max
d

(

T

i
⁢
d

⁢

c

i
⁢
d

wt

)

❘
"[RightBracketingBar]"

+

σ

max
m

max
n

❘
"[LeftBracketingBar]"

c

i
⁢
d

wt

-

max
d

(

T

i
⁢
d

⁢

c

i
⁢
d

wt

)

❘
"[RightBracketingBar]"

❘
"[LeftBracketingBar]"

c

i
⁢
d

wt

-

max
d

(

T

i
⁢
d

⁢

c

i
⁢
d

wt

)

❘
"[RightBracketingBar]"

+

σ

max
m

max
n

❘
"[LeftBracketingBar]"

c

i
⁢
d

wt

-

max
d

(

T

i
⁢
d

⁢

c

i
⁢
d

wt

)

❘
"[RightBracketingBar]"

{SIM}^{wt} = \sum_{d = 1}^{n} T_{id} ❘ V_{i d}^{wt} ❘, where i = 1, 2, 3 \dots,

m, d = 1, 2, 3 \dots, n, w = 1, 2, 3, \dots, l

where

$V_{id}^{wt}$

represents a correlation coefficient between the w^thlandscape node in the residential outdoor environment and a most ideal node with respect to the d^thenvironmental element of the i^thlevel; σ represents a resolution coefficient, indicating the importance of max calculation.

By combining the first aspect, in some implementations of the first aspect, the method further includes: the encoding of the GPS positions and change states of the landscape nodes in the residential outdoor environment is fused with the levels and class names of the residential outdoor environmental elements, landscape approach motivation change states, and GPS position attributes. The organizational structure of a geographical concept set of the residential outdoor environment is preset as a K-level structure tree. When 1<k≤K, the maximum number of branches included in the k^thlevel is defined as

$D_{k}^{\max},$

and the code of the classification level structure tree has

$K + \sum_{k = 2}^{K} D_{k}^{\max} bits .$

By combining the first aspect, in some implementations of the first aspect, the method further includes: the process of building a distribution adjacency network of the landscape nodes based on the initial landscape approach motivation decision-making matrix of the landscape nodes includes:

- building a decision-making matrix S of the distribution adjacency network based on the evolution state and similarity state results of landscape approach motivation of the environmental elements, where the distribution adjacency network C=(w, e) consists of w landscape nodes and e edges, and is an undirected network of the landscape nodes.

By combining the first aspect, in some implementations of the first aspect, the method further includes: calculating, based on actual geographical distances between the landscape nodes in the residential outdoor environment, local adjacency indexes CWU-Call number

$G E_{q}^{t} and {GS}_{q}^{t}$

of the evolution state and similarity state of environmental element landscape approach motivation of the landscape node q in the t^thperiod as follows:

$G E_{q}^{t} = \frac{\sum_{q} T_{q h}^{t} (d) \cdot {EVO}_{q}^{t} - {\overline{EVO}}^{t} \cdot \sum_{q} T_{q h}^{t} (d)}{S_{_EVO} \sqrt{\frac{w \cdot \sum_{q} T_{q h}^{2} (d) - {(\sum_{q} T_{q h}^{t} (d))}^{2}}{w - 1}}}$

${GS}_{q}^{t} = \frac{\sum_{q} T_{q h}^{t} (d) \cdot {SIM}_{q}^{t} - {\overline{SIM}}^{t} \cdot \sum_{q} T_{q h}^{t} (d)}{S_{_SIM} \sqrt{\frac{w \cdot \sum_{q} T_{q h}^{2} (d) - {(\sum_{q} T_{q h}^{t} (d))}^{2}}{w - 1}}}$

where

${EVO}_{q}^{t} and {SIM}_{q}^{t}$

represent evolution state and similarity state attribute values of the q^thlandscape node in the residential outdoor environment in the t^thperiod, respectively;

$T_{q h}^{t} (d)$

represents a spatial weight between the landscape node q and the landscape node h in the t^thperiod, determined by the actual geographical distance between the landscape nodes; EVO^tand SIM^trepresent average values of the evolution state and similarity state attribute values of all the landscape nodes, respectively; S_{_EVO}and S_{_SIM}represent standard deviations of the evolution state and similarity state attribute values of all the landscape nodes, respectively; w represent the quantity of the landscape nodes;

- loading the geographical concept set of the residential outdoor environment and the local adjacency indexes of the landscape nodes into the decision-making matrix S of the distribution adjacency network, and calculating a driving degree

$S_{q h}^{t}$

of the edge connecting the landscape node q and the landscape node h in the t^thperiod, where q≤w, h≤w:

$S E_{q h}^{t} = \frac{w_{q h}^{t} {EVO}_{q}^{t} / \sum_{q \neq h} {EVO}_{q}^{t} + w_{q h}^{t} {EVO}_{q}^{t} / \sum_{q \neq h} {EVO}_{q}^{t}}{G E_{q}^{t} / \sum_{q = 1}^{w} w_{q h}^{t} + G E_{q}^{t} / \sum_{h = 1}^{w} w_{q h}^{t}}$

${SS}_{q h}^{t} = \frac{w_{q h} {SIM}_{q}^{t} / \sum_{q \neq h} {SIM}_{q}^{t} + w_{q h} {SIM}_{q}^{t} / \sum_{q \neq h} {SIM}_{q}^{t}}{G S_{q}^{t} / \sum_{q = 1}^{w} w_{q h}^{t} + G S_{q}^{t} / \sum_{h = 1}^{w} w_{q h}^{t}}$

$S_{q h}^{t} = T_{_SE} {SE}_{q h}^{t} + T_{_SS} S S_{q h}^{t}$

where

${SE}_{qh}^{t} and {SS}_{qh}^{t}$

represent the driving degrees of the adjacency distribution network in the landscape approach motivation evolution state and similarity state between the landscape nodes q and h in the residential outdoor environment in the t^thperiod, respectively;

$w_{q h}^{t}$

represents a non-standardized symmetric spatial interaction matrix between the landscape nodes q and h in the t^thperiod; T_{_SE}and T_{_SS}represent equal distance weights of the driving degrees of the adjacency distribution network in the evolution state and similarity state, respectively.

By combining the first aspect, in some implementations of the first aspect, the method further includes: the process of performing environmental classification decision-making processing on the distribution adjacency network of the landscape nodes to obtain residential outdoor environmental decision-making results includes:

- analyzing evolution state and similarity state data of the landscape approach motivation information of the residential outdoor environment, calculating intra-class sums of squares of deviations WSS and inter-class sums of squares of deviations BSS of the evolution state and similarity state, selecting the level with minimum WSS and maximum BSS as an optimal level, and dividing the evolution state and similarity state into f levels to obtain a level pattern of totally f²landscape approach motivation change states of the residential outdoor environment;
- generating, based on the level pattern of change states of the residential outdoor environment, a decision-making network representing the adjacent spatial distribution of the landscape nodes in the residential outdoor environment, with the driving degrees

$S_{q h}^{t}$

of the landscape approach motivation change states of the landscape nodes and their corresponding environmental elements as edge weights, the adjacency between landscape nodes as the basis for connecting landscape nodes, and the geographical relationship between engineering drawing elements and the driving degrees of the residential outdoor environment nodes;

- analyzing spatial effect intensities of the landscape approach motivation distribution adjacency network of the landscape nodes in the residential outdoor environment, to obtain residential outdoor environment landscape nodes to be improved as follows:

${SEI}^{t} = \sum_{h \neq q} \frac{{AS}_{qh}^{t}}{{NOD}_{q}^{t}}$

where SEI^trepresents the spatial effect intensities of the landscape nodes in the residential outdoor environment in the t^thperiod;

$A S_{q h}^{t}$

represents a driving degree assignment for the landscape approach motivation change state of the landscape nodes q and h in the t^thperiod; when

$S_{qh}^{t} > 0, {AS}_{qh}^{t} = 1; when S_{qh}^{t} < 0, {AS}_{qh}^{t} = 0; {NOD}_{q}^{t}$

represents the number of edges connecting the q^thlandscape node with other nodes in the t^thperiod;

- normalizing the values of spatial effect intensities, calculating intra-class sums of squares of deviations WSS and inter-class sums of squares of deviations BSS of the normalized spatial effect intensities, selecting the level with minimum WSS and maximum BSS as an optimal level to obtain u levels of the spatial effect intensities, and assigning the u levels as u_i=j, for j=1, 2, . . . , u; meanwhile, assigning the level pattern of f²residential outdoor environment change states of the landscape nodes as g_i=i, for i=1, 2, . . . , f², to obtain a renewal and governance priority Pr_{_nod}of the landscape nodes as follows:

$\Pr_{_nod} = 𝓊_{i} \cdot g_{i} .$

In a second aspect, in order to achieve the above objective, the present invention discloses a decision-making system for a residential outdoor environment based on landscape approach motivation, including:

- a data translation module, configured to obtain engineering drawing data and material test data of the residential outdoor environment, and perform a landscape approach motivation test on the material test data of the residential outdoor environment to obtain environmental element data passing the test, where the engineering drawing data of the residential outdoor environment includes plane vector data and GPS positioning data of the residential outdoor environment, and the material test data includes standard image data of residential outdoor environmental elements;
- a state recognition module, configured to delineate landscape nodes in the residential outdoor environment based on the engineering drawing data of the residential outdoor environment and the environmental element data passing the test, and determine evolution and similarity change states of the landscape nodes based on landscape approach motivation information of the environmental elements corresponding to the landscape nodes in the residential outdoor environment;
- a network building module, configured to generate an initial landscape approach motivation decision-making matrix of the landscape nodes and their corresponding environmental elements based on the GPS positions and change states of the landscape nodes in the residential outdoor environment, encode discrete variables of the GPS positions and change states of the landscape nodes in the residential outdoor environment, and build a distribution adjacency network of the landscape nodes based on the initial landscape approach motivation decision-making matrix of the landscape nodes; and
- an environmental decision-making module, configured to perform environmental classification decision-making processing on the distribution adjacency network of the landscape nodes to obtain residential outdoor environmental decision-making results.

Beneficial effects of the present invention are as follows:

The present invention:

(1) In response to the problems of subjectivity and cumbersome processing of cognitive data measurement methods for residential outdoor environments, the present invention combines brain cognitive test analysis technology and records electroencephalogram (EEG) signals of instantaneous stimulus of residential environmental elements in real time. By setting cognitive response determines rules for EEG “approach-avoidance”, required environmental element landscape approach motivation cognitive feature information is extracted from complex and varied EEG signals, to achieve objective collection and rational analysis of environmental cognitive data, thereby improving the accuracy of cognitive feature extraction of the residential outdoor environment.

(2) In response to the differences, fluidity, and insufficient evolutionary analysis of residential outdoor environmental elements, the present invention, based on obtaining the results of environmental element landscape approach motivation indicators, comprehensively considers spatio-temporal range data of elements, establishes an initial landscape approach motivation decision-making matrix of landscape nodes in the residential outdoor environment and their corresponding environmental elements, and determines evolution and similarity change states of the landscape nodes, thereby achieving analysis on the spatio-temporal change states of multi-dimensional environmental elements, and ensuring the efficiency and accuracy of matrix operation of residential outdoor environmental elements.

(3) In response to the problems of low integration, low spatial resolution, and lack of application docking for environmental renewal and governance in residential landscape approach motivation network decision-making, the present invention maps the spatio-temporal change states of approach motivation of the landscape nodes to a geographical space, combines the decision-making matrix of the distribution adjacency network to generate a residential outdoor environmental decision-making network, determines the level pattern and spatial effect intensities of the change states of the landscape nodes, and provides renewal and governance priorities for the residential outdoor environment landscape nodes, thereby improving the capabilities of accurate decision-making and integrated analysis of the residential outdoor environment.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to describe the technical solutions in the embodiments of the present invention or in the prior art more clearly, the following briefly introduces the accompanying drawings required in the description of the embodiments or the prior art. Apparently, those of ordinary skill in the art can derive other drawings from the accompanying drawings without any creative effort.

FIG. 1 is a schematic flowchart of a method of the present invention;

FIGS. 2A-2D are development trend diagrams of a landscape approach motivation evolution state of a residential outdoor environment according to an embodiment of the present invention;

FIGS. 3A-3D are development trend diagrams of a landscape approach motivation similarity state of the residential outdoor environment according to an embodiment of the present invention;

FIGS. 4A-4C are diagrams showing residential outdoor environmental decision-making networks in three periods of summer, autumn, and winter according to an embodiment of the present invention; and;

FIG. 5 is a schematic structural diagram of a system of the present invention.

DETAILED DESCRIPTION OF THE EMBODIMENTS

The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings therein. Apparently, the described embodiments are only some of the embodiments of the present invention, not all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without any creative efforts fall within the scope of protection of the present invention.

First Embodiment

As shown in FIG. 1, a decision-making method for a residential outdoor environment based on landscape approach motivation includes the following steps:

S101: Obtain engineering drawing data and material test data of the residential outdoor environment, and perform a landscape approach motivation test on the material test data of the residential outdoor environment to obtain environmental element data passing the test, where the engineering drawing data of the residential outdoor environment includes plane vector data and GPS positioning data of the residential outdoor environment, and the material test data includes standard image data of residential outdoor environmental elements;

The process of performing a landscape approach motivation test on the material test data of the residential outdoor environment includes:

- idle state debugging, Stroop effect testing, and random playback of standard images of the residential outdoor environmental elements;

A hypothesis test is performed on Stroop effect testing results to obtain fitted landscape approach motivation evaluation indicators, and the fitted landscape approach motivation evaluation indicators are normalized to ultimately obtain the environmental element data passing the test;

Specifically, the present invention will be further elaborated through examples:

Frontal lobe EEG signals of the standard images of the residential outdoor environmental elements are obtained, the EEG signals are accurately cut according to environmental element classes and collection periods, a data cleaning process of band-pass filtering, noise artifact removal, ICA analysis, and spectral analysis is used, and a significance test and an internal consistency reliability test are performed to calculate power spectral densities of frontal lobe EEG α, θ, and γ rhythm bands of environmental elements passing the tests.

According to the principle of asymmetry in the frontal lobe of the brain, from a ratio of the power spectral densities of the EEG α, θ, and γrhythm bands, landscape approach motivation disturbance indicators (x₁; x₂; x₃) that can represent landscape “approach-avoidance” response are generated, calculated as follows:

$x_{1} = \frac{1}{n} \sum_{n} \frac{{PSD}_{left} (α) - {PSD}_{right} (α)}{{PSD}_{left} (α) + {PSD}_{right} (α)}$

$x_{2} = \frac{1}{n} PSD (θ)$

$x_{3} = \frac{1}{n} \sum_{n} \frac{\log ({PSD}_{left} (γ)) - \log ({PSD}_{right} (γ))}{\log ({PSD}_{left} (γ)) + \log ({PSD}_{right} (γ))}$

where n represents the number of EEG channels in the frontal lobe of the brain; PSD(α), PSD(θ), and PSD(γ) represent the power spectral densities of the α, θ, and γ rhythm bands, respectively; left and right represent the EEG channels on the left and right of the frontal lobe of the brain, respectively.

A hypothesis test (t-value and two-tailed p-value) and statistical verification of confidence intervals and effect sizes are performed on the Stroop effect testing results, to screen valid landscape approach motivation disturbance indicator values. On this basis, a change baseline of the landscape approach motivation is calibrated by stable EEG signals obtained in the idle state debugging phase, to obtain fitted landscape approach motivation evaluation indicators AW, specifically:

$A W = a ❘ x_{1_{-} post} - x_{1_{-} p r e} ❘ + b ❘ x_{2_{-} post} - x_{2_{-} p r e} ❘ + c ❘ x_{3_{-} post} - x_{3_{-} p r e} ❘ + φ$

where x_{1_pre}, x_{2_pre}, x_{3_pre}, x_{1_post}, x_{2_post}, and x_{3_post}represent average values of landscape approach disturbance indicators in the last 1 minute of the 30-second idle state debugging phase and the random playback phase of residential outdoor environmental element materials, respectively; a, b, and c represent coefficients of the landscape approach disturbance indicators in a linear combination, and φ represents a random error; the larger the value of AW, the stronger the cognitive motivation to approach a landscape environment.

The value of the landscape approach motivation evaluation indicator is normalized to obtain residential outdoor environmental element structural information

$R_{i d}^{t} = {r_{id}^{t}, i = 1, 2, 3 \dots, m, d = 1, 2, 3 \dots, n}$

in the t^thperiod, totally including m levels and n elements. A calculation formula for normalizing residential outdoor environmental elements is as follows:

$A W_{i}^{t} = \frac{r_{i}^{t} - μ_{i}^{t}}{S_{i}^{t}}$

where

$A W_{i}^{t}$

represents a landscape approach motivation evaluation indicator value normalized from the i^thclass of residential outdoor environmental elements in the t^thperiod;

$r_{i}^{t}$

represents a original landscape approach motivation evaluation indicator value of the i^thclass of environmental elements in the t^thperiod;

$μ_{i}^{t}$

represents an average value or original landscape approach motivation evaluation indicators of the i^thclass of environmental elements in the t^thperiod;

$S_{i}^{t}$

represents a standard deviation of the original landscape approach motivation evaluation indicators of the i^thclass of environmental elements in the t^thperiod.

In this embodiment, the plane vector data of the residential outdoor environment is organized using AutoCAD software, where the plane vector data includes a residential red line range, geographic elevation data, road streamline data, green planting data, hard space data, building vector data, and GPS positioning data of residential outdoor environmental elements. Meanwhile, based on the current construction status of the residential outdoor environment and the demand for landscape renewal and governance, the system collects image data of environmental elements in summer, autumn, and winter. The actual pixel percentage of the environmental elements represented by the residential outdoor environment images is >70%, which ensures that the image standards of the environmental elements in different periods are consistent. An element level system including landscape function, form, and monomer is built, where functional level elements include distribution space, recreational garden, sports field, and featured platform; morphological level elements include enclosure, sequence, shape, and symbols; monomer level elements include vegetation, water body, structure, and facility.

Instantaneous stimulation EEG data of the residential outdoor environmental elements is processed using an Eegomylab portable EEG collection system and an MNE-Python EEG cleaning toolkit, where the P values of all the environmental elements passing the significance test are <0.05; the internal consistency Cronbach a of each class of elements is >0.65. Based on subjects and the number of trials of test materials, it is proved using G*Power software that the instantaneous stimulation results of the environmental elements satisfy the significance P-value ≤0.05, statistical power >0.5, confidence interval not including zero, and effect size Cohen's d≥0.2. The recognition of landscape approach motivation changes in the residential outdoor environmental elements can be continued.

S102: Delineate landscape nodes in the residential outdoor environment based on the engineering drawing data of the residential outdoor environment and the environmental element data passing the test, and determine evolution and similarity change states of the landscape nodes based on landscape approach motivation information of the environmental elements corresponding to the landscape nodes in the residential outdoor environment;

S103: Generate an initial landscape approach motivation decision-making matrix of the landscape nodes and their corresponding environmental elements based on the GPS positions and change states of the landscape nodes in the residential outdoor environment, encode discrete variables of the GPS positions and change states of the landscape nodes in the residential outdoor environment, and build a distribution adjacency network of the landscape nodes based on the initial landscape approach motivation decision-making matrix of the landscape nodes;

The process of generating an initial landscape approach motivation decision-making matrix of the landscape nodes and their corresponding environmental elements is as follows:

Based on structural landscape approach motivation information R_idof the residential outdoor environmental elements in the landscape nodes, an initial decision-making matrix Z^wtis constituted for the landscape approach motivation evaluation indicators of w landscape nodes A_l={a_w, w=1, 2, 3, . . . , l} in the t^thperiod, and probability matrices P^wtof the initial decision-making matrix Z^wtare calculated sequentially as follows:

$P^{wt} = {(p_{id}^{wt})}_{m \times n} = \frac{Z^{wt}}{\sum_{j = 1}^{w} Z^{wt}},$

$where$

$i = 1, 2, 3 \dots, m, d = 1, 2, 3 \dots, n$

where

$p_{id}^{wt}$

represents the uncertainty or landscape approach motivation information of the w^thlandscape node on the n^thresidential outdoor environmental element of the m^thlevel in the t^thperiod.

Based on the principle of information entropy, an entropy value E_idof each residential outdoor environmental element and a utility value O_idof its landscape approach motivation information to obtain a weight T_idof the environmental element, specifically:

$O_{id} = 1 - E_{id} = 1 + \frac{1}{\ln (w)} \sum_{t = 1}^{w} p_{id}^{t} \ln p_{id}^{t}, 0 \leq E_{id} \leq 1$

$T_{id} = \frac{O_{id}}{\sum_{d}^{w} O_{id}}, 0 \leq T_{id} \leq 1$

where

$p_{id}^{t}$

represents an amount of landscape approach motivation information of the n^thresidential outdoor environmental element in the t^thperiod; w represents a quantity of the landscape nodes in the residential outdoor environment.

The change state of the landscape approach motivation information of the environmental element corresponding to the landscape node in the residential outdoor environment is decomposed into an evolution state and a similarity state, wherein the evolution state EVO^wtrepresents a degree of dominance of the residential outdoor environmental element, reflects a vector space distance between the environmental element and its optimal value, and is calculated as follows:

$B^{wt} = \sum_{i = 1}^{m} \sum_{d = 1}^{n} T_{id} \cdot b_{id}^{wt}, where i = 1, 2, 3 \dots, m, d = 1, 2, 3 \dots, n$

$c_{id}^{wt} = \frac{b_{id}^{wt}}{b_{id}^{- wt}}$

${EVO}^{wt} = \frac{\sqrt{\sum_{i = 1}^{m} \sum_{d = 1}^{n} (T_{id} c_{id}^{wt} - \max_{d} (T_{id} c_{id}^{wt}))}}{\max (\sqrt{\sum_{i = 1}^{m} \sum_{d = 1}^{n} (T_{id} c_{id}^{wt} - \max_{d} T_{id} c_{id}^{wt})})}$

where B^wtrepresents a comprehensive landscape approach motivation variable of the residential outdoor environmental elements;

$b_{id}^{wt}$

represents a landscape approach motivation evaluation indicator of the d^thenvironmental element of the i^thlevel corresponding to the w^thlandscape node in the residential outdoor environment in the t^thperiod,

$c_{id}^{wt}$

represents an evolution coefficient of the landscape approach motivation evaluation indicators.

The similarity state SIM^wtrepresents an associated development trend of the residential outdoor environmental elements, reflects a collaborative change trend of the environmental elements, and is calculated as follows:

$V_{id}^{wt} = \frac{\min_{m} \min_{n} ❘ c_{id}^{wt} - \max_{d} (T_{id} c_{id}^{wt}) ❘ + σ \max_{m} \max_{n} ❘ c_{id}^{wt} - \max_{d} (T_{id} c_{id}^{wt}) ❘}{❘ c_{id}^{wt} - \max_{d} (T_{id} c_{id}^{wt}) ❘ + σ \max_{m} \max_{n} ❘ c_{id}^{wt} - \max_{d} (T_{id} c_{id}^{wt}) ❘}$

${SIM}^{wt} = \sum_{d = 1}^{n} T_{id} ❘ V_{id}^{wt} ❘, where i = 1, 2, 3 \dots, m, d = 1, 2, 3 \dots, n, w = 1, 2, 3, \dots, l$

where

In this embodiment, kernel density analysis is performed on the GPS positioning data of the residential outdoor environmental elements using the GIS software, to recognize high-frequency and multi-type environmental element aggregation areas. Residents' use frequency and density in the residential outdoor environment are obtained using map punctuation. 15 representative landscape nodes are selected based on the superposition of the high-frequency and multi-type environmental element aggregation areas and the residents' use frequency and density, and with reference to the results of the aforementioned landscape approach motivation test of the environmental elements.

Weights of the residential outdoor environmental elements in 3 periods of summer, autumn, and winter are calculated by the initial landscape approach motivation decision-making matrix of the residential outdoor environmental elements (Table 1), where weight coefficients of the functional level in the 3 periods are 27.52%, 34.45%, and 41.42%, respectively; weight coefficients of the morphological level in the 3 periods are 36.31%, 30.37%, and 24.73%, respectively; weight coefficients of the individual level in the 3 periods are 36.18%, 35.19%, and 33.86%, respectively. The resolution coefficient σ is set to 0.5, and the evolution state and similarity state of landscape approach motivation of the environmental elements are then identified to obtain a dynamic development trend of the landscape nodes in the residential outdoor environment (FIG. 2 and FIG. 3).

TABLE 1

Weights of residential outdoor environmental elements

Weight coefficient (%)

Summer
Autumn
Winter

(May
(September
(December

to
to
to

Level
Element
August)
November)
March)

Function
Distribution
8.20
3.24
8.55

space

Recreation
6.74
12.08
11.70

garden

Sports field
6.03
5.73
15.25

Featured
6.55
13.40
5.92

platform

Morphology
Enclosure
7.75
4.18
5.97

Sequence
12.12
11.22
7.47

Shape
8.63
8.49
5.84

Symbol
7.81
6.48
5.45

Individual
Vegetation
12.44
9.30
5.01

Water body
8.40
10.21
10.18

Building
3.70
4.89
4.71

Facility
11.64
10.79
13.96

The encoding of the GPS positions and change states of the landscape nodes in the residential outdoor environment is fused with the levels and class names of the residential outdoor environmental elements, landscape approach motivation change states, and GPS position attributes. The organizational structure of a geographical concept set of the residential outdoor environment is preset as a K-level structure tree. When 1<k≤K, the maximum number of branches included in the k^thlevel is defined as

$D_{k}^{\max},$

and the code of the classification level structure tree has

$K + \sum_{k = 2}^{K} D_{k}^{\max} bits .$

Element attributes are converted into visual variables by conditional determination according to attribute thresholds of the levels and classes of the residential outdoor environmental elements, and an objective mapping relationship of the real residential outdoor environment is output by drawing parameters. The visual variables are vector symbols of graphic elements, including lines, ellipses, polygons, etc. Each type of graphic elements has its drawing parameters

In this embodiment, the structure tree of the geographical concept set of the residential outdoor environment is designed to four levels by One-Hot encoding, where the first level represents landscape nodes, the second and third levels represent element levels and classes, and the fourth level represents landscape approach motivation change states with time series, including an evolution state and a similarity state. On this basis, based on a visualization objective of a decision-making network for the residential outdoor environment, a comparison table of drawing parameters and visual variable conversion formats of element attributes of the residential outdoor environment (Table 2) is created.

TABLE 2

Comparison table of drawing parameters

and visual variable conversion formats

Applicable

Element
Drawing
Visual

graphic
Conversion

attribute
parameter
variable
Scope
element
method

Change state
PenColor
color
type
round
step gradient

Driving
length
length
[−1, 1]
line
class

degree

segment
mapping

Name
font size
size
—
character
—

Quantity
FillColor
color
[0, 100]
ellipse
step gradient

Geographical
—
—
WGS84
—
linear scale

coordinates

A decision-making matrix S of the distribution adjacency network is built based on the landscape nodes and the evolution state and similarity state results of landscape approach motivation of the landscape nodes and their corresponding environmental elements. The distribution adjacency network C=(w, e) consists of w landscape nodes and e edges, and is an undirected network of the landscape nodes.

Based on actual geographical distances between the landscape nodes in the residential outdoor environment, local adjacency indexes

${GE}_{q}^{t} and {GS}_{q}^{t}$

of the evolution state and similarity state of environmental element landscape approach motivation of the landscape node q in the t^thperiod are calculated, specifically:

${GE}_{q}^{t} = \frac{\sum_{q} T_{qh}^{t} (d) \cdot {EVO}_{q}^{t} - {\overline{EVO}}^{t} \cdot \sum_{q} T_{qh}^{t} (d)}{S_{- EVO} \sqrt{\frac{w \cdot \sum_{q} T_{qh}^{2} (d) - {(\sum_{q} T_{qh}^{t} (d))}^{2}}{w - 1}}}$

${GS}_{q}^{t} = \frac{\sum_{q} T_{qh}^{t} (d) \cdot {SIM}_{q}^{t} - {\overline{SIM}}^{t} \cdot \sum_{q} T_{qh}^{t} (d)}{S_{- SIM} \sqrt{\frac{w \cdot \sum_{q} T_{qh}^{2} (d) - {(\sum_{q} T_{qh}^{t} (d))}^{2}}{w - 1}}}$

where

${EVO}_{q}^{t} and {SIM}_{q}^{t}$

represent evolution state and similarity state attribute values of the q^thlandscape node in the residential outdoor environment in the t^thperiod, respectively;

$T_{qh}^{t} (d)$

represents a spatial weight between the landscape node q and the landscape node h in the t^thperiod, determined by the actual geographical distance between the landscape nodes; EVO^tand SIM^trepresent average values of the evolution state and similarity state attribute values of all the landscape nodes, respectively; S_{_EVO}and S_{_SIM}represent standard deviations of the evolution state and similarity state attribute values of all the landscape nodes, respectively; w represent the quantity of the landscape nodes;

When

$G E_{q}^{t} or {GS}_{q}^{t} > 0,$

the adjacent areas of the landscape nodes have a high aggregation distribution of evolution state or similarity state attribute values of environmental elements; when

$G E_{q}^{t} or {GS}_{q}^{t} > 0,$

the adjacent areas of the landscape nodes have low aggregation of evolution state or similarity state attribute values of environmental elements.

The geographical concept set of the residential outdoor environment and the local adjacency indexes of the landscape nodes are loaded into the decision-making matrix S of the distribution adjacency network, and the driving degree

$S_{q h}^{t}$

of the edge connecting the landscape node q and the landscape node h in the t^thperiod is calculated, where q≤w, h≤w:

${SE}_{q h}^{t} and {SS}_{q h}^{t}$

represent the driving degrees of the adjacency distribution network in the landscape approach motivation evolution state and similarity state between the landscape nodes q and h in the residential outdoor environment in the t^thperiod, respectively;

When

$S_{q h}^{t} > 0,$

the landscape approach motivation change state distribution network between the landscape nodes q and h has a relatively high driving degree, indicating that the landscape nodes q and h contribute significantly to the spatial pattern of landscape approach motivation in a common adjacency area of the residential outdoor environment, when

$S_{q h}^{t} < 0,$

the landscape approach motivation change state distribution network between the landscape nodes q and h has a relatively low driving degree, indicating that the landscape nodes q and h contribute little to the spatial pattern of landscape approach motivation in the common adjacency area of the residential outdoor environment.

S104: Perform environmental classification decision-making processing on the distribution adjacency network of the landscape nodes to obtain residential outdoor environmental decision-making results.

Retrospective analysis is performed on evolution state and similarity state data of landscape approach motivation of the residential outdoor environment, intra-class sums of squares of deviations WSS and inter-class sums of squares of deviations BSS of the evolution state and similarity state are calculated, the level with minimum WSS and maximum BSS is selected as an optimal level, and the evolution state and similarity state are divided into f levels to obtain a level pattern of totally f²change states of the residential outdoor environment.

Based on the level pattern of change states of the residential outdoor environment, a decision-making network representing the adjacent spatial distribution of the landscape nodes in the residential outdoor environment is generated, with the driving degrees

Spatial effect intensities of the landscape approach motivation distribution adjacency network of the landscape nodes in the residential outdoor environment are analyzed, to obtain residential outdoor environment landscape nodes to be improved as follows:

${SEI}^{t} = \sum_{h \neq q} \frac{A S_{q h}^{t}}{{NOD}_{q}^{t}}$

where SEI^trepresents the spatial effect intensities of the landscape nodes in the residential outdoor environment in the t^thperiod;

$A S_{q h}^{t}$

represents a driving degree assignment for the landscape approach motivation change state of the landscape nodes q and h in the t^thperiod; when

$S_{q h}^{t} > 0, A S_{q h}^{t} = 1; when S_{q h}^{t} < 0, A S_{q h}^{t} = 0; {NOD}_{q}^{t}$

represents the number of edges connecting the q^thlandscape node with other nodes in the t^thperiod;

The values of spatial effect intensities are normalized, intra-class sums of squares of deviations WSS and inter-class sums of squares of deviations BSS of the normalized spatial effect intensities are calculated, the level with minimum WSS and maximum BSS is selected as an optimal level to obtain u levels of the spatial effect intensities, and the u levels are assigned as u_i=j, for j=1, 2, . . . , u. Meanwhile, the level pattern of f²residential outdoor environment change states of the landscape nodes is assigned as g_i=i, for i=1, 2, . . . , f², to obtain a renewal and governance priority Pr_{_nod}of the landscape nodes as follows:

$\Pr_{_nod} = u_{i} \cdot g_{i} .$

In this embodiment, the landscape approach motivation change states in the residential outdoor environment are divided into 3 levels, where the evolution state includes 3 levels: low [0, 0.332), medium [0.332, 0.523), and high [0.523, 1]; the similarity state includes 3 levels: low [0, 0.566), medium [0.566, 0.702), and high [0.702,1]; totally 9 residential outdoor environment change state level standards and corresponding 5 combination patterns are obtained (Table 3). By implementing the decision-making network, the distribution adjacency network of the landscape nodes in the residential outdoor environment is drawn (