CN112215265A - Method and device for determining quantitative relation between passenger transport hub shift and passenger flow - Google Patents
Method and device for determining quantitative relation between passenger transport hub shift and passenger flow Download PDFInfo
- Publication number
- CN112215265A CN112215265A CN202011016362.1A CN202011016362A CN112215265A CN 112215265 A CN112215265 A CN 112215265A CN 202011016362 A CN202011016362 A CN 202011016362A CN 112215265 A CN112215265 A CN 112215265A
- Authority
- CN
- China
- Prior art keywords
- data
- neurons
- passenger
- algorithm
- hidden layer
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/251—Fusion techniques of input or preprocessed data
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Traffic Control Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a method and a device for determining the quantitative relation between passenger transport hub shift and passenger flow, comprising the following steps: step 10, preprocessing data; step 20, setting network parameters; step 30, calculating the number of the optimal neurons of the hidden layer; step 40: and outputting the passenger transport hub shift and passenger flow quantitative relational expression, and restoring the output data into actual data. The method can be used for simulating based on mass passenger transport hub data, and compared with the traditional data fitting method, the method has the advantages of higher efficiency and more accurate result: based on the processed passenger transport hub data, a calculation method of the number of the neurons in the hidden layer is provided, the number of the neurons in the hidden layer is measured by increasing the operation times of the algorithm and based on the concept of the fitting coefficient, the influence of random errors of the operation of the algorithm on the result is reduced, and the accuracy of the simulation result of the algorithm is improved.
Description
Technical Field
The invention relates to a method and a device for determining a quantitative relation model between passenger transport hub shift and passenger flow, belonging to the field of passenger transport and data fusion.
Background
The passenger transport hub is usually located in the national or regional political, economic and cultural center, is an important component of a transportation system, is a junction of transportation lines of a transportation network with different transportation modes, is an important place for passenger flow distribution, and has a promoting effect on the connection between regions and the development of cities. With the rapid development of economy in China, the scale of the passenger transportation hub is continuously enlarged, the number of the passenger transportation hub is continuously increased, the passenger flow volume is increasingly expanded, and certain pressure is brought to the sustainable development of the passenger transportation hub. The traditional passenger transport hub comprises a railway hub, a highway hub, an aviation hub, a public transportation hub, a track transfer hub and the like, the passenger transport hub has various types, and how to quickly and accurately find the quantitative relation between the hub shift and the passenger flow becomes a research hotspot in recent years. Different passenger transport hubs have different positions and functions in cities, and have larger passenger flow space-time distribution characteristics and flow difference. Aiming at a certain specific passenger transport hub type, the traditional data fitting method, such as linear fitting, polynomial fitting, logarithmic fitting, power function fitting and the like, is difficult to accurately reflect the quantitative relation between the shift and the passenger flow; aiming at different types of passenger transportation hubs, different methods are needed to research the quantitative relation between the shift and the passenger flow, and the method has the defects of low fitting precision and high labor cost.
With the rapid development of machine learning, a large number of artificial intelligence algorithms, such as a random forest algorithm, a K nearest neighbor algorithm, a markov algorithm, a bayesian algorithm, a neural network algorithm and the like, have good performances of rapidness, accuracy and the like in solving complex problems, and are widely applied in various fields. The BP neural network algorithm is a supervised learning algorithm, the basic structure of the BP neural network algorithm is composed of nonlinear change units, the BP neural network algorithm has strong nonlinear mapping capability, and can approach any function theoretically. Parameters such as the number of layers of the network, the number of processing units of each layer, the learning coefficient of the network and the like can be set according to specific conditions, and the method has the characteristics of high flexibility, accurate learning result and the like.
In summary, in the current research on the quantitative relationship between the shift of different types of passenger transportation hubs and the passenger flow, the conventional method cannot accurately represent the relationship between the shift of different types of passenger transportation hubs and the passenger flow, and is not beneficial to the sustainable development of the passenger transportation hubs. In the current big data era, based on massive passenger transport hub data, the quantitative relation between the shift of different types of passenger transport hubs and passenger flow is learned through a set of general artificial intelligence algorithm, and basis and convenience can be provided for the construction, development and improvement of the passenger transport hubs. The BP neural network algorithm is an important component of an artificial intelligence algorithm, and can obtain the quantitative relation between the hub shift and the passenger flow by training and learning in a human thinking simulation mode based on mass passenger transport hub data.
Disclosure of Invention
The invention aims to provide a method and a device for determining the quantitative relation between a passenger transportation hub shift and passenger flow based on an improved BP neural network, so as to solve the problems of high efficiency, universality and accuracy of solving the quantitative relation between the passenger transportation hub shift and the passenger flow.
The invention discloses a method for determining the quantitative relation between passenger transport hub shift and passenger flow based on an improved BP neural network, which is characterized by comprising the following steps:
step 10, data preprocessing, specifically: based on the original passenger transport hub shift and passenger flow data, carrying out normalization processing on the data by using a minimum and maximum standardization method;
step 20, setting network parameters, wherein the network parameters comprise a network layer number, a network transfer function, an algorithm training function and an algorithm stopping condition;
step 30, calculating the number of the best neurons of the hidden layer, specifically: initializing the range of the number of the neurons according to an empirical method, and setting the maximum operation times of the algorithm under each neuron number; the passenger transport hub shift and passenger flow data are in one-to-one correspondence, and the data are randomly divided into experimental data and test data; constructing a neural network based on experimental data, and calculating a fitting coefficient based on a residual square sum and a total dispersion square sum between actual output data and expected output data of test data; calculating the mean value of the fitting coefficient by using an arithmetic mean method aiming at each neuron, and calculating the variance of the fitting coefficient based on the mean value; setting the minimum value of the fitting coefficient mean value, and screening and reserving the number of the neurons with the fitting coefficient mean value larger than the minimum value; respectively carrying out normalization processing on the mean value and the variance of the fitting coefficient corresponding to the reserved neuron number by adopting a mean value normalization method; distributing weights for the mean value and the variance of the processed fitting coefficients, and setting the weighted sum of the two parts as an evaluation function value of the neuron; evaluating the number of the neurons corresponding to the maximum value of the function value as the optimal number of the neurons of the hidden layer;
step 40: outputting a passenger transport hub shift and passenger flow quantitative relational expression, and restoring output data into actual data, specifically: based on the set network layer number, the network transfer function, the algorithm training function, the algorithm stopping condition, the number of the hidden layer optimal neurons and the preprocessed passenger transport hub shift and passenger flow data, operating the BP neural network algorithm and outputting a passenger transport hub shift and passenger flow quantitative relational expression; based on the data preprocessing method, the output data is restored to actual data by using an inverse normalization method.
Preferably, in the step 10, the normalization process is performed by using the following formula:
wherein x' represents passenger transport hub shift data after normalization, x represents passenger transport hub shift data before normalization, and xminRepresenting minimum passenger hub shift data, x, before normalizationmaxRepresenting maximum passenger hub shift data before normalization, y' representing passenger flow data after normalization, y representing passenger flow data before normalization, yminRepresents the minimum traffic data before normalization, ymaxRepresenting the maximum guest data before normalization.
Preferably, in said step 20,
for the number of network layers: the input layer, the hidden layer and the output layer are all set to be single layers, and the number of network layers is 3;
for the network transfer function: and (3) adopting a Logsig function as a transfer function between the input layer and the hidden layer, wherein the value range of an output value is (0, 1):
a Purelin function is adopted as a transfer function between the hidden layer and the output layer, where w represents a weight, b represents a threshold:
y '═ w logsig (x') + b (equation four)
Training a function for the algorithm: adopting a tranlmm function as a training function of the algorithm;
for the algorithm stop condition: taking the maximum iteration time ite, the training target tar and the verification time ver as judgment basis for stopping the algorithm, and stopping the algorithm when one of the conditions meets the requirement; otherwise, continuing to circulate; when the running times of the algorithm reach the maximum iteration time ite, stopping the algorithm; when the Mean Square Error (MSE) between actual output data and expected output data is smaller than a preset training target tar, stopping the algorithm; and when the verification times of the algorithm reach the preset verification times ver, stopping the algorithm.
Preferably, the step 30 specifically includes the following steps:
step 301, initialize the range of hidden layer neuron number [ L ] according to experience1，L2]The number of the hidden layer neurons is related to the number of the input layer neurons and the number of the output layer neurons, and the calculation method comprises the following steps:
wherein, l represents the number of neurons in the hidden layer, in represents the number of neurons in the input layer, out represents the number of neurons in the output layer, and a represents a random integer in a certain range [ int1, int2 ];
step 302, setting the maximum operation times N of the algorithm under each neuron number;
step 303, based on the preprocessed passenger transportation hub shift and passenger flow data, corresponding the passenger transportation hub shift to the passenger flow data one, randomly dividing per1 percent of the passenger transportation hub shift and the passenger flow data into experimental data, and dividing per2 percent of the passenger transportation hub shift and the passenger flow data into test data, wherein:
per1+ per2 ═ 100 (formula six)
Step 304, training a neural network based on experimental data, inputting test data to calculate and record a fitting coefficient between actual output data and expected output data of an output layer based on the trained neural network:
wherein j represents an index of the number of times of operation of the algorithm,
representing the fitting coefficient of the j-th operation result under the number of l neurons in the hidden layer, I representing the number of test data, I representing the index of the test data, and outputiIndicates the ith actual output data, targetiIndicating the i-th desired output data,
representing a desired output data mean;
increasing the arithmetic operation times until the maximum operation times N is reached, and calculating and recording the mean value and the variance of all fitting coefficients:
wherein the content of the first and second substances,
represents the mean value of the fitting coefficient under the number of l neurons in the hidden layer,
representing the variance of the fitting coefficient under the number of l neurons in the hidden layer, and J representing the operation times of the algorithm;
step 305, at [ L1，L2]Traversing the number of neurons in the hidden layer within the range, repeating the training process and the testing process in the step 304, and calculating and recording the mean value and the variance of the fitting coefficients corresponding to the number of the neurons in the hidden layer;
step 306, setting the minimum Min of the fitting coefficient mean value, and screening and reserving the number of the neurons with the fitting coefficient mean value larger than the minimum Min;
step 307, normalizing the mean and variance of the fitting coefficients corresponding to the number of the reserved neurons by using a mean normalization method:
wherein R islExpressing the mean value of the normalized fitting coefficient under the number of l neurons in the hidden layer, SlRepresents the normalized fitting coefficient variance under the number of l neurons in the hidden layer,
representing the number of neurons remaining after the screening of step 306;
step 308, setting an evaluation function based on the normalized mean and variance of the fitting coefficients, and calculating the evaluation function value of the number of neurons in the hidden layer:
f(Rl)＝βRl+γSl(formula twelve)
β + γ ═ 1 (thirteen formula)
Wherein, f (R)l) Expressing evaluation function values corresponding to the number of l neurons in the hidden layer, wherein beta and gamma respectively express the weight of the mean value and the variance of the normalized fitting coefficient;
step 309, calculating the evaluation function value of each hidden layer neuron, comparing the evaluation function values corresponding to different neuron numbers of the hidden layer, and the neuron number with the maximum evaluation function value is the optimum neuron number l of the hidden layerb。
Preferably, the step 40 specifically includes the following steps:
step 401, based on the set number of network layers, network transfer function, algorithm training function, algorithm stopping condition, number of hidden layer optimal neurons and the preprocessed passenger transport hub shift and passenger flow data, simulating by using a BP neural network algorithm, and outputting a passenger transport hub shift and passenger flow quantitative relational expression:
wherein y' represents the output passenger flow data, p represents the index of the number of the neurons in the hidden layer, and wpRepresenting the connection weight of the input layer to the pth neuron of the hidden layer, bpRepresenting the threshold of connection of the input layer to the pth neuron of the hidden layer, WpRepresenting the connection weight from the pth neuron of the hidden layer to the output layer, and B representing the connection threshold from the hidden layer to the output layer;
step 402, performing inverse normalization on the output data based on the normalization method of the minimum-maximum normalization, and reducing the output data into actual data:
y″′＝y″(ymax-ymin)+ymin(formula fifteen)
Wherein y' ″ represents the actual passenger flow data corresponding to the output passenger flow data after inverse normalization.
The invention discloses a device for determining the quantitative relation between passenger transport hub shift and passenger flow based on an improved BP neural network, which is characterized by comprising the following components:
the data preprocessing module is used for preprocessing data, and specifically comprises: based on the original passenger transport hub shift and passenger flow data, carrying out normalization processing on the data by using a minimum and maximum standardization method;
the network parameter setting module is used for setting network parameters, wherein the network parameters comprise a network layer number, a network transfer function, an algorithm training function and an algorithm stopping condition;
the hidden layer optimal neuron number calculating module is used for calculating the number of the hidden layer optimal neurons, and specifically comprises the following steps: initializing the range of the number of the neurons according to an empirical method, and setting the maximum operation times of the algorithm under each neuron number; the passenger transport hub shift and passenger flow data are in one-to-one correspondence, and the data are randomly divided into experimental data and test data; constructing a neural network based on experimental data, and calculating a fitting coefficient based on a residual square sum and a total dispersion square sum between actual output data and expected output data of test data; calculating the mean value of the fitting coefficient by using an arithmetic mean method aiming at each neuron, and calculating the variance of the fitting coefficient based on the mean value; setting the minimum value of the fitting coefficient mean value, and screening and reserving the number of the neurons with the fitting coefficient mean value larger than the minimum value; respectively carrying out normalization processing on the mean value and the variance of the fitting coefficient corresponding to the reserved neuron number by adopting a mean value normalization method; distributing weights for the mean value and the variance of the processed fitting coefficients, and setting the weighted sum of the two parts as an evaluation function value of the neuron; evaluating the number of the neurons corresponding to the maximum value of the function value as the optimal number of the neurons of the hidden layer;
the output and reduction module is used for outputting a passenger transport hub shift and passenger flow quantitative relational expression and reducing the output data into actual data, and specifically comprises the following steps: based on the set network layer number, the network transfer function, the algorithm training function, the algorithm stopping condition, the number of the hidden layer optimal neurons and the preprocessed passenger transport hub shift and passenger flow data, operating the BP neural network algorithm and outputting a passenger transport hub shift and passenger flow quantitative relational expression; based on the data preprocessing method, the output data is restored to actual data by using an inverse normalization method.
The invention discloses a system for determining the quantitative relation between passenger transport hub shift and passenger flow based on an improved BP neural network, which is characterized by comprising the following steps: a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method of any one of the above when executing the program.
The computer-readable storage medium of the present invention, on which a computer program is stored, is characterized in that the program, when executed by a processor, implements the method of any one of the above.
Compared with the prior art, the invention has the advantages that: the method and the device for determining the quantitative relation between the passenger transport hub shift and the passenger flow based on the improved BP neural network can be used for simulating based on mass passenger transport hub data, and compared with a traditional data fitting method, the method and the device have the advantages that the efficiency is higher and the result is more accurate by utilizing mass data: based on the processed passenger transport hub data, a calculation method of the number of the neurons in the hidden layer is provided, the number of the neurons in the hidden layer is measured by increasing the operation times of the algorithm and based on the concept of the fitting coefficient, the influence of random errors of the operation of the algorithm on the result is reduced, and the accuracy of the simulation result of the algorithm is improved. In addition, the method has strong universality and is suitable for calculating the quantitative relation between the shift and the passenger flow of different passenger transport hubs.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
Fig. 1 is a flow chart of a method for determining the quantitative relationship between the shift of the passenger transportation hub and the passenger flow based on the improved BP neural network.
Detailed Description
Example one
As shown in fig. 1, the method for determining the quantitative relationship between the shift of the passenger transportation hub and the passenger flow based on the improved BP neural network of the present invention comprises the following steps:
step 10, data preprocessing, specifically: based on the original passenger transport hub shift and passenger flow data, the data is normalized by using a minimum and maximum normalization method.
Wherein, in the step 10, the normalization processing is performed by using the following formula:
wherein x' represents passenger transport hub shift data after normalization, x represents passenger transport hub shift data before normalization, and xminRepresenting minimum passenger hub shift data, x, before normalizationmaxRepresenting maximum passenger hub shift data before normalization, y' representing passenger flow data after normalization, y representing passenger flow data before normalization, yminRepresents the minimum traffic data before normalization, ymaxRepresenting the maximum guest data before normalization.
And 20, setting network parameters, wherein the network parameters comprise a network layer number, a network transfer function, an algorithm training function and an algorithm stopping condition.
Wherein, for the number of network layers: the input layer, the hidden layer and the output layer are all set to be single layers, and the number of network layers is 3;
for the network transfer function: and (3) adopting a Logsig function as a transfer function between the input layer and the hidden layer, wherein the value range of an output value is (0, 1):
a Purelin function is adopted as a transfer function between the hidden layer and the output layer, where w represents a weight, b represents a threshold:
y '═ w logsig (x') + b (equation four)
Training a function for the algorithm: adopting a tranlmm function as a training function of the algorithm;
for the algorithm stop condition: taking the maximum iteration time ite, the training target tar and the verification time ver as judgment basis for stopping the algorithm, and stopping the algorithm when one of the conditions meets the requirement; otherwise, continuing to circulate; when the running times of the algorithm reach the maximum iteration time ite, stopping the algorithm; when the Mean Square Error (MSE) between actual output data and expected output data is smaller than a preset training target tar, stopping the algorithm; and when the verification times of the algorithm reach the preset verification times ver, stopping the algorithm.
Step 30, calculating the number of the best neurons of the hidden layer, specifically: initializing the range of the number of the neurons according to an empirical method, and setting the maximum operation times of the algorithm under each neuron number; the passenger transport hub shift and passenger flow data are in one-to-one correspondence, and the data are randomly divided into experimental data and test data; constructing a neural network based on experimental data, and calculating a fitting coefficient based on a residual square sum and a total dispersion square sum between actual output data and expected output data of test data; calculating the mean value of the fitting coefficient by using an arithmetic mean method aiming at each neuron, and calculating the variance of the fitting coefficient based on the mean value; setting the minimum value of the fitting coefficient mean value, and screening and reserving the number of the neurons with the fitting coefficient mean value larger than the minimum value; respectively carrying out normalization processing on the mean value and the variance of the fitting coefficient corresponding to the reserved neuron number by adopting a mean value normalization method; distributing weights for the mean value and the variance of the processed fitting coefficients, and setting the weighted sum of the two parts as an evaluation function value of the neuron; and the number of the neurons corresponding to the maximum value of the evaluation function value is the optimal number of the neurons of the hidden layer.
Wherein, the step 30 specifically comprises the following steps:
step 301, initialize the range of hidden layer neuron number [ L ] according to experience1，L2]The number of the hidden layer neurons is related to the number of the input layer neurons and the number of the output layer neurons, and the calculation method comprises the following steps:
wherein, l represents the number of neurons in the hidden layer, in represents the number of neurons in the input layer, out represents the number of neurons in the output layer, and a represents a random integer in a certain range [ int1, int2 ];
step 302, setting the maximum operation times N of the algorithm under each neuron number;
step 303, based on the preprocessed passenger transportation hub shift and passenger flow data, corresponding the passenger transportation hub shift and the passenger flow data one to one, randomly dividing per1 percent of the passenger transportation hub shift and the passenger flow data into experimental data, and dividing per2 percent of the passenger transportation hub shift and the passenger flow data into test data, wherein:
per1+ per2 ═ 100 (formula six)
Step 304, training a neural network based on experimental data, inputting test data to calculate and record a fitting coefficient between actual output data and expected output data of an output layer based on the trained neural network:
wherein j represents an index of the number of times of operation of the algorithm,
representing the fitting coefficient of the j-th operation result under the number of l neurons in the hidden layer, I representing the number of test data, I representing the index of the test data, and outputiIndicates the ith actual output data, targetiIndicating the i-th desired output data,
representing a desired output data mean;
increasing the arithmetic operation times until the maximum operation times N is reached, and calculating and recording the mean value and the variance of all fitting coefficients:
wherein the content of the first and second substances,
represents the mean value of the fitting coefficient under the number of l neurons in the hidden layer,
determining the variance of the fitting coefficient under the number of l neurons in the hidden layer, wherein J represents the operation times of the algorithm;
step 305, at [ L1，L2]Traversing the number of neurons in the hidden layer within the range, repeating the training process and the testing process in the step 304, and calculating and recording the mean value and the variance of the fitting coefficients corresponding to the number of the neurons in the hidden layer;
step 306, setting the minimum Min of the fitting coefficient mean value, and screening and reserving the number of the neurons with the fitting coefficient mean value larger than the minimum Min;
step 307, normalizing the mean and variance of the fitting coefficients corresponding to the number of the reserved neurons by using a mean normalization method:
wherein R islExpressing the mean value of the normalized fitting coefficient under the number of l neurons in the hidden layer, SlRepresents the normalized fitting coefficient variance under the number of l neurons in the hidden layer,
representing the number of neurons remaining after the screening of step 306;
step 308, setting an evaluation function based on the normalized mean and variance of the fitting coefficients, and calculating the evaluation function value of the number of neurons in the hidden layer:
f(Rl)＝βRl+γSl(formula twelve)
β + γ ═ 1 (thirteen formula)
Wherein, f (R)l) Expressing evaluation function values corresponding to the number of l neurons in the hidden layer, wherein beta and gamma respectively express the weight of the mean value and the variance of the normalized fitting coefficient;
step 309, calculating the evaluation function value of each hidden layer neuron, comparing the evaluation function values corresponding to different neuron numbers of the hidden layer, and the neuron number with the maximum evaluation function value is the optimum neuron number l of the hidden layerb。
Step 40: outputting a passenger transport hub shift and passenger flow quantitative relational expression, and restoring output data into actual data, specifically: based on the set network layer number, the network transfer function, the algorithm training function, the algorithm stopping condition, the number of the hidden layer optimal neurons and the preprocessed passenger transport hub shift and passenger flow data, operating the BP neural network algorithm and outputting a passenger transport hub shift and passenger flow quantitative relational expression; based on the data preprocessing method, the output data is restored to actual data by using an inverse normalization method.
Wherein, the step 40 specifically comprises the following steps:
step 401, based on the set number of network layers, network transfer function, algorithm training function, algorithm stopping condition, number of hidden layer optimal neurons and the preprocessed passenger transport hub shift and passenger flow data, simulating by using a BP neural network algorithm, and outputting a passenger transport hub shift and passenger flow quantitative relational expression:
wherein y' represents the output passenger flow data, p represents the index of the number of the neurons in the hidden layer, and wpRepresenting the connection weight of the input layer to the pth neuron of the hidden layer, bpRepresenting the threshold of connection of the input layer to the pth neuron of the hidden layer, WpRepresenting the connection weight from the pth neuron of the hidden layer to the output layer, and B representing the connection threshold from the hidden layer to the output layer;
step 402, performing inverse normalization on the output data based on the normalization method of the minimum-maximum normalization, and reducing the output data into actual data:
y″′＝y″(ymax-ymin)+ymin(formula fifteen)
Wherein y' ″ represents the actual passenger flow data corresponding to the output passenger flow data after inverse normalization.
Example two
The invention also provides a device for determining the quantitative relation between the passenger transport hub shift and the passenger flow based on the improved BP neural network, which comprises the following steps:
the data preprocessing module is used for preprocessing data, and specifically comprises: based on the original passenger transport hub shift and passenger flow data, carrying out normalization processing on the data by using a minimum and maximum standardization method;
the network parameter setting module is used for setting network parameters, wherein the network parameters comprise a network layer number, a network transfer function, an algorithm training function and an algorithm stopping condition;
the hidden layer optimal neuron number calculating module is used for calculating the number of the hidden layer optimal neurons, and specifically comprises the following steps: initializing the range of the number of the neurons according to an empirical method, and setting the maximum operation times of the algorithm under each neuron number; the passenger transport hub shift and passenger flow data are in one-to-one correspondence, and the data are randomly divided into experimental data and test data; constructing a neural network based on experimental data, and calculating a fitting coefficient based on a residual square sum and a total dispersion square sum between actual output data and expected output data of test data; calculating the mean value of the fitting coefficient by using an arithmetic mean method aiming at each neuron, and calculating the variance of the fitting coefficient based on the mean value; setting the minimum value of the fitting coefficient mean value, and screening and reserving the number of the neurons with the fitting coefficient mean value larger than the minimum value; respectively carrying out normalization processing on the mean value and the variance of the fitting coefficient corresponding to the reserved neuron number by adopting a mean value normalization method; distributing weights for the mean value and the variance of the processed fitting coefficients, and setting the weighted sum of the two parts as an evaluation function value of the neuron; evaluating the number of the neurons corresponding to the maximum value of the function value as the optimal number of the neurons of the hidden layer;
the output and reduction module is used for outputting a passenger transport hub shift and passenger flow quantitative relational expression and reducing the output data into actual data, and specifically comprises the following steps: based on the set network layer number, the network transfer function, the algorithm training function, the algorithm stopping condition, the number of the hidden layer optimal neurons and the preprocessed passenger transport hub shift and passenger flow data, operating the BP neural network algorithm and outputting a passenger transport hub shift and passenger flow quantitative relational expression; based on the data preprocessing method, the output data is restored to actual data by using an inverse normalization method.
EXAMPLE III
The invention also provides a system for determining the quantitative relation between the passenger transport hub shift and the passenger flow based on the improved BP neural network, which is characterized by comprising the following steps: a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method of any one of the above when executing the program.
Example four
The invention also proposes a computer-readable storage medium, on which a computer program is stored, characterized in that the program, when executed by a processor, implements the method of any one of the preceding claims.
In order to more intuitively explain the flow and the result of the research on the quantitative relation between the shift and the passenger flow of the passenger transport hub by using the method and the device, the specific implementation process is explained by taking the flight and passenger flow data of the capital international airport as an example.
The specific contents are as follows:
1. and (4) preprocessing flight and passenger flow data.
And counting flight and passenger flow data of the capital international airport for 3 months by taking half an hour as a unit, and respectively carrying out normalization processing on the flight and passenger flow data.
| Partially normalized flight data | Partially normalized passenger flow data |
| 0.264151 | 0.285213 |
| 0.396226 | 0.623731 |
| 0.339623 | 0.506648 |
| 0.09434 | 0.037813 |
| 0.339623 | 0.492659 |
| 0.188679 | 0.218474 |
| 0.132075 | 0.080715 |
| 0.433962 | 0.543359 |
2. And setting parameters of the BP neural network.
Setting 4 parameters: network layer number, network transfer function, algorithm training function and algorithm stopping condition.
3. And calculating the optimal neuron number of the hidden layer.
The number of neurons of the input layer is 1, the number of neurons of the output layer is 1, the range of random integers is set to be [1,10], and the range of the number of neurons of the hidden layer is set to be [3,12 ]; taking 90% of the preprocessed data as experimental data and 10% of the preprocessed data as test data, running the algorithm for 100 times under each neuron number, and recording the mean and variance of the fitting coefficient between the actual output data and the expected output data of the test data; keeping the number of neurons with the fitting coefficient mean value larger than 0.5, respectively carrying out normalization processing on the fitting coefficient mean value and the variance corresponding to the number of the kept neurons, distributing weight 0.5 to the normalized mean value and distributing weight 0.5 to the normalized variance, and carrying out weighted summation to obtain an evaluation function value;
| number of neurons in hidden layer | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 |
| Evaluation function value | 0.618 | 1.202 | 1.578 | 2.194 | 2.194 | 0.530 | 0.629 | 0.507 | 0.921 | 0.539 |
The optimal number of neurons is 7.
4. And (4) outputting the flight and passenger flow quantitative relational expression and restoring the output data into actual data.
Inputting the preprocessed flight and passenger flow data, network parameters and the number of the hidden layer optimal neurons, and operating a BP neural network algorithm to obtain a flight and passenger flow quantitative relation expression.
And restoring the passenger flow data output by the flight and passenger flow quantitative relational expression into actual passenger flow data.
It should be understood that the above-mentioned embodiments are merely preferred examples of the present invention, and not restrictive, but rather, all the changes, substitutions, alterations and modifications that come within the spirit and scope of the invention as described above may be made by those skilled in the art, and all the changes, substitutions, alterations and modifications that fall within the scope of the appended claims should be construed as being included in the present invention.
Claims (8)
1. A method for determining the quantitative relation between passenger transport hub shift and passenger flow based on an improved BP neural network is characterized by comprising the following steps:
step 10, data preprocessing, specifically: based on the original passenger transport hub shift and passenger flow data, carrying out normalization processing on the data by using a minimum and maximum standardization method;
step 20, setting network parameters, wherein the network parameters comprise a network layer number, a network transfer function, an algorithm training function and an algorithm stopping condition;
step 30, calculating the number of the best neurons of the hidden layer, specifically: initializing the range of the number of the neurons according to an empirical method, and setting the maximum operation times of the algorithm under each neuron number; the passenger transport hub shift and passenger flow data are in one-to-one correspondence, and the data are randomly divided into experimental data and test data; constructing a neural network based on experimental data, and calculating a fitting coefficient based on a residual square sum and a total dispersion square sum between actual output data and expected output data of test data; calculating the mean value of the fitting coefficient by using an arithmetic mean method aiming at each neuron, and calculating the variance of the fitting coefficient based on the mean value; setting the minimum value of the fitting coefficient mean value, and screening and reserving the number of the neurons with the fitting coefficient mean value larger than the minimum value; respectively carrying out normalization processing on the mean value and the variance of the fitting coefficient corresponding to the reserved neuron number by adopting a mean value normalization method; distributing weights for the mean value and the variance of the processed fitting coefficients, and setting the weighted sum of the two parts as an evaluation function value of the neuron; evaluating the number of the neurons corresponding to the maximum value of the function value as the optimal number of the neurons of the hidden layer;
step 40: outputting a passenger transport hub shift and passenger flow quantitative relational expression, and restoring output data into actual data, specifically: based on the set network layer number, the network transfer function, the algorithm training function, the algorithm stopping condition, the number of the hidden layer optimal neurons and the preprocessed passenger transport hub shift and passenger flow data, operating the BP neural network algorithm and outputting a passenger transport hub shift and passenger flow quantitative relational expression; based on the data preprocessing method, the output data is restored to actual data by using an inverse normalization method.
2. The method for determining the quantitative relationship between the shift and the passenger flow of the passenger transportation hub based on the improved BP neural network as claimed in claim 1, wherein: in the step 10, the normalization process is performed using the following formula:
wherein x' represents the passenger transport hub shift data after normalization, and x represents the normalized dataPrevious passenger transport hub shift data, xminRepresenting minimum passenger hub shift data, x, before normalizationmaxRepresenting maximum passenger hub shift data before normalization, y' representing passenger flow data after normalization, y representing passenger flow data before normalization, yminRepresents the minimum traffic data before normalization, ymaxRepresenting the maximum guest data before normalization.
3. The method for determining the quantitative relationship between the shift and the passenger flow of the passenger transportation hub based on the improved BP neural network as claimed in claim 1, wherein: in the step 20, it is assumed that,
for the number of network layers: the input layer, the hidden layer and the output layer are all set to be single layers, and the number of network layers is 3;
for the network transfer function: and (3) adopting a Logsig function as a transfer function between the input layer and the hidden layer, wherein the value range of an output value is (0, 1):
a Purelin function is adopted as a transfer function between the hidden layer and the output layer, where w represents a weight, b represents a threshold:
y '═ w logsig (x') + b (equation four)
Training a function for the algorithm: adopting a tranlmm function as a training function of the algorithm;
for the algorithm stop condition: taking the maximum iteration time ite, the training target tar and the verification time ver as judgment basis for stopping the algorithm, and stopping the algorithm when one of the conditions meets the requirement; otherwise, continuing to circulate; when the running times of the algorithm reach the maximum iteration time ite, stopping the algorithm; when the Mean Square Error (MSE) between actual output data and expected output data is smaller than a preset training target tar, stopping the algorithm; and when the verification times of the algorithm reach the preset verification times ver, stopping the algorithm.
4. The method for determining the quantitative relationship between the shift of passenger transportation hub and the passenger flow based on the improved BP neural network as claimed in claim 1, wherein said step 30 specifically comprises the steps of:
step 301, initialize the range of hidden layer neuron number [ L ] according to experience1，L2]The number of the hidden layer neurons is related to the number of the input layer neurons and the number of the output layer neurons, and the calculation method comprises the following steps:
wherein, l represents the number of neurons in the hidden layer, in represents the number of neurons in the input layer, out represents the number of neurons in the output layer, and a represents a random integer in a certain range [ int1, int2 ];
step 302, setting the maximum operation times N of the algorithm under each neuron number;
step 303, based on the preprocessed passenger transportation hub shift and passenger flow data, corresponding the passenger transportation hub shift and the passenger flow data one to one, randomly dividing per1 percent of the passenger transportation hub shift and the passenger flow data into experimental data, and dividing per2 percent of the passenger transportation hub shift and the passenger flow data into test data, wherein:
per1+ per2 ═ 100 (formula six)
Step 304, training a neural network based on experimental data, inputting test data to calculate and record a fitting coefficient between actual output data and expected output data of an output layer based on the trained neural network:
wherein j represents an index of the number of times of operation of the algorithm,
the fitting coefficient of the j-th operation result under the number of l neurons of the hidden layer is represented,i denotes the number of test data, I denotes the test data index, outputiIndicates the ith actual output data, targetiIndicating the i-th desired output data,
representing a desired output data mean;
increasing the arithmetic operation times until the maximum operation times N is reached, and calculating and recording the mean value and the variance of all fitting coefficients:
wherein the content of the first and second substances,
represents the mean value of the fitting coefficient under the number of l neurons in the hidden layer,
representing the variance of the fitting coefficient under the number of l neurons in the hidden layer, and J representing the operation times of the algorithm;
step 305, at [ L1，L2]Traversing the number of neurons in the hidden layer within the range, repeating the training process and the testing process in the step 304, and calculating and recording the mean value and the variance of the fitting coefficients corresponding to the number of the neurons in the hidden layer;
step 306, setting the minimum Min of the fitting coefficient mean value, and screening and reserving the number of the neurons with the fitting coefficient mean value larger than the minimum value;
step 307, normalizing the mean and variance of the fitting coefficients corresponding to the number of the reserved neurons by using a mean normalization method:
wherein R islExpressing the mean value of the normalized fitting coefficient under the number of l neurons in the hidden layer, SlRepresents the normalized fitting coefficient variance under the number of l neurons in the hidden layer,
representing the number of neurons remaining after the screening of step 306;
step 308, setting an evaluation function based on the normalized mean and variance of the fitting coefficients, and calculating the evaluation function value of the number of neurons in the hidden layer:
f(Rl)＝βRl+γSl(formula twelve)
β + γ ═ 1 (thirteen formula)
Wherein, f (R)l) Expressing evaluation function values corresponding to the number of l neurons in the hidden layer, wherein beta and gamma respectively express the weight of the mean value and the variance of the normalized fitting coefficient;
step 309, calculating the evaluation function value of each hidden layer neuron, comparing the evaluation function values corresponding to different neuron numbers of the hidden layer, and the neuron number with the maximum evaluation function value is the optimum neuron number l of the hidden layerb。
5. The method for determining the quantitative relationship between the shift of passenger transportation hub and the passenger flow based on the improved BP neural network as claimed in claim 1, wherein the step 40 specifically comprises the following steps:
step 401, based on the set number of network layers, network transfer function, algorithm training function, algorithm stopping condition, number of hidden layer optimal neurons and the preprocessed passenger transport hub shift and passenger flow data, simulating by using a BP neural network algorithm, and outputting a passenger transport hub shift and passenger flow quantitative relational expression:
wherein y' represents the output passenger flow data, p represents the index of the number of the neurons in the hidden layer, and wpRepresenting the connection weight of the input layer to the pth neuron of the hidden layer, bpRepresenting the threshold of connection of the input layer to the pth neuron of the hidden layer, WpRepresenting the connection weight from the pth neuron of the hidden layer to the output layer, and B representing the connection threshold from the hidden layer to the output layer;
step 402, performing inverse normalization on the output data based on the normalization method of the minimum-maximum normalization, and reducing the output data into actual data:
y″′＝y″′(ymax-ymin)+ymin(formula fifteen)
Wherein y' ″ represents the actual passenger flow data corresponding to the output passenger flow data after inverse normalization.
6. An apparatus for determining the quantitative relationship between passenger transportation hub shift and passenger flow based on an improved BP neural network, the apparatus comprising:
the data preprocessing module is used for preprocessing data, and specifically comprises: based on the original passenger transport hub shift and passenger flow data, carrying out normalization processing on the data by using a minimum and maximum standardization method;
the network parameter setting module is used for setting network parameters, wherein the network parameters comprise a network layer number, a network transfer function, an algorithm training function and an algorithm stopping condition;
the hidden layer optimal neuron number calculating module is used for calculating the number of the hidden layer optimal neurons, and specifically comprises the following steps: initializing the range of the number of the neurons according to an empirical method, and setting the maximum operation times of the algorithm under each neuron number; the passenger transport hub shift and passenger flow data are in one-to-one correspondence, and the data are randomly divided into experimental data and test data; constructing a neural network based on experimental data, and calculating a fitting coefficient based on a residual square sum and a total dispersion square sum between actual output data and expected output data of test data; calculating the mean value of the fitting coefficient by using an arithmetic mean method aiming at each neuron, and calculating the variance of the fitting coefficient based on the mean value; setting the minimum value of the fitting coefficient mean value, and screening and reserving the number of the neurons with the fitting coefficient mean value larger than the minimum value; respectively carrying out normalization processing on the mean value and the variance of the fitting coefficient corresponding to the reserved neuron number by adopting a mean value normalization method; distributing weights for the mean value and the variance of the processed fitting coefficients, and setting the weighted sum of the two parts as an evaluation function value of the neuron; evaluating the number of the neurons corresponding to the maximum value of the function value as the optimal number of the neurons of the hidden layer;
the output and reduction module is used for outputting a passenger transport hub shift and passenger flow quantitative relational expression and reducing the output data into actual data, and specifically comprises the following steps: based on the set network layer number, the network transfer function, the algorithm training function, the algorithm stopping condition, the number of the hidden layer optimal neurons and the preprocessed passenger transport hub shift and passenger flow data, operating the BP neural network algorithm and outputting a passenger transport hub shift and passenger flow quantitative relational expression; based on the data preprocessing method, the output data is restored to actual data by using an inverse normalization method.
7. A system for determining the quantitative relationship between passenger transport hub shift and passenger flow based on an improved BP neural network is characterized by comprising the following components: memory, processor and computer program stored on the memory and executable on the processor, which when executed by the processor implements the method of any one of claims 1 to 5.
8. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the method according to any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011016362.1A CN112215265B (en) | 2020-09-24 | 2020-09-24 | Method and device for determining quantitative relation between shifts and passenger flow of passenger transport hub |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011016362.1A CN112215265B (en) | 2020-09-24 | 2020-09-24 | Method and device for determining quantitative relation between shifts and passenger flow of passenger transport hub |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112215265A true CN112215265A (en) | 2021-01-12 |
| CN112215265B CN112215265B (en) | 2023-07-11 |
Family
ID=74052500
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011016362.1A Active CN112215265B (en) | 2020-09-24 | 2020-09-24 | Method and device for determining quantitative relation between shifts and passenger flow of passenger transport hub |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112215265B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103729688A (en) * | 2013-12-18 | 2014-04-16 | 北京交通大学 | Section traffic neural network prediction method based on EMD |
| CN105719025A (en) * | 2016-01-26 | 2016-06-29 | 华北电力大学（保定） | Prediction method for corrosion rate of Q235 galvanized steel grounding net of transformer station |
| CN107292387A (en) * | 2017-05-31 | 2017-10-24 | 汪薇 | A kind of method that honey quality is recognized based on BP |
-
2020
- 2020-09-24 CN CN202011016362.1A patent/CN112215265B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103729688A (en) * | 2013-12-18 | 2014-04-16 | 北京交通大学 | Section traffic neural network prediction method based on EMD |
| CN105719025A (en) * | 2016-01-26 | 2016-06-29 | 华北电力大学（保定） | Prediction method for corrosion rate of Q235 galvanized steel grounding net of transformer station |
| CN107292387A (en) * | 2017-05-31 | 2017-10-24 | 汪薇 | A kind of method that honey quality is recognized based on BP |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112215265B (en) | 2023-07-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109947086B (en) | Mechanical fault migration diagnosis method and system based on counterstudy | |
| CN111539515B (en) | Complex equipment maintenance decision method based on fault prediction | |
| CN106874581B (en) | Building air conditioner energy consumption prediction method based on BP neural network model | |
| CN111860982A (en) | Wind power plant short-term wind power prediction method based on VMD-FCM-GRU | |
| CN109829577B (en) | Rail train running state prediction method based on deep neural network structure model | |
| CN106920007B (en) | PM based on second-order self-organizing fuzzy neural network2.5Intelligent prediction method | |
| CN112489497B (en) | Airspace operation complexity evaluation method based on deep convolutional neural network | |
| CN113553755B (en) | Power system state estimation method, device and equipment | |
| Fang et al. | Failure mode and effects analysis: an integrated approach based on rough set theory and prospect theory | |
| CN115564114A (en) | Short-term prediction method and system for airspace carbon emission based on graph neural network | |
| CN112465199A (en) | Airspace situation evaluation system | |
| CN111415008B (en) | Ship flow prediction method based on VMD-FOA-GRNN | |
| CN114169502A (en) | Rainfall prediction method and device based on neural network and computer equipment | |
| CN113591402A (en) | Digital power transformer health state fuzzy comprehensive evaluation method | |
| CN112215265A (en) | Method and device for determining quantitative relation between passenger transport hub shift and passenger flow | |
| CN115907079B (en) | Airspace traffic flow prediction method based on attention space-time diagram convolutional network | |
| CN114363004B (en) | Risk assessment method, risk assessment device, computer equipment and storage medium | |
| CN114036991B (en) | Method and device for determining relation between passenger flow of passenger transportation hub and passenger flow of traffic connection mode | |
| CN114386631A (en) | Method and device for rush repair of power load fault and electronic equipment | |
| Li et al. | Application of combination forecasting model in aircraft failure rate forecasting | |
| Feng et al. | A physics-based PSO-BPNN model for civil aircraft noise assessment | |
| CN111832834A (en) | Airport pavement service performance prediction method and device and electronic equipment | |
| CN112580797A (en) | Incremental learning method of multi-mode multi-label prediction model | |
| CN113449258B (en) | Quality evaluation method and device for intelligent electric meter, terminal equipment and storage medium | |
| Sun | [Retracted] The Legal Regulation of Artificial Intelligence and Edge Computing Automation Decision‐Making Risk in Wireless Network Communication |
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 |