CN113343565A - Neighborhood effect mode construction and CA simulation method and system considering spatial heterogeneity - Google Patents
Neighborhood effect mode construction and CA simulation method and system considering spatial heterogeneity Download PDFInfo
- Publication number
- CN113343565A CN113343565A CN202110599588.7A CN202110599588A CN113343565A CN 113343565 A CN113343565 A CN 113343565A CN 202110599588 A CN202110599588 A CN 202110599588A CN 113343565 A CN113343565 A CN 113343565A
- Authority
- CN
- China
- Prior art keywords
- neighborhood
- land
- type
- area
- constraint
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000000694 effects Effects 0.000 title claims abstract description 90
- 238000010276 construction Methods 0.000 title claims abstract description 49
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000004088 simulation Methods 0.000 title claims abstract description 20
- 230000001413 cellular effect Effects 0.000 claims abstract description 60
- 230000001788 irregular Effects 0.000 claims abstract description 10
- 238000005259 measurement Methods 0.000 claims abstract description 8
- 230000009466 transformation Effects 0.000 claims abstract description 5
- 238000005192 partition Methods 0.000 claims description 27
- 238000006243 chemical reaction Methods 0.000 claims description 23
- 238000004422 calculation algorithm Methods 0.000 claims description 18
- 238000011160 research Methods 0.000 claims description 15
- 238000004364 calculation method Methods 0.000 claims description 12
- 238000004590 computer program Methods 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 10
- 230000006870 function Effects 0.000 claims description 9
- 230000008030 elimination Effects 0.000 claims description 4
- 238000003379 elimination reaction Methods 0.000 claims description 4
- 230000011218 segmentation Effects 0.000 claims description 4
- 238000000638 solvent extraction Methods 0.000 claims description 4
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 238000009826 distribution Methods 0.000 abstract description 19
- 230000003993 interaction Effects 0.000 abstract description 15
- 230000008859 change Effects 0.000 abstract description 11
- 230000002452 interceptive effect Effects 0.000 abstract description 4
- 230000004069 differentiation Effects 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 10
- 239000000872 buffer Substances 0.000 description 7
- 238000000926 separation method Methods 0.000 description 5
- 230000002776 aggregation Effects 0.000 description 4
- 238000012935 Averaging Methods 0.000 description 3
- 238000005054 agglomeration Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 238000006073 displacement reaction Methods 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000004888 barrier function Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000009028 cell transition Effects 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000007704 transition Effects 0.000 description 2
- 241000872198 Serjania polyphylla Species 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000005311 autocorrelation function Methods 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 230000008614 cellular interaction Effects 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000010206 sensitivity analysis Methods 0.000 description 1
- 230000011664 signaling Effects 0.000 description 1
- 238000011524 similarity measure Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Medical Informatics (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Software Systems (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a neighborhood effect mode construction and CA simulation method and system considering spatial heterogeneity.A neighborhood range of a central cell is obtained by introducing a Delaunay triangulation network clustering method under multiple constraints, and a final constraint type cell multi-order irregular neighborhood is constructed by combining a spatial neighborhood relationship and a neighborhood range identified by a triangulation network; and the similarity and difference of the cellular transformation rules in different neighborhoods are mined by clustering the distance attenuation sequences of the spatial measurement factors of the historical data and establishing a neighborhood effect mode library. The invention adopts heterogeneous rules to define the adjacent relation of the cells, considers the space distribution complexity characteristics, embodies the difference of the interaction range of the land parcel and can truly reflect the land utilization change rule. The land use interactive differentiation rule implied in the vector space is mined from historical data, so that numerical calibration with huge workload is avoided, and the problem of subjectivity and one-sidedness is not easy to occur.
Description
Technical Field
The invention relates to the field of cellular automata and land utilization research, in particular to a neighborhood effect mode construction and CA simulation method and system considering spatial heterogeneity.
Background
The Cellular Automata (CA) model is a common land use change simulation model, can simulate a macroscopic complex dynamic pattern based on microscopic simple cellular interaction, and therefore has an important role in the field of land use change or urban growth. In the cellular automaton model, the neighborhood refers to a range capable of affecting the central cell, and the research content includes neighborhood definition and neighborhood effect modeling. Different from the homogeneous neighborhood of the grid cellular automata, the vector cellular automata has the advantages that due to the fact that the sizes and the shapes of cellular units are different, the number and the range of the neighborhoods of the cellular units are possibly different, the shapes are not symmetrical, the definitions of the neighborhoods are complex and diverse, and therefore the fact that the irregular neighborhoods of the cellular units of the vector cellular automata are divided reasonably is beneficial to carrying out urban land utilization change simulation more scientifically.
The neighborhood definition includes both proximity identification and range division. The current definition mode of the proximity relation of the vector cellular automaton mainly comprises the following steps: (1) and (3) a definition method based on the spatial adjacency relation. (2) And (3) a definition method based on the space distance. (3) A definition method based on cellular space clustering. According to the cellular neighborhood, various neighborhoods are constructed in two ways of spatial or non-spatial expansion: (1) the spatial expansion may form a variety of neighborhoods, restricted topology neighborhoods, buffer neighborhoods (centroid buffers, boundary buffers), truncated buffer neighborhoods (centroid truncated buffers, boundary truncated buffers), multi-order neighborhoods, wide-area neighborhoods, and the like. (2) The non-spatially extended representation is a neighborhood defined based on a promotional relationship proposed by Moreno. Research shows that according to neighborhood sensitivity analysis of different adjacency relation types and different spatial distance measurement modes, the neighborhood definition method of the centroid interception buffer area is superior to other methods in city expansion simulation.
The neighborhood effect is one of key problems to be considered for establishing a cellular automata model of a complex system, refers to the attraction and repulsion action of local land utilization interaction, and is the expression of local influence. Because the existing vector cellular automata model is modeled based on the assumption of spatial homogeneity, the adjacent relation of cells is generally defined by adopting a consistent rule, the spatial distribution complexity characteristics such as uneven density distribution and various forms of geographic entities are not considered, the difference of land parcel interaction ranges is not reflected, meanwhile, the cell conversion probability and the complex land use interaction of the comprehensive influence of multiple land categories are not concerned enough, so that the change rule of land use cannot be truly reflected, and a plurality of problems to be deeply researched urgently exist, such as: (1) the vector cell proximity relation identification rule does not take into account the complexity of the land use spatial distribution. (2) The vector cell neighborhood range does not take into account the difference in local interaction range. (3) The vector cellular neighborhood effect influence calculation does not consider the diversity of the land utilization space interaction mode.
Disclosure of Invention
The purpose of the invention is as follows: aiming at the problems, the invention aims to provide a neighborhood effect mode construction and CA simulation method and system considering spatial heterogeneity, and the division of irregular cell neighborhoods of a vector cellular automaton is carried out by introducing a Delaunay triangulation network clustering method under multiple constraints; and the similarity and difference of the cellular transformation rules in different neighborhoods are mined by clustering the distance attenuation sequences of the spatial measurement factors of the historical data and establishing a neighborhood effect mode library. The method can take the spatial distribution complexity characteristics of uneven density distribution, different forms and the like of geographic entities into consideration, reflect the difference of interaction ranges of land parcels and truly reflect the land utilization change rule.
The technical scheme is as follows: in order to achieve the purpose, the invention adopts the following technical scheme:
a neighborhood effect mode construction method considering spatial heterogeneity comprises the following steps:
(1) acquiring basic geographic information data and land utilization data, and performing large land block segmentation and small land block elimination on the non-construction land according to a set threshold;
(2) extracting a landform unit centroid to construct a Delaunay triangulation network, and deleting edges of the triangulation network by using edge length constraint to simplify, so that the simplified edges of the triangulation network identify the spatial proximity relation of cells of the cellular automaton model; wherein each ground unit corresponds to a cell in the cellular automaton model;
(3) clustering the Delaunay triangulation network by using boundary, area and compactness constraints and attribute similarity constraints to obtain the neighborhood range of each central cell;
(4) constructing a final constraint type cellular multi-order irregular neighborhood by combining the spatial neighborhood identified by the Delaunay triangulation network and a neighborhood range;
(5) using the vector neighborhood index as a spatial measurement index, and introducing a distance attenuation function based on a neighborhood order to construct neighborhood index distance attenuation sequences of different land types in the neighborhood of each central cell;
(6) clustering neighborhood index distance attenuation sequences of all central cells in historical data to construct a neighborhood effect mode library, and replacing Euclidean distance with DTW distance in the clustering process.
Preferably, in the step (1), a mesh having a mesh size equal to an average area of the construction sites is used to divide the non-construction sites having an area larger than a set first threshold; and deleting the plots with the areas smaller than the set second threshold value to eliminate noise points and avoid overlong calculation time of the cellular automaton model.
Preferably, the boundary constraint in step (3) specifies a neighborhood without considering the neighbor relation of the physical barrier; area constraint refers to an area index
Not greater than a set area threshold, wherein AzjRepresenting a parcel z in partition ijA μ represents the average area of each land of the partition, NiRepresenting the number of blocks in partition i; compactness constraint refers to compactness
Not less than a set compactness threshold, wherein SZiIs the sum of the areas of the blocks within partition i after merging, SCiIs the area of the minimum circumscribed circle after the combination of the blocks in the partition i; attribute similarity constraint refers to attribute similarity
Is not more than the set attribute similarity threshold value, Sm,nIs the semantic distance of the type of land to which the block m and the block n in the partition i belong.
Preferably, the triangulation method under multiple constraints in step (3) includes the following steps:
(3.1) inputting an area threshold, a compactness threshold, an attribute similarity threshold and a triangular network which is established according to land utilization data and used for finishing side length constraint processing and physical obstacle processing;
(3.2) constructing a seed point list from the land parcel in a random manner; the seed points are used as the starting points of the clustering clusters, and other unprocessed points are continuously merged according to constraint conditions;
(3.3) taking out a seed point, if the seed point is not processed, calculating the compactness of each neighbor after being added into the cluster, sequencing the neighbors from large to small according to the compactness, and continuing the step (4); otherwise, the seed point is removed from the seed point list, and the next seed point is selected;
(3.4) selecting the neighbor rho with the highest compactness, respectively calculating the area index, the compactness and the attribute similarity, and judging whether the area index is not larger than the area threshold value input by the user, whether the compactness is not smaller than the compactness threshold value input by the user and whether the attribute similarity is not larger than the attribute similarity threshold value input by the user;
(3.5) if the neighbor rho meets all the conditions, adding the neighbor rho into the cluster, updating the neighbor again, and repeating the steps (3.4) and (3.5); otherwise, continuing to judge the next neighbor;
(3.6) if all neighbors are processed, the subclass clustering is completed, the seed points are taken out, and the steps (3.3) - (3.6) are repeated
And (3.7) finishing processing all the seed points, finishing clustering, and outputting a partitioning result to obtain a neighborhood range constraint layer.
Preferably, in the step (5), the neighborhood exponential distance decay sequence of ground type k in the neighborhood of the central cell c is represented as:
Scenec,k＝{vNI1,vNI2,…,vNIn}
vNI therein1,vNI2,…,vNInRespectively representing a 1 st order neighborhood index, a 2 nd order neighborhood index, … … and an n th order neighborhood index, wherein the n th order neighborhood index is composed of the ratio of the sum of the areas of the cells of the land type k in the n th order neighborhood of the central cell c to the sum of the areas of all the cells.
Preferably, the DBSCAN algorithm based on the DTW distance is adopted in the step (6) to cluster the neighborhood exponential distance attenuation sequences; and adding a constraint window when calculating the DTW distance of the two sequences in the clustering process.
Preferably, the neighborhood effect pattern library constructed in the step (6) is expressed as:
wherein,
in order to be able to use the number of right-of-way type,
representing all neighborhood effect mode conversion potentials of central cells with the land type of x and the neighborhood land type of k in a research area, eta representing the number of curve types clustered by neighborhood effect modes with the land type of x, g representing one curve type clustered by the neighborhood effect modes,
wherein, LUx-＞yRepresents the center cell with the ground type shifted from x to y in the g-th neighborhood effect pattern type, count represents the count function,
indicating the set of neighborhood exponential distance decay sequences with the center cell land type x and the neighborhood land type k in the g-th neighborhood effect mode type, LUxTo represent
The central unit cell in (1).
The land use cellular automaton model simulation method based on the neighborhood effect mode construction method considering spatial heterogeneity comprises the following steps:
for a central cellular with the land type x and a neighborhood distance attenuation sequence with the neighborhood land type k, finding a sequence most similar to the sequence in a neighborhood effect mode library to obtain the type of a neighborhood effect mode, and calculating the conversion probability of the central cellular to the land type y in the mode
Wherein g' is a curve type corresponding to the most similar sequence of the neighborhood effect mode library;
the conversion probability under the influence of the neighborhood space scene is calculated by the following formula:
wherein
The number of land types;
the conversion probability under the influence of the local neighborhood is calculated by the following formula:
Plocal|x-＞y＝Pscene|x-＞y×PvNI|x-＞y
wherein alpha isx,y,dRepresenting the area of a ground-type y cell within the d-order neighborhood of a center cell of ground-type x, αx,dRepresenting the area of all cells in the d-order neighborhood of the central cell with the right type x;
determining local constraint P in cellular automata model by finding the most similar sequence in a neighborhood effect pattern librarylocal|x-＞yTo determine whether the central cell is transformed and the direction of the transformation, to perform land use simulation using the cellular automaton model.
Based on the same inventive concept, the neighborhood effect mode construction system considering the spatial heterogeneity provided by the invention comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, and when the computer program is loaded to the processor, the neighborhood effect mode construction method considering the spatial heterogeneity is realized.
Based on the same inventive concept, the land use cellular robot model simulation system provided by the invention comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein the computer program realizes the land use cellular robot model simulation method when being loaded to the processor.
Has the advantages that: compared with the prior art, the invention has the following advantages:
1. vector cellular heterogeneous multi-level neighborhood delineation accounting for spatial heterogeneity. The existing vector cellular automata model is based on the assumption of space homogeneity, neighborhood planning with a fixed order, such as one-order or three-order, is generally adopted when the neighborhood planning is carried out, and the planning method does not consider the difference of uneven density distribution of geographic entities and interaction range of land parcels, and is not sufficient for paying attention to the cell conversion probability and the complex land utilization interaction which are comprehensively influenced by various land parcels. The heterogeneous multi-order neighborhood division method provided by the invention considers the spatial heterogeneity, divides different multi-order irregular neighborhoods of each vector cell by taking the range of the cluster as a boundary on the basis of the Delaunay triangulation network clustering result under the multiple constraints of the land mass, can consider the uneven distribution of the density of geographic entities, and can reflect the difference of the interactive influence range of land utilization more truly.
2. And establishing a neighborhood effect mode library based on a clustering algorithm. The existing model calibration, expert knowledge assessment and empirical analysis based on mathematics and statistics establish a neighborhood effect mode library, the former has huge workload, and the latter is easy to have the problems of subjectivity and one-sidedness. Firstly, on the basis of a constructed constrained multi-order neighborhood, a vector neighborhood index is used as a space measurement index for describing space homogeneity and space heterogeneity of a specific landscape to express neighborhood influence; then, considering that the neighborhood influence has heterogeneity in different distances, introducing a distance attenuation function to calculate neighborhood exponential distance attenuation sequences of various land types around the central plot; secondly, clustering is carried out by using a DBSAN algorithm according to the neighborhood index distance attenuation sequence data of different places, and a parameter combination with the optimal clustering result is selected by using a contour coefficient, so that various neighborhood distribution modes of cell neighborhoods are condensed, and the cell neighborhood distribution modes are called neighborhood effect modes; and finally, counting the conversion probability corresponding to different neighborhood effect modes, and constructing a neighborhood effect mode library to lay a foundation for calculating the neighborhood influence in the next step.
3. Center cell similarity matching is performed based on the improved DTW distance. To calculate the neighborhood impact of a central cell, a similarity measure of the DTW distance in the signaling is introduced into the cell neighborhood transformation rules. Because the distance dimension of the neighborhood exponential distance decay curve is actually the order of the neighborhood rather than the absolute distance, and the final matching result may have semantic errors due to excessive stretching and displacement, the method adds a constraint window on the basis of the conventional DTW distance algorithm, thereby solving the problem of sequence deformation, reducing the calculation amount and improving the algorithm efficiency. Through the improved DTW distance, the similarity between the neighborhood exponential distance decay sequence curve of the central cell and the neighborhood effect mode curve is measured, so that the neighborhood effect mode curve most similar to the central cell decay sequence curve is obtained, and the conversion probability of different land use types of the central cell is obtained.
Drawings
FIG. 1 is a flowchart of the irregular neighborhood partition and influence calculation of a cell according to an embodiment of the present invention.
FIG. 2 is a schematic diagram of a large plot partitioning according to an embodiment of the present invention; wherein, (a) indicates a plot to be divided; (b) schematically the block segmentation results.
FIG. 3 is a schematic diagram of Delaunay triangulation network clustering under side length constraint in the embodiment of the present invention; wherein, (a) is an initial triangulation network; (b) after the long edge is deleted.
FIG. 4 is a schematic diagram of boundary constraints in an embodiment of the present invention; wherein, (a) is before the boundary constraint process; (b) after the boundary constraint processing.
FIG. 5 is a diagram of a hierarchical structure of land based semantics in an embodiment of the present invention.
Fig. 6 is a flowchart of the Delaunay triangulation algorithm under multiple constraints in the embodiment of the present invention.
FIG. 7 is a flow chart of constrained multi-level neighborhood construction according to an embodiment of the present invention.
FIG. 8 is a diagram illustrating a constrained multi-level neighborhood in an embodiment of the invention.
FIG. 9 is a schematic diagram of a sample data neighborhood space scenario in an embodiment of the present invention; wherein, (a) is a land use status map; (b) the method is a multi-order neighborhood schematic diagram, wherein black represents a 1-order neighborhood, and black with a light degree represents a 2-order neighborhood, and so on; (c) is the change in the neighborhood space scene vNI for that point.
Fig. 10 is a DTW path diagram of a constrained window in an embodiment of the present invention.
Fig. 11 is a flow chart of neighborhood exponential distance decay sequence clustering in the embodiment of the present invention.
Fig. 12 is a schematic diagram of cell transition probability under the influence of a neighborhood space scenario in the embodiment of the present invention.
FIG. 13 is a schematic diagram illustrating a block transform probability calculation under local constraints according to an embodiment of the present invention.
Detailed Description
The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings and specific embodiments.
In order to make up for the deficiency of the existing vector cellular automaton cellular neighborhood effect model research and fully consider spatial heterogeneity, the embodiment of the invention provides a neighborhood effect model construction method taking spatial heterogeneity into consideration by taking a vector cellular automaton neighborhood as a research object. The specific process is shown in fig. 1, and mainly comprises the following steps:
s1: pre-processing of basic geographic information data and land use data
In the research, the Jiangyin city is selected as a research area, and the required data mainly comprises land utilization data and basic geographic information data of the Jiangyin city. Wherein the land use data comprises three years 2007, 2012, and 2017. The land types mainly comprise residential land (R), commercial land (C), industrial land (M), public service land (A), transportation land (S), village construction land (V), agricultural land (F) and other land (O), and are respectively correspondingly coded as 1-8. The basic geographic data mainly comprises data of roads of all levels, administrative divisions, block divisions, water systems, gradients and the like, and can be flexibly selected and adjusted according to application requirements. The data used are summarized in table 1:
table 1 data description of the model
The raw vector land use data is simply pre-processed, including large block segmentation and small block elimination by mean and standard deviation. The urban land utilization change process is a random and dispersed process, and in order to avoid that the model precision is too low due to too great differences of land sizes, large non-construction land is segmented, the specific method is as follows:
(1) calculating the mean value mu according to the whole non-construction landiAnd standard deviation σi
(2) To make the area larger than mui+2σiIs identified as a plot to be segmented
(3) Calculating the average area mu of the construction landjUsing a grid size of μ as shown in FIG. 2jThe grid divides the land blocks to be divided
The elimination of the tiny patches is to eliminate noise points and avoid too long model running time caused by too many unimportant tiny patches to be calculated. Similar to the above-described large parcel identification approach, the small parcel can be identified using the mean and standard deviation definition thresholds in combination with the actual situation.
S2: delaunay triangulation network recognition cellular neighborhood based on space position constraint
The Delaunay triangulation network is composed of a series of non-overlapping triangles, and can effectively depict city space structures and express spatial proximity relations. The side length of the triangulation network stores rich information, and the average value and the standard deviation of the whole side length of the triangulation network are used as important attributes of the triangulation network, so that the whole spatial distribution characteristics of a data set can be well reflected. The triangular network is constructed by taking the mass center of the plot as a node, so that the urban space can be converted into a 'point-line' abstract space of a graph, the topological relation of the space is expressed by the graph, and the triangular network is subjected to edge deletion simplification by utilizing the edge length constraint criterion. The side length constraint comprises integral side length constraint and local side length constraint, after the integral long side constraint is defined through a mean value and a standard deviation, a basic form of a cluster can be obtained after a long side is deleted, but some local long sides still exist, so that the clustering effect is not ideal, and therefore after the triangular network is roughly divided, the local side length constraint is respectively applied to the sub-clusters. Finally, the sides of the triangle are used as the basis for identifying the adjacent relationship, as shown in FIG. 3. The nodes store information such as land utilization types and adjacent plots, and mainly store information related to neighborhood judgment, such as edge lengths and node information at two ends.
S3: clustering by utilizing Delaunay triangulation network under multiple constraints to obtain neighborhood range
And dividing the interaction range of the land parcel by utilizing a Delaunay triangulation algorithm under multiple constraints, and further establishing a constraint type multi-order neighborhood to lay a foundation for next neighborhood effect influence calculation.
To ensure uniformity within a sub-region, boundary continuity is an important condition. Therefore, an index design clustering algorithm which can make the sub-regions as compact and similar as possible is selected. In the embodiment, the constraint condition is divided into a space constraint index and an attribute constraint index, wherein the former comprises boundary, area and compactness constraint, and the latter refers to the type semantic similarity of the use place.
(1) Boundary constraint
Since the interaction between the plots is interrupted by physical obstacles such as blocking of rivers, canyons, major roads, etc., it is necessary to exclude the influence of this portion when defining the neighborhood, as shown in fig. 4.
(2) Area index
The area index is used for generating partitions with more consistent unit sizes, and the index is defined as the following formula (1):
in the formula, AzjRepresents a land z in the partition ijA μ represents the average area of each land of the partition, NiRepresenting the number of blocks within a partition. The smaller the area indicator, the closer the size of the plot in the partition.
(3) Degree of compactness of shape
Compactness is often used to reflect city space morphology, and higher value thereof means that city functional areas are more compactly distributed, which is more beneficial to city management. The embodiment uses the compactness index to constrain the shape of the partitions to produce more regular partitions while avoiding destroying the topological relationship of the plot itself. The plots in the sub-areas are first merged and then defined as the ratio of the actual area of the combined shape to its minimum circumscribed circle, expressed as equation (2):
in the formula, RiIs the compactness of partition i, SZiIs the sum of the areas, SC, after the plots are mergediIs the area of its smallest circumscribed circle. Relative to the one-dimensional compactness index, RiThe index value is an index value under two-dimensional measurement, and actually reflects the similarity degree of the combined shape of the partition sub-units and the reference shape. The closer the value is to 1, the more regular the representative shape is, the more similar it is to the reference shape, and conversely, the more different it is, the more dispersed the distribution thereof is.
(4) Similarity of attributes
The land use situation in the real world is complicated, any possible land use type can appear around a land, and the land use properties of a plurality of land use units contained in a natural partition can be different. Therefore, in addition to the geometric constraint, the land property thereof needs to be synthesized, and the property similarity is defined, so that the partitions are not only regular in shape and uniform in size, but also similar or compatible in land property. According to the classification of urban land and the standard of planning construction land (GB50137-2011), the urban construction land is divided into three types, namely large, medium and small, and according to different researches or applications, the small types under the same medium type can be selectively combined, and the combination of the large types needs to consider the mutual relation between land types. Such as residential land and commercial sibling urban construction land, and have suction among each other, so can merge; the residential land and the industrial land have repulsive action in a short distance and cannot be combined. The present embodiment combines the industrial sites and the logistics storage sites of the 8-site type, and combines them into the public service sites, considering that the public utility service sites are small in the research area and belong to the same service type as the public management and public service sites. The merged land use type code and semantic hierarchy are as shown in FIG. 5.
The semantic hierarchy chart actually implies semantic distances between the types of places. Setting the semantic distance of the same type of land as 1; the residential land, the commercial land and the public service land belong to urban construction land, are closely connected with one another, are close in position in the figure, are far away from one another in semantic distance, and are set to be 2; industrial land, transportation land and residential land, commercial land and public service land belong to urban construction land, agricultural land and other land types such as mountain land belong to non-construction land, are closely related to each other and are set to be 3; although village construction land and urban construction land belong to the same construction land, due to urban and rural difference, the relation is general and is set as 4, and although the village construction land and agricultural land have different land properties, the semantic distance is set as 4 due to the same rural area; the difference of land property between the construction land and the non-construction land is obvious, direct connection is less, and the semantic distance is set to be 5. The resulting semantic distances between the destination types are shown in table 2.
Table 2 land type semantic distance
For a certain partition i, the attribute similarity is defined as the average semantic distance between all the plots in the range according to the semantic distance. Firstly, establishing a semantic distance matrix among all the land parcels, and then averaging all the values in the matrix, as shown in formula (3):
in the formula, SiIs the attribute similarity index value of partition i, Sm,nIs the semantic distance, N, of the type of land to which the plot m and the plot N belongiIs the number of plots for partition i.
According to the boundary limit, area, compactness and attribute similarity indexes, a triangulation algorithm (as shown in figure 6) under space and attribute multiple constraints is designed, and further the infinite extension of the neighborhood is limited. The specific clustering process is as follows:
(1) inputting data: the method comprises the steps of establishing and completing a triangulation network of side length constraint processing and physical barrier processing according to land utilization data by using an area threshold, a compactness threshold and an attribute similarity threshold.
(2) A seed point list is constructed from the plot in a random manner. And the seed points are used as the starting points of the clustering clusters, and other unprocessed points are continuously merged according to the constraint conditions.
(3) Taking out a seed point, if the seed point is not processed, calculating the compactness of each neighbor after being added into the cluster, sequencing the neighbors from large to small according to the compactness, and continuing the step (4); otherwise, the seed point is removed from the seed point list, and the next seed point is selected.
(4) Selecting the neighbor rho with the highest compactness, respectively calculating the area index, the compactness and the attribute similarity, and judging whether the area index is not larger than the area threshold value input by the user, whether the compactness is not smaller than the compactness threshold value input by the user and whether the attribute similarity is not larger than the attribute similarity threshold value input by the user. (constraint conditions)
(5) If the rho point meets all the conditions, adding the rho point into the cluster, updating the neighborhood again, and repeating the steps (4) and (5); otherwise, the next neighborhood is continuously judged.
(6) If all neighbors are processed and no point satisfying the condition exists, the subclass clustering is completed, the seed point is taken out, and the steps (3) to (6) are repeated
(7) And finishing the processing of all the seed points, finishing the clustering, and outputting a partitioning result, namely a neighborhood range constraint layer.
S4: constructing heterogeneous constrained cellular multi-order irregular neighborhood
By constructing the constrained multi-order neighborhood, the effective information of the neighborhood is fully acquired, and meanwhile, the neighborhood is prevented from expanding outwards without limit, and the main construction steps are shown in fig. 7.
The construction of the heterogeneous constrained neighborhood mainly comprises the following three steps:
the first step is as follows: and determining the connection relation between the points based on the triangulation network spatial clustering under the side length constraint, and further identifying the adjacent relation of the cells of the cell automaton model.
(1) Extracting a geocellular centroid to construct a Delaunay triangulation network;
(2) adding integral long side and local long side constraints to the triangular net;
(3) and checking the topological relation to obtain the adjacent relation.
The second step is that: and (4) dividing the neighborhood range by utilizing a triangulation algorithm under multiple constraints of space and attributes.
(1) Determining an area index, compactness and attribute similarity threshold;
(2) superposing boundary limiting layers such as main roads, rivers, administrative division layers and the like, and preliminarily dividing a data set;
(3) clustering a plurality of clusters to obtain a neighborhood range image layer.
The third step: the final constrained cell multi-order irregular neighborhood constructed by combining the spatial neighborhood identified by the Delaunay triangulation and the neighborhood range is shown in FIG. 8. The central unit cell in the figure respectively searches the 1 st order neighborhood and the 2 nd order neighborhood … … through the connection relation of the points until the boundary of the neighborhood range is touched, and the search is stopped. Because the neighborhood ranges are different, each cell has neighborhoods with different orders, so that the range difference of land use interaction is reflected.
S5: construction of neighborhood exponential distance decay sequences
The neighborhood characteristics and the measurement thereof are the basis for expressing the neighborhood effect, and the calculation of the influence of the neighborhood effect is a key component of a cellular automaton model. Spatial metrics are a method for describing spatial homogeneity and spatial heterogeneity of a specific landscape, and have an important role in analyzing land use spatial structure and distribution patterns. Therefore, a distance attenuation function is introduced and described using the concept of "neighborhood space scenario".
(1) Spatial metric index
Vector Neighborhood Index (NI) was used as a spatial metric Index. Firstly, the ring zones are established around the central unit cell, then the NI index is calculated for each ring zone, and finally the NI index of each ring zone is synthesized to establish a neighborhood rule. As in equation (4):
in the formula, alphac,k,dIs the cell radius of cIs the sum of the areas of land utilization types k in d neighborhoods, alphac,dIs the sum of the neighborhood areas within the neighborhood of c cell radius d.
(2) Neighborhood space scenario
In order to better express the attraction and repulsion effect between land utilization, the embodiment introduces a distance attenuation function to construct a neighborhood exponential distance attenuation sequence of various land uses around a central plot, and describes the neighborhood characteristics by using a concept of "neighborhood space scene".
Scenec,k＝{vNI1,vNI2,…,vNIn} (5)
Wherein, Scenec,kThe space distribution characteristic of the plot type k for the neighborhood space scene of the central cell c is represented, and the neighborhood index of the plot type k is { vNI }1,vNI2,…,vNInTherein vNI1Representing a first order neighborhood index, vNI2Representing the second order neighborhood index, and so on, the sequence is shown in a curved form. A neighborhood space scenario for a certain sample of data is schematically shown in fig. 9.
S6: calculation of neighborhood effect patterns
In the embodiment, a local typical neighborhood effect mode is condensed by clustering neighborhood exponential distance decay sequences in historical data, and corresponding conversion probability is recorded, so that a land utilization interactive differentiation rule implied in a vector space is mined.
The "distance dimension" of the neighborhood exponential distance decay curve of the present embodiment is actually the order of the neighborhood rather than the absolute distance, a certain degree of stretching or displacement has little influence on the analysis of the model, but excessive stretching and displacement may cause semantic errors in the final result of matching. In contrast, in the embodiment, a constraint window is added on the basis of the conventional DTW distance algorithm, so that the problem of sequence deformation can be solved, the calculation amount can be reduced, and the algorithm efficiency can be improved. For two curves R ═ R (R)1，…，rp)，S＝(s1，…，sq) P, q > 1, modified DTW distance as in equation (6):
where w represents the size of the constraint window and DTW (u, v) represents the point-by-point matching of the two curves to ruAnd svMinimum cumulative distance of points, dist (r)u,sv) Represents a point ruAnd point svThe euclidean distance between them. As can be seen from equation (6), the constraint window can be actually understood as a matching point index constraint, i.e. the index distance between the current point and the matching point cannot exceed w. The window constraint of size 3 is shown in fig. 10, where the gray part represents the limit range of the point matching path of the DTW algorithm and the black square part represents the best matching path.
In this embodiment, a neighborhood exponential distance decay sequence clustering algorithm is designed by combining DTW similarity and DBSCAN clustering, and a unique ID is assigned to each distance decay sequence, so that sequence data can be easily associated with a point set in an original DBSCAN algorithm. In the clustering process, the Euclidean distance between two points is replaced by the DTW distance of the distance attenuation sequence, and the minimum search radius corresponds to the minimum similarity.
Inputting a sequence data set L, parameters MinPts and Eps, and outputting a clustering result Cluster array, wherein the value of the clustering result Cluster array represents the clustering Cluster to which each sequence curve belongs. For ease of understanding, the following description is still in terms of a "set of points", LhRepresenting a point in the data set, actually corresponding to the curve ID. The algorithm flow is as follows:
(1) traverse the data set L if the current point LhCan reach the number of points NEps (L)h) If the minimum Pts is larger than the minimum Pts, selecting the point as a seed point, and setting the cluster as g;
(2) centering on the seed point, if the similarity between two points (DTW distance between the two curves represented) is less than Eps, the point is merged to cluster g; merging the points which meet the minimum density constraint, namely core points, into the sub data set S;
(3) traversing the subdata set S, and repeating the step (2) by taking the current point as a center until no point to be processed exists in the subdata set;
(4) and (3) repeating the steps (1), (2) and (3) until no point to be processed exists in the data set L.
The specific implementation details of the above algorithm are shown in fig. 11.
In order to evaluate the scientificity and rationality of clustering combination and obtain the optimal clustering parameters under the condition of lack of experimental knowledge, the embodiment evaluates the result by using the contour coefficients, firstly finds the approximate range of the parameters through large-range iteration, then further reduces the iteration within the range, and selects the parameter combination which enables the contour coefficients of the clustering result to be maximum as the optimal parameter combination. The degree of agglomeration and the degree of separation of the clusters are two important indexes for describing clustering results, wherein the degree of agglomeration expresses the degree of tightness of the samples, and the degree of separation expresses the degree of separation of the samples. In general, more classification numbers means less degree of aggregation and greater degree of separation. The contour coefficient is established based on the degree of agglomeration and the degree of separation, and the clustering result is evaluated without relying on prior knowledge.
For a sample L in a data set L of N sample datahThe contour coefficient is defined as formula (7):
in the formula, bhIs LhAverage distance to samples in the nearest neighboring cluster (this example refers to DTW distance between sequences), ahIs LhAverage distance to other sample points within the cluster. ShThe method can be used for evaluating whether the cluster where one point is located at present is suitable, the value of the cluster is between-1 and 1, the closer to 1 represents that the distance from the sample to each point in the cluster is far less than the distance from the sample to other clusters, and the better the effect is; conversely, a closer to-1 indicates a poorer effect.
For a certain clustering process, the contour coefficients of the entire data set are defined as formula (8):
on the basis of the congealed non-construction land cell neighborhood effect mode, corresponding conversion probability is counted according to historical conversion conditions and recorded in a neighborhood effect mode library. The following describes specific details of calculating neighborhood effect patterns for a certain type of land use. Recording that the central cell set in the research area is C, the land type is x, the land type set in the neighborhood of the central cell is K, the order set of the neighborhood index is D, eta represents the number of curve types clustered by using a neighborhood effect mode with the land type being z, and the clustering of vNI distance attenuation sequences of i types of lands in non-construction neighborhoods in the research area is expressed as formulas (9) to (12):
Ix,k,Lx,k＝Cluster(Scenec,k|c∈C,k∈K) (9)
in the formula, Scenec,kA neighborhood exponential distance attenuation sequence representing the land type k in the neighborhood of the central cell c; i isx,kRepresenting a set of neighborhood exponential distance attenuation sequences with the central cellular land type of x and the neighborhood land type of k in the clustering cluster; l isx,kExpressing a set obtained by averaging all orders of neighborhood indexes of a neighborhood index distance attenuation sequence with the central cellular land type of x and the neighborhood land type of k in a cluster, and L of different land utilization typesx,kForming a neighborhood effect mode library with the land type of x;
is d-order neighborhood index of one cell in the g-th type clustered by neighborhood effect mode with the type x (subscript number represents the sequence of the cell in the g-th type)Number); avg is an averaging function that averages over unequal length sequences. Further, the probability of urging the middle cell to turn from the right type x to the left type y, which is calculated from the g-th type clustered by the neighborhood effect pattern with the right type x, can be obtained by formula (13):
in the formula, LUx-＞yRepresents the center cell with the ground type shifted from x to y in the g-th neighborhood effect pattern type, count represents the count function,
representing the set of neighborhood exponential distance decay sequences with the center cell land type x and the neighborhood land type k in the g-th neighborhood effect mode type, LUxTo represent
The central unit cell in (1). The neighborhood effect mode conversion potential of a central cell with a right-type x and a neighborhood right-type k-turn right-type y in the study area is represented as formula (14):
thus, the influence of the neighborhood space scene on the central cell is calculated. Note the book
For the number of land types, a case base containing all neighborhood effect modes and corresponding conversion probabilities is established as formula (15):
the neighborhood effect mode and the case library of the corresponding conversion probability thereof established through the steps can truly reflect the change rule of land utilization, and the cellular automaton model for land utilization can be accurately simulated on the basis of the change rule of land utilization.
The land use cellular automaton model simulation method provided by the embodiment of the invention specifically comprises the steps of inputting a central cell with a land use type x and a neighborhood distance attenuation sequence of a neighborhood land use type k in an application stage, searching a sequence most similar to the sequence in a case library, and outputting the conversion probability of the central cell in a land use distribution mode
Wherein g' is the type corresponding to the most similar sequence of the neighborhood effect mode library, and the conversion probability under the influence of the neighborhood space scene is obtained through a formula (17).
As shown in fig. 12, the cell transition probability under the influence of the neighborhood space scenario is represented by the combined influence of the type interactions of the respective places. The bold curve in the figure represents the curve in the neighborhood effect pattern library that is most similar to the distribution pattern of the current cellular sites.
The neighborhood space scenario is actually a spatial distribution pattern that describes different land utilizations, emphasizing the effect of its "density" decay trend on the central cell. Considering that the autocorrelation function of the similar neighborhoods is still one of the non-negligible influence factors, the local neighborhood influence is jointly expressed by the ground vector neighborhood index in combination with the neighborhood space scene and the conversion direction as the formula (18):
Plocal|x-＞y＝Pscene|x-＞y×PvNI|x-＞y (18)
in the formula, alphax,y,dRepresenting the area of a ground-type y cell within the d-order neighborhood of a center cell of ground-type x, αx,dRepresenting the area of all cells in the d-order neighborhood of the central cell of type x.
FIG. 13 demonstrates the calculation of the probability of a transition of a near factor for an agricultural site. According to vNI distance attenuation sequence data of the land type of the neighborhood of the plot, the most similar judgment results are respectively R-1, C-0, M-0, A-1, S-0, V-4 and F-2, wherein a letter R, C, M, A, S, V, F represents a neighborhood land type code; the numbers represent different neighborhood effect patterns. If the transition probabilities recorded in the neighborhood effect pattern library are tabulated in the figure, the probability of the plot turning to residential land under the influence of the neighborhood space scene is 0.333+0.170+0.195+0.174+0.174+0.200+ 0.173-1.419, and then the plot is multiplied by the occupancy ratio of 0.423 in the common 3-step neighborhood, so that the probability of turning to residential land under the influence of local constraint is 0.600, and the probabilities of turning to other land types are analogized in turn.
Determining local constraint P in cellular automata model by finding a sequence most similar to the sequence in a neighborhood effect pattern librarylocal|x-＞yAnd determining whether the central cell is converted and the direction of the conversion, thereby performing land use simulation using the cellular automaton model.
Based on the same inventive concept, the neighborhood effect mode construction system considering spatial heterogeneity provided by the embodiments of the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the computer program is loaded into the processor, the neighborhood effect mode construction method considering spatial heterogeneity is implemented.
Based on the same inventive concept, the land use cellular automaton model simulation system provided by the embodiment of the invention comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein the computer program realizes the land use cellular automaton model simulation method when being loaded to the processor.
In summary, the invention is based on the modeling of the spatial heterogeneity hypothesis, adopts the heterogeneous rule to define the neighborhood relationship of the cells, considers the spatial distribution complexity characteristics of uneven density distribution, various shapes and the like of geographic entities, embodies the difference of interaction ranges of land parcels, and can truly reflect the land utilization change rule. According to the method, the local typical neighborhood effect mode is compiled from historical data, and the corresponding conversion probability is recorded, so that the land utilization interactive differentiation rule hidden in the vector space is mined, and thus, the numerical calibration with huge workload is avoided, and the problems of subjectivity and one-sidedness are not easy to occur. The method is objective, and simultaneously considers the characteristic that the neighborhood effects in different areas have differences. Based on the method, the division and influence calculation of the irregular neighborhood of the vector cellular automaton can be carried out under the condition of considering the spatial heterogeneity, and the theory and method research of the vector cellular automaton on the interaction of the complex neighborhood and the land utilization is perfected and improved, so that the cellular automaton is promoted to have more geographical characteristics in the urban research field.
Claims (10)
1. A neighborhood effect mode construction method considering spatial heterogeneity is characterized by comprising the following steps:
(1) acquiring basic geographic information data and land utilization data, and performing large land block segmentation and small land block elimination on the non-construction land according to a set threshold;
(2) extracting a landform unit centroid to construct a Delaunay triangulation network, and deleting edges of the triangulation network by using edge length constraint to simplify, so that the simplified edges of the triangulation network identify the spatial proximity relation of cells of the cellular automaton model; wherein each ground unit corresponds to a cell in the cellular automaton model;
(3) clustering the Delaunay triangulation network by using boundary, area and compactness constraints and attribute similarity constraints to obtain the neighborhood range of each central cell;
(4) constructing a final constraint type cellular multi-order irregular neighborhood by combining the spatial neighborhood identified by the Delaunay triangulation network and a neighborhood range;
(5) using the vector neighborhood index as a spatial measurement index, and introducing a distance attenuation function based on a neighborhood order to construct neighborhood index distance attenuation sequences of different land types in the neighborhood of each central cell;
(6) clustering neighborhood index distance attenuation sequences of different land types of each cell in historical data to construct a neighborhood effect mode library, and using DTW distance to replace Euclidean distance in the clustering process.
2. The neighborhood effect model construction method considering spatial heterogeneity according to claim 1, wherein in the step (1), a mesh with a mesh size equal to an average area of construction lands is used, and non-construction lands with an area larger than a set first threshold are segmented; and deleting the plots with the areas smaller than the set second threshold value to eliminate noise points and avoid overlong calculation time of the cellular automaton model.
3. The neighborhood effect model construction method considering spatial heterogeneity according to claim 1, wherein the boundary constraint in step (3) defines a neighborhood without considering the neighbor relation of the physical obstacles; area constraint refers to an area index
Not greater than a set area threshold, wherein AzjRepresenting a parcel z in partition ijA μ represents the average area of each land of the partition, NiRepresenting the number of blocks in partition i; compactness constraint refers to compactness
Not less than a set compactness threshold, wherein SZiIs the sum of the areas of the blocks within partition i after merging, SCiIs the area of the minimum circumscribed circle after the combination of the blocks in the partition i; attribute similarity constraint refers to attribute similarity
Not greater than setThreshold of similarity of attributes, Sm，nIs the semantic distance of the type of land to which the block m and the block n in the partition i belong.
4. The neighborhood effect pattern construction method considering spatial heterogeneity according to claim 3, wherein said triangulation method under multiple constraints in step (3) comprises the following steps:
(3.1) inputting an area threshold, a compactness threshold, an attribute similarity threshold and a triangular network which is established according to land utilization data and used for finishing side length constraint processing and physical obstacle processing;
(3.2) constructing a seed point list from the land parcel in a random manner; the seed points are used as the starting points of the clustering clusters, and other unprocessed points are continuously merged according to constraint conditions;
(3.3) taking out a seed point, if the seed point is not processed, calculating the compactness of each neighbor after being added into the cluster, sequencing the neighbors from large to small according to the compactness, and continuing the step (4); otherwise, the seed point is removed from the seed point list, and the next seed point is selected;
(3.4) selecting the neighbor rho with the highest compactness, respectively calculating the area index, the compactness and the attribute similarity, and judging whether the area index is not larger than the area threshold value input by the user, whether the compactness is not smaller than the compactness threshold value input by the user and whether the attribute similarity is not larger than the attribute similarity threshold value input by the user;
(3.5) if the neighbor rho meets all the conditions, adding the neighbor rho into the cluster, updating the neighbor again, and repeating the steps (3.4) and (3.5); otherwise, continuing to judge the next neighbor;
(3.6) if all neighbors are processed, the subclass clustering is completed, the seed points are taken out, and the steps (3.3) - (3.6) are repeated
And (3.7) finishing processing all the seed points, finishing clustering, and outputting a partitioning result to obtain a neighborhood range constraint layer.
5. The method for constructing neighborhood effect patterns considering spatial heterogeneity according to claim 1, wherein in said step (5), the neighborhood exponential distance decay sequence of type k is used in the neighborhood of the central cell c as:
Scenec，k＝{vNI1，vNI2，…，vNIn}
vNI therein1，vNI2，…，vNInRespectively representing a 1 st order neighborhood index, a 2 nd order neighborhood index, … … and an n th order neighborhood index, wherein the n th order neighborhood index is the ratio of the sum of the areas of the cells of the land type k in the n th order neighborhood of the central cell c to the sum of the areas of all the cells.
6. The neighborhood effect mode construction method considering spatial heterogeneity according to claim 1, wherein in the step (6), a DBSCAN algorithm based on DTW distance is adopted to cluster neighborhood exponential distance decay sequences; and adding a constraint window when calculating the DTW distance of the two sequences in the clustering process.
7. The neighborhood effect pattern construction method considering spatial heterogeneity according to claim 1, wherein the neighborhood effect pattern library constructed in the step (6) is expressed as:
wherein,
in order to be able to use the number of right-of-way type,
representing all neighborhood effect mode conversion potentials of central cells with the land type of x and the neighborhood land type of k in a research area, eta representing the number of curve types clustered by neighborhood effect modes with the land type of x, g representing one curve type clustered by the neighborhood effect modes,
wherein, LUx-＞yRepresents the central cell with the ground type shifted from x to y in the curve type of the g-th neighborhood effect pattern, count represents the count function,
the curve type of the g-th neighborhood effect mode represents a set of neighborhood exponential distance attenuation sequences with the land type x of the central cell and the land type k of the neighborhood, LUxTo represent
The central unit cell in (1).
8. The land use cellular automaton model simulation method based on the neighborhood effect model construction method considering spatial heterogeneity according to any one of claims 1 to 7, comprising the steps of:
for a central cellular with the land type x and a neighborhood distance attenuation sequence with the neighborhood land type k, finding a sequence most similar to the sequence in a neighborhood effect mode library to obtain the type of a neighborhood effect mode, and calculating the conversion probability of the central cellular to the land type y in the mode
Where g' is the most similar of the neighborhood effect pattern libraryThe curve types corresponding to the sequences;
the conversion probability under the influence of the neighborhood space scene is calculated by the following formula:
wherein
The number of land types;
the conversion probability under the influence of the local neighborhood is calculated by the following formula:
Plocal|x-＞y＝Pscene|x-＞y×PvNI|x-＞y
wherein alpha isx，y，dRepresenting the area of a ground-type y cell within the d-order neighborhood of a center cell of ground-type x, αx，dRepresenting the area of all cells in the d-order neighborhood of the central cell with the right type x;
determining local constraint P in cellular automata model by finding the most similar sequence in a neighborhood effect pattern librarylocal|x-＞yTo determine whether the central cell is transformed and the direction of the transformation, to perform land use simulation using the cellular automaton model.
9. A neighborhood effect pattern construction system taking into account spatial heterogeneity, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the computer program when loaded into the processor implements the neighborhood effect pattern construction method taking into account spatial heterogeneity according to any one of claims 1 to 7.
10. A land use cellular automaton model simulation system comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the computer program, when loaded into the processor, implements the land use cellular automaton model simulation method of claim 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110599588.7A CN113343565B (en) | 2021-05-31 | 2021-05-31 | Neighborhood effect mode construction and CA simulation method and system considering spatial heterogeneity |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110599588.7A CN113343565B (en) | 2021-05-31 | 2021-05-31 | Neighborhood effect mode construction and CA simulation method and system considering spatial heterogeneity |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113343565A true CN113343565A (en) | 2021-09-03 |
| CN113343565B CN113343565B (en) | 2024-01-09 |
Family
ID=77472481
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110599588.7A Active CN113343565B (en) | 2021-05-31 | 2021-05-31 | Neighborhood effect mode construction and CA simulation method and system considering spatial heterogeneity |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113343565B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2024103770A1 (en) * | 2022-11-14 | 2024-05-23 | 华为技术有限公司 | Space division method and apparatus, and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102163224A (en) * | 2011-04-06 | 2011-08-24 | 中南大学 | Adaptive spatial clustering method |
| US20170103172A1 (en) * | 2015-10-07 | 2017-04-13 | The Arizona Board Of Regents On Behalf Of The University Of Arizona | System And Method To Geospatially And Temporally Predict A Propagation Event |
| CN112434957A (en) * | 2020-11-27 | 2021-03-02 | 广东电网有限责任公司肇庆供电局 | Cellular automaton-based distribution network line inspection area grid division method |
| US20210150086A1 (en) * | 2019-11-18 | 2021-05-20 | Sidewalk Labs LLC | Methods, systems, and media for generative urban design where maps are generated using fractals |
-
2021
- 2021-05-31 CN CN202110599588.7A patent/CN113343565B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102163224A (en) * | 2011-04-06 | 2011-08-24 | 中南大学 | Adaptive spatial clustering method |
| US20170103172A1 (en) * | 2015-10-07 | 2017-04-13 | The Arizona Board Of Regents On Behalf Of The University Of Arizona | System And Method To Geospatially And Temporally Predict A Propagation Event |
| US20210150086A1 (en) * | 2019-11-18 | 2021-05-20 | Sidewalk Labs LLC | Methods, systems, and media for generative urban design where maps are generated using fractals |
| CN112434957A (en) * | 2020-11-27 | 2021-03-02 | 广东电网有限责任公司肇庆供电局 | Cellular automaton-based distribution network line inspection area grid division method |
Non-Patent Citations (9)
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2024103770A1 (en) * | 2022-11-14 | 2024-05-23 | 华为技术有限公司 | Space division method and apparatus, and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113343565B (en) | 2024-01-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109272170B (en) | A kind of traffic zone dividing system based on Louvain algorithm | |
| CN109493119B (en) | POI data-based urban business center identification method and system | |
| Estivill-Castro et al. | Multi-level clustering and its visualization for exploratory spatial analysis | |
| CN116911055B (en) | Digital twinning-based intelligent city planning management system | |
| CN107330734A (en) | Business address system of selection based on Co location patterns and body | |
| CN107818338B (en) | Method and system for building group mode identification facing map synthesis | |
| CN113158084B (en) | Method, device, computer equipment and storage medium for processing movement track data | |
| CN113343565A (en) | Neighborhood effect mode construction and CA simulation method and system considering spatial heterogeneity | |
| WO2024103770A1 (en) | Space division method and apparatus, and storage medium | |
| CN111310340B (en) | Urban area interaction abnormal relation identification method and equipment based on human movement | |
| Dahal | Effect of different distance measures in result of cluster analysis | |
| Bembenik et al. | Methods for mining co–location patterns with extended spatial objects | |
| Wang et al. | Using clustering methods in geospatial information systems | |
| CN112084609B (en) | Power supply partition dividing method for power industry | |
| Parchitelli et al. | A pre-process clustering methods for the waste collection problem | |
| Goudarzi et al. | A hybrid spatial data mining approach based on fuzzy topological relations and MOSES evolutionary algorithm | |
| CN118229699B (en) | Road point cloud segmentation method based on weighted rejection strategy | |
| Salman et al. | Enhancing Clustering technique to plan social infrastructure services | |
| CN118093673B (en) | Mapping data processing method | |
| Ren et al. | Adaptive road candidates search algorithm for map matching by clustering road segments | |
| Bae et al. | SD-Miner: A spatial data mining system | |
| Varoudis et al. | Beyond Axial Lines: High-Resolution Geometric Analysis of London's Urban Fabric Using LIDAR Scans | |
| Whelan et al. | Comparing two density-based clustering methods for reducing very large spatio-temporal dataset | |
| Liu | Pattern Recognition Method of Urban Road Network Based on Support Vector Machine Algorithm | |
| AZRI | VORONOI CLASSIFIED AND CLUSTERED CONSTELLATION DATA STRUCTURE FOR THREE-DIMENSIONAL URBAN BUILDINGS |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |