CN101089884A - Human body data space dynamic modeling method - Google Patents

Abstract

A space-dynamic modeling method of population data includes normalizing influence factor data, inputting influence factor normalization value as GP to form search space for quickly solving out model structure with optimum adaptability and carrying out quick precise optimization on individuality to copied by utilizing modified GA in selection-copying operation process of GA each generation individuality.

Description

A kind of human body data space dynamic modeling method

Technical field

The present invention relates to a kind of human body data space dynamic modeling method, the people information that can be used in the projects such as Environmental Health diagnosis of risk, lose from natural calamity assessment and on-the-spot sample survey accurately obtains.

Background technology

The population growth has caused immense pressure for global resources, environmental bearing capacity: plough and forest land area is die-offed, ecological diversity destroys seriously, and the human survival condition goes from bad to worse etc.In time obtain on the different scale accurate population space distribution and change information thereof for solving these societies, economy and environmental problem, the integrated management ability that improves population, resource and environment has significance.And demographic data normally adds up step by step according to administrative unit and gather.This statistical method often causes in the research population different with the space cell yardstick that other data are depended on, and makes to be fused between data to be a difficult problem.In addition because Increase of population and migration also need a large amount of energy and financial resources to keep the real-time of people information.Therefore very necessity is carried out spatialization with the census data, simulates the process of real spatial distribution state of population and dynamic transition by modeling.After first density of population isogram in 1857 produced, population spatial distribution research developed rapidly.The population spatial distribution research method is divided into two big classes: face interpolation and curved surface modeling.The face interpolation is that demographic data is changed in different unit, face territory, and purpose is that census data-switching with source region (source zone) is on target area (target zone).The population distribution curved surface modeling then is to utilize suitable formula census data to be assigned in the graticule mesh system of a rule to go, and each graticule mesh in the system has all comprised the population estimation value of its ad-hoc location.

Though the population spatialization method can provide a large amount of population space distributions and change information, all there is the mathematical relation between a difficult problem-searching factor of influence and the demographic data inevitably in they.Be to realize the population spatialization in a lot of at home and abroad these type of researchs by the method for setting up linearity between the factor and the population or nonlinear regression model (NLRM).Progressively return is to ask for one of regression model method the most commonly used.It is a kind of calculation method of being permitted of " having into has ", importance according to variable is selected significant variable one by one, but also some variable of considering selected regression equation might be selected into and loses original importance along with other variablees thereafter, in time these variablees are rejected away from regression equation, final regression equation only keeps important variable.Though the modeling method that stepwise regression analysis etc. are commonly used is simple to operate, the result is convenient to explain, requires pre-defined model structure and model parameter, and this is difficult to determine often.In addition, also have the ripe model of some other subjects, for example the Gravity Models on the physics successfully is used for population spatial distribution.The prerequisite that Gravity Models uses in the population spatialization be the hypothesis people all be tending towards or near the life of the place in city, not so there is good the connection in its life zone with the down town; Even in the rural area, dense population areas is many near the main line of communication, and the closer to the city, the density of population is more than innerland height.Though this modeling method just can obtain population on each graticule mesh with this as long as Gravity Models omit inching, selecting which population distribution influence factor variable input model equally also is an individual difficult problem.If a certain factor variable in the collective analysis model is often introduced bias easily.

On the whole, the modeling method of above-mentioned relevant population spatial distribution has that precision is low, the deficiency of model optimization weak effect.

Summary of the invention

Technology of the present invention is dealt with problems and is: overcome that existing population data space modeling accuracy is low, the deficiency of model optimization weak effect, a kind of human body data space dynamic modeling method genetic programming algorithm is nested the basis with the improvement genetic algorithm on is provided, this method has the modeling accuracy height, the advantage that model optimization is effective.

Technical solution of the present invention is: a kind of human body data space dynamic modeling method, and its characteristics are: at first handle the n kind nature that the population space distribution is exerted an influence and the raw data of socio-economic factor, draw the normalized value of these data; After with the influence factor normalized value as genetic programming algorithm (genetic programming, the search volume is formed in GP) input, comes rapid solving to have the population spatial distribution model of optimal adaptation degree; In GP algorithm individual choice replicate run of per generation process, utilize to improve genetic algorithm (genetic algorithms GA) treats and duplicates individuality and be optimized, and finally realizes the purpose of human body data space dynamic modeling, and its concrete steps are as follows:

(1) utilizes the GIS technology to obtain the raw data that population distribution is had the nature and the socio-economic factor of influence, these data are carried out normalized;

(2) initialization genetic programming algorithm and improvement genetic algorithm parameter;

(3) with the input of the normalized value of various influence factors, form the search volume, find the solution population spatial distribution model with optimal adaptation degree as genetic programming algorithm; In genetic programming algorithm individual choice replicate run of per generation process, utilize to improve genetic algorithm and treat and duplicate individuality and carry out quick accurate optimization, finally realize human body data space dynamic modeling.

Described initialization genetic programming algorithm parameter has population scale, i.e. number of individuals GP_Size, sample size GP_N in the population, genetic algebra GP_Gen, maximum degree of depth Max_Dep, maximum degree of depth Max_CDep, the crossover probability GP_P of intersecting of generating _cWith variation probability GP_P _mInitialization improves genetic algorithm parameter population scale, i.e. number of individuals GA_Size in the population, number of samples GA_N and genetic algebra GA_Gen; The method rapid solving that is nested with genetic programming algorithm and improvement genetic algorithm has the population spatial distribution model of optimal adaptation degree then, that is:

A. at first determine the basic composition unit in the genetic programming algorithm search volume, comprise the basic operator (f of n kind influence factor normalized value ₁, f ₂..., f _n) and the elementary arithmetic operational symbol, form GP_Size individuality at random by the basic composition unit again.These individualities all are the alternativess of population spatial distribution model, i.e. possible mathematical relation expression formula between population distribution and the input influence factor variable generally has following functional form:

POPU＝f(X ₁f ₁，X ₂f ₂，...，X _nf _n) (1)

POPU is the population data variable in the formula; (f ₁, f ₂..., f _n) and (X ₁, X ₂..., X _n) be respectively all kinds of input influence factor variablees and coefficient thereof.

B. calculate at individual Kpid (i) _ GP (all the sample calculation theoretical values of 1≤i≤GP_Size) and the coefficient of determination between the measured value, with its as t generation (fitness BsJi that should individuality among 1≤t≤GP_Gen) (and i, t) _ GP, its computing formula is:

$\mathrm{BsJi}(i,t)\_\mathrm{GP}=\frac{\underset{j=1}{\overset{\mathrm{GP}\_N}{\mathrm{\Σ}}}(P\_\mathrm{GP}\left(j\right)-\stackrel{\‾}{P})(P^{\′}\_\mathrm{GP}\left(j\right)-\stackrel{\‾}{P^{\′}})}{\sqrt{\underset{j=1}{\overset{\mathrm{GP}\_N}{\mathrm{\Σ}}}{(P\_\mathrm{GP}\left(j\right)-\stackrel{\‾}{P})}^2\underset{j=1}{\overset{\mathrm{GP}\_N}{\mathrm{\Σ}}}{(P^{\′}\_\mathrm{GP}\left(j\right)-\stackrel{\‾}{P^{\′}})}^2}}---\left(2\right)$

In the formula

With

Be respectively the mean value of all sample measured values and theory of computation value; P ' _ GP (j) is individual Kpid (i) _ GP in the sample j (theory of computation value of 1≤j≤GP_N); P_GP (j) is the measured value of sample j;

C. according to the determined fitness of formula (2), take to compete selection strategy and select to duplicate individuality, promptly from colony, choose one group of individuality at random, relatively should organize each member's fitness to produce new individuality, select actual best individual Kpid (BesOpt) _ GP, i.e. POPU=f (XB ₁f ₁, XB ₂f ₂..., XB _nf _n), after utilize improving genetic algorithm it is optimized quickly and accurately, the individuality that duplicates after the optimization is the poorest to replace this group.Genetic algorithm is as follows to the step that optimum individual in the group of individuals is optimized with improving:

A. at first model structure is differentiated, then no longer optimized if belong to the structure of having optimized.

B. adopt the real coding mode, directly use the type real of real data in the sample, according to model coefficient (X to be optimized in the global error model _B1, X _B2..., X _Bn), generate GA_Size individuality in the initial population at random;

C. set up the fitness evaluation function that improves genetic algorithm, calculating is at individual Kpid (i) _ GA (summation of variance between the theory of computation value of all samples of 1≤i≤GA_Size) and the measured value, with it as t generation (fitness BsJi (i that should individuality among 1≤t≤GA_Gen), t) _ and GA, computing formula is:

$\mathrm{BsJi}(i,t)\_\mathrm{GA}=\frac{1}{\frac{\mathrm{GA}\_\mathrm{Size}\×\underset{j=1}{\overset{\mathrm{GA}\_\mathrm{GN}}{\mathrm{\Σ}}}{(P^{\′}\_\mathrm{GA}\left(j\right)-P\_\mathrm{GA}\left(j\right))}^2}{\underset{i=1}{\overset{\mathrm{GA}\_\mathrm{Size}}{\mathrm{\Σ}}}\underset{j=1}{\overset{\mathrm{GA}\_\mathrm{GN}}{\mathrm{\Σ}}}{(P^{\′}\_\mathrm{GA}\left(j\right)-P\_\mathrm{GA}\left(j\right))}^2}+{10}^{-10}}---\left(2\right)$

P_GA in the formula (j) is individual Kpid (i) _ GA in the sample j (theory of computation value of 1≤j≤GA_N); P_GA (j) is the measured value of sample j.

D. according to each individual pairing fitness in the determined fitness evaluation function calculation of formula (2) colony.It by ascending sort, is won bad (fitness is big) by excellent (fitness is little) and eliminate, adopt the ratio preference pattern to select to the individuality after the ordering and duplicate.May cause the inconsistent problem of the individual number of new and old population when the fitness ratio rounds in calculating, all individual number differences before and after duplicating are also sorted, successively the bigger individuality of loss being added 1 is 0 up to difference.Choose two point of crossing by waiting to intersect on the individuality subsequently, exchange two and wait to intersect that part realizes individual interlace operation between the individual point of crossing at each.Mutation operation adopts multistage variation, variation probability GA_P _mAlso be in the uncertain value of (0,0.1) between one.

E. to the operation of the repeating step c of filial generation colony, carry out new round genetic evolution process, when reaching the genetic algebra t=GA_Gen that sets or fitness best values and equal preset value, then the best individuality of fitness is an optimum individual, is optimum solution.

D. the new individuality after genetic programming algorithm duplicates selection exchanges and mutation operation.

E. generate degree of depth Max_Dep, maximum degree of depth Max_CDep, the crossover probability GP_P of intersecting with maximum _cWith variation probability GP_P _mBe the constraint condition of genetic programming algorithm operation, circular flow step C, D, when genetic algebra t=GP_Gen or fitness best values equal preset value, optimum individual Kpid (the Best) _ GP of gained, i.e. POPU=f (X _E1f ₁, X _E2f ₂..., X _Enf _n), just for having the population spatial distribution model of optimal adaptation degree; Wherein: f ₁, f ₂..., f _nBe respectively the basic operator of n kind influence factor normalized value, X _E1, X _E2..., X _EnBe respectively the pairing coefficient of basic operator.

The present invention's advantage compared with prior art is: the present invention has overcome shortcoming low to the population spatial distribution modeling accuracy traditionally, the model optimization weak effect, genetic programming algorithm and improvement genetic algorithm are nested, be incorporated in the modeling and optimization of population spatial distribution, with genetic program design optimization model structure, with the genetic algorithm optimization model parameter, successfully realized the robotization of population spatial distribution modeling process.

Description of drawings

Fig. 1 is the process flow diagram of a kind of human body data space dynamic modeling method of the present invention.

Embodiment

As shown in Figure 1, specific implementation method of the present invention is as follows:

1, for the precision and the efficient of enhancement algorithms, the parameter of genetic programming algorithm and improvement genetic algorithm is carried out initialization; The parameter of initialization genetic programming algorithm mainly contains population scale GP_Size (〉=10), sample size GP_N (〉=100), genetic algebra GP_Gen (〉=200), the maximum degree of depth Max_Dep (≤15) of generation, maximum degree of depth Max_CDep (≤10), the crossover probability GP_P of intersecting _c(≤1) and variation probability GP_P _m(≤1); Initialization improves genetic algorithm parameter population scale, i.e. number of individuals GA_Size (〉=10) in the population, number of samples GA_N (〉=100), genetic algebra GA_Gen (〉=200).

2, utilize that GIS technique computes such as spatial analysis goes out the gradient, river, means of transportation, cover in the soil and the primitive attribute value of the five kinds of population distribution influence factors in contiguous villages and small towns, and these values are carried out normalized.The formula of normalized is:

$f_{\mathrm{jk}}=\frac{{\mathrm{org}}_{\mathrm{jk}}}{\sqrt{\underset{j=1}{\overset{\mathrm{GP}\_N}{\mathrm{\Σ}}}{\left({\mathrm{org}}_{\mathrm{jk}}\right)}^2}}---\left(1\right)$

Org in the formula _JkAnd f _JkRepresent sample j (1≤j≤GP_N) original value and the normalized value of k (1≤k≤5) class influence factor attribute respectively.

3, finish top two steps after, have the population spatial distribution model of optimal adaptation degree at last with genetic programming algorithm and the Evolutionary Modeling algorithm rapid solving that is nested of improvement genetic algorithm, its step is as follows:

(1) elder generation determines the basic composition unit in the genetic programming algorithm search volume, comprises the basic operator (f of five kinds of influence factor normalized values ₁, f ₂, f ₃, f ₄, f ₅) and the elementary arithmetic operational symbol+,-, * ,/, ln (), exp () }; The back forms GP_Size individuality at random by the basic composition unit.Individuality is carried out standardization processing, and all coefficient entries all are positioned at the right of operational symbol, help the identification to similar model structure like this.

(2) calculate individual Kpid (i) _ GP (all the sample calculation theoretical values of 1≤i≤GP_Size) and the coefficient of determination between the measured value, with its as t generation (fitness BsJi that should individuality among 1≤t≤GP_Gen) (and i, t) _ GP, its computing formula is:

$\mathrm{BsJi}(i,t)\_\mathrm{GP}=\frac{\underset{j=1}{\overset{\mathrm{GP}\_N}{\mathrm{\Σ}}}(P\_\mathrm{GP}\left(j\right)-\stackrel{\‾}{P})(P^{\′}\_\mathrm{GP}\left(j\right)-\stackrel{\‾}{P^{\′}})}{\sqrt{\underset{j=1}{\overset{\mathrm{GP}\_N}{\mathrm{\Σ}}}{(P\_\mathrm{GP}\left(j\right)-\stackrel{\‾}{P})}^2\underset{j=1}{\overset{\mathrm{GP}\_N}{\mathrm{\Σ}}}{(P^{\′}\_\mathrm{GP}\left(j\right)-\stackrel{\‾}{P^{\′}})}^2}}---\left(2\right)$

In the formula

With

Be respectively the mean value of all sample measured values and theory of computation value; P ' _ GP (j) is individual Kpid (i) _ GP in the sample j (theory of computation value of 1≤j≤GP_N); P_GP (j) is the measured value of sample j;

(3) according to the determined fitness of formula (2), take to compete selection strategy and select to duplicate individuality, promptly from colony, select one group of individuality at random, relatively should organize each member's fitness to produce new individuality, select actual best individual Kpid (BesOpt) _ GP, i.e. POPU=f (X _B1f ₁, X _B2f ₂X _B3f ₃, X _B4f ₄, X _B5f ₅) after utilize improving genetic algorithm it is optimized quickly and accurately, the individuality that duplicates after the optimization is the poorest to replace this group.Same individuality need to prove that contemporary individuality is that choosing of putting back to arranged, so may repeatedly be chosen or duplicate.Genetic algorithm is as follows to the step that optimum individual in the group of individuals is optimized with improving:

1. at first model structure is differentiated, then no longer optimized if belong to the structure of having optimized.

2. adopt the real coding mode, directly use the type real of real data in the sample, according to model coefficient (X to be optimized in the global error model _B1, X _B2, X _B3, X _B4, X _B5), generate GA_Size individuality in the initial population at random;

3. set up the fitness evaluation function that improves genetic algorithm, calculating is at individual Kpid (i) _ GA (summation of variance between the theory of computation value of all samples of 1≤i≤GA_Size) and the measured value, with it as t generation (fitness BsJi (i that should individuality among 1≤t≤GA_Gen), t) _ and GA, computing formula is:

$\mathrm{BsJi}(i,t)\_\mathrm{GA}=\frac{1}{\frac{\mathrm{GA}\_\mathrm{Size}\×\underset{j=1}{\overset{\mathrm{GA}\_\mathrm{GN}}{\mathrm{\Σ}}}{(P^{\′}\_\mathrm{GA}\left(j\right)-P\_\mathrm{GA}\left(j\right))}^2}{\underset{i=1}{\overset{\mathrm{GA}\_\mathrm{Size}}{\mathrm{\Σ}}}\underset{j=1}{\overset{\mathrm{GA}\_\mathrm{GN}}{\mathrm{\Σ}}}{(P^{\′}\_\mathrm{GA}\left(j\right)-P\_\mathrm{GA}\left(j\right))}^2}+{10}^{-10}}---\left(3\right)$

P ' in the formula _ GA (j) is individual Kpid (i) _ GA in the sample j (theory of computation value of 1≤j≤GA_N); P_GA (j) is the measured value of sample j.

4. calculate each individual pairing fitness in the colony according to the determined fitness of formula (3).It by ascending sort, is won bad (fitness is big) by excellent (fitness is little) and eliminate, adopt the ratio preference pattern to select to the individuality after the ordering and duplicate.May cause the inconsistent problem of the individual number of new and old population when the fitness ratio rounds in calculating, all individual number differences before and after duplicating are also sorted, successively the bigger individuality of loss being added 1 is 0 up to difference.Choose two point of crossing by waiting to intersect on the individuality subsequently, exchange two and wait to intersect that part realizes individual interlace operation between the individual point of crossing at each.The way that fixing crossover probability is set with simple generic algorithm is different, the crossover probability GA_P of used improvement genetic algorithm _cIt is a random value that is positioned between (0.8,1).Because all individualities all show as one 5 dimensional vector, therefore the individuality after guaranteeing variation still under the prerequisite in the hunting zone, is carried out individual variation with the method for giving selected individual plus noise.Mutation operation adopts multistage variation, variation probability GA_P _mAlso be in the uncertain value of (0,0.1) between one.

5. to the operation 4. of filial generation colony repeating step, carry out new round genetic evolution process, when reaching the genetic algebra t=GA_Gen that sets or fitness best values and equal preset value, then the best individuality of fitness is an optimum individual, is optimum solution.

(4) the new individuality after genetic programming algorithm duplicates selection exchanges and mutation operation.Interlace operation is exactly the point of crossing of two individualities of picked at random, exchanges the following subtree in these two point of crossing then mutually and generates two new individualities.After mutation operation then was the change point and subordinate branch subtree thereof of selected parent individuality at random, the deletion catastrophe point replaced it with its subordinate branch subtree again.

(5) generate degree of depth Max_Dep, maximum degree of depth Max_CDep, the crossover probability GP_P of intersecting with maximum _cWith variation probability GP_P _mBe the constraint condition of genetic programming algorithm operation, circular flow step (3), (4), when genetic algebra t=GP_Gen or fitness best values equal preset value, optimum individual Kpid (the Best) _ GP of gained, i.e. POPU=f (X _E1f ₁, X _E2f ₂, X _E3f ₃, X _E4f ₄, X _E5f ₅) for having the population spatial distribution model of optimal adaptation degree; Wherein: f ₁, f ₂, f ₃, f ₄, f ₅Be respectively the basic operator of five kinds of influence factor normalized values, X _E1, X _E2, X _E3, X _E4, X _E5Be respectively the pairing coefficient of basic operator.

The content that is not described in detail in the instructions of the present invention belongs to this area professional and technical personnel's known prior art.

Claims (9)

1, a kind of human body data space dynamic modeling method is characterized in that may further comprise the steps:

(1) utilizes the GIS technology to obtain the raw data that population distribution is had the nature and the socio-economic factor of influence, these data are carried out normalized;

(2) initialization genetic programming algorithm and improvement genetic algorithm parameter;

(3) with the input of the relevant raw data behind step (1) normalized value as genetic programming algorithm, form the search volume, find the solution population spatial distribution model with optimal adaptation degree; In genetic programming algorithm individual choice replicate run of per generation process, utilize to improve genetic algorithm and treat and duplicate individuality and carry out quick accurate optimization, finally realize human body data space dynamic modeling.

2, a kind of human body data space dynamic modeling method according to claim 1, it is characterized in that: initialization genetic programming algorithm parameter has population scale in the described step (2), i.e. number of individuals GP_Size, sample size GP_N in the population, genetic algebra GP_Gen, maximum degree of depth Max_Dep, maximum degree of depth Max_CDep, the crossover probability GP_P of intersecting of generating _cWith variation probability GP_P _m

3, a kind of human body data space dynamic modeling method according to claim 1, it is characterized in that: initialization improves genetic algorithm parameter in the described step (2) population scale, i.e. number of individuals GA_Size in the population, number of samples GA_N and genetic algebra GA_Gen.

4, a kind of human body data space dynamic modeling method according to claim 1 is characterized in that: the method for building up of population spatial distribution model is as follows in the described step (3):

(1) at first determines basic composition unit in the genetic programming algorithm search volume, comprise the basic operator (f of n kind influence factor normalized value ₁, f ₂..., f _n) and the elementary arithmetic operational symbol, form GP_Size individuality at random by the basic composition unit again;

(2) calculate (all the sample calculation theoretical values of 1≤i≤GP_Size) and the coefficient of determination between the measured value at individual Kpid (i) _ GP, with it as t generation (fitness BsJi (i that should individuality among 1≤t≤GP_Gen), t) _ and GP, t is a genetic algebra, its computing formula is:

$\mathrm{BsJi}(i,t)\_\mathrm{GP}=\frac{\underset{j=1}{\overset{\mathrm{GP}\_N}{\mathrm{\Σ}}}(P\_\mathrm{GP}\left(j\right)-\stackrel{\‾}{P})(P^{\′}\_\mathrm{GP}\left(j\right)-\stackrel{\‾}{P^{\′}})}{\sqrt{\underset{j=1}{\overset{\mathrm{GP}\_N}{\mathrm{\Σ}}}{(P\_\mathrm{GP}\left(j\right)-\stackrel{\‾}{P})}^2\underset{j=1}{\overset{\mathrm{GP}\_N}{\mathrm{\Σ}}}{(P^{\′}\_\mathrm{GP}\left(j\right)-\stackrel{\‾}{P^{\′}})}^2}}---\left(2\right)$

In the formula

With

Be respectively the mean value of all sample measured values and theory of computation value; P ' _ GP (j) is individual Kpid (i) _ GP in the sample j (theory of computation value of 1≤j≤GP_N); P_GP (j) is the measured value of sample j;

(3) according to the determined fitness of formula (2), take to compete selection strategy and select to duplicate individuality, promptly from colony, choose one group of individuality at random, relatively should organize each member's fitness to produce new individuality, select actual best individual Kpid (BesOpt) _ GP, i.e. POPU=f (X _B1f ₁, X _B2f ₂..., X _Bnf _n), utilize improving genetic algorithm again the coefficient of individual the inside is optimized quickly and accurately, the individuality that duplicates after the optimization is the poorest to replace this group;

(4) the new individuality after genetic programming algorithm duplicates selection intersects and mutation operation;

(5) generate degree of depth Max_Dep, maximum degree of depth Max_CDep, the crossover probability GP_P of intersecting with maximum _cWith variation probability GP_P _mBe the constraint condition of genetic programming algorithm operation, circular flow step (3), (4), when genetic algebra t=GP_Gen or fitness best values equal preset value, optimum individual Kpid (the Best) _ GP of gained, i.e. POPU=f (X _E1f ₁, X _E2f ₂..., X _Enf _n), for having the population spatial distribution model of optimal adaptation degree.

5, according to claim 1 or 4 described a kind of human body data space dynamic modeling methods, it is characterized in that: it is as follows to the step that optimum individual in the group of individuals is optimized that described employing improves genetic algorithm:

(1) at first model structure is differentiated, then no longer optimized if belong to the structure of having optimized;

(2) adopt the real coding mode, directly use the type real of real data in the sample, according to model coefficient (X to be optimized in the global error model _B1, X _B2..., X _Bn), generate GA_Size individuality in the initial population at random;

(3) set up the fitness evaluation function that improves genetic algorithm, calculating is at individual Kpid (i) _ GA (1≤i≤GA_Size), the theory of computation value of all samples and measured value between the summation of variance, with its as t generation (1≤≤ fitness BsJi (i that should individuality in GA_Gen), t) _ and GA, computing formula is:

$\mathrm{BsJi}(i,t)\_\mathrm{GA}=\frac{1}{\frac{\mathrm{GA}\_\mathrm{Size}\×\underset{j=1}{\overset{\mathrm{GA}\_\mathrm{GN}}{\mathrm{\Σ}}}{(P^{\′}\_\mathrm{GA}\left(j\right)-P\_\mathrm{GA}\left(j\right))}^2}{\underset{i=1}{\overset{\mathrm{GA}\_\mathrm{Size}}{\mathrm{\Σ}}}\underset{j=1}{\overset{\mathrm{GA}\_\mathrm{GN}}{\mathrm{\Σ}}}{(P^{\′}\_\mathrm{GA}\left(j\right)-P\_\mathrm{GA}\left(j\right))}^2}+{10}^{-10}}---\left(3\right)$

P ' in the formula _ GA (j) is individual Kpid (i) _ GA in the sample j (theory of computation value of 1≤j≤GA_N); P_GA (j) is the measured value of sample j;

(4) according to each individual pairing fitness in the determined fitness evaluation function calculation of formula (3) colony, to its press fitness little-big ascending sort, adopting the ratio preference pattern to select to the individuality after the ordering duplicates, all individual number differences before and after duplicating are also sorted, successively the bigger individuality of loss being added 1 is 0 up to difference, choose two point of crossing by waiting to intersect on the individuality subsequently at each, exchange two and wait to intersect that part realizes individual interlace operation between the individual point of crossing, mutation operation adopts multistage variation;

(5) to the operation of filial generation colony repeating step (4), carry out new round genetic evolution process, when reaching the genetic algebra t=GA_Gen that sets or fitness best values and equal preset value, then the best individuality of fitness is an optimum individual, is optimum solution.

6, a kind of human body data space dynamic modeling method according to claim 1 is characterized in that: described variation probability GA_P _mBe in the uncertain value of (0,0.1) between one.

7, a kind of human body data space dynamic modeling method according to claim 1 is characterized in that: described crossover probability GA_P _cIt is a random value that is positioned between (0.8,1).

8, a kind of human body data space dynamic modeling method according to claim 1 is characterized in that: the formula that the population distribution influence factor data normalization in the described step (1) is handled is:

$f_{\mathrm{jk}}=\frac{{\mathrm{org}}_{\mathrm{jk}}}{\sqrt{\underset{j=1}{\overset{\mathrm{GP}\_N}{\mathrm{\Σ}}}{\left(\mathrm{or}{g}_{\mathrm{jk}}\right)}^2}}---\left(1\right)$

Org in the formula _JkAnd f _JkRepresent sample j (k of 1≤j≤GP_N) (1≤j≤n) original value and the normalized value of class influence factor attribute respectively.

9, a kind of human body data space dynamic modeling method according to claim 4, it is characterized in that: in the described step (1) the individual GP_Size that is formed at random by the basic composition unit is carried out standardization processing, all coefficient entries all are positioned at the right of operational symbol, help the identification to similar model structure like this.

Images