CN102737346A - Assessment factor rating method for land resources based on ant colony clustering model - Google Patents

Abstract

The invention relates to an assessment factor rating method for land resources based on an ant colony clustering model. The method comprises the steps of inputting indexes of scale of each assessment factor; establishing a two-dimensional planar grid according to assessment factor objects, and randomly distributing each assessment factor object on the two-dimensional planar grid; setting artificial ants and randomly distributing the artificial ants on the two-dimensional planar grid; and iteratively picking up and dropping the assessment factor objects in probability through the artificial ants so as to obtain a rating result of the assessment factor objects. The assessment factor rating method for the land resources based on the ant colony clustering model can automatically rate the assessment factor objects and provide technical support for rational use of the land resources.

Description

Evaluation for Soil Resources factor rank division methods based on the ant colony clustering model

Technical field

The invention belongs to land resource automatic Evaluation technical field, particularly relate to a kind of Evaluation for Soil Resources factor rank division methods based on the ant colony clustering model.

Background technology

(1) land resource quality assessment technology

Along with the arrival of 21 century, many global problems such as the population problem of face of mankind, grain security problem are increasingly serious.China is as populous nation, and land resource is rare relatively, how rationally to utilize the soil, realizes that sustainable using of land resources is the current problem that presses for solution.Therefore, adopting the technological means and the method for science that the quality of land resource is estimated, is the necessary means that promotes that land resource is rationally utilized.

Definition according to the FAO of FAO (Food and Agriculture Organization of the United Nation); Evaluation for Soil Resources is meant the process that the soil is assessed to the specific effect of utilizing mode to show; Comprise that aspect attributes such as form to the soil, soil, vegetation, weather carry out quality comprehensive and identify, thereby distinguish and more different soil utilizes mode to suitability degree that evaluation objective showed.Relevant document: [1] FAO.Land Evaluation.Towards a revised framework.2007..Using for reference on the basis of advanced foreign technology, China has formed that the soil that comprises to farming land and construction land grades, the land valuation system that meets current national conditions demand of deciding grade and level, appraisal, appraisal of land suitability, land consolidation utilization evaluation, land deterioration evaluation etc.

When estimating the land resource quality; The land resource quality there is the object such as various factors entity such as business service center, country fair, iirigation water source, road network of appreciable impact, is defined as " the Evaluation for Soil Resources factor " (abbreviating " the evaluation factor " as) or " Evaluation for Soil Resources index ".Because estimating the difference of the scale (or size) of the factor will directly cause estimating function and the effect that the factor possesses and have difference; It also exists tangible different to the spatial dimension of the influence of land quality and intensity, when estimating factorial analysis, also need treat with a certain discrimination.For example in the resource evaluation of urban land; Service facilities such as the business service center that size is different with scale, hospital; The function that it possessed is also different with all kinds of services that can provide, and then spacial influence scope, the mode of action and the intensity of land resource quality is also had nothing in common with each other.Therefore, when carrying out Evaluation for Soil Resources, need it be divided into some ranks, respectively the spacial influence scope and the action intensity of the evaluation factor of different stage be analyzed according to estimating factor scale.Relevant document: [2] State Administration for Quality Supervision and Inspection and Quarantine. urban land deciding grade and level rules (GB/T 18507-2001) ［ S ］ .2001 that grades; Relevant document: [3] Ministry of Land and Resources. farming land deciding grade and level rules (TD/T 1005-2004) ［ S ］ .2003.

(2) land resource quality assessment factor rank partitioning technology

Yet the method that at present relevant land valuation factor rank is divided mainly is the evaluation factor scaled index according to the band classification, artificially the land valuation factor is divided into some ranks by the land valuation expert according to self experience.(annotate: scaled index is used to reflect the influence degree of each factor object to the land resource quality, and scaled index is big more, and then its intensity to the land resource quality influence is big more.Scaled index is usually according to calculating by the reaction evaluating factor a series of index comprehensives big or small, scale; Its computing method all have comparatively detailed elaboration in relevant land valuation standard criterion; Like urban land deciding grade and level rules, the farming land deciding grade and level rules etc. of grading, not focal point of the present invention.) its shortcoming is that mainly subjectivity is too strong, the science of results depends critically upon expertise.Therefore, for the more scientific and rational division land valuation factor, and then, must develop more rational land valuation factor rank division methods for land valuation factor space coverage and intensive analysis provide basic data support accurately.

(3) ant colony clustering model

The ant colony clustering model is that a kind of graveyard organizational behavior to ant is carried out the intelligent algorithm that computer simulation realizes, he can realize the automatic cluster to high dimensional data, therefore is applied to the data mining field.In natural ant population; Single ant can be judged in the travelling process and currently runs into the difference of object and surroundings and make corresponding movement: will the object different with surroundings " pick up " and move, simultaneously the object similar with surroundings " put down ".This " picking up " and " putting down " behavior through ant have finally realized analogical object is classified as one type, and different objects are divided into the purpose of multiclass.This mechanism of ant colony clustering model extremely approaches the hierarchical policy of Evaluation for Soil Resources factor object, is about to the close object of scaled index and is divided into one-level, and different objects then is divided in the different ranks.The correlation technique appearance that the ant colony clustering model is applied to the land resource quality assessment is not arranged at present as yet.

Summary of the invention

To the limitation of estimating factor rank division methods in the existing Evaluation for Soil Resources, invent the Evaluation for Soil Resources factor rank division methods based on the ant colony clustering model of a kind of intellectuality, robotization.The influence of estimating factor pair land resource quality for scientific and reasonable research and analysis provides basic basis accurately, for reasonable, sustainable use land resource provide the technical method support.

Technical scheme of the present invention is a kind of Evaluation for Soil Resources factor rank division methods based on the ant colony clustering model, may further comprise the steps:

Step 1, the scaled index of factor object is respectively estimated in input;

Step 2 according to the quantity N that estimates factor object, makes up the two dimensional surface grid of a big or small N * N, and each evaluation factor object is assigned randomly on the two dimensional surface grid;

Step 3 is provided with the only artificial ant of M, and the only artificial ant of M is assigned randomly on the two dimensional surface grid, and it is 0 that iterations is set, and the status indication of every artificial ant is non-loaded;

Step 4 is carried out following substep to every artificial ant,

Step 4.1, artificial ant is moved a grid towards direction at random;

Step 4.2 judges in the current grid at artificial ant place whether the factor of evaluation object is arranged, and then directly gets into step 4.3 if having, and does not then return execution in step 4.1 if having, and estimates factor object up to running into, and gets into step 4.3;

Step 4.3 judges whether the state of artificial ant is non-loaded,

If load is arranged, artificial ant is by probability P _dPut down the evaluation factor object of load, estimate factor object if artificial ant is successfully put down, artificial ant state is changed to non-loaded, otherwise artificial ant keeps the evaluation factor object of load; Said probability P _dSimilarity according to scaled index between the evaluation factor object of grid around in the evaluation factor object of artificial ant present load and the artificial ant visual range is confirmed;

If non-loaded, artificial ant is by probability P _pPick up current grid inner evaluation factor object, estimate factor object, artificial ant state has been changed to load if artificial ant is successfully picked up; Said probability P _pSimilarity according to scaled index between the evaluation factor object of grid around in evaluation factor object in the current grid and the artificial ant visual range is confirmed;

Step 5, all artificial ants move a grid towards at random direction respectively, and wherein state is that loaded artificial ant is by probability P _dPut down and estimate factor object, iterations adds 1; Estimate factor object if certain only artificial ant is successfully put down, change to artificial ant state non-loaded; Said probability P _dSimilarity according to scaled index between the evaluation factor object of grid around in the evaluation factor object of artificial ant present load and the artificial ant visual range is confirmed;

Step 6 is returned step 4 and is repeated, and reaches preset iterations n until iterations, obtains estimating factor object classification results.

And, in step 4.3 and the step 5, probability P _dComputing formula following,

$P_d={\left(\frac{{\mathrm{\λ}}_2}{{\mathrm{\γ}}_2+{\mathrm{\λ}}_2}\right)}^2$

In the formula, γ ₂Be one greater than 0 parameter, λ ₂Be in evaluation factor object and the artificial ant visual range of artificial ant present load around the similarity of scaled index between the evaluation factor object of grid;

${\mathrm{\λ}}_2=\max \{0,\frac{1}{S^2}\underset{i\∈\mathrm{Neighbor}}{\mathrm{\Σ}}(1-|\frac{{\mathrm{SI}}_o-{\mathrm{SI}}_i}{\mathrm{\γ}}\left|\right)\}$

In the formula, γ be one (0-1] parameter, s is the parameter of number of grid around being used for confirming in the artificial ant visual range, SI _iBe the scaled index of grid inner evaluation factor object i around certain in the artificial ant visual range, Neighbor is set of the evaluation factor object of grid around all in the artificial ant visual range, SI _oIt is the scaled index of the evaluation factor object of artificial ant present load.

And, in the step 4.3, probability P _pComputing formula following,

$P_p={\left(\frac{{\mathrm{\γ}}_1}{{\mathrm{\γ}}_1+{\mathrm{\λ}}_1}\right)}^2$

In the formula, γ ₁Be one greater than 0 parameter; λ ₁Be in current grid inner evaluation factor object and the artificial ant visual range around the similarity of scaled index between the evaluation factor object of grid;

${\mathrm{\λ}}_1=\max \{0,\frac{1}{S^2}\underset{i\∈\mathrm{Neighbor}}{\mathrm{\Σ}}(1-|\frac{{\mathrm{SI}}_{\mathrm{cur}}-{\mathrm{SI}}_i}{\mathrm{\γ}}\left|\right)\}$

In the formula, γ be one (0-1] parameter, s is the parameter of number of grid around being used for confirming in the artificial ant visual range, SI _iBe the scaled index of grid inner evaluation factor object i around certain in the artificial ant visual range, Neighbor is set of the evaluation factor object of grid around all in the artificial ant visual range, SI _CurIt is the scaled index of current grid inner evaluation factor object.

The present invention introduces the ant colony clustering model and estimates factor object classification field, and combines to estimate the basic characteristics of factor object classification, designs suitable ant colony clustering model.Advantage of the present invention: have simple, robotization, intelligent characteristics generally; Various evaluation factor rank division methods with respect to widespread use in actual engineering at present; The problem that the present invention mainly solves has: (1) makes full use of the ant colony clustering model algorithm in the advantage of clustering problem aspect finding the solution, and the ant colony clustering model is introduced estimated finding the solution of factor rank partition problem; (2), made up and be applicable to the ant colony clustering model of estimating the division of factor rank according to the characteristics of estimating factor rank partition problem; (3) the present invention can estimate the classification of factor object automatically, for the factor impact analysis of land resource quality assessment provides science, basic basis accurately, and then technical support is provided for the reasonable utilization that promotes land resource.

Description of drawings

Fig. 1 is the process principle figure of the embodiment of the invention;

Fig. 2 for the embodiment of the invention will estimate factor object at random be assigned to the synoptic diagram on the two dimensional surface grid;

Fig. 3 for the embodiment of the invention with artificial ant at random be assigned to the synoptic diagram on the two dimensional surface grid;

The synoptic diagram that Fig. 4 creeps in plane grid for artificial ant in the embodiment of the invention;

Fig. 5 is the direction synoptic diagram of in plane grid, creeping of artificial ant in the embodiment of the invention, and wherein Fig. 5 a is the situation that ant has 8 directions to creep, and Fig. 5 b is the situation that ant has 5 directions to creep, and Fig. 5 c is the situation that ant has 3 directions to creep;

Fig. 6 is the visual range synoptic diagram of artificial ant in plane grid in the embodiment of the invention, the visual range of ant when wherein Fig. 6 a is s=1, the visual range of ant when Fig. 6 b is s=3, the visual range of ant when Fig. 6 c is s=5.

Embodiment

Specify technical scheme of the present invention below in conjunction with accompanying drawing and embodiment.

Like Fig. 1, the land resource quality assessment factor rank division methods based on the ant colony clustering model of embodiment of the invention design is mainly constantly creeped through ant and is picked up factor object, creeps and put down factor object, obtains hierarchy.Can adopt computer software technology to realize automatic operational scheme by those skilled in the art during practical implementation.The realization flow of embodiment may further comprise the steps:

Step 1, the scaled index of factor object is respectively estimated in input.

Embodiment imports a certain type of scaled index of estimating the factor (like business service center).Wherein scaled index is used to reflect the scale of this evaluation factor, is obtained by a series of indexs (like teacher's number of school, floor area of building etc.) COMPREHENSIVE CALCULATING usually, and the non-focal point of the present invention of its computing method specifically can be referring to relevant land valuation rules.Can be designated as SI, wherein the span of SI is the real number of [0-100].

Step 2 according to the quantity N that estimates factor object, makes up the two dimensional surface grid of a big or small N * N, and each evaluation factor object is assigned randomly on the two dimensional surface grid.

Each estimate the locus of the factor can number represent with ranks (Row, Col).The evaluation factor object of embodiment distributes realizes seeing that accompanying drawing 2:N factor object to be allocated is respectively factor object 1, factor object 2, factor object 3, factor object 4... factor object N; For after each estimates factor object distribution ranks at random number, the distribution on the two dimensional surface grid is as shown in the figure.

During practical implementation, estimate adoptable data structure of factor object such as following table:

The data structure of factor object

Field name	Field type	Explanation
			ID	int	Object number
SI	float	Scaled index
			Row	int	The row of object in grid number
Col	int	The row of object in grid number

Step 3 is provided with the only artificial ant of M, and the only artificial ant of M is assigned randomly on the two dimensional surface grid, and it is 0 that iterations is set, and the status indication of every artificial ant is non-loaded.

Can the only artificial ant of M be assigned on the two dimensional surface grid through Random assignment ranks number.Wherein, the value of M must be less than the value of N.Divide this particular problem according to estimating factor rank, suggestion M desirable [0.05-0.2] times N.The artificial ant of embodiment is distributed and realizes seeing that accompanying drawing 3:M artificial ant is designated as ant 1, ant 2... ant M respectively, for behind every ant Random assignment ranks number, has obtained Random assignment and has estimated the two dimensional surface grid behind factor object and the artificial ant.

During practical implementation, adoptable data structure of artificial ant such as following table:

The data structure of artificial ant

Field name	Field type	Explanation
			ID	int	Object number
FID	int	The current factor object of bearing of ant
			Row	int	The row of object in grid number
Col	int	The row of object in grid number

Step 4, every artificial ant is carried out following substep:

Step 4.1, artificial ant is moved a grid towards direction at random.

For certain only artificial ant, its moving direction in the two dimensional surface grid has 8 kinds maybe.Therefore,, can generate an integer random number between [1-8] at random,, and adjust its row column number value by its moving direction of random number decision that generates moving should the manual work ant time.The mode that artificial ant is moved in the enforcement sees that 4,8 kinds of ant moving directions of accompanying drawing define and corresponding ranks computing formula is following:

1:Col=Col-1,Row=Row-1

2:Col=Col,Row=Row-1

3:Col=Col+1,Row=Row-1

4:Col=Col+1,Row=Row

5:Col=Col+1,Row=Row+1

6:Col=Col,Row=Row+1

7:Col=Col-1,Row=Row+1

8:Col=Col-1,Row=Row

Must guarantee that ant is mobile in grid plan; The ranks that are ant number must be positioned between [1-N]; Therefore the direction ratio of when ant is in the edge, can creeping is less, as ant among Fig. 5 a have that 8 directions can be creeped, ant has that 5 directions can be creeped among Fig. 5 b, ant has 3 directions to creep among Fig. 5 c.

Step 4.2 judges in the current grid at artificial ant place whether the factor of evaluation object is arranged, and then directly gets into step 4.3 if having, and does not then return execution in step 4.1 if having, and estimates factor object up to running into, and gets into step 4.3.

If do not estimate factor object in the current grid at artificial ant place, that can not stop moves always, estimates factor object up to running into.

Step 4.3 judges whether the state of artificial ant is non-loaded.

Because to put down load is to be undertaken by probability, remove for the first time the execution in step 4.3, all have part in the only artificial ant of M and have load, so this determining step is set.

If load is arranged, artificial ant is by probability P _dPut down the evaluation factor object of load, estimate factor object if artificial ant is successfully put down, artificial ant state is changed to non-loaded, otherwise artificial ant keeps the evaluation factor object of load; Said probability P _dSimilarity according to scaled index between the evaluation factor object of grid around in the evaluation factor object of artificial ant present load and the artificial ant visual range is confirmed.Artificial ant is according to probability P _dPut down estimate factor object concrete realization can for, calculate P _dAnd produce [0-1] random number, if this random number is less than or equal to P _d, ant is just put down this object of load, otherwise does not put down.

Among the embodiment, probability P _dComputing formula following,

$P_d={\left(\frac{{\mathrm{\λ}}_2}{{\mathrm{\γ}}_2+{\mathrm{\λ}}_2}\right)}^2$

In the formula, γ ₂Be one and rule of thumb be provided with by the domain expert usually, to estimating this particular problem of factor object classification, γ greater than 0 parameter ₂The suggestion span be [0.4-0.6].λ ₂Be in evaluation factor object and the artificial ant visual range of artificial ant present load around the similarity of scaled index between the evaluation factor object of grid:

${\mathrm{\λ}}_2=\max \{0,\frac{1}{S^2}\underset{i\∈\mathrm{Neighbor}}{\mathrm{\Σ}}(1-|\frac{{\mathrm{SI}}_o-{\mathrm{SI}}_i}{\mathrm{\γ}}\left|\right)\}$

In the formula, γ be one (0-1] parameter, s is the parameter of number of grid around being used for confirming in the artificial ant visual range, SI _iBe the scaled index of grid inner evaluation factor object i around certain in the artificial ant visual range, Neighbor is set of the evaluation factor object of grid around all in the artificial ant visual range, SI _oIt is the scaled index of the evaluation factor object of artificial ant present load.

If non-loaded, artificial ant is by probability P _pPick up current grid inner evaluation factor object, estimate factor object, artificial ant state has been changed to load if artificial ant is successfully picked up; Said probability P _pSimilarity according to scaled index between the evaluation factor object of grid around in evaluation factor object in the current grid and the artificial ant visual range is confirmed.The evaluation factor object that artificial ant is picked up will move along with artificial ant together.Artificial ant is according to probability P _pPick up current grid inner evaluation factor object concrete realization can for, calculate P _pAnd produce [0-1] random number, if this random number is less than or equal to P _p, ant is just picked up this object, otherwise does not pick up.So P _pNear 1, the possibility that ant is picked up object is just big more more.

Among the embodiment, probability P _pComputing formula following,

$P_p={\left(\frac{{\mathrm{\γ}}_1}{{\mathrm{\γ}}_1+{\mathrm{\λ}}_1}\right)}^2$

In the formula, γ ₁Be one and rule of thumb be provided with in advance by the domain expert usually, to estimating this particular problem of factor object classification, γ greater than 0 parameter ₁The suggestion span be [0.4-0.6]; λ ₁Be in current grid inner evaluation factor object and the artificial ant visual range around the similarity of scaled index between the evaluation factor object of grid;

${\mathrm{\λ}}_1=\max \{0,\frac{1}{S^2}\underset{i\∈\mathrm{Neighbor}}{\mathrm{\Σ}}(1-|\frac{{\mathrm{SI}}_{\mathrm{cur}}-{\mathrm{SI}}_i}{\mathrm{\γ}}\left|\right)\}$

In the formula, γ be one (0-1] parameter, s is the parameter of number of grid around being used for confirming in the artificial ant visual range.SI _iBe the scaled index of grid inner evaluation factor object i around certain in the artificial ant visual range, Neighbor is set of the evaluation factor object of grid around all in the artificial ant visual range, SI _CurIt is the scaled index of current grid inner evaluation factor object.

γ rule of thumb is provided with by the domain expert usually, and according to the distributed area of estimating factor scaled index, the span of suggestion γ is [0.4-0.9].The s value rule of thumb is provided with by the domain expert according to the quantity of N usually, and the suggestion value is the odd number between [3-9], and s was greater than 1 o'clock, and number of grid was s around artificial ant visual range was interior ²-1.N numerical value is big more, and it is big more to establish s.Usually s is big more, can access classification results preferably, but also can increase calculated amount simultaneously.Shown in Fig. 6 a, the visual range of ant is the current grid at ant self place during s=1; Shown in Fig. 6 b, grid around the visual range of ant comprises on every side 8 during s=3; Shown in Fig. 6 c, grid around the visual range of ant comprises on every side 24 during s=5.

Step 5, all artificial ants move a grid towards at random direction respectively, and wherein state is that loaded artificial ant is by probability P _dPut down and estimate factor object, iterations adds 1; Estimate factor object if certain only artificial ant is successfully put down, change to artificial ant state non-loaded; Said probability P _dSimilarity according to scaled index between the evaluation factor object of grid around in the evaluation factor object of artificial ant present load and the artificial ant visual range is confirmed.

Among the embodiment, the mode that artificial ant is moved is consistent with step 4.1, probability P _dCalculating consistent with step 4.3.

Step 6 is returned step 4 and is repeated, and reaches preset iterations n until iterations, obtains estimating factor object classification results.When N is big more, the value of n is also big more as required during practical implementation.According to problem scale, span can be at 200-10000 usually.

When initial, estimate factor pair as if dispersion stochastic distribution on the two dimensional surface grid of N * N for N.After M ant of process creeps, estimate factor object for N and can on grid, assemble in heaps.Each gathering center is exactly a rank.What gathering centers are statistics have, just to what ranks should be arranged.For example, 100 objects are creeped after the cluster through ant, possibly on grid, occur 7 accumulation area, and then correspondence has 7 ranks.Along with the increase of iterations, the number of assembling the center is fewer and feweri.Carry out through iteration, obtain final classification results.

Specific embodiment described herein only is that the present invention's spirit is illustrated.Person of ordinary skill in the field of the present invention can make various modifications or replenishes or adopt similar mode to substitute described specific embodiment, but can't depart from spirit of the present invention or surmount the defined scope of appended claims.

Claims (3)

1. the Evaluation for Soil Resources factor rank division methods based on the ant colony clustering model is characterized in that, may further comprise the steps:

Step 1, the scaled index of factor object is respectively estimated in input;

Step 2 according to the quantity N that estimates factor object, makes up the two dimensional surface grid of a big or small N * N, and each evaluation factor object is assigned randomly on the two dimensional surface grid;

Step 3 is provided with the only artificial ant of M, and the only artificial ant of M is assigned randomly on the two dimensional surface grid, and it is 0 that iterations is set, and the status indication of every artificial ant is non-loaded;

Step 4 is carried out following substep to every artificial ant,

Step 4.1, artificial ant is moved a grid towards direction at random;

Step 4.2 judges in the current grid at artificial ant place whether the factor of evaluation object is arranged, and then directly gets into step 4.3 if having, and does not then return execution in step 4.1 if having, and estimates factor object up to running into, and gets into step 4.3;

Step 4.3 judges whether the state of artificial ant is non-loaded,

If load is arranged, artificial ant is by probability P _dPut down the evaluation factor object of load, estimate factor object if artificial ant is successfully put down, artificial ant state is changed to non-loaded, otherwise artificial ant keeps the evaluation factor object of load; Said probability P _dSimilarity according to scaled index between the evaluation factor object of grid around in the evaluation factor object of artificial ant present load and the artificial ant visual range is confirmed;

If non-loaded, artificial ant is by probability P _pPick up current grid inner evaluation factor object, estimate factor object, artificial ant state has been changed to load if artificial ant is successfully picked up; Said probability P _pSimilarity according to scaled index between the evaluation factor object of grid around in evaluation factor object in the current grid and the artificial ant visual range is confirmed;

Step 5, all artificial ants move a grid towards at random direction respectively, and wherein state is that loaded artificial ant is by probability P _dPut down and estimate factor object, iterations adds 1; Estimate factor object if certain only artificial ant is successfully put down, change to artificial ant state non-loaded; Said probability P _dSimilarity according to scaled index between the evaluation factor object of grid around in the evaluation factor object of artificial ant present load and the artificial ant visual range is confirmed;

Step 6 is returned step 4 and is repeated, and reaches preset iterations n until iterations, obtains estimating factor object classification results.

2. according to claim 1 based on the Evaluation for Soil Resources factor rank division methods of ant colony clustering model, it is characterized in that: in step 4.3 and the step 5, probability P _dComputing formula following,

$P_d={\left(\frac{{\mathrm{\λ}}_2}{{\mathrm{\γ}}_2+{\mathrm{\λ}}_2}\right)}^2$

In the formula, γ ₂Be one greater than 0 parameter, λ ₂Be in evaluation factor object and the artificial ant visual range of artificial ant present load around the similarity of scaled index between the evaluation factor object of grid;

${\mathrm{\λ}}_2=\max \{0,\frac{1}{S^2}\underset{i\∈\mathrm{Neighbor}}{\mathrm{\Σ}}(1-|\frac{{\mathrm{SI}}_o-{\mathrm{SI}}_i}{\mathrm{\γ}}\left|\right)\}$

In the formula, γ be one (0-1] parameter, s is the parameter of number of grid around being used for confirming in the artificial ant visual range, SI _iBe the scaled index of grid inner evaluation factor object i around certain in the artificial ant visual range, Neighbor is set of the evaluation factor object of grid around all in the artificial ant visual range, SI _oIt is the scaled index of the evaluation factor object of artificial ant present load.

3. according to claim 1 based on the Evaluation for Soil Resources factor rank division methods of ant colony clustering model, it is characterized in that: in the step 4.3, probability P _pComputing formula following,

$P_p={\left(\frac{{\mathrm{\γ}}_1}{{\mathrm{\γ}}_1+{\mathrm{\λ}}_1}\right)}^2$

In the formula, γ ₁Be one greater than 0 parameter; λ ₁Be in current grid inner evaluation factor object and the artificial ant visual range around the similarity of scaled index between the evaluation factor object of grid;

${\mathrm{\λ}}_1=\max \{0,\frac{1}{S^2}\underset{i\∈\mathrm{Neighbor}}{\mathrm{\Σ}}(1-|\frac{{\mathrm{SI}}_{\mathrm{cur}}-{\mathrm{SI}}_i}{\mathrm{\γ}}\left|\right)\}$

In the formula, γ be one (0-1] parameter, s is the parameter of number of grid around being used for confirming in the artificial ant visual range, SI _iBe the scaled index of grid inner evaluation factor object i around certain in the artificial ant visual range, Neighbor is set of the evaluation factor object of grid around all in the artificial ant visual range, SI _CurIt is the scaled index of current grid inner evaluation factor object.

Images