiment of the present invention, the LVQ neural network algorithm is a LVQ1 algorithm, and the implementation steps thereof are as follows:
step 1: initializing weight W between input layer and competition layeri,jAnd learning rate η (η > 0);
step 2: converting the input vector X to (X)1,x2,...,xR)TImporting an input layer, R being the number of input elements, and calculating the distance d between neurons of the competition layer and an input vector:
wherein S islThe number of competitive neurons;
step 3: selecting the neuron with the shortest distance to the input vector if djAt a minimum, the class label that labels the output layer neurons to which it is connected is Cj;
Step 4: let the class label corresponding to the input vector be CxIf C is presentj=CxThen, the weight is adjusted as follows:
Wij-new=Wij-old+η(x-Wij-old)
otherwise, the weight is adjusted as follows:
Wij-new=Wij-old-η(x-Wij-old)
step 5: jump to Step2 and repeat the execution until the set iteration number or the error precision requirement is met.
In an embodiment of the present invention, the step S5 is specifically implemented as follows:
step 1: let genetic algebra t equal to 0, initialize population Q (t)0) And each individual of the initial population is observed once to obtain a state p (t);
step 2: the fitness function of the LVQ neural network is defined according to the distance principle, the average distance from a random individual in a population to a sample point of an input layer is specifically adopted, and the calculation mode is as follows:
wherein, FtRepresenting a set of elements belonging to class k, NkNumber of elements, x, representing class kjA training sample input vector representing an LVQ neural network;
the fitness of the individual is defined as:
step 3: the termination condition of the iterative computation is:
wherein N is the number of sample input vectors;
if D < epsilon, then end;
step 4: and (3) performing observation once on each individual in the population Q (t), then calculating the fitness of each state, updating the population Q (t) by using a quantum rotating gate, recording the best individual and the fitness thereof, and returning to Step 2.
Compared with the prior art, the invention has the following beneficial effects: the method has better accuracy and real-time performance in short-time traffic flow prediction based on the QGA-LVQ neural network, and the reliability of the short-time traffic flow prediction is verified by test results on a plurality of data sets.
Drawings
FIG. 1 is a diagram of dynamic path selection based on an artificial neural network.
Fig. 2 is a coordinate diagram of the quantum revolving door.
Fig. 3 is a view of the structure of the LVQ neural network.
FIG. 4 is a QGA-LVQ neural network-based short-term traffic flow prediction.
Detailed Description
The technical scheme of the invention is specifically explained in the following by combining the attached drawings.
The invention provides a smart city traffic flow prediction method based on a quantum genetic optimization LVQ neural network, which comprises the following steps:
s1, carrying out wavelet noise reduction on the traffic flow data acquired by the ground detector, and removing interference noise signals;
step S2, dividing the denoised data into a training sample part and a test sample part;
step S3, normalizing the two divided samples;
step S4, initializing LVQ neural network algorithm parameters and QGA algorithm parameters;
s5, inputting a training sample, and optimizing the weight of the LVQ neural network algorithm parameter through a QGA algorithm;
and step S6, inputting the test sample to the optimized LVQ neural network for traffic flow short-term prediction.
In this embodiment, the LVQ neural network includes an input layer, a competition layer and an output layer, wherein the competition layer is responsible for completing classification of input vectors in neurons of the input layer.
In this embodiment, the LVQ neural network algorithm is a LVQ1 algorithm, and the implementation steps thereof are as follows:
step 1: initializing weight W between input layer and competition layeri,jAnd learning rate η (η > 0);
step 2: input vector X is equal to (X)1,x2,...,xR)TImporting an input layer, wherein R is the number of input elements, and calculating the distance d between neurons of the competition layer and an input vector:
wherein S islThe number of competitive neurons;
step 3: selecting the neuron with the shortest distance to the input vector if djAt a minimum, the class label that labels the output layer neuron to which it is connected is Cj;
Step 4: let the class label corresponding to the input vector be CxIf C is presentj=CxThen, the weight is adjusted as follows:
Wij-new=Wij-old+η(x-Wij-old)
otherwise, the weight is adjusted as follows:
Wij-new=Wij-old-η(x-Wij-old)
step 5: jump to Step2 and repeat the execution until the set iteration number or the error precision requirement is met.
In this embodiment, the step S5 is specifically implemented as follows:
step 1: let genetic algebra t equal to 0, initialize population Q (t)0) And each individual in the initial population is observed once to obtain a state p (t);
step 2: a fitness function of the LVQ neural network is defined according to a distance principle, the average distance from a random individual in a population to a sample point of an input layer is specifically adopted, and the calculation mode is as follows:
wherein, FtRepresenting a set of elements belonging to class k, NkNumber of elements representing class k, xjA training sample input vector representing an LVQ neural network;
the fitness of the individual is defined as:
step 3: the termination condition of the iterative computation is:
wherein N is the number of sample input vectors;
if D < epsilon, then end;
step 4: and (3) performing observation once on each individual in the population Q (t), then calculating the fitness of each state, updating the population Q (t) by using a quantum rotating gate, recording the best individual and the fitness thereof, and returning to Step 2.
The following explains a specific implementation process of the present invention.
1. Quantum genetic algorithm (Quantum genetic algorithm, QGA)
Compared with the traditional genetic algorithm, the quantum genetic algorithm has the advantages of high convergence rate and strong global optimization capability. Some known quantum algorithms are faster than conventional general purpose computers in processing problems, and thus complete a very complicated solution in a reasonable time.
A) Quantum bit encoding
The definitions |0 > and |1 > in the two-dimensional complex vector space represent two different qubit states, and in addition to the two states, the state of a qubit can be a superposition between the two states. As a minimum unit of information, the state of one qubit can be represented as:
where α and β each represent a complex number, called the probability magnitude of the correlation, and satisfy the condition | α2+|β|2=1。
In the quantum genetic algorithm, chromosomes are encoded by using quantum bits and quantum superposition states, and the encoding mode of each quantum chromosome is as follows:
wherein t represents the population generation number, the quantum population of the t generation is represented asm represents the number of quanta and n represents the population size. Further, the following normalization condition needs to be satisfied:
B) quantum revolving door
As the most basic operation step, qubits use quantum gates to perform state transitions by matrix transformation in order to complete population evolution. Qubit operation typically employs a quantum rotating gate, which is defined as follows:
the population evolution process is as follows:
wherein, thetaiIndicating the rotation angle, theta needs to be specified according to the adjustment ruleiAngle and direction of (d). The coordinate diagram of the quantum rotary gate is shown in fig. 2.
C) Quantum crossing and mutation
The quantum crossing operation is a full interference crossing operation based on the coherent properties of the quantum. For quantum crossing operations, each quantum chromosome in the population needs to be crossed. If the population number is 6 and the chromosome length is 7, a cross-over procedure for diagonal rearrangement and combination is given in Table 2.
TABLE 2 Total interference Cross
In the quantum mutation operation, the quantum genetic algorithm uses a quantum mutation operator U (omega (delta theta)i) To implement update optimization.
ω(Δθi)=f(αi,βi)*Δθi (7)
Wherein, f (alpha)i,βi) Indicating the direction of rotation, Δ θiThe rotation is shown, and delta represents an adjustment factor, which is generally small.
2. Short-term traffic flow prediction basic algorithm based on QGA-LVQ neural network
A) Structure of LVQ neural network
In consideration of the real-time property of short-time traffic flow prediction, an LVQ neural network which is an extension of an SOM self-organizing network is selected to realize the prediction. The LVQ neural network can supervise learning of the clustering centers to classify the input data, and can already approach most of the short-term traffic systems. The short-time traffic flow prediction is performed by using a three-layer LVQ neural network, which is divided into an input layer, a competition layer and an output layer as shown in FIG. 3. Wherein, the competition layer is mainly responsible for completing the classification of the input vectors in the neurons of the input layer.
The LVQ network algorithm is divided into 2 types: LVQ1 and LVQ 2. LVQ1 is more common, so the LVQ1 algorithm is used herein, with the following steps:
step 1: initializing weight W between input layer and competition layeri,jAnd a learning rate η (η > 0);
step 2: input vector X is equal to (X)1,x2,...,xR)TImporting an input layer, R being the number of input elements, and calculating the distance d between neurons of the competition layer and an input vector:
wherein S islThe number of competitive neurons;
step 3: selecting the competition layer neuron with the shortest distance with the input vector if djAt a minimum, the class label that labels the output layer neuron to which it is connected is Cj;
Step 4: let the class label corresponding to the input vector be CxIf C isj=CxThen, the weight is adjusted as follows:
Wij-new=Wij-old+η(x-Wij-old) (9)
otherwise, the weight is adjusted as follows:
Wij-new=Wij-old-η(x-Wij-old) (10)
step 5: and jumping to Step2 to repeat execution until the set iteration number or error precision requirement is met.
B) Predictive algorithm flow
The invention mainly predicts the short-term traffic flow of a single time point, and the research idea is to select an initial value for the LVQ neural network through a quantum genetic algorithm, and finally the population gradually converges to an optimal solution so as to achieve the aim of further improving the classification accuracy. The QGA-LVQ neural network prediction steps are as follows:
step 1: let genetic algebra t equal to 0, initialize population Q (t)0) And each individual of the initial population is observed once to obtain a state p (t);
step 2: a fitness function of the LVQ neural network is defined according to a distance principle, the average distance from a random individual in a population to a sample point of an input layer is specifically adopted, and the calculation mode is as follows:
wherein, FtRepresenting a set of elements belonging to class k, NkNumber of elements representing class k, xjA training sample input vector representing an LVQ neural network;
the fitness of the individual is defined as:
step 3: the termination condition of the iterative computation is:
wherein N is the number of sample input vectors;
if D < epsilon, then end;
step 4: and (3) performing observation once on each individual in the population Q (t), then calculating the fitness of each state, updating the population Q (t) by using a quantum rotating gate, recording the best individual and the fitness thereof, and returning to Step 2.
4. Results and analysis of the experiments
4.1 Experimental data set and environmental parameter configuration
To verify the feasibility of the proposed method, simulation experiments were performed using 5 urban network traffic data sets, wherein data sets dataset1, dataset2 and dataset3 were all from the data of the deluge traffic data research laboratory, university of minnesota, usa, 1440 groups of data and 864 groups of data, respectively. The other two sets of dataset4 and dataset5 are from the PeMS system, and are 864 sets of data. All data were time spaced 5 minutes apart and all traffic flow data was normalized.
The simulation experiment environment is configured as follows: MATLAB 2012a simulation software, Window7 operating system, Intel core i58400CPU, 8GB memory. The experimental parameter configuration is as follows: population size was 50, maximum number of iterations was 200, chromosome length was 10, ε was 0.01, α was 0.5, β was 0.5. The number of the neurons of an input layer, a competition layer and an output layer in the LVQ neural network is 180, 180 multiplied by 5 and 5 respectively, and the learning rate eta is 0.1.
4.2 Experimental evaluation index
To verify the convergence and optimization of the algorithm proposed herein, 4 evaluation indices were used for performance analysis [7-9], including Mean Absolute Percentage Error (MAPE), Equality Coefficient (EC), Mean Absolute Error (MAE), and Root Mean Square Error (RMSE). The specific calculation formula of each index is as follows:
the mean absolute percentage error is:
the equalization coefficients are:
the mean absolute error is:
the root mean square error is:
wherein N represents the number of test samples, Yp(t) represents the predicted output value of the QGA-LVQ neural network at time t, YrAnd (t) is expressed as an actual traffic flow numerical value.
4.3, analysis of results
FIG. 4 is a short-term traffic flow prediction from QGA-LVQ neural networks on dataset1 data sets. Table 3 is the results of 10 runs of short-term traffic flow prediction based on QGA-LVQ neural networks. As can be seen from fig. 4 and table 3, the predicted values of the proposed QGA-LVQ neural network are almost consistent with the actual traffic flow, verifying its feasibility.
TABLE 3 short term traffic flow prediction performance results based on QGA-LVQ neural network
The QGA-LVQ neural network, GA-BP neural network of the present invention were mapped on dataset1 dataset[9]Sum wavelet neural network[8]Comparative experimental analysis was performed as shown in table 4. Compared with other two neural network prediction models, the QGA-LVQ has certain advantages on 5 indexes, and particularly the MAPE is reduced by about 40%. Although the QGA-LVQ neural network has longer running time than the wavelet neural network, the convergence speed is faster because the primitive and the whole structure of the wavelet neural network are determined according to the wavelet analysis theory, but the QGA-LVQ neural network is still shortened by about 7 seconds compared with the GA-BP, and the real-time performance is stronger.
TABLE 4 comparative analysis of the predicted performance results of three neural networks
TABLE 5 prediction of QGA-LVQ neural networks on 5 datasets
Table 5 shows the predicted results of the QGA-LVQ neural network on 5 data sets, each of which was averaged 10 times. Through the analysis of the experimental results, the QGA-LVQ neural network provided by the method is stable on each data set, good in reliability and good in convergence rate and prediction accuracy.
The invention provides a QGA-LVQ neural network-based urban traffic network flow prediction method. By applying the quantum genetic algorithm with better global solution, the problems that the LVQ neural network is sensitive to the initial weight and is easy to fall into a local minimum value are solved, and the convergence speed is improved. On 5 general traffic flow data sets, comparative simulation experiments were performed with traffic flow predictions based on GA-BP neural networks and wavelet neural networks. The result shows that the short-term traffic flow prediction based on the QGA-LVQ neural network has better accuracy and real-time performance, and the reliability of the short-term traffic flow prediction is verified by the test results on a plurality of data sets.
The above are preferred embodiments of the present invention, and all changes made according to the technical scheme of the present invention that produce functional effects do not exceed the scope of the technical scheme of the present invention belong to the protection scope of the present invention.
Claims (3)
1. A smart urban traffic flow prediction method based on a quantum genetic optimization LVQ neural network is characterized by comprising the following steps:
step S1, carrying out wavelet noise reduction on the traffic flow data acquired by the ground detector, and removing interference noise signals;
step S2, dividing the traffic flow data after noise reduction into a traffic flow data training sample part and a traffic flow data testing sample part;
step S3, normalizing the divided two traffic flow data samples;
step S4, initializing LVQ neural network algorithm parameters and QGA algorithm parameters;
s5, inputting traffic flow data training samples, and optimizing weights of LVQ neural network algorithm parameters through a QGA algorithm;
s6, inputting traffic flow data test samples to the optimized LVQ neural network, and predicting the short-time traffic flow of a single time point;
the step S5 is specifically implemented as follows:
step 1: let genetic algebra t equal to 0, initialize population Q (t)0) And each individual in the initial population is observed once to obtain a state p (t);
step 2: a fitness function of the LVQ neural network is defined according to a distance principle, the average distance from a random individual in a population to a sample point of an input layer is specifically adopted, and the calculation mode is as follows:
wherein, FtRepresenting a set of elements belonging to class k, NkNumber of elements representing class k, xjRepresenting a traffic flow data training sample input vector of the LVQ neural network;
the fitness of the individual is defined as:
step 3: the termination condition of the iterative computation is:
wherein N is the number of input vectors of the traffic flow data sample;
if | D | < ε, then end;
step 4: and (3) performing observation once on each individual in the population Q (t), calculating the fitness of each state, updating the population Q (t) by using a quantum revolving gate, recording the best individual and the fitness thereof, and returning to Step 2.
2. The quantum genetic optimization-based smart city traffic flow prediction method based on the LVQ neural network of claim 1, wherein the LVQ neural network comprises an input layer, a competition layer and an output layer, wherein the competition layer is responsible for completing the classification of input vectors in neurons of the input layer.
3. The intelligent city traffic flow prediction method based on the quantum genetic optimization LVQ neural network of claim 1, wherein the LVQ neural network algorithm is LVQ1 algorithm, and the implementation steps are as follows:
step 1: initializing weight W between input layer and competition layerijAnd learning rate eta, eta > 0;
step 2: input vector X is equal to (X)1,x2,...,xR)TImporting an input layer, R being the number of input elements, and calculating the distance d between neurons of the competition layer and the input vectorj:
Wherein S islThe number of competitive neurons;
step 3: selecting the competition layer neuron with the shortest distance with the input vector if djAt a minimum, the class label that labels the output layer neuron to which it is connected is Cj;
Step 4: let the class label corresponding to the input vector be CxIf C isj=CxThen, the weight is adjusted as follows:
Wij-new=Wij-old+η(X-Wij-old)
otherwise, the weight is adjusted as follows:
Wij-new=Wij-old-η(X-Wij-old)
step 5: jump to Step2 and repeat the execution until the set iteration number or the error precision requirement is met.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911085728.8A CN110852505B (en) | 2019-11-08 | 2019-11-08 | Smart city traffic flow prediction method based on quantum genetic optimization LVQ neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911085728.8A CN110852505B (en) | 2019-11-08 | 2019-11-08 | Smart city traffic flow prediction method based on quantum genetic optimization LVQ neural network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110852505A CN110852505A (en) | 2020-02-28 |
CN110852505B true CN110852505B (en) | 2022-07-15 |
Family
ID=69598596
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911085728.8A Active CN110852505B (en) | 2019-11-08 | 2019-11-08 | Smart city traffic flow prediction method based on quantum genetic optimization LVQ neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110852505B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116417150A (en) * | 2023-04-27 | 2023-07-11 | 国科赛赋河北医药技术有限公司 | Gene therapy drug toxicity long-term risk assessment method and application |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106910337A (en) * | 2017-01-19 | 2017-06-30 | 长安大学 | A kind of traffic flow forecasting method based on glowworm swarm algorithm Yu RBF neural |
CN107145941A (en) * | 2017-04-12 | 2017-09-08 | 西北农林科技大学 | The real-time dynamic acquisition method of light requirement based on optimal light quality and photon flux density |
CN108597246A (en) * | 2017-12-11 | 2018-09-28 | 武汉大学 | A method of Path selection real time problems are solved to avoid local congestion |
CN109818775A (en) * | 2018-12-14 | 2019-05-28 | 南昌大学 | Short-term network method for predicting based on adaptive differential evolution algorithm Optimization of Wavelet neural network |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2016127918A1 (en) * | 2015-02-13 | 2016-08-18 | 北京嘀嘀无限科技发展有限公司 | Transport capacity scheduling method and system |
-
2019
- 2019-11-08 CN CN201911085728.8A patent/CN110852505B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106910337A (en) * | 2017-01-19 | 2017-06-30 | 长安大学 | A kind of traffic flow forecasting method based on glowworm swarm algorithm Yu RBF neural |
CN107145941A (en) * | 2017-04-12 | 2017-09-08 | 西北农林科技大学 | The real-time dynamic acquisition method of light requirement based on optimal light quality and photon flux density |
CN108597246A (en) * | 2017-12-11 | 2018-09-28 | 武汉大学 | A method of Path selection real time problems are solved to avoid local congestion |
CN109818775A (en) * | 2018-12-14 | 2019-05-28 | 南昌大学 | Short-term network method for predicting based on adaptive differential evolution algorithm Optimization of Wavelet neural network |
Non-Patent Citations (2)
Title |
---|
Application of Quantum Genetic Optimization of LVQ Neural Network in Smart City Traffic Network Prediction;Fuquan Zhang 等;《IEEE Access》;20200603;第8卷;全文 * |
基于AGA-LVQ神经网络的软件可靠性预测模型研究;乔辉 等;《计算机科学》;20130131;第40卷(第1期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN110852505A (en) | 2020-02-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Yang et al. | Wavelet neural network with improved genetic algorithm for traffic flow time series prediction | |
Xiang et al. | A clustering-based surrogate-assisted multiobjective evolutionary algorithm for shelter location problem under uncertainty of road networks | |
Yu et al. | Error correction method based on data transformational GM (1, 1) and application on tax forecasting | |
Feng et al. | Parallel cooperation search algorithm and artificial intelligence method for streamflow time series forecasting | |
Lv et al. | [Retracted] An Economic Forecasting Method Based on the LightGBM‐Optimized LSTM and Time‐Series Model | |
CN110929958A (en) | Short-term traffic flow prediction method based on deep learning parameter optimization | |
Tian et al. | An adaptive ensemble predictive strategy for multiple scale electrical energy usages forecasting | |
Peng et al. | An automatic hyperparameter optimization DNN model for precipitation prediction | |
Mi et al. | A dynamic ensemble deep deterministic policy gradient recursive network for spatiotemporal traffic speed forecasting in an urban road network | |
Chen et al. | The optimal design and application of LSTM neural network based on the hybrid coding PSO algorithm | |
Ma et al. | Fuzzy hybrid framework with dynamic weights for short‐term traffic flow prediction by mining spatio‐temporal correlations | |
CN110852505B (en) | Smart city traffic flow prediction method based on quantum genetic optimization LVQ neural network | |
CN114708479A (en) | Self-adaptive defense method based on graph structure and characteristics | |
CN114461791A (en) | Social text sentiment analysis system based on deep quantum neural network | |
CN108631817A (en) | A method of Frequency Hopping Signal frequency range prediction is carried out based on time frequency analysis and radial neural network | |
Xing et al. | RL-GCN: Traffic flow prediction based on graph convolution and reinforcement learning for smart cities | |
Qu et al. | Improving parking occupancy prediction in poor data conditions through customization and learning to learn | |
Mead | Hybrid CNN and LSTM Model (HCLM) for Short-Term Traffic Volume Prediction | |
Chang et al. | Enhanced road information representation in graph recurrent network for traffic speed prediction | |
Huo et al. | Tensor-based cooperative control for large scale multi-intersection traffic signal using deep reinforcement learning and imitation learning | |
Li et al. | A data-driven rutting depth short-time prediction model with metaheuristic optimization for asphalt pavements based on RIOHTrack | |
Bai | Aga-lstm: An optimized lstm neural network model based on adaptive genetic algorithm | |
Liu et al. | Research on routing optimisation of logistics distribution vehicle based on cloud model | |
Wang et al. | A proactive decision support system for online event streams | |
Poczeta et al. | Structure optimization and learning of fuzzy cognitive map with the use of evolutionary algorithm and graph theory metrics |
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 |