CN106951581A - Commercial complex simulator - Google Patents
Commercial complex simulator Download PDFInfo
- Publication number
- CN106951581A CN106951581A CN201710054024.9A CN201710054024A CN106951581A CN 106951581 A CN106951581 A CN 106951581A CN 201710054024 A CN201710054024 A CN 201710054024A CN 106951581 A CN106951581 A CN 106951581A
- Authority
- CN
- China
- Prior art keywords
- simulation
- data
- consumer
- model
- result
- 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
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
Abstract
A kind of Commercial Complex simulator, including presentation layer, functional layer and data Layer.Presentation layer includes function menu, fast operating area, Drawing zone and instant window.The module that functional layer includes has：Input basic data, setting Discrete Choice Model, conditions setting, the simulation of operation individual, simulation result analysis and analog result visualization model, Discrete Choice Model is set including the use of existing model and estimation new model, input basic data, conditions setting, the output of setting Discrete Choice Model are all connected with running individual analog module, the output of the individual analog module of operation is connected to simulation result analysis and analog result visualization model, and the individual analog module of operation includes simulation spatial behavior, time-space behavior, consumer behavior and dynamic response/queuing.Data Layer includes basic data, preference pattern data, simulation setting and boundary condition data and result data module.
Description
Technical field
The invention belongs to commercial space planning and designing technical field, more particularly to Commercial Complex simulator.
Background technology
With the development of China's economic society, consumer demand level is also improved constantly, and Commercial Complex is used as one kind
Emerging commercial space development model, shows from powerful competitiveness and vast potential for future development.Commercial Complex is with business
Industry function is leading, integrates a variety of functions such as retail, food and drink, leisure, amusement, culture, traffic, unitized overall development, uniformly
The emerging large scale business facility managed, be managed collectively, its appearance reflects the necessity of commercial space development.With traditional business
Industry facility (such as department store, supermarket, commercial street) is compared, the characteristics of Commercial Complex shows many new, such as function and industry
State is more combined that various, space utilization is more intensive, focus on the development of vertical direction spatial, space form and is not limited to line style and cleverer
Various, stronger tune time effect etc. outside Spatial Dimension living.
The features described above of Commercial Complex has catered to the consumption pattern and theory of New Times, thus is developed rapidly,
The fast development situation that fully under way, blowout is built just is being presented.However, from the point of view of the Commercial Complex put into effect, this
Novel facility also exposes problems while bringing various convenient to consumer, including consumer environment it is crowded it is chaotic,
The industry situation stand-by period is long for emerging service intensity, facility has hidden danger etc. with not enough, personal security is built.It is comprehensive to business above
Common consumer's negative experience is summarized in vivo, it is clear that these problems all with Commercial Complex function, the planning in space
Design is closely related.On the other hand, from the visual angle of exploitation, subject of operation and planning and designing personnel, how more reasonably
Function proportioning and space configuration are carried out, so as to improve consumption experience, attracts passenger flow, drives the generation of more consumption activities, be urgent
The problem of being essential to be solved.
The key for solving these problems is before building or reconstructing Commercial Complex, to the passenger flow in Commercial Complex
Situation has a perspective assurance.If this point can be accomplished, it is possible to tried one's best by planning and designing means and avoid causing
The problem of influence consumer experience such as crowded, long, potential safety hazard of queuing up, while administrator can be allowed more reasonably to coordinate
Spatial relationship, optimizes space layout, enables preferably to meet consumer demand, balances benefits of different parties, creates maximum value.
Although the problem of existing research is to Commercial Complex have many concerns, but is ground mostly using the method for qualitative research
The person of studying carefully is more from theory, experience and value judgement, proposes to " Commercial Complex should be such as He Jianshe " this problem
N-person game n.Only a few studies attempt to analyze the Trip distribution of Commercial Complex, but these researchs are general using set
The combined data of aspect, the method such as utilization space syntax, regression analysis, correlation analysis carries out general explanation to Trip distribution, and
Without predictive ability truly.Especially, the obvious precision of collection meter method that these quantitative studies are used is not enough.
There is a small number of American-European and Japan research to be developed by many agent's technologies (multi agent system, MAS)
For the individual analog platform in shopping center, but consumer space's behavioral mechanism for being relied on of these platforms is all typically all base
In Deterministic rules.These requirements of existing individual analog platform to data are very high, are typically necessary each consumption of precognition
The cmplete block design of person, such as shopping list (shopping list), schedule table (agenda), further in accordance with such plan
Carry out with default rule and simulate.In addition, the individual analog platform of existing foreign countries lacks to the reliability and precision of analog result
Weary checking, is not also illustrated to its application value in planning and designing mostly.
The content of the invention
It is also the method route of complete set it is an object of the invention to provide Commercial Complex simulator so that planning is set
Meter personnel and the exploitation network operator of Commercial Complex can utilize the spatial behavior mechanism rule of consumer, comprehensive for some business
Fit placement scheme carries out reliable, high-precision consumer's individual space Behavior modeling prediction, obtains in Commercial Complex
The analog results such as the Trip distribution in portion space, thus provide decision support for the space configuration optimization of Commercial Complex.
A kind of Commercial Complex simulator, including presentation layer, functional layer and data Layer,
Presentation layer includes function menu, fast operating area, Drawing zone and instant window,
The module that functional layer includes has：Input basic data, setting Discrete Choice Model, conditions setting, operation
Body simulation, simulation result analysis and analog result visualization model, setting Discrete Choice Model is including the use of existing model and estimates
New model is counted, input basic data, the output of the setting Discrete Choice Model of conditions setting are all connected with running individual simulation
Module, the output of the individual analog module of operation is connected to simulation result analysis and analog result visualization model, the individual mould of operation
Intending module includes simulation spatial behavior, time-space behavior, consumption and dynamic response/queuing,
Data Layer includes basic data, preference pattern data, simulation setting and boundary condition data and result data module,
Basic data includes scheme and subregion, and preference pattern data include built-in default models parameter and self-definition model ginseng
Number, simulation setting and boundary condition data include simulation setup parameter, entrance distribution, all kinds of Annual distributions and consumption distribution, knot
Fruit data include analog result and results of comparison, and alternatively model data, simulation are set and side consumer behaviour measured data
The input of boundary's condition data and result data module,
The running of Commercial Complex simulator includes：
User is by function menu and fast operating area input data and completes command operating, from Drawing zone and instant window
Middle acquisition result；
Internally functional layer, it is necessary first to user input basic data, while also requiring that user setting reflection consumer is empty
Between active mechanism Discrete Choice Model, model both can utilize built-in default models, can also by user provide actual measurement number
New model according to estimates；
After the preparation of basic data and model has been completed, user can set boundary condition (the simulation people of simulation
Number, entrance distribution, number realization etc.), individual simulation is then run, the content of simulation includes spatial behavior, time behavior, consumption
Behavior, queuing behavior；
After simulation is completed, user by simulation result analysis and analog result visualization model result is carried out analysis and
Output, and drawing and result being visualized by way of video；
Data in data Layer include the scheme subregion in basic data and explain variable-value, Discrete Choice Model data
In built-in default models parameter and user-defined model parameter, simulation setting and boundary condition data in simulation setting ginseng
Number, consumer entrance distribution, consumer reach Annual distribution, in the residence time destribution of each cell, between each cell
Traffic time distribution, each cell consumption proportion and the amount of consumption be distributed；Analog result and results of comparison in result data.
Discrete Choice Model explains the mechanism of consumer space's behavior behind to measure, and the spatial behavior of consumer is regarded
For each continuously selection course：
If a certain consumer has K stop activity P after entrance E entrance during migration1,P2,…,PK-1,PK, most
The whereabouts left afterwards is designated as H, its path " E → P1→P2→…→PK-1→PK→ H " is split as following K+1 OD (origin-
Destination, origin and destination) combination, reflect its each step wherefrom come, where：
" from E to P1", " from P1To P2" ..., " from PK-1To PK", " from PKTo H ", each OD combinations are one in fact
Secondary selection：According to current scene and location, where next step is selected in many alternate items；
The space of Commercial Complex is divided into several space cells, then these space cells are exactly consumer space's choosing
The alternate item selected, in addition, when each step is selected, consumer also has a special alternate item --- ending activity, from
Commercial Complex is opened, is referred to as going home below；
If the actual observation data in certain Commercial Complex consumer path, individual path is split as to select data,
To estimate Discrete Choice Model, in simulation process, continuously selected using the Discrete Choice Model prediction each step of consumer
Behavior, will predict the outcome " assembling " into its individual path, re-forms other spatial statisticses results.
The concrete form of Discrete Choice Model is used as using MNL models (Multinomial Logit).
Described individual simulation is using the Behavior modeling of consumer's individual space as representative, including step：
From consumer since entrance, select probability is calculated with Discrete Choice Model；
Trade-off decision is performed with Monte Carlo simulation, record result is to path, if going home then end simulation；
If not going home, then individual data items are updated for selection next time, select probability are then calculated again,
In Monte Carlo simulation, provided with n alternate item, its select probability is respectively Pk(k=1,2 ..., n), then can be
0-1 probability interval is divided into n sections：
Wherein i-th section isThereby guarantee that this n section length correspond respectively to n alternate item selection it is general
Rate, if for example, having 4 alternate items A, B, C, D, its select probability is respectively 0.4,0.1,0.2,0.3, then can be the general of 0-1
Rate interval is divided into 4 sections --- and [0,0.4], (0.4,0.5], (0.5,0.7], (0.7,1], three waypoints are added up by following
Mode is obtained：
P (A)=0.4, P (A)+P (B)=0.4+0.1=0.5,
P (A)+P (B)+P (C)=0.4+0.1+0.2=0.7,
Thereby guarantee that the length of this 4 sections corresponds respectively to the select probability of 4 alternate items, and then generate a 0-1
Uniform random number, the section fallen into according to it determine which corresponding alternate item be selected.
In addition to the spatial behavior simulation on basis, individual behavior simulation also includes time-space behavior and consumer behavior is simulated, its
In,
Each consumer that is modeled as of time dimension increases timeline, on the premise of known consumer's entry time, root
Go to traffic time that next destination is spent and stay time in this place every time according to consumer, obtain its activity every time
The beginning and end moment, and its state in which of any time and position, entry time, residence time, traffic time
Distribution all provides actual observation data by user and estimated by Commercial Complex simulator, or in the situation for lacking observed data
Under directly inputted by user；
Consumption imitation is each space cell setting one by way of actual observation data estimation or user are directly inputted
Individual amount of consumption distribution, it is contemplated that can all have a large amount of customers not consumed in many cells, if not considering these 0 yuan in advance
Element will cause special distributional pattern, and can not reproduce the phenomenon do not consumed in simulations,
Therefore the non-consumption proportion of each cell, then the sample (amount of consumption to consuming are set first>0) setting consumption
Volume is distributed, and in simulations, when consumer have selected a space cell carry out activity, simulation system is primarily based on the cell
Non- consumption proportion, by monte carlo method predict consumer whether consume, only when there occurs consumption, then be based on
The consumption distribution of the cell generates this movable amount of consumption at random.
The present invention has the flexibility of height, can be completely to obtain in the case of only one Commercial Complex scheme
Quite high-precision analog result.It is of the invention then there is high accuracy and high by the result with the contrast verification of actual observation data
Reliability, and there is obvious decision support to act on planning and designing practice.
Brief description of the drawings
The basic framework of Fig. 1 Commercial Complex simulators (CCSIM) of the present invention.
The basic procedure that individual behavior is simulated in Fig. 2 present invention.
In Fig. 3 embodiment of the present invention, the simulation precision figure of individual activity number of times.
In Fig. 4 embodiment of the present invention, the simulation precision figure of distribution of visitors is stopped.
In Fig. 5 embodiment of the present invention, the simulation precision figure of OD distributions.
In Fig. 6 embodiment of the present invention, the characteristic index comparative result figure of Scenario Simulating.
Embodiment
The present invention, which has been developed, forms an end-user program --- CCSIM (commercial complex
Simulator, Commercial Complex simulator).Invention and the basic framework of program are as shown in Figure 1.
It will be noted from fig. 1 that user by function menu and fast operating area input data and can complete command operating,
Result is obtained from Drawing zone and instant window.On internally the task of functional layer is realized, it is necessary first to the number of user input basis
According to；The Discrete Choice Model of user's setting reflection consumer space's active mechanism is also required simultaneously, and model both can be built in
Default models, can also by user provide measured data estimation new model；Complete the beam worker of basic data and model
After work, user can set the boundary condition (simulation number, entrance distribution, number realization etc.) of simulation, then run individual mould
Intend, the content of simulation includes spatial behavior, time behavior, consumer behavior, queuing behavior etc.；After simulation is completed, user can be with
Result is analyzed and exported, and drawing and result being visualized by way of video.The process realized in function
In be related to volume of data storage, management, exchange, these data include basic data in scheme subregion and explanatory variable take
Value；Built-in default models parameter and user-defined model parameter in Discrete Choice Model data；Simulation setting and perimeter strip
Number of packages according in simulation setup parameter, consumer entrance distribution, consumer reach Annual distribution, in the stop of each cell
Between distribution, between each cell traffic time distribution, each cell consumption proportion and the amount of consumption be distributed；In result data
Analog result and results of comparison etc..Notice self-definition model parameter in Fig. 1, all kinds of Annual distributions, consumption distribution, results of comparison
Wire be dotted line, it is not the necessary data of simulation and forecast to represent these, and they can be provided extra consumer by user
Behavior measured data is estimated.If user did investigation and obtained such data, then it is possible thereby to provide more
Plus agreeing with the model and simulating boundary condition of specific cases so that precision of prediction is higher；If user's data without as, this
Invention and CCSIM programs can also obtain analog result as high-quality as possible.
The main method that the present invention is used is that Discrete Choice Model and individual behavior are simulated, wherein, the central inventive of the latter
It is Monte Carlo simulation.They are discussed in detail respectively below.
Discrete Choice Model
The housing choice behavior of consumer
The main purpose of Discrete Choice Model (discrete choice model) is to explain that consumer is empty with metering model
Between behavior behind mechanism, the spatial behavior of consumer is considered as continuously selection course again and again by it：If a certain consumption
Person has K stop activity P after entrance E (entrance) entrance during migration1,P2,…,PK-1,PK, finally leave
Whereabouts is designated as H (home), then its path " E → P1→P2→…→PK-1→PK→ H " obviously can be split as following K+1
OD (origin-destination, origin and destination) combine, reflect its each step wherefrom come, where：" from E to P1", " from
P1To P2" ..., " from PK-1To PK", " from PKTo H ", each OD combinations are once to select in fact：According to current scene
And location, where select next step in many alternate items.Needed before using this technology analysis Commercial Complex
Space be divided into several space cells, then these space cells be exactly consumer space selection alternate item, in addition,
When each step is selected, consumer also has a special alternate item --- ending activity, Commercial Complex is left, below by it
Referred to as go home.By this mode, if the actual observation data in certain Commercial Complex consumer path, so that it may by individual
Path is split as selecting data；Conversely, in simulation process, can be connected using each step of Discrete Choice Model prediction consumer again
Continuous housing choice behavior, will predict the outcome " assembling " into its individual path, re-forms other spatial statisticses results.
The general principle of Discrete Choice Model
The measuring principle of Discrete Choice Model is briefly introduced below based on the specific linguistic context of the present invention.The model is based on random
Utility theory (random utility theory), it is assumed that for each selection main body (i.e. consumer) n, each
Alternate item i corresponds to certain effectiveness Uni, the effectiveness is constituted by fixing with random two parts, as shown in formula 2.1：
Uni=Vni+εni (2.1)
In above formula, VniFor fixed effectiveness, εniFor Random utility.Fixed effectiveness can be solved by independent variable in total utility
The systematization composition released.However, although the explanation strengths to selection result can as far as possible be improved by the meticulous selection of independent variable,
But due to the extremely complicated property of consumer space's behavior in itself in reality, during necessarily having some influent factors due to analysis
Accept or reject, be difficult to accurately the reason such as to estimate and be missed outside, consider further that the presence of measurement error, these nonsystematics, therefore
It is also that unknowable part just constitutes Random utility.Usually, fixed effectiveness uses the form that respective linear variable displacement is combined, such as
X in formula 2.2, formulaniFor the value vector of each independent variable, β is the parameter vector for characterizing each independent variable weighing factor.
Vni=β 'xni (2.2)
Discrete Choice Model follows maximization of utility it is assumed that i.e. consumer will select total utility highest option, such as formula
2.3.If for example, consumer have selected some space cell, the total utility of the space cell in certain once migration decision-making
Should be all higher compared with other all space cells, it also should be higher than that and go home；If selected for going home, then the total utility now gone home should
Higher than all space cells.In formula 2.3, although fixed effectiveness Vni、VnjCan be uniquely true by the specific level of each independent variable
It is fixed, but Random utility εni、εnjSize be difficult to clearly obtain, can only be described by statistical distribution.Therefore, consume
Person n chooses the generation of this event of alternate item i and otherwise random, its probability P in certain decision-makingniAs shown in formula 2.4.
It can be seen that, PniCalculating depend on εniDistribution, this by cause according to Random utility be distributed difference it is assumed that
Discrete Choice Model has diversified forms, it is necessary to specific setting again.The present invention allows any type of setting in theory, as
Multinomial Logit mode (multinomial logit model, MNL), nesting Logit models (nested logit
Model, NL) etc..
Concrete form：MNL models
MNL models are most simple, conventional Discrete Choice Models, it is assumed that Random utility εniObey independent identically distributed sweet
Boolean is distributed (Gumbel distribution).After having such hypothesis, the select probability P in formula 2.4niWith regard to available
The probability density function of independence and the Clarence Gamble distribution of Random utility is calculated, and obtains the closing form of shape such as formula 2.5.
MNL models have the advantages that technical threshold is low, are easily achieved, form is succinct, result is sane.Due to these advantages,
This technology is as the built-in model form in CCSIM, while being disappeared there is provided one group based on Shanghai City Wujiao Court WanDa Plaza
The simplification version model result of expense person's behavior investigation data estimation (only retains necessary explanatory variable, to ensure the general of the model
Property), so that user directly uses in no measured data.
Supplementary form：Nested Logit models
Because many Commercial Complexes are the space environment of many building multilayers, so may go out when dividing space cell
Existing multiple cells are located at the scene in same building, for example：If considering influence of the floor to consumer behaviour, it is necessary to will
Each layer of same building is divided into an independent space cell.Now, alternately between each space cell of item just
The dependency relation of complexity occurs：Typically there is stronger correlation, this and MNL moulds between cell inside same building
The theory hypothesis of type are separated to a certain extent to be closed.If improving the precision of analysis and simulation, it can now select more
The nested Logit models become more meticulous.Alternate item is divided into several subsets by the model --- the space cell category of same building
In a subset, a nested tree structure of layering is thus form, consumer first selects the big option of building positioned at upper strata
(minor matters point), reselection is located at the space cell (leaf node) of the building interior of lower floor.From subset Bm(m=1 ..., M)
Alternate item i be chosen non-similarity factor λ in the probability such as formulas 2.6 chosen of main body n, formulam(0<λm<1) subset B is reflectedmIt is interior
The size of portion's correlation, λmBigger, correlation is smaller.
Because nested Logit models are complex, it is often more important that its model specification is dependent on user to specific business
The space of synthesis case is divided, so being difficult to set up certain default setting.Therefore, without using the model in CCSIM
Form.Although the more general MNL models of the model are more superior in theoretical science, but if user wishes to use the mould
Type, then need voluntarily to complete the data acquisition of correlation, modeling, subsequently simulate work.By practice test, general MNL models
Very high level has been reached on precision of prediction.
The selection of explanatory variable
User is allowed neatly to select explanatory variable according to the actual conditions of itself case in CCSIM, but if user is certainly
Define explanatory variable, it is necessary to gather corresponding consumer space's behavioral data, set up Discrete Choice Model and variation coefficient is entered
Row estimation (can be automatically performed by CCSIM).Users for the ease of those shortage data use, built-in simplification in CCSIM
The model of version, only remains the general explanatory variable that nearly all case can all be used, with general to greatest extent
Property.These explanatory variables include：1) total area of business (unit of each space cell：m2), 2) between the cell of each two space
Plan range (unit：M) it is, 3) vertical apart from (unit：Layer), 4) consumer is to the familiarity of each space cell, 5) consumer
To the number of times of visiting of each space cell, the accumulation number of activities that 6) consumer has currently participated in this activity.For this 6
Individual variable, user need to only provide first 3 or 4 (depending on whether have it is reliable be familiar with degrees of data depending on), latter 2 then in simulations oneself
Dynamic generation, without user's processing.
Individual behavior is simulated
The basic procedure of consumer's individual space Behavior modeling system proposed by the present invention is as shown in Figure 2.Each consumer
Entrance be given, need constantly to select after entrance, when selecting each time, simulation system will collect the situation of presence
The value of lower every explanatory variable, and calculate selection respective to each alternate item generally using consumer space's action selection model
Rate；Be distributed according to current select probability, simulation system will determine final selection result using monte carlo method, and by its
It recorded the tail end in the individual path of the consumer；Said process is to complete once to select, and now simulation system can examine this
As a result whether it is last alternate item --- go home, if going home, then the individual path simulation of the consumer is declaration knot
Beam, if it is not, then the space cell that simulation system can just select it is as its newest current location, updates every solution
The value of variable is released, for example, the distance to other space cells is updated according to current location, by the visiting of current location cell
Number of times adds 1,1 accumulation number of activities of increase etc., it is noted herein that, if consumer is in the first time selection of entrance
Have selected and go home, then be considered as it is illogical and need reselect, the fact that certain occur possibility it is extremely low (time
The effectiveness of family is minimum when starting), it is a kind of theoretic supplement；After individual data items renewal is completed, simulation system will weight
Its select probability to each alternate item is newly calculated, the selection result of a new round is generated and recorded using monte carlo method, such as
This circulation is performed, untill consumer selection is gone home.
As shown in Fig. 2 the key in flow is the calculating of select probability in decision-making each time and the determination of selection result.Before
Person depends on Discrete Choice Model, and the latter then depends on Monte-carlo Simulation Method (Monte Carlo
simulation).Although probability is with inherent uncertainty --- the selected machine of high probability option in once selecting
Can be bigger, but possible " lucky " is low probability option to result on the contrary.Monte Carlo simulation is by the side by being continuously generated random number
Formula, reasonably does deterministic prediction according to probabilistic probability, and its reasonability is presented as that " probability " repeatedly occurs with event
When " frequency " be consistent.Its application mode in this technique is briefly introduced with an example below：Provided with 4 alternate items
A, B, C, D, its select probability are respectively 0.4,0.1,0.2,0.3, then 0-1 probability interval can be divided into 4 sections --- [0,
0.4], (0.4,0.5], (0.5,0.7], (0.7,1], three waypoints are obtained by following cumulative mode：P (A)=0.4, P
(A)+P (B)=0.4+0.1=0.5, P (A)+P (B)+P (C)=0.4+0.1+0.2=0.7, it is possible thereby to ensure this 4 sections
Length correspond respectively to the select probabilities of 4 alternate items, and then generate 0-1 uniform random number, fallen according to it
The section entered determines which corresponding alternate item is selected, it is clear that, although random number, which may be fallen on, in once attempting appoints
In what section, but if there is enough trials, then fall into 4 sections frequency should with their own length so that with choosing
The distribution for selecting probability is agreed.Shaked the elbows due to being similar in this method based on random number, therefore be assigned Meng Teka
The name of Lip river (world-renowned gambling city).
Above is to the individual simulation process of single consumer.With same method mould can be carried out to many consumers
Intend, the individual path for obtaining everyone carries out collect statistics as baseline results, then to it, obtains more meaningful analysis knot
Really, such as individual mean activity number of times, the spatial distribution of stop person-time.So far, the process of a sunykatuib analysis is simply completed,
Because Monte Carlo simulation has randomness, in order to obtain sufficiently stable result, multiple sunykatuib analysis is generally required, specifically
Number realization is determined according to actual conditions.
Basic procedure above describes consumer space's Behavior modeling of core the most, when research of the invention also increases
Between dimension, carry out time-space behavior simulation, and consider consumption of the consumer in each stop, carry out consumer behavior simulation,
Its method is introduced briefly below.
The simulation of time dimension mainly increases timeline for each consumer, in the premise of known consumer's entry time
Under, go to traffic time that next destination is spent and stay time in this place every time according to consumer, obtain its every
Secondary movable beginning and end moment, and its state in which of any time and position.Entry time, residence time, traffic
The distribution of time can provide actual observation data by user and be estimated by CCSIM, or in the situation for lacking observed data
Under directly inputted by user.Random number is constantly extracted in being distributed in simulation process dependent on these as during the admission of prediction
Between, the residence time, traffic time.In addition it may be noted that when considering the simulation of time dimension, user can set up one in advance
Discrete Choice Model with time correlation key element so that the row of some explanation key elements only within the specific period to consumer
To make a difference, these can also be automatically performed in CCSIM.
Consumer behavior simulation is relatively simple, similar with the analog form of stay time：By actual observation data estimation or
The mode that user directly inputs is each one amount of consumption distribution of space cell setting.Different is, it is envisioned that a lot
Can all there are a large amount of customers not consumed in cell, if not considering that these 0 elements will cause special distribution shape in advance
State, and the phenomenon do not consumed can not be reproduced in simulations, therefore, this technology requirement sets the non-consumption ratio of each cell first
Example, then the sample (amount of consumption to consuming>0) setting amount of consumption distribution.In simulations, whenever consumer have selected a sky
Between cell carry out activity when, simulation system is primarily based on the non-consumption proportion of the cell, pass through monte carlo method prediction consumption
Whether person consumes, only when there occurs consumption, then the consumption distribution based on the cell generates disappearing for this activity at random
Take volume.
Below by a specific Commercial Complex case --- exemplified by the Wujiao Court WanDa Plaza of Shanghai City, examine calligraphy or painting model hair
Bright precision.We obtain the individual path of 323 consumers of Wujiao Court Wanda by consumer behaviour investigation, based on the data
Discrete Choice Model is established, and baseline situation is simulated using the model.So-called baseline situation (base
Scenario be) not change any condition in reality, by identical consumer (323 samples, it is known that it is respective enter
Mouthful, and to the familiarity of each cell) be input in the configuration of current space environment, the result obtained by comparative simulation with
Actual observation result, assesses the systematic error of simulation.Research carries out error evaluation in 3 aspects：Consumer's individual activity number of times
It is the scalar (scalar) of only 1 value；The stop distribution of visitors of each space cell is the one-dimensional vector for having 45 values,
It is also most concerned content of the invention；The OD distributions of each minizone are 45*45 two-dimensional matrixs.Can be pre- from the complexity of result
Phase, the size of above-mentioned tripartite's surface error can be raised successively, be specifically described below.In view of the randomness in monte carlo method,
The number of times of simulation is set as 300 times.
Fig. 3 shows 323 consumer's individual activity number of times average values, standard deviation, median, the actual observations of maximum
As a result (thick line), the single result fluctuation (fine rule) simulated every time, and the simulation of preceding n times average result gradually convergent situation
(thick line, line tail is the average result of all 300 simulations).It can be seen that, although the result of single simulation has in various degree
Fluctuation, but average result has tended to convergence when less than 50 times, and hereafter gap is stable to be between thick line and thick line
System error.The actual result of 323 individual activity number of times averages is 4.45 times, and the average result of 300 simulations is 4.56 times, phase
Only have 2.49% to error rate, show very high precision level；The averaging analog result of individual activity number of times standard deviation
It is lower slightly compared with actual result, but error level is equally very good；The actual result of individual middle position number of activities is 4 times, and single
Analog result is then 4 times or 5 times, with being actually sufficiently close to；Finally, the actual result of individual maximum activity number of times is 12 times,
Higher than 9.34 times of averaging analog result, the result carries bigger contingency in itself, also repeatedly occurs in being simulated at 300 times
The maximum activity number of times of 12 times.In summary compare, simulation system error very little on individual mean activity number of times has very
Good performance.
Fig. 4 (a) and Fig. 4 (b) respectively illustrate the contrast of the mean space distribution of actual spatial distribution and 300 simulations,
Each point in figure represents the stop activity of primary consumer, it can be seen that the actual distributional pattern with simulation closely,
It is directly perceived to reflect that analog result still has performance well in this aspect.In order to more accurately compare two distribution it is similar
Property, actual broken line graph and scatter diagram with simulation average result is respectively illustrated in Fig. 4 (c) and Fig. 4 (d)：Two in broken line graph
Although bar line is deviated considerably from a small number of cells, entirety closely, has accurately held essential characteristic；Presented in scatter diagram
Obvious linear trend, and Trendline is diagonal (y=x), the actual coefficient correlation with simulation is up to 0.95,1%
Under level significantly.Contrast can be with judgement more than summarizing：Simulation system has in the overall space distribution of prediction consumer activities
Very high precision, it is as a result very reliable.
Fig. 5 (a) and Fig. 5 (b) respectively illustrate the contrast of the average OD distributions of actual OD distributions and 300 simulations.Due to
There is the traffic connection between substantial amounts of OD seldom, in order to show facility, real OD data are normalized in 0-10 interval,
Only main OD of the value more than 1 is to being just revealed.It can be seen that, it is actual with analog result still with higher similar
Property, simulation system can effectively hold OD distribution basic configuration, but with the relatively simple stop distribution of visitors phase in Fig. 4
Than the error simulated in this aspect is significantly increased.Further estimate the actual coefficient correlation with simulation OD values, it is found that it declines
To 0.76 (p<0.01), it was confirmed that although analog result is consistent with actual trend, the precision in many details has
Declined.Even so, it is contemplated that the complexity of problem is greatly promoted, the present invention has reached preferable water under the conditions of existing
It is flat.
The present invention can be used for guiding plan design, below still by taking the Wujiao Court WanDa Plaza of Shanghai City as an example.Contemplate and pass through one
A little planning and designing strategies change its function and space configuration, Scenario Simulating are carried out to the environment after change, to predict and quantify
The potential effect of these strategies is evaluated, so as to provide decision support for planning design work.
The big problem of Wujiao Court Wanda is the skewness weighing apparatus of consumer between each space cell, and some cells are excessively
It is crowded, and other cells then lack popularity.Cell in view of those shortage popularities is to pay relatively high-rented non-mostly
Main force shop, the equilibrium degree for improving consumer's distribution not only contributes to these shops, is also that overall Commercial Complex is wished to
Result.Therefore, it can to improve the harmonious of overall distribution, the current activity for lacking popularity place of increase stop person-time as
The target 1 of planning and designing measure and target 2.On the other hand, the mean activity number of times of single consumer is considered as Commercial Complex
The important indicator of vigor, if the index is improved, then same amount of consumer will bring more activities, bring more business
Machine.Therefore, it can using improve consumer mean activity number of times as planning and designing measure target 3.
Here by taking 3 scenes as an example.Scene 1 is that the offside in existing subway gateway sets a new subway to come in and go out
Mouthful, because subway is the main traffic mode that consumer reaches Wujiao Court Wanda, this measure is directed primarily to balance Trip distribution；Scene
2 be retail shop's brand in the not high building (special power fashion is converged) of the current popularity of upgrading, consumer is had it very high
Familiarity, it is intended to improve the popularity of the building；Scene 3 is then a pure spatial strategy, in 3 floor of 5 buildings of Wujiao Court Wanda
Between aerial vestibule is set so that consumer can more easily building upper strata between move, this measure is wished by getting through
Contact to promote the migration of consumer, improve the mean activity number of times of individual.
This 3 scenes are simulated respectively, and calculate corresponding index.Space Gini coefficient is to estimate each cell activity
The lack of uniformity (target 1) of distribution, and total visiting number of times and the number of activities per capita of single consumer that special power fashion is converged are then
For target 2 and target 3, as a result such as Fig. 6.It can be seen that, increase the product that new gateway (scene 1) and the special power fashion of upgrading are converged
Board (scene 2) can significantly decrease space Gini coefficient, but increase aerial vestibule (scene 3) to this index without notable
Influence.If improving the activity person-time that the not high special power fashion of current popularity is converged, its brand of directly upgrading in 3 kinds of scenes
(scene 2) is undoubtedly maximally effective.Finally, for improving in the mean activity number of times of single consumer this target, 3 kinds of scenes
Only set up aerial vestibule (scene 3) to be achieved, and increase gateway (scene 1), the special power fashion remittance brand (scene 2) of lifting
To this all without remarkable result.
Claims (6)
1. a kind of Commercial Complex simulator, including presentation layer, functional layer and data Layer,
Presentation layer includes function menu, fast operating area, Drawing zone and instant window,
The module that functional layer includes has：Input basic data, setting Discrete Choice Model, conditions setting, the individual mould of operation
Plan, simulation result analysis and analog result visualization model, setting Discrete Choice Model are new including the use of existing model and estimation
Model, input basic data, the output of the setting Discrete Choice Model of conditions setting are all connected with running individual analog module,
The output of the individual analog module of operation is connected to simulation result analysis and analog result visualization model, the individual analog module of operation
Including simulation spatial behavior, time-space behavior, consumption and dynamic response/queuing,
Data Layer includes basic data, preference pattern data, simulation setting and boundary condition data and result data module,
Basic data includes scheme and subregion, and preference pattern data include built-in default models parameter and self-definition model parameter,
Simulation setting and boundary condition data include simulation setup parameter, entrance distribution, all kinds of Annual distributions and consumption distribution, number of results
According to including analog result and results of comparison, alternatively model data, simulation are set and perimeter strip consumer behaviour measured data
The input of number of packages evidence and result data module,
The running of Commercial Complex simulator includes：
User is by function menu and fast operating area input data and completes command operating, is obtained from Drawing zone and instant window
Obtain result；
Internally functional layer, it is necessary first to user input basic data, while also requiring that user setting reflection consumer space lives
The Discrete Choice Model of mechanism, model can both utilize built-in default models, can also provide measured data by user and estimate
Count new model；
After the preparation of basic data and model has been completed, the boundary condition that user can set simulation (simulation number, enters
Mouthful distribution, number realization etc.), then run individual simulation, the content of simulation including spatial behavior, time behavior, consumer behavior,
Queuing behavior；
After simulation is completed, user carries out analysis to result and defeated by simulation result analysis and analog result visualization model
Go out, and drawing and result being visualized by way of video；
Data in data Layer include the scheme subregion in basic data and explained in variable-value, Discrete Choice Model data
Built-in default models parameter and user-defined model parameter, simulation setting and the simulation setup parameter in boundary condition data,
The entrance distribution of consumer, the Annual distribution of consumer's arrival, the residence time destribution in each cell, the friendship between each cell
Logical Annual distribution, consumption proportion and the amount of consumption distribution in each cell；Analog result and results of comparison in result data.
2. Commercial Complex simulator as claimed in claim 1, it is characterised in that metering mould is used as using Discrete Choice Model
Type explains the mechanism of consumer space's behavior behind, and the spatial behavior of consumer is considered as and continuously selected again and again
Journey：
If a certain consumer has K stop activity P after entrance E entrance during migration1,P2,…,PK-1,PK, finally from
The whereabouts opened is designated as H, its path " E → P1→P2→…→PK-1→PK→ H " is split as following K+1 OD (origin-
Destination, origin and destination) combination, reflect its each step wherefrom come, where：
" from E to P1", " from P1To P2" ..., " from PK-1To PK", " from PKTo H ", each OD combinations are once to select in fact
Select：According to current scene and location, where next step is selected in many alternate items；
The space of Commercial Complex is divided into several space cells, then these space cells are exactly what consumer space selected
Alternate item, in addition, when each step is selected, consumer also has a special alternate item --- ending activity, leave business
Industry synthesis, is referred to as going home below；
If the actual observation data in certain Commercial Complex consumer path, individual path is split as to select data, is used to
Discrete Choice Model is set up, in simulation process, the continuous housing choice behavior of each step of consumer is predicted using Discrete Choice Model,
" assembling " will be predicted the outcome into its individual path, other spatial statisticses results are re-formed.
3. Commercial Complex simulator as claimed in claim 2, it is characterised in that selection MNL models (Multinomial
Logit) as the concrete form of Discrete Choice Model.
4. Commercial Complex simulator as claimed in claim 1, it is characterised in that described individual simulation is with consumer's individual
Spatial behavior is modeled as representing, including step：
From consumer since entrance, select probability is calculated with Discrete Choice Model；
Trade-off decision is performed with Monte Carlo simulation, record result is to path, if going home then end simulation；
If not going home, then individual data items are updated for selection next time, select probability is then calculated again.
5. Commercial Complex simulator as claimed in claim 4, it is characterised in that in Monte Carlo simulation, alternative provided with n
, its select probability is respectively Pk(0-1 probability interval, n), then can be divided into n sections by k=1,2 ...：Wherein i-th section is
Thereby guarantee that the length of this n section corresponds respectively to the select probability of n alternate item, provided with 4 alternate items A, B, C, D, its
Select probability is respectively 0.4,0.1,0.2,0.3, then 0-1 probability interval can be divided into 4 sections --- and [0,0.4], (0.4,
0.5], (0.5,0.7], (and 0.7,1], three waypoints are obtained by following cumulative mode：
P (A)=0.4, P (A)+P (B)=0.4+0.1=0.5,
P (A)+P (B)+P (C)=0.4+0.1+0.2=0.7,
And then 0-1 uniform random number is generated, the section fallen into according to it determines which corresponding alternate item is chosen
In.
6. Commercial Complex simulator as claimed in claim 4, it is characterised in that also simulate and consume row including time-space behavior
To simulate, wherein,
Time dimension is modeled as each consumer increase timeline, on the premise of known consumer's entry time, according to disappearing
Expense person goes to the traffic time that next destination spent and stay time in this place every time, obtain its every time activity open
Begin and finish time, and its state in which of any time and position, entry time, residence time, the distribution of traffic time
All by user provide actual observation data estimated by Commercial Complex simulator, or shortage observed data in the case of by
User directly inputs；
Consumer behavior simulation is each space cell setting one by way of actual observation data estimation or user are directly inputted
Individual amount of consumption distribution,
Provided with there is a large amount of customers not consumed in multiple cells, if it is special not consider that these 0 elements will cause in advance
Distributional pattern, and can not reproduce the phenomenon do not consumed in simulations,
The non-consumption proportion of each cell, then the sample (amount of consumption to consuming are set first>0) setting amount of consumption distribution,
In simulations, when consumer have selected a space cell carry out activity, simulation system is primarily based on not disappearing for the cell
Taking ratio, predicting whether consumer consumes by monte carlo method, only when there occurs consumption, then based on the cell
Consumption distribution generate this movable amount of consumption at random.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710054024.9A CN106951581B (en) | 2017-01-24 | 2017-01-24 | Commercial complex simulator |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710054024.9A CN106951581B (en) | 2017-01-24 | 2017-01-24 | Commercial complex simulator |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106951581A true CN106951581A (en) | 2017-07-14 |
| CN106951581B CN106951581B (en) | 2023-06-02 |
Family
ID=59466180
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710054024.9A Active CN106951581B (en) | 2017-01-24 | 2017-01-24 | Commercial complex simulator |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106951581B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110502569A (en) * | 2019-08-16 | 2019-11-26 | 浙江财经大学 | A kind of standard well screen based on Discrete Choice Model selects visual analysis method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2003281348A (en) * | 2002-03-25 | 2003-10-03 | Yunitekku:Kk | Market area analyzing system, method, program and recording medium |
| CN102792327A (en) * | 2010-02-04 | 2012-11-21 | 宝洁公司 | Method for conducting consumer research |
| CN105320653A (en) * | 2014-05-27 | 2016-02-10 | 杭州中瑞思创科技股份有限公司 | Consuming behavior pattern collecting system and method |
-
2017
- 2017-01-24 CN CN201710054024.9A patent/CN106951581B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2003281348A (en) * | 2002-03-25 | 2003-10-03 | Yunitekku:Kk | Market area analyzing system, method, program and recording medium |
| CN102792327A (en) * | 2010-02-04 | 2012-11-21 | 宝洁公司 | Method for conducting consumer research |
| CN105320653A (en) * | 2014-05-27 | 2016-02-10 | 杭州中瑞思创科技股份有限公司 | Consuming behavior pattern collecting system and method |
Non-Patent Citations (2)
| Title |
|---|
| 王灿 等: "基于消费者行为的商业综合体空间特征与评价——以上海五角场万达广场为例", 《2015中国城市规划年会》 * |
| 王灿 等: "离散选择模型研究进展", 《地理科学进展》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110502569A (en) * | 2019-08-16 | 2019-11-26 | 浙江财经大学 | A kind of standard well screen based on Discrete Choice Model selects visual analysis method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106951581B (en) | 2023-06-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Simmonds | The design of the DELTA land-use modelling package | |
| Lam et al. | An activity-based time-dependent traffic assignment model | |
| Krafta | Modelling intraurban configurational development | |
| Cole | Synergy and congestion in the tourist destination life cycle | |
| Tang et al. | A hybrid algorithm for urban transit schedule optimization | |
| Pagliara et al. | The state-of-the-art in building residential location models | |
| CN108537691A (en) | A kind of region visit intelligent management system and method | |
| US20110022428A1 (en) | Modelling a transport market | |
| Rasouli et al. | Assessment of model uncertainty in destinations and travel forecasts of models of complex spatial shopping behaviour | |
| CN102750411A (en) | Urban dynamic micro-simulation method based on multi-agent discrete choice model | |
| Pfaffenbichler | The strategic, dynamic and integrated urban land use and transport model MARS (Metropolitan Activity Relocation Simulator) | |
| CN105956694A (en) | Heterogeneous data source integrated modeling and optimizing method for interior space value of commercial real estate | |
| Yuan et al. | How to mitigate theme park crowding? A prospective coordination approach | |
| Vandet et al. | Optimal placement and sizing of charging infrastructure for EVs under information-sharing | |
| Yang et al. | Multi-objective based demand response strategy optimization considering differential demand on reliability of power system | |
| CN106951581A (en) | Commercial complex simulator | |
| Kıyıldı et al. | The capacity analysis of the check-in unit of Antalya Airport using the fuzzy logic method | |
| Abolhasani et al. | A collective decision-making framework for simulating urban land-use planning: An application of game theory with event-driven actors | |
| Li et al. | Dynamic pricing, vehicle relocation and staff rebalancing for station-based one-way electric carsharing systems considering nonlinear charging profile | |
| Habib et al. | Modeling of job mobility and location choice decisions | |
| Levinson | An evolutionary transportation planning model: Structure and application | |
| Lombardi et al. | Application of the Analytic Network Process and the Multi-modal framework to an urban upgrading case study | |
| Kobayashi | An activity model: a demand model for transportation | |
| von Winterfeldt | 11 Decisions with Multiple Stakeholders and Conflicting Objectives | |
| Bargur et al. | A comprehensive approach to the planning of the tourism industry |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20180806 Address after: No. 1239, Siping Road, Yangpu District, Shanghai Applicant after: TONGJI University Applicant after: SHANGHAI TONGJI URBAN PLANNING & DESIGN INSTITUTE Address before: No. 1239, Siping Road, Yangpu District, Shanghai Applicant before: Tongji University |
|
| TA01 | Transfer of patent application right | ||
| CB02 | Change of applicant information |
Address after: No. 1239, Siping Road, Yangpu District, Shanghai Applicant after: TONGJI University Applicant after: SHANGHAI TONGJI URBAN PLANNING & DESIGN INSTITUTE Co.,Ltd. Address before: No. 1239, Siping Road, Yangpu District, Shanghai Applicant before: Tongji University Applicant before: Shanghai Tongji Urban Planning & Design Institute |
|
| CB02 | Change of applicant information | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |