CN111126827A - Input-output accounting model construction method based on BP artificial neural network - Google Patents
Input-output accounting model construction method based on BP artificial neural network Download PDFInfo
- Publication number
- CN111126827A CN111126827A CN201911320387.8A CN201911320387A CN111126827A CN 111126827 A CN111126827 A CN 111126827A CN 201911320387 A CN201911320387 A CN 201911320387A CN 111126827 A CN111126827 A CN 111126827A
- Authority
- CN
- China
- Prior art keywords
- input
- output
- data
- neural network
- artificial neural
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0637—Strategic management or analysis, e.g. setting a goal or target of an organisation; Planning actions based on goals; Analysis or evaluation of effectiveness of goals
- G06Q10/06375—Prediction of business process outcome or impact based on a proposed change
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Abstract
The application discloses input-output accounting model construction method based on BP artificial neural network, it is big to the input-output table acquisition degree of difficulty of city scale, prior art calculates complicacy and consuming time and power scheduling problem, this application uses province level input-output table as the basis, utilizes BP artificial neural network model, constructs input-output BP artificial neural network model to obtain the input-output table of city scale, provide solid data basis for the relevant research work of city scale hydroenergy coupled system.
Description
Technical Field
The application relates to the technical field of environmental ecology, in particular to a method for constructing an input-output accounting model based on a BP artificial neural network, which is used for analyzing environmental problems related to aspects such as energy, industrial structures, water resources and the like.
Background
With the rapid growth of population, the continuous acceleration of urbanization process and the driving of climate change, the global demand for water, energy and food will increase day by day, and it is expected that by year 2050, the global demand for water, energy and food will increase by more than 50% than in year 2015, which will cause huge pressure on the existing water, energy and food systems. More importantly, increasingly complex environmental issues will lead to more serious water, food and energy safety issues, thereby posing a threat to human life safety. Therefore, more and more researchers begin to take Water, energy and food as a Whole (WEFN) to perform related research, and the main research contents include food shortage, energy safety, Water resource shortage, industrial structure adjustment, climate change and the like, which will lay a solid foundation for green development of China and even global society.
There are many relevant research methods for water energy and grain coupling systems, the main methods include general equilibrium model, ecological network analysis, comprehensive index evaluation, life cycle evaluation, input and output analysis, and the like, and meanwhile, the research scale mainly relates to the aspects of the world, the country, the region, the city, the family, and the like. Most researchers mainly use input-output models to research the problems of water energy and grain coupling systems in national and regional scales, however, the research on the problems of the water energy and grain coupling systems in the urban scale is relatively less.
The Input-output (IO) model mainly uses an IO table to analyze the system relevance between different industries in the economic system. At present, the compilation of the input-output table generally involves complex processes of department data statistics, data integration, compilation and the like, and the compilation work is mainly completed by the national statistical bureau and provincial statistical departments. At present, the input-output tables available to domestic researchers are mainly of national scale and provincial scale, and the relevant data of the input-output tables of urban scale are relatively few and the acquisition difficulty is high. Therefore, many research teams at home and abroad plan to obtain the urban-scale input-output table by using the provincial input-output table through a series of complex economic calculations, but the calculation process is often complex, time-consuming and labor-consuming. Therefore, the difficulty in acquiring city scale input-output table data greatly limits the research on related environmental problems, and further has a certain barrier effect on city development.
Disclosure of Invention
Aiming at the problems that the acquisition difficulty of an input-output table of the urban scale is high, the calculation is complex and time and labor are consumed in the prior art, the application aims to provide an input-output accounting model construction method based on a BP artificial neural network, so that the input-output table of the urban scale is obtained, and a solid data base is provided for relevant research work of a water and grain coupling system of the urban scale.
In order to realize the task, the following technical scheme is adopted in the application:
in a first aspect, the application provides a method for constructing an input-output accounting model based on a BP artificial neural network, comprising the following steps:
Acquiring input-output table data of different years from various government statistical data and constructing a data set;
The BP artificial neural network model comprises an input layer, a hidden layer and an output layer, wherein the input layer indexes are parameters related to the output layer indexes, and comprise intermediate use/final use/import/other, town/rural population, industry GDP, final consumption expenditure, capital consumption sum and a state/provincial input-output table; the number of the neurons of the hidden layer is as follows:
wherein n is the number of neurons in the input layer, m is the number of neurons in the output layer, and a is [1,10 ]]A constant between; the number of the neurons of the output layer is 1, and the indexes of the output layer correspond to the indexes of the corresponding input layer; when the input layer index is the industry A of the national input-output table, the output layer index is the industry A of the provincial input-output table;
The training process comprises a data input process and a function mapping relation of the input and output layers of the neural network is sought; wherein, from the data obtained in step 1, a data set corresponding to the data of the input-output table of the previous years is selected and input into the established BP artificial neural network model, and the corresponding relationship of the data set is as follows: a Chinese input-output table, a regional input-output table, a provincial input-output table and an urban input-output table; sequentially inputting the data set into a BP artificial neural network model, training the BP artificial neural network model by using the data set, thereby establishing a function mapping relation between an input layer and an output layer, and simultaneously storing the determined function mapping relation;
The difference between the simulation value and the true value is judged by adopting the root mean square error so as to calibrate and verify the trained BP artificial neural network model, and the specific formula is as follows:
wherein RMSE is the root mean square error, Xobs，iAs an observed value, Xmodel，iIs an analog value, and n is an analog number;
selecting a data set established by the data of the input-output table of the rest year from the data obtained in the step 1 as output layer data, simulating by using the trained BP artificial neural network model to obtain corresponding output layer data, then simulating by using the output layer data and the corresponding provincial/urban scale input-output table through a genetic algorithm, a principal component analysis method and Monte Carlo, and finally obtaining an optimal output layer index through n iterations, thereby obtaining an optimal BP artificial neural network model;
Furthermore, the various government statistical data include China's annual statistics, China's regional input-output table, provinces 'annual statistics, and provinces' regional input-output table.
Further, the input-output table data of different years, wherein the different years are 1990, 1992, 1995, 1997, 2000, 2002, 2005, 2007, 2010, 2012, and 2015.
Further, in step 3, when training the BP artificial neural network model, a data set is established by using input-output table data in 1990, 1992, 1995, 1997, 2000, 2002, 2005 and 2007.
Further, in step 4, data sets established by the data of the input-output tables in 2010, 2012 and 2015 are selected as output layer data, and the trained BP artificial neural network model is used for simulation to obtain corresponding output layer data.
In a second aspect, the present application provides a back-propagation (BP) artificial neural network-based input-output accounting model, including:
the data collection module is used for acquiring input-output table data of different years from various government statistical data and constructing a data set;
the network model establishing module is used for establishing a BP artificial neural network model, and comprises an input layer, a hidden layer and an output layer, wherein the input layer indexes are parameters related to the output layer indexes, including intermediate use/final use/import/other, town/rural population, industry GDP, final consumption expense, capital consumption total amount and a state/provincial input-output table; the number of the neurons of the hidden layer is as follows:
wherein n is the number of neurons in the input layer, m is the number of neurons in the output layer, and a is [1,10 ]]A constant between; the number of the neurons of the output layer is 1, and the indexes of the output layer correspond to the indexes of the corresponding input layer; when the input layer index is the industry A of the national input-output table, the output layer index is the industry A of the provincial input-output table;
and the network model training module is used for selecting a data set corresponding to the input-output table data of the previous years from the data obtained in the step 1 and inputting the data set into the established BP artificial neural network model, wherein the corresponding relation of the data set is as follows: a Chinese input-output table, a regional input-output table, a provincial input-output table and an urban input-output table; sequentially inputting the data set into a BP artificial neural network model, training the BP artificial neural network model by using the data set, thereby establishing a function mapping relation between an input layer and an output layer, and simultaneously storing the determined function mapping relation;
the network model optimization module is used for judging the difference between the simulation value and the true value by adopting the root mean square error so as to calibrate and verify the trained BP artificial neural network model, and the specific formula is as follows:
wherein RMSE is the root mean square error, Xobs，iAs an observed value, Xmodel，iIs an analog value, and n is an analog number; selecting a data set established by the data of the input-output table of the rest year from the data obtained in the step 1 as output layer data, simulating by using the trained BP artificial neural network model to obtain corresponding output layer data, then simulating by using the output layer data and the corresponding provincial/urban scale input-output table through a genetic algorithm, a principal component analysis method and Monte Carlo, and finally obtaining an optimal output layer index through n iterations, thereby obtaining an optimal BP artificial neural network model;
the application module is used for inputting the provincial input-output table and related index data into the constructed optimal BP artificial neural network model, and the data of the output layer is the data of the corresponding city scale input-output table; and (4) compiling the city scale input-output table by utilizing general equilibrium analysis.
In a third aspect, the present application provides a terminal device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the processor implements the steps of the method for constructing the input-output accounting model based on the BP artificial neural network according to the first aspect.
In a fourth aspect, the present application provides a computer-readable storage medium, where a computer program is stored, and when executed by a processor, the computer program implements the steps of the method for building a BP artificial neural network-based input-output accounting model according to the first aspect.
The application has the following technical characteristics:
1. the method mainly utilizes the autonomous learning ability of the BP artificial neural network, associates with the storage function and rapidly seeks the ability of an optimized solution, converts the complicated calculation and reasoning process of an input-output table into a simple and convenient function mapping relation, and has the advantages of simplicity, high efficiency, time saving, labor saving and the like.
2. The method is based on the BP artificial neural network model, and on the basis of the provincial input-output table, the constructed input-output BP artificial neural network model (IO-BP-ANN) is used for obtaining the input-output table of the city scale, so that a foundation is laid for relevant city scale input-output analysis work.
Drawings
Fig. 1 is a schematic flowchart of a method for constructing an input-output accounting model based on a BP artificial neural network according to an embodiment of the present application;
FIG. 2 is a schematic structural diagram of a BP artificial neural network established in the present application;
FIG. 3 is a diagram illustrating the detailed classification of an input-output table;
fig. 4 is a schematic structural diagram of an input-output accounting model based on a BP artificial neural network according to an embodiment of the present application;
fig. 5 is a schematic diagram of a terminal device according to an embodiment of the present application.
Detailed Description
The input-output model is mainly based on an input-output table. The IO table adopted in China is a value type IO table, and data in the table reflects the currency flow between different industries in an economic system. The data of the IO table is typically divided into 4 quadrants. The first quadrant reflects the intermediate use of products of other industries in various industries, the second quadrant reflects the destination (including consumption, capital formation, export and the like) of final products of various industries, the third quadrant reflects the initial investment of various industries, and the fourth quadrant reflects the redistribution of national income. The input-output table is an important component of a new national economic accounting system and is also an important basis for carrying out macroscopic economic management in China. The input-output table can provide abundant and comprehensive basic data for deeply researching national economy comprehensive balance, industrial structure adjustment and compiling long-distance planning and macroscopic economy prediction.
Artificial Neural Networks (ANN) appeared after the 40 th 20 th century, which are formed by connecting a plurality of neurons with adjustable connection weights, have the characteristics of large-scale parallel processing, distributed information storage, good self-organizing self-learning capability and the like, and are increasingly widely applied to the fields of information processing, pattern recognition, intelligent control, system modeling and the like. Particularly, an Error back-propagation (BP) algorithm can approximate any continuous function, and has strong nonlinear mapping capability, and parameters such as the number of middle 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, so that the method has great flexibility, and is one of the most widely applied neural network models at present. The BP artificial neural network (BP-ANN) is a multilayer feedforward network trained according to error back propagation (error back propagation for short), the algorithm is called BP algorithm, the basic idea is a gradient descent method, a gradient search technology is utilized, the mean square error of an actual output value and an expected output value of the network is expected to be minimum, and a model topological structure comprises an input layer, a hidden layer and an output layer.
The application provides a method for constructing an input-output accounting model based on a BP artificial neural network, which converts a provincial scale input-output table into an urban scale input-output table, and as shown in figure 1, comprises the following steps:
And acquiring input-output table data of different years from various government statistical data and constructing corresponding data sets.
The basic data in the application are from the annual book of Chinese statistics, the input-output table of China,
Relevant government statistical data such as 'Hebei province statistical annual book', 'Guangdong province statistical annual book' and 'Hebei province regional input-output table', and the like, wherein the data age limit is 1990, 1992, 1995, 1997, 2000, 2002, 2005, 2007, 2010, 2012, 2015 and the like. The detailed classification of intermediate use, end use, import and other categories of the input-output table consumption end is shown in fig. 3.
Firstly, constructing a basic structure of a BP artificial neural network model: an input layer, a hidden layer, and an output layer.
The input layer indexes are mainly parameters related to the output layer indexes, such as: intermediate use/end use/import/other, town/rural population, industry GDP, end consumption expenditure, capital consumption headquarters, and national/provincial input-output tables, etc., the specific target design requires the user to set it as needed and further calibration of the model.
The hidden layer is one or more layers of neuron structures between the input layer and the output layer, and the neuron structures are not directly connected with the outside, but the change of the state of the neuron structures can affect the relation between the input and the output. In the network design process, the determination of the neuron number of the hidden layer is very important. The excessive number of neurons in the hidden layer can increase the network calculation amount and easily generate the over-fitting problem; if the number of the neurons is too small, the network performance is affected and the expected effect cannot be achieved. The number of hidden layer neurons in the network has a direct link to the complexity of the real problem, the number of neurons in the input and output layers, and the setting of the expected error. At present, there is no clear formula for determining the number of neurons in the hidden layer, and only some empirical formulas are used, and the final determination of the number of neurons still needs to be determined according to experience and multiple experiments. The following empirical formula is referred to herein for the problem of choosing the number of hidden layer neurons:
wherein n is the number of neurons in the input layer, m is the number of neurons in the output layer, and a is [1,10 ]]Constant in between.
The output layer is mainly the design purpose of the neural network, and the number of the neurons is mainly set according to the research content. In this application, the number of neurons in the output layer is 1, and the index corresponds to the corresponding index in the input layer, for example, when the index in the input layer is the industry a of the national input-output table, that is, the agriculture, forestry, animal husbandry and fishery products and services, the index in the output layer is the industry a of the provincial input-output table: agriculture, forestry, animal husbandry and fishery products and services, and so on.
The network model training process is mainly a data input process and seeks a function mapping relation specific to the input and output layers of the neural network.
In the present application, data sets corresponding to the input-output table data in 1990, 1992, 1995, 1997, 2000, 2002, 2005 and 2007 are input into the established BP artificial neural network model, and the corresponding relationship of the data sets is as follows: the method comprises the steps of inputting data sets into BP artificial neural network models in sequence according to a Chinese input-output table and a regional input-output table (Hebei province, Guangdong province, Henan province and the like), a province-level input-output table and a city-level input-output table (Shijiazhuang city, Tangshan city, Qinhuang island city and the like), so that a function mapping relation between an input layer and an output layer is established, and meanwhile, the determined function mapping relation is automatically stored in a network.
The main purpose of the calibration, verification and sensitivity analysis of the model is to enable the established model to better simulate the real situation, and the simulation result of the model is more accurate.
The calibration and verification of the model mainly compare the error magnitude between actual data and simulation data so as to judge the accuracy of the constructed model. In the present application, the Root Mean Square Error (RMSE) is used to determine the difference between the simulated value and the actual value, and the specific formula is:
wherein RMSE is the root mean square error, Xobs，iAs an observed value, Xmodel，iIs an analog value, and n is an analog number.
In addition, the value of RMSE is the only evaluation criterion, the smaller the value, the higher the model accuracy. Since the general RMSE method only optimizes one single index parameter, several methods must be used to weight multiple parameter indexes, and the present application further introduces Principal Component Analysis (PCA) and Genetic Algorithm (GA) to gradually select and optimize the input layer indexes.
The principal component analysis method is a multivariate statistical method for examining the correlation among a plurality of variables, and researches how to disclose the internal structure among the plurality of variables through a few principal components, namely, deriving the few principal components from the original variables so that the few principal components keep the information of the original variables as much as possible and are not correlated with each other. The genetic algorithm is a calculation model for simulating the natural selection and genetic mechanism of Darwinian biological evolution theory in the biological evolution process, is a method for searching an optimal solution by simulating the natural evolution process, and a target function (fitting optimality function) is a root mean square error. Secondly, sensitivity analysis is a method for researching and analyzing the sensitivity degree of the state or output change of a system (or a model) to the change of system parameters or ambient conditions, sensitivity analysis is often utilized in an optimization method to research the stability of an optimal solution when original data is inaccurate or changed, and the sensitivity analysis can also determine which parameters have larger influence on the system or the model, in the application, multi-parameter sensitivity analysis based on Monte Carlo simulation is adopted, the sensitivity of each parameter is given while comprehensively considering the model operation results for multiple times (such as N times), moreover, the measurement of the sensitivity is not to compare the output change value with the parameter change value, but to classify the objective function values of the model results for N times according to a defined objective function value (fitting optimal function), then two groups of accumulated frequencies are calculated, a determination is made as to the sensitivity of each parameter accordingly.
According to the method, a data set corresponding to input-output table data of the rest of years (namely 2010, 2012, 2015 and the like) is used as input layer data, corresponding output layer data is obtained through simulation of a BP artificial neural network model constructed after sample training, then the output layer data (namely a simulation value) and a corresponding provincial/urban scale input-output table (namely a true value) are used, an optimal input layer index system is finally obtained through a genetic algorithm, principal component analysis and Monte Carlo simulation for n times of iteration processes, and an optimal BP artificial neural network model (IO-BP-ANN) is established.
Inputting a provincial input-output table (such as the Guangdong province input-output table) and related index data into the constructed IO-BP-ANN model, inputting corresponding urban scale input-output table data into an output layer, and then using general equilibrium analysis to finally finish the compilation of the urban scale input-output table.
According to another aspect of the present application, there is provided a BP artificial neural network-based input-output accounting model 1, as shown in fig. 4, including:
the data collection module 11 is used for acquiring input-output table data of different years from various government statistical data and constructing a data set;
the network model establishing module 12 is used for establishing a BP artificial neural network model, and comprises an input layer, a hidden layer and an output layer, wherein the input layer indexes are parameters related to the output layer indexes, including intermediate use/final use/import/other, town/rural population, industry GDP, final consumption expense, capital consumption total and state/provincial input-output tables; the number of the neurons of the hidden layer is as follows:
wherein n is the number of neurons in the input layer, m is the number of neurons in the output layer, and a is [1,10 ]]A constant between; the number of the neurons of the output layer is 1, and the indexes of the output layer correspond to the indexes of the corresponding input layer; when the input layer index is the industry A of the national input-output table, the output layer index is the industry A of the provincial input-output table;
a network model training module 13, configured to select, from the data obtained in step 1, a data set corresponding to the data of the input-output table in the previous years, and input the data set into the established BP artificial neural network model, where the corresponding relationship of the data set is: a Chinese input-output table, a regional input-output table, a provincial input-output table and an urban input-output table; sequentially inputting the data set into a BP artificial neural network model, training the BP artificial neural network model by using the data set, thereby establishing a function mapping relation between an input layer and an output layer, and simultaneously storing the determined function mapping relation;
the network model optimization module 14 is configured to determine a difference between the simulated value and the true value by using a root mean square error, so as to calibrate and verify the trained BP artificial neural network model, and the specific formula is as follows:
wherein RMSE is the root mean square error, Xobs，iAs an observed value, Xmodel，iIs an analog value, and n is an analog number; selecting a data set established by the data of the input-output table of the rest year from the data obtained in the step 1 as output layer data, simulating by using the trained BP artificial neural network model to obtain corresponding output layer data, then simulating by using the output layer data and the corresponding provincial/urban scale input-output table through a genetic algorithm, a principal component analysis method and Monte Carlo, and finally obtaining an optimal output layer index through n iterations, thereby obtaining an optimal BP artificial neural network model;
the application module 15 is used for inputting the provincial input-output table and related index data into the constructed optimal BP artificial neural network model, and the data of the output layer is the data of the corresponding city scale input-output table; and (4) compiling the city scale input-output table by utilizing general equilibrium analysis.
It should be noted that the specific execution steps of the modules are the same as the corresponding steps in the foregoing method embodiments, and are not described herein again.
Referring to fig. 5, an embodiment of the present application further provides a terminal device 2, where the terminal device 2 may be a computer or a server; the method comprises a memory 22, a processor 21 and a computer program 23 stored in the memory 22 and executable on the processor, wherein the processor 21 implements the steps of the above-mentioned input-output accounting model construction method based on the BP artificial neural network when executing the computer program 23, for example, the steps 1 to 5 in the foregoing embodiment.
The computer program 23 may also be divided into one or more modules/units, which are stored in the memory 22 and executed by the processor 21 to accomplish the present application. One or more modules/units may be a series of computer program instruction segments capable of performing specific functions, where the instruction segments are used to describe an execution process of the computer program 23 in the terminal device 2, for example, the computer program 23 may be divided into an obtaining module, an identifying module, and a displaying module, and functions of each module are described in the foregoing description, and are not repeated.
The implementation of the present application provides a computer-readable storage medium, which stores a computer program, and the computer program, when executed by a processor, implements the steps of the above-mentioned input-output accounting model construction method based on the BP artificial neural network, for example, steps 1 to 5 in the foregoing embodiments.
The integrated modules/units, if implemented in the form of software functional units and sold or used as separate products, may be stored in a computer readable storage medium. Based on such understanding, all or part of the flow in the method of the embodiments described above can be realized by a computer program, which can be stored in a computer readable storage medium and can realize the steps of the embodiments of the methods described above when the computer program is executed by a processor. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer readable medium may include: any entity or device capable of carrying computer program code, recording medium, U.S. disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution media, and the like.
The above embodiments are only used to illustrate the technical solutions of the present application, and not to limit the same; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present application and are intended to be included within the scope of the present application.
Claims (8)
1. A method for constructing an input-output accounting model based on a BP artificial neural network is characterized by comprising the following steps:
step 1, data collection
Acquiring input-output table data of different years from various government statistical data and constructing a data set;
step 2, establishing a BP artificial neural network model
The BP artificial neural network model comprises an input layer, a hidden layer and an output layer, wherein the input layer indexes are parameters related to the output layer indexes, and comprise intermediate use/final use/import/other, town/rural population, industry GDP, final consumption expenditure, capital consumption sum and a state/provincial input-output table; the number of the neurons of the hidden layer is as follows:
wherein n is the number of neurons in the input layer, m is the number of neurons in the output layer, and a is [1,10 ]]A constant between; the number of the neurons of the output layer is 1, and the indexes of the output layer correspond to the indexes of the corresponding input layer; when the input layer index is the industry A of the national input-output table, the output layer index is the industry A of the provincial input-output table;
step 3, training BP artificial neural network model
The training process comprises a data input process and a function mapping relation of the input and output layers of the neural network is sought; wherein, from the data obtained in step 1, a data set corresponding to the data of the input-output table of the previous years is selected and input into the established BP artificial neural network model, and the corresponding relationship of the data set is as follows: a Chinese input-output table, a regional input-output table, a provincial input-output table and an urban input-output table; sequentially inputting the data set into a BP artificial neural network model, training the BP artificial neural network model by using the data set, thereby establishing a function mapping relation between an input layer and an output layer, and simultaneously storing the determined function mapping relation;
step 4, model calibration, verification and sensitivity analysis
The difference between the simulation value and the true value is judged by adopting the root mean square error so as to calibrate and verify the trained BP artificial neural network model, and the specific formula is as follows:
wherein RMSE is the root mean square error, Xobs，iAs an observed value, Xmodel，iIs an analog value, and n is an analog number;
selecting a data set established by the data of the input-output table of the rest year from the data obtained in the step 1 as output layer data, simulating by using the trained BP artificial neural network model to obtain corresponding output layer data, then simulating by using the output layer data and the corresponding provincial/urban scale input-output table through a genetic algorithm, a principal component analysis method and Monte Carlo, and finally obtaining an optimal output layer index through n iterations, thereby obtaining an optimal BP artificial neural network model;
step 5, inputting the provincial input-output table and related index data into the constructed optimal BP artificial neural network model, wherein the output layer data is the corresponding city scale input-output table data; and (4) compiling the city scale input-output table by utilizing general equilibrium analysis.
2. The method of claim 1, wherein the various government statistical data include "yearbook for statistics in china", table of input and output in china ", table of statistical yearbook for statistics in each province, and table of input and output in each province.
3. The method for constructing the input-output accounting model based on the BP artificial neural network as claimed in claim 1, wherein the input-output table data of different years are 1990, 1992, 1995, 1997, 2000, 2002, 2005, 2007, 2010, 2012 and 2015.
4. The method for constructing an input-output accounting model based on the BP artificial neural network as claimed in claim 1, wherein the step 3 is a data set established by input-output table data in 1990, 1992, 1995, 1997, 2000, 2002, 2005 and 2007 when training the BP artificial neural network model.
5. The method for constructing an input-output accounting model based on the BP artificial neural network as claimed in claim 1, wherein in step 4, the data sets established by the input-output table data in 2010, 2012 and 2015 are selected as output layer data, and the trained BP artificial neural network model is used for simulation to obtain the corresponding output layer data.
6. An input-output accounting model based on a BP artificial neural network is characterized by comprising the following components:
the data collection module is used for acquiring input-output table data of different years from various government statistical data and constructing a data set;
the network model establishing module is used for establishing a BP artificial neural network model, and comprises an input layer, a hidden layer and an output layer, wherein the input layer indexes are parameters related to the output layer indexes, including intermediate use/final use/import/other, town/rural population, industry GDP, final consumption expense, capital consumption total amount and a state/provincial input-output table; the number of the neurons of the hidden layer is as follows:
wherein n is the number of neurons in the input layer, m is the number of neurons in the output layer, and a is [1,10 ]]A constant between; the number of the neurons of the output layer is 1, and the indexes of the output layer correspond to the indexes of the corresponding input layer; when the input layer index is the industry A of the national input-output table, the output layer index is the industry A of the provincial input-output table;
and the network model training module is used for selecting a data set corresponding to the input-output table data of the previous years from the data obtained in the step 1 and inputting the data set into the established BP artificial neural network model, wherein the corresponding relation of the data set is as follows: a Chinese input-output table, a regional input-output table, a provincial input-output table and an urban input-output table; sequentially inputting the data set into a BP artificial neural network model, training the BP artificial neural network model by using the data set, thereby establishing a function mapping relation between an input layer and an output layer, and simultaneously storing the determined function mapping relation;
the network model optimization module is used for judging the difference between the simulation value and the true value by adopting the root mean square error so as to calibrate and verify the trained BP artificial neural network model, and the specific formula is as follows:
wherein RMSE is the root mean square error, Xobs，iAs an observed value, Xmodel，iIs an analog value, and n is an analog number; selecting a data set established by the data of the input-output table of the rest year from the data obtained in the step 1 as output layer data, simulating by using the trained BP artificial neural network model to obtain corresponding output layer data, then simulating by using the output layer data and the corresponding provincial/urban scale input-output table through a genetic algorithm, a principal component analysis method and Monte Carlo, and finally obtaining an optimal output layer index through n iterations, thereby obtaining an optimal BP artificial neural network model;
the application module is used for inputting the provincial input-output table and related index data into the constructed optimal BP artificial neural network model, and the data of the output layer is the data of the corresponding city scale input-output table; and (4) compiling the city scale input-output table by utilizing general equilibrium analysis.
7. A terminal device comprising a memory, a processor and a computer program stored in said memory and executable on said processor, characterized in that the steps of the method according to any of claims 1 to 6 are implemented when the computer program is executed by the processor.
8. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911320387.8A CN111126827A (en) | 2019-12-19 | 2019-12-19 | Input-output accounting model construction method based on BP artificial neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911320387.8A CN111126827A (en) | 2019-12-19 | 2019-12-19 | Input-output accounting model construction method based on BP artificial neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111126827A true CN111126827A (en) | 2020-05-08 |
Family
ID=70500381
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911320387.8A Pending CN111126827A (en) | 2019-12-19 | 2019-12-19 | Input-output accounting model construction method based on BP artificial neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111126827A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112990651A (en) * | 2021-01-15 | 2021-06-18 | 嘉兴学院 | Construction method of basin environment resource input-output model |
| CN114330937A (en) * | 2022-03-15 | 2022-04-12 | 广东工业大学 | Implicit carbon emission accounting method, device and storage medium |
-
2019
- 2019-12-19 CN CN201911320387.8A patent/CN111126827A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112990651A (en) * | 2021-01-15 | 2021-06-18 | 嘉兴学院 | Construction method of basin environment resource input-output model |
| CN114330937A (en) * | 2022-03-15 | 2022-04-12 | 广东工业大学 | Implicit carbon emission accounting method, device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | A seasonal GM (1, 1) model for forecasting the electricity consumption of the primary economic sectors | |
| CN102183621B (en) | Aquaculture dissolved oxygen concentration online forecasting method and system | |
| CN103577694B (en) | Aquaculture water quality short-time combination forecast method on basis of multi-scale analysis | |
| CN109214449A (en) | A kind of electric grid investment needing forecasting method | |
| CN107169621A (en) | A kind of Dissolved Oxygen in Water Forecasting Methodology and device | |
| CN103218669A (en) | Intelligent live fish cultivation water quality comprehensive forecasting method | |
| CN110751318A (en) | IPSO-LSTM-based ultra-short-term power load prediction method | |
| CN109934422A (en) | Neural network wind speed prediction method based on time series data analysis | |
| CN109214591B (en) | Method and system for predicting aboveground biomass of woody plant | |
| CN107329511A (en) | Based on the vegetable aquaculture luminous environment high efficiency regulatory method and system between suitable root warm area | |
| CN111292124A (en) | Water demand prediction method based on optimized combined neural network | |
| CN110163444A (en) | A kind of water demand prediction method based on GASA-SVR | |
| CN111126827A (en) | Input-output accounting model construction method based on BP artificial neural network | |
| CN113554213A (en) | Natural gas demand prediction method, system, storage medium and equipment | |
| CN109754122A (en) | A kind of Numerical Predicting Method of the BP neural network based on random forest feature extraction | |
| CN113361761A (en) | Short-term wind power integration prediction method and system based on error correction | |
| CN106296434A (en) | A kind of Grain Crop Yield Prediction method based on PSO LSSVM algorithm | |
| CN111914488B (en) | Data area hydrologic parameter calibration method based on antagonistic neural network | |
| CN114119273A (en) | Park comprehensive energy system non-invasive load decomposition method and system | |
| CN114154401A (en) | Soil erosion modulus calculation method and system based on machine learning and observation data | |
| CN110264010B (en) | Novel rural power saturation load prediction method | |
| CN107808245A (en) | Based on the network scheduler system for improving traditional decision-tree | |
| CN116258060A (en) | Soil testing formula fertilization method based on machine learning | |
| CN110298762A (en) | A kind of family ranch culture efficiency analysis system | |
| Gao et al. | Establishment of economic forecasting model of high-tech industry based on genetic optimization neural network |
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 |