测绘学报
測繪學報
측회학보
Acta Geodaetica et Cartographica Sinica
2015年
10期
1160-1166,1176
,共8页
尹章才%孙华涛%陈雪菲%刘清全
尹章纔%孫華濤%陳雪菲%劉清全
윤장재%손화도%진설비%류청전
概率时间地理%马尔科夫链%定向移动%棱柱体
概率時間地理%馬爾科伕鏈%定嚮移動%稜柱體
개솔시간지리%마이과부련%정향이동%릉주체
probabilistic time geography%Markov chains%directed movements%prism
概率时间地理是经典时间地理基于概率的一种扩展,它采用概率描述移动对象在可达位置的非等可能性。已有的概率时间地理是基于正态分布或布朗桥的,其方差与移动速度无关或随移动速度的增大而发散,因而难以兼顾应用针对性和稳定性。本文提出了一种基于马尔科夫链的概率时间地理方法。首先,构建中间关于两边的双向条件马尔科夫链,它在移动速度足够大时的极限可视为布朗桥,因而具有稳定性数字特征。然后,建立定向移动到马尔科夫链的映射关系,主要是根据定向移动的时空位置、移动速度等信息建立马尔科夫链的步长、状态空间和转移矩阵,这样马尔科夫链与移动速度有关。最后,利用双向马尔科夫链连续计算定向移动在任意时刻的概率分布云,其方差的针对性和稳定性在实例中进行了验证。
概率時間地理是經典時間地理基于概率的一種擴展,它採用概率描述移動對象在可達位置的非等可能性。已有的概率時間地理是基于正態分佈或佈朗橋的,其方差與移動速度無關或隨移動速度的增大而髮散,因而難以兼顧應用針對性和穩定性。本文提齣瞭一種基于馬爾科伕鏈的概率時間地理方法。首先,構建中間關于兩邊的雙嚮條件馬爾科伕鏈,它在移動速度足夠大時的極限可視為佈朗橋,因而具有穩定性數字特徵。然後,建立定嚮移動到馬爾科伕鏈的映射關繫,主要是根據定嚮移動的時空位置、移動速度等信息建立馬爾科伕鏈的步長、狀態空間和轉移矩陣,這樣馬爾科伕鏈與移動速度有關。最後,利用雙嚮馬爾科伕鏈連續計算定嚮移動在任意時刻的概率分佈雲,其方差的針對性和穩定性在實例中進行瞭驗證。
개솔시간지리시경전시간지리기우개솔적일충확전,타채용개솔묘술이동대상재가체위치적비등가능성。이유적개솔시간지리시기우정태분포혹포랑교적,기방차여이동속도무관혹수이동속도적증대이발산,인이난이겸고응용침대성화은정성。본문제출료일충기우마이과부련적개솔시간지리방법。수선,구건중간관우량변적쌍향조건마이과부련,타재이동속도족구대시적겁한가시위포랑교,인이구유은정성수자특정。연후,건립정향이동도마이과부련적영사관계,주요시근거정향이동적시공위치、이동속도등신식건립마이과부련적보장、상태공간화전이구진,저양마이과부련여이동속도유관。최후,이용쌍향마이과부련련속계산정향이동재임의시각적개솔분포운,기방차적침대성화은정성재실례중진행료험증。
Probabilistic time geography (PTG) is suggested as an extension of (classical ) time geography ,in order to present the uncertainty of an agent located at the accessible position by probability . This may provide a quantitative basis for most likely finding an agent at a location .In recent years ,PTG based on normal distribution or Brown bridge has been proposed ,its variance ,however ,is irrelevant with the agent’s speed or divergent with the increase of the speed;so they are difficult to take into account application pertinence and stability .In this paper , a new method is proposed to model PTG based on Markov chain .Firstly ,a bidirectional conditions Markov chain is modeled ,the limit of which ,when the movingspeedislargeenough,canberegardedastheBrownbridge,thushasthecharacteristicsofdigital stability .Then ,the directed movement is mapped to Markov chains .The essenti al part is to build step length ,the state space and transfer matrix of Markov chain according to the space and time position of directional movement , movement speed information , to make sure the Markov chain rel ated to the movement speed .Finally ,calcul ating continuously the probability distribution of the directed movement at any time by the Markov chains ,it can be get the possibility of an agent located at the accessible position . Experimental results show that ,the vari ance based on Markov chains not only is rel ated to speed ,but also is tending towards stability with increasing the agent’s maximum speed .