CN104318099B - The mobile analogue experiment method of dynamic point on two-dimensional random road network - Google Patents

Abstract

The invention discloses the mobile analogue experiment method of dynamic point on a kind of two-dimensional random road network, concrete steps are as follows: random number generation module generates random number; MBM sets up probabilistic model framework, and the information of path model is saved to file; Computing module sets up probabilistic model, adopts time queue algorithm to calculate and simulates the real-time status of each moment Moving Objects, being preserved by operation result with document form; Analysis module reads the feature of data to random motion and analyzes and add up, and is shown in interface by analysis result by the mode of figure.The invention solves existing mobility model and cannot simulate the problem of random motion object in the status information in each moment, is a kind of inspiration platform of many bunches of markov chains, can flexible configuration, for various problems provides experimental situation.

Description

The mobile analogue experiment method of dynamic point on two-dimensional random road network

Technical field

The invention belongs to the mobile analogue experiment method technical field of dynamic point, be specifically related to the mobile analogue experiment method of dynamic point on a kind of two-dimensional random road network.

Background technology

Classic random walk theory, appears in much mathematics and physical model, mainly considers the random walk on simple but unconfined figure.For lattice point figure, if allow particle move about down indefinite duration, is particle with probability 1 return to origin? can infinite repeatedly return to origin? nineteen twenty-one, P ó lya demonstrates and works as n=1, when 2, particle repeatedly turns back to starting point so that probability 1 is infinite, but as n>3, particle only limitedly repeatedly can turn back to starting point.

Random walk on UNICOM's non-directed graph, namely reversible Markov chain, and the inner link of electric network, and the successful Application of matrix analysis and harmonic analysis method, make it become constitutional diagram public opinion-medium most study in recent years, one of problem that achievement is the abundantest.Reversible Markov chain, has application at multiple fields.Towards the research with maneuver point movement, mainly concentrate on the foundation of location index model.Assuming that object does arbitrary motion in two-dimensional space, there is following index structure in difference as required: current and future location information successively for mobile object, creates the model that a class carries out information location management; Along with people are to the raising of past phenomenon attention rate, the model that can process mobile object historical position information has had certain development; As development trend in recent years, can process that mobile object is gone over simultaneously, the model of current and future location information etc. also arises at the historic moment, this extreme enrichment application of dynamic some movement.

In the research of mobility model, the Spatio-Temporal Data Model for Spatial of two-dimensional network mobile object is started late.Existing Spatio-Temporal Data Model for Spatial is mainly conceived to the motion state of record move object, as the record set of joining day index in road net data model.This research field of time-space network mobile object is significant for network classical analysis.In practical application, the motor pattern of mobile object can be divided into unrestricted motion (if boats and ships are in large marine traveling), constrained motion (motion as pedestrian) and the motion (as train, automobile move along permanent haulage line in certain region) in fixed network, and wherein fixed network motion is pattern the most general in application.When the motor pattern of mobile object is defined as the unrestricted motion in fixed network, its movement locus can be understood as one save over, current and Future Information, motor development and in the past irrelevant markov chain.

The mobile simulated experiment of dynamic point at present in fixed network, setting two dimensional path net is fixed mostly, less to the research of environmental change and portability.The simulated experiment of dynamic some movement is more with random walk research, more from index angle, study towards dynamic point location, provide the Platform Designing of a good simulated experimental environments less, the status information of random motion object in each moment when environmental baseline is changeable, cannot be simulated.

Summary of the invention

The object of the present invention is to provide the mobile analogue experiment method of dynamic point on a kind of two-dimensional random road network, solve existing mobility model and cannot simulate the problem of random motion object in the status information in each moment.

The technical solution adopted in the present invention is, the mobile analogue experiment method of dynamic point on two-dimensional random road network, and based on the mobile simulation experiment platform of dynamic point on two-dimensional random road network, concrete steps are as follows:

1st step: random number generation module generates random number, for MBM and computing module provide random data source after receiving order;

2nd step: MBM obtains the data source of random number generation module, sets up probabilistic model framework, and is encapsulated in the mode of class by probabilistic model framework, be kept in internal memory, the information of path model is saved to file simultaneously;

3rd step: computing module is responsible for data operation when running, random motion model is set up according to the probabilistic model framework that the 2nd step is set up, adopt time queue algorithm to calculate and simulate the real-time status of each moment Moving Objects, operation result random is in real time encapsulated in the mode of class simultaneously, be kept in internal memory, and by data output interface, operation result preserved with document form;

4th step: the file output that the file preserve the 2nd step and the 3rd step are preserved is to analysis module, analysis module reads data, whole random motion process can be reappeared, and the feature of random motion is analyzed and added up, preserve analysis result by the mode of file, by the mode of figure, real-time random motion model, the dynamic some state in each moment and analysis result are shown.

Feature of the present invention is also,

In 1st step, random number generation module comprises order receiving interface, random number generator and data transmission interface, and described random number generator generates a healthy and strong random number by calling CryptGenRandom function.

The flow process that 2nd step MBM sets up probabilistic model framework is:

Step 2.1: first arrange modeling parameters, then reads modeling parameters, according to modeling scope creation random node position;

Step 2.2: adopt Waxman modeling method to be create random walk between random node, internodal path meets Poisson distribution;

Step 2.3: carry out continuity testing to the random walk that step 2.2 creates by width first traversal, if do not have isolated node, performs step 2.4; If there is isolated node, then return step 2.2, re-create random walk; If step 2.2 repeatedly after still there is isolated node, then return step 2.1, reset modeling parameters;

Step 2.4: the establishment of entering action limit, arrange each dynamic some parameter according to simulation demand, all model datas are saved to external file, and modeling terminates.

Waxman modeling method, shown in (1):

$P(u,v)={\mathrm{\αe}}^{-\frac{d}{\left(\mathrm{\β}L\right)}}---\left(1\right)$

Wherein P (u, v) is that node u is directly connected probability with node v, and modeling parameters α >0, β <=1, d are the distances between summit u and vertex v, and L is the distance of lie farthest away in all summits in plane; α value is larger, and in figure, limit is more; β value is larger, and in figure, long limit is larger than the ratio of minor face, and Waxman thinks that the connection probability between node is relevant to its distance, and out-degree frequency obeys Poisson distribution, and distance is nearer, and probability is larger.

The concrete operation flow process of the 3rd step computing module is:

Step 3.1: computing module obtains random motion model,

Step 3.2: according to the time queue of random motion model creation, all dynamic nodes are joined in time queue, and its working time is initialized as 0;

Step 3.3: judge if run duration transfinites, then to terminate the T.T. restriction that dynamic some run duration sets when whether exceeding initial creation model computing, perform step 3.7; If run duration does not transfinite, then carry out step 3.4;

Step 3.4: queue computing time team head move a little by generation state, namely whether dynamic point moves and direction of motion at subsequent time, and according to dynamic next moment state, time when calculating dynamic some state changes next time, upgrades the state of dynamic;

$x_{ii}\left(t\right)=x_{{\mathrm{al}}_i}+\left(v_{ci}{t}^{\&prime;}\right)cos\&lsqb;\arctan \&lsqb;\frac{y_{{\mathrm{bl}}_i}-y_{{\mathrm{al}}_i}}{x_{{\mathrm{bl}}_i}-x_{{\mathrm{al}}_i}}\&rsqb;\&rsqb;---\left(2\right)$

$y_{ii}\left(t\right)=y_{{\mathrm{al}}_i}+\left(v_{ci}{t}^{\&prime;}\right)sin\&lsqb;\arctan \&lsqb;\frac{y_{{\mathrm{bl}}_i}-y_{{\mathrm{al}}_i}}{x_{{\mathrm{bl}}_i}-x_{{\mathrm{al}}_i}}\&rsqb;\&rsqb;---\left(3\right)$

$t^{\&prime;}=\left\{\begin{array}{cc}{\mathrm{\&Delta;t}}_m& (v_i=v_{ci})\\ {\mathrm{\&Delta;t}}_{m-1}& (v_i=0,m>l)\\ 0& (v_i=0,m=l)\end{array}\right.---\left(4\right)$

${\mathrm{\&Delta;t}}_m=\left\{\begin{array}{cc}\frac{{\mathrm{\&zeta;}}_l}{v_i}& v_i=v_{ci}\\ t^{\&prime;}-\underset{p=1}{\overset{i-1}{\&Sigma;}}{\mathrm{\&Delta;t}}_p& v_i=0\end{array}\right.---\left(5\right)$

In formula, l _irepresent the path at the current place of dynamic point; with represent l _ithe two-end-point in path; (x _ii(t), y _ii(t)) represent that dynamic point is at l _ithe coordinate at path place; T' is the interval sampling time; v _cifor being randomly assigned to the speed of dynamic point; Δ t _mabout the segmentation function in interval sampling time t'; v _irepresent dynamic some movement velocity, Δ t _m-1be that one refers to, refer to and decile division is carried out to t'; M is the demarcation interval number to t'; Wherein, discrete random variable ζ _l={ l ₁..., l _i..., l _krepresent path l _ilength, wherein, k represents the number in path; Δ t _prepresent the time of p paths of passing by; P represents the number of times through path;

The position of dynamic some subsequent time movement is obtained by formula (2) and formula (3); Position according to current location and subsequent time can way to acquire length l _i; The Δ t of subsequent time is upgraded by formula (4) and formula (5) _m;

Step 3.5: according to the state upgrading rear dynamic point, queue update time, is about to upgrade rear dynamic point according to its Δ t _mthe size of value, reinserts in time queue by ascending order, ensures that the dynamic point next time upgraded is arranged in the forefront of time queue;

Step 3.6: return step 3.3;

Step 3.7: finally operation result random is in real time encapsulated in the mode of class, be kept in internal memory, and by data output interface, operation result is preserved with document form.

The operational process comprising random dynamic point to the analysis of the feature of random motion in 4th step graphically reappears, the running orbit analysis of dynamic point, the probability of occurrence statistics of dynamic point in whole model platform, the tracking relationship analysis between multiple dynamic point, mobile under the state of the dynamic and stalic state conversion of dynamic point are analyzed.

The mobile simulation experiment platform of dynamic point on two-dimensional random road network comprises random number generation module and the MBM be connected with random number generation module respectively and computing module, and MBM is all connected with analysis module with computing module.

The invention has the beneficial effects as follows: the mobile analogue experiment method of dynamic point on two-dimensional random road network of the present invention, employing random number drives, create random walk model, and by time queue algorithm, simulation is with the state of maneuver point in each moment, solve existing mobility model and cannot simulate the problem of random motion object in the status information in each moment, adopt OO thought, package path model and dynamic some object, increase the reusability of code, adopt document form storing path model, dynamic dotted state data, analyze and statistics, and final display analysis result to graphically, it is a kind of inspiration platform of many bunches of markov chains, can flexible configuration, for various problems provides experimental situation.

Accompanying drawing explanation

Fig. 1 is the mobile simulation experiment platform framework map of dynamic point on two-dimensional random road network;

Fig. 2 is the dynamic point mobile analogue experiment method MBM process flow diagram on two-dimensional random road network of the present invention;

Fig. 3 is the operational flowchart of the mobile analogue experiment method computing module of dynamic point on two-dimensional random road network of the present invention;

Fig. 4 is the layering logic layers figure of the mobile analogue experiment method of dynamic point on two-dimensional random road network of the present invention;

Fig. 5 is the graph of a relation of dynamic point mobile analogue experiment method data-driven layer on two-dimensional random road network of the present invention and operation layer;

Fig. 6 is the dynamic point mobile analogue experiment method probabilistic model framework file structure figure on two-dimensional random road network of the present invention;

Fig. 7 is the dynamic point mobile analogue experiment method probabilistic model statistics file structural drawing on two-dimensional random road network of the present invention.

Embodiment

Below in conjunction with the drawings and specific embodiments, the present invention is described in detail.

The mobile analogue experiment method of dynamic point on two-dimensional random road network of the present invention, based on the mobile simulation experiment platform of dynamic point on two-dimensional random road network, as shown in Figure 1, comprise random number generation module and the MBM be connected with random number generation module respectively and computing module, MBM is all connected with analysis module with computing module, and concrete steps are as follows:

1st step: random number generation module generates random number, for MBM and computing module provide random data source after receiving order; Random number generation module comprises order receiving interface, random number generator and data transmission interface, random number generator uses CryptographicServiceProvider (CSP) technology, a healthy and strong random number is generated, for MBM and computing module provide random data source by calling CryptGenRandom function; Random number generation module, as platform base, for MBM and computing module provide random number to drive, ensures the randomness of motion model;

2nd step: MBM obtains the data source of random number generation module, sets up probabilistic model framework, comprising model node parameter, dynamic some parameter and path model, as shown in Figure 2, the idiographic flow that MBM sets up probabilistic model framework is:

Step 2.1: arrange modeling parameters by optimum configurations interface, is arranged key parameter during modeling, comprising size, the quantity of model interior joint, α and the β parameter etc. of generation pass employing Waxman modeling method of model; Read modeling parameters, according to modeling scope creation random node position;

Step 2.2: adopt Waxman modeling method to be create random walk between random node, but there is path between not all node, according to Waxman modeling method, internodal path will meet Poisson distribution, namely more likely have Path Connection between the node that distance is nearer, the curve of distribution is controlled by parameter alpha and β;

Step 2.3: because whole model must ensure by arbitrary node, other any node can be arrived, and the path using Waxman modeling method to generate only ensures to meet Poisson distribution, can not ensure that each node can be communicated with, so after generation pass, by width first traversal, continuity testing is carried out to the random walk that step 2.2 creates: if there is no isolated node, perform step 2.4; If there is isolated node, then return step 2.2, use Waxman modeling method to re-create random walk; If step 2.2 repeatedly after still there is isolated node, then returning step 2.1, by revising α and β parameter change probability distribution, resetting modeling parameters;

Waxman modeling method is a kind of probabilistic model, as the basic algorithms most in use that network topology generates, be commonly used to produce random network, can ensure that the coordinates measurement of two-dimensional random road network meets Poisson distribution, Waxman modeling method, shown in (1):

$P(u,v)={\mathrm{\&alpha;e}}^{-\frac{d}{\left(\mathrm{\&beta;}L\right)}}---\left(1\right)$

Wherein P (u, v) is that node u is directly connected probability with node v, and modeling parameters α >0, β <=1, d are the distances between summit u and vertex v, and L is the distance of lie farthest away in all summits in plane; α value is larger, and in figure, limit is more; β value is larger, and in figure, long limit is larger than the ratio of minor face, and Waxman thinks that the connection probability between node is relevant to its distance, and out-degree frequency obeys Poisson distribution, and distance is nearer, and probability is larger;

Step 2.4: after path model creates and terminates, enter the establishment of action limit, need to arrange the parameter such as the reference position of each dynamic point (i.e. mobile object), the translational speed of dynamic point, model parameter flexible design, the concrete mobile object of simulating as required is arranged, such as, if desired simulate Dynamic Vehicle Routing Problems, can arrange demand for dynamic point, be vehicle set load, finally all model datas are saved to external file, modeling terminates;

Probabilistic model framework is encapsulated in the mode of class, be kept in internal memory, the information of path model be saved to the file of XML format simultaneously, make model have reproducibility, repeatedly can carry out random motion experiment in same model, and supply a model data for experimental analysis;

3rd step: computing module is responsible for data operation when running, random motion model is set up according to the probabilistic model framework that the 2nd step adopts Waxman modeling method to set up, adopt time queue algorithm to calculate and simulate the real-time status of each moment Moving Objects, as shown in Figure 3, concrete operation flow process is:

Step 3.1: computing module obtains random motion model;

Step 3.2: according to the time queue of random motion model creation, all dynamic nodes are joined in time queue, and its working time is initialized as 0;

Step 3.3: judge if run duration transfinites, then to terminate the T.T. restriction that dynamic some run duration sets when whether exceeding initial creation model computing, perform step 3.7; If run duration does not transfinite, then carry out step 3.4;

Step 3.4: queue computing time team head move a little by generation state, namely whether dynamic point moves and direction of motion at subsequent time, and according to dynamic next moment state, time when calculating dynamic some state changes next time, upgrades the state of dynamic;

$x_{ii}\left(t\right)=x_{{\mathrm{al}}_i}+\left(v_{ci}{t}^{\&prime;}\right)cos\&lsqb;arctan\&lsqb;\frac{y_{{\mathrm{bl}}_i}-y_{{\mathrm{al}}_i}}{x_{{\mathrm{bl}}_i}-x_{{\mathrm{al}}_i}}\&rsqb;\&rsqb;---\left(2\right)$

$y_{ii}\left(t\right)=y_{{\mathrm{al}}_i}+\left(v_{ci}{t}^{\&prime;}\right)sin\&lsqb;arctan\&lsqb;\frac{y_{{\mathrm{bl}}_i}-y_{{\mathrm{al}}_i}}{x_{{\mathrm{bl}}_i}-x_{{\mathrm{al}}_i}}\&rsqb;\&rsqb;---\left(3\right)$

$t^{\&prime;}=\left\{\begin{array}{cc}{\mathrm{\&Delta;t}}_m& (v_i=v_{ci})\\ {\mathrm{\&Delta;t}}_{m-1}& (v_i=0,m>l)\\ 0& (v_i=0,m=l)\end{array}\right.---\left(4\right)$

${\mathrm{\&Delta;t}}_m=\left\{\begin{array}{cc}\frac{{\mathrm{\&zeta;}}_l}{v_i}& v_i=v_{ci}\\ t^{\&prime;}-\underset{p=1}{\overset{i-1}{\&Sigma;}}{\mathrm{\&Delta;t}}_p& v_i=0\end{array}\right.---\left(5\right)$

In formula, l _irepresent the path at the current place of dynamic point; with represent l _ithe two-end-point in path; (x _ii(t), y _ii(t)) represent that dynamic point is at l _ithe coordinate at path place; T' is the interval sampling time; v _cifor being randomly assigned to the speed of dynamic point; Δ t _mabout the segmentation function in interval sampling time t'; v _irepresent dynamic some movement velocity, Δ t _m-1be that one refers to, refer to and decile division is carried out to t'; M is the demarcation interval number to t'; Wherein, discrete random variable ζ _l={ l ₁..., l _i..., l _krepresent path l _ilength, wherein, k represents the number in path; Δ t _prepresent the time of p paths of passing by; P represents the number of times through path;

The run duration of dynamic point carries out time poll with t', and the motion conditions inside each t' time is by Δ t _mjudge, concrete determination methods is: the position being obtained the movement of dynamic some subsequent time by formula (2) and formula (3); Position according to current location and subsequent time can way to acquire length l _i; The Δ t of subsequent time is upgraded by formula (4) and formula (5) _m;

Step 3.5: according to the state upgrading rear dynamic point, queue update time, is about to upgrade rear dynamic point according to its Δ t _mthe size of value, reinserts in time queue by ascending order, ensures that the dynamic point next time upgraded is arranged in the forefront of time queue;

Step 3.6: return step 3.3;

Step 3.7: finally operation result random is in real time encapsulated in the mode of class, be kept in internal memory, and by data output interface, operation result is preserved with XML and TXT document form;

Time queue is the core of whole experimental technique, its propelling represents the motion of all dynamic points in the whole model space, time queue requires that all objects in queue all must realize time interface, all dynamic points can join in time queue, time queue is an orderly queue, the moment sequence that all objects change according to state, state changes the moment and the immediate dynamic point of current time comes queue foremost, become next processed dynamic point, after dynamic point is processed, calculate the moment that next state changes, and it is inserted in time queue in order again, experimentally requirement can be that sampling time point is added in time queue, and the insertion frequency representative of sampled point sampling interval, being also sample frequency, can arranging when inputting experiment parameter,

4th step: the XML file that the 2nd step is obtained, XML and the TXT file output that 3rd step obtains is to analysis module, during owing to generating a random motion example, might not need to carry out data analysis, or also do not know what kind of data analysis this carries out, so in the present invention, simulation and the data analysis of random motion example are what to separate, namely while simulation random motion, do not carry out data analysis, but by file output interface, data are saved, the data file of preserving is XML format or TXT form, file content can be read, by the reading of data file, whole random motion process can be reappeared, also can be used as the data source of analysis module simultaneously, operation result is carried out to the analysis of different angles, the operational process comprised with maneuver point graphically reappears, the running orbit analysis of dynamic point, the probability of occurrence statistics of dynamic point in whole model platform, tracking relationship analysis between multiple dynamic point, the mobile analysis of dynamic point under the state of the dynamic and stalic state conversion, analysis result is preserved by the mode of file, by the mode of figure by real-time random motion model, the dynamic some state in each moment and analysis result are shown in interface.

The logical organization of the mobile simulation experiment platform of dynamic point on two-dimensional random road network:

As shown in Figure 4, simulation experiment platform is divided into three levels from logical level: data-driven layer (corresponding random number generation module), operation layer (corresponding MBM and computing module) and analysis layer (correspondence analysis module).

Data-driven layer provides data-driven interface, and operation layer obtains random number by this interface and drives, and carries out stochastic arithmetic; Operation layer primary responsibility sets up probabilistic model, enters action limit real-time operation, and by output interface, preserves real-time random data, for analysis layer provides data source; Analysis layer obtains data by real-time random data source, and analyzes it, and analysis result is shown in interface to graphically the most at last.

Based on data-driven layer, operation layer is core, and the relation between them is mainly reflected in the establishment of real-time random motion model, and the simulation of the state of enforcement.As shown in Figure 5, data-driven layer is associated by command interface and data-interface with between operation layer, and interface is made up of a prescription method, the reading of data and order be conveyed through interface to perform.Use interface can reduce the degree of coupling between each module, increases the reusability of code.Data-driven layer have employed event synchronization mechanism, and synchronization only performs a request, controls the unicity of reading and writing data, ensure that the quality generating random number.

Model utilizes OO thought, the feature of abstract model, and by the succession of class, increase the reusability of code, in model structure, the inheritance of class is:

A. node base class

Node base class is the basis of whole model, can derive stationary nodes and dynamic node class by node base class.Comprise privately owned attribute in node base class: index, index is that the overall situation of each node uniquely indicates, retrieve, preserve or reading model time use;

B. stationary nodes class

Stationary nodes Similar integral is from node base class, and representing two end points of a paths in a model, is also the tie point between mulitpath.The position of stationary nodes, adjacent path is contained in this type of, and the attribute such as adjacent node.Because the model related to is in two-dimentional theorem in Euclid space, so position is a two-dimensional coordinate value.Adjacent path, referring to present node is the path of beginning or end, and each stationary nodes must have at least one adjacent path, to ensure that this stationary nodes can not become isolated node.Adjacent node, another stationary nodes that the adjacent path referring to present node has, because each stationary nodes has at least one adjacent path, so also at least have an adjacent node simultaneously.

Comprise in stationary nodes class and add paths, delete path, and preserve the method for nodal information.When modeling, by the method added paths with delete path, create whole model; After modeling terminates, by preserving the method for nodal information, nodal information is saved to hard disk.

C. time interface

Time interface defines the time method of dynamic node, comprises the method calculating subsequent time and the method upgrading current time.All dynamic nodes all must realize this interface.

D. dynamic node class

Dynamic node Similar integral, from node base class, achieves time interface simultaneously, and represent the Moving Objects on path model that moves about in a model, experimental difference, can derive different Moving Objects.Contain the speed of dynamic node in this type of, the node that sets out, destination node, the attribute such as path, place.Speed, refers to the movement velocity of dynamic node, and this attribute produces at random when modeling, and lower velocity limit and the upper limit are specified when arranging modeling parameters.Set out node, refers to dynamic node in current kinetic process, by which stationary nodes.Destination node, refers to dynamic node in current kinetic process, to which stationary nodes moves.Path, place, refers to present node and just moves on which paths.

Comprise the method for preserving nodal information in dynamic node class, and achieve in time interface the method calculating subsequent time and the method upgrading current time.

E. class of paths

Class of paths represents the access path between two stationary nodes in a model.The attributes such as the length in path, end points, index are contained in this type of.Length, refers to the Euclidean distance between the two-end-point of path.End points, refers to two stationary nodes that path connects.Index, the overall situation referring to path uniquely indicates, retrieve, preserve or reading model time use.

Comprise in class of paths and path end points, calculating path length are set, and the method for storing path information.Method when modeling by arranging path end points Makes Path; Modeling terminates the method for rear use storing path information, and nodal information is saved to hard disk.The method of calculating path length will use when testing and running.

F. path model class

Path model class saves static objects all in model, the object just no longer changed after namely creating during modeling.This type of contains stationary nodes set and set of paths.

The communication method using commons such as removing, establishment are comprised in path model class.Can be all static objects of model creation by creation method; Static objects all in model can be removed by sweep-out method.

When model of creation static object, may produce the isolated point not having to connect, this type of is traveled through all stationary nodes and path by the private method of breadth first traversal, when finding to occur isolated point, removes all static objects, again modeling.

G. Instance Interface

A real-time random motion model experiment examples of platforms must have two methods: the method for initialization example and the method for running example, Instance Interface is to this has been definition.Real-time random motion model experiment examples of platforms all must realize this interface.

H. example class

Example class contains static data and the dynamic data of a real-time random motion model experiment, comprising path model object, dynamic point, current time node, and time queue.Path model object, carries out modeling by the initialized method of example, for whole example provides static data support.Dynamic point creates and initialization when modeling, when example runs, along with the passing of time queue, calculates the real-time status of each dynamic point.Current time node, when example runs, represents the current time point be in.Time queue, the core of real-time random motion model experiment platform, all dynamic points are all according to time queue campaign.

MBM Makes Path after model terminates, can the information of path model be saved in XML file, for analysis module provides the support of static objects data, when computing module calculates the state of each dynamic point, the state of the current dynamic point of real-time preservation is in XML file, and file layout as shown in Figure 6; While calculating each dynamic dotted state, all dynamic points are added up, also the accessed situation of all static dynamic points is added up, as path and the accessed number of times of stationary nodes, after all computings terminate, statistical information is saved to text, file layout as shown in Figure 7.

The inheritance of file output interface: file output class achieves file output interface, wherein contains two methods of preserving file, thereafter in derivative various file output class, has all made carbon copies this two methods; The derived class of file output class comprises: parameter output class, time output class, path model output class and motor point output class; Each derived class carries out file storage for different objects, and file can be stored as XML format or TXT form.

The random motion that the present invention be directed in two-dimentional theorem in Euclid space on random UNICOM figure designs, so for different random motion models, different data analysing methods may be taked, and the data analysis of different angles is carried out for same group of data or same example.For the uncertainty of data analysis module, the present invention is that data analysis module provides interface.In the interface by the reading to data file, can obtain the status information of whole random motion example, the analytical approach then in calling interface, to data analysis.Interface uses the form of dynamic link library, is separated by data analysis module, reduces the degree of coupling of intermodule, make analysis module that different language can be adopted to write with the mobile simulation experiment platform of whole dynamic point.

The mobile analogue experiment method of dynamic point on two-dimensional random road network of the present invention, use Object--oriented method, route in model, node and Moving Objects are encapsulated in class, by abstract, the encapsulation of class, inherit, the characteristic such as polymorphic, carry out data operation, for model provides external standard interface, increase the extensibility of model, in model, data acquisition XML format and text formatting are preserved, for data analysis provides data source, a kind of inspiration platform of many bunches of markov chains, can flexible configuration, for various problems provides experimental situation.

Claims (1)

1. the mobile analogue experiment method of dynamic point on two-dimensional random road network, is characterized in that, based on the mobile simulation experiment platform of dynamic point on two-dimensional random road network, concrete steps are as follows:

1st step: random number generation module generates random number, for MBM and computing module provide random data source after receiving order;

2nd step: MBM obtains the data source of random number generation module, sets up probabilistic model framework, and is encapsulated in the mode of class by probabilistic model framework, be kept in internal memory, the information of path model is saved to file simultaneously;

3rd step: computing module is responsible for data operation when running, random motion model is set up according to the probabilistic model framework that the 2nd step is set up, adopt time queue algorithm to calculate and simulate the real-time status of each moment Moving Objects, operation result random is in real time encapsulated in the mode of class simultaneously, be kept in internal memory, and by data output interface, operation result preserved with document form;

4th step: the file output that the file preserve the 2nd step and the 3rd step are preserved is to analysis module, analysis module reads data, whole random motion process can be reappeared, and the feature of random motion is analyzed and added up, preserve analysis result by the mode of file, by the mode of figure, real-time random motion model, the dynamic some state in each moment and analysis result are shown;

The mobile simulation experiment platform of dynamic point on described two-dimensional random road network comprises random number generation module and the MBM be connected with random number generation module respectively and computing module, and MBM is all connected with analysis module with computing module; Described random number generation module comprises order receiving interface, random number generator and data transmission interface, and described random number generator generates a healthy and strong random number by calling CryptGenRandom function;

The flow process that described 2nd step MBM sets up probabilistic model framework is:

Step 2.1: first arrange modeling parameters, then reads modeling parameters, according to modeling scope creation random node position;

Step 2.2: adopt Waxman modeling method to be create random walk between random node, internodal path meets Poisson distribution;

Step 2.3: carry out continuity testing to the random walk that described step 2.2 creates by width first traversal, if do not have isolated node, performs step 2.4; If there is isolated node, then return step 2.2, re-create random walk; If step 2.2 repeatedly after still there is isolated node, then return step 2.1, reset modeling parameters;

Step 2.4: the establishment of entering action limit, arrange each dynamic some parameter according to simulation demand, all model datas are saved to external file, and modeling terminates;

Described Waxman modeling method, shown in (1):

$P(u,v)={\mathrm{\&alpha;e}}^{-\frac{d}{\left(\mathrm{\&beta;}L\right)}}---\left(1\right)$

Wherein P (u, v) is that node u is directly connected probability with node v, and modeling parameters α >0, β <=1, d are the distances between summit u and vertex v, and L is the distance of lie farthest away in all summits in plane; α value is larger, and in figure, limit is more; β value is larger, and in figure, long limit is larger than the ratio of minor face, and Waxman thinks that the connection probability between node is relevant to its distance, and out-degree frequency obeys Poisson distribution, and distance is nearer, and probability is larger;

The concrete operation flow process of described 3rd step computing module is:

Step 3.1: computing module obtains random motion model,

Step 3.2: according to the time queue of random motion model creation, all dynamic nodes are joined in time queue, and its working time is initialized as 0;

Step 3.3: judge if run duration transfinites, then to terminate the T.T. restriction that dynamic some run duration sets when whether exceeding initial creation model computing, perform step 3.7; If run duration does not transfinite, then carry out step 3.4;

Step 3.4: queue computing time team head move a little by generation state, namely whether dynamic point moves and direction of motion at subsequent time, and according to dynamic next moment state, time when calculating dynamic some state changes next time, upgrades the state of dynamic;

$x_{ii}\left(t\right)=x_{{\mathrm{al}}_i}+\left(v_{ci}{t}^{\&prime;}\right)cos\&lsqb;arctan\&lsqb;\frac{y_{{\mathrm{bl}}_i}-y_{{\mathrm{al}}_i}}{x_{{\mathrm{bl}}_i}-x_{{\mathrm{al}}_i}}\&rsqb;\&rsqb;---\left(2\right)$

$y_{ii}\left(t\right)=y_{{\mathrm{al}}_i}+\left(v_{ci}{t}^{\&prime;}\right)\sin \&lsqb;arctan\&lsqb;\frac{y_{{\mathrm{bl}}_i}-y_{{\mathrm{al}}_i}}{x_{{\mathrm{bl}}_i}-x_{{\mathrm{al}}_i}}\&rsqb;\&rsqb;---\left(3\right)$

$t^{\&prime;}=\left\{\begin{array}{cc}{\mathrm{\&Delta;t}}_m& (v_i=v_{ci})\\ {\mathrm{\&Delta;t}}_{m-1}& (v_i=0,m>l)\\ 0& (v_i=0,m=l)\end{array}\right.---\left(4\right)$

${\mathrm{\&Delta;t}}_m=\left\{\begin{array}{cc}\frac{{\mathrm{\&zeta;}}_l}{v_i}& v_i=v_{ci}\\ t^{\&prime;}-\underset{p=1}{\overset{i-1}{\&Sigma;}}{\mathrm{\&Delta;t}}_p& v_i=0\end{array}\right.---\left(5\right)$

In formula, l _irepresent the path at the current place of dynamic point; with represent l _ithe two-end-point in path; (x _ii(t), y _ii(t)) represent that dynamic point is at l _ithe coordinate at path place; T ' is the interval sampling time; v _cifor being randomly assigned to the speed of dynamic point; Δ t _mabout the interior segmentation function of interval sampling time t '; v _irepresent dynamic some movement velocity, Δ t _m-1be that one refers to, refer to and decile division is carried out to t '; M is the demarcation interval number to t '; Wherein, discrete random variable ζ _l={ l ₁..., l _i..., l _krepresent path l _ilength, wherein, k represents the number in path; Δ t _prepresent the time of p paths of passing by; P represents the number of times through path;

The position of dynamic some subsequent time movement is obtained by formula (2) and formula (3); Position according to current location and subsequent time can way to acquire length l _i; The Δ t of subsequent time is upgraded by formula (4) and formula (5) _m;

Step 3.5: according to the state upgrading rear dynamic point, queue update time, is about to upgrade rear dynamic point according to its Δ t _mthe size of value, reinserts in time queue by ascending order, ensures that the dynamic point next time upgraded is arranged in the forefront of time queue;

Step 3.6: return step 3.3;

Step 3.7: finally operation result random is in real time encapsulated in the mode of class, be kept in internal memory, and by data output interface, operation result is preserved with document form;

The operational process comprising random dynamic point to the analysis of the feature of random motion in described 4th step graphically reappears, the running orbit analysis of dynamic point, the probability of occurrence statistics of dynamic point in whole model platform, the tracking relationship analysis between multiple dynamic point, mobile under the state of the dynamic and stalic state conversion of dynamic point are analyzed.