周周记：数学建模学习（35）

摘要：Share interest, spread happiness, increase knowledge, leave a beautiful!

分享兴趣，传播快乐，增长见闻，留下美好！

亲爱的您，这里是LearningYard新学苑。

今天小编给您带来

【周周记：数学建模学习（35）】

欢迎您的访问！

Share interest, spread happiness, increase knowledge, leave a beautiful!

Dear, this is LearningYard New Academy.

Today, the editor brings you

"Weekly Diary: Learning Mathematical Modeling (35)"

Welcome to your visit!

马尔可夫算法原理及案例分析

Principles and Case Analysis of Markov Algorithms

1.马尔可夫链的数学定义和核心概念

1.Markov Chain Mathematical Definition and Core Concepts

马尔可夫链是一个随机过程 Xn，其中 Xn 表示在时间 n 的状态。它包含以下几个核心要素：

A Markov chain is a stochastic process Xn, where Xn represents the state at time n. It includes the following core elements:

状态空间 S：所有可能状态的集合，例如 S = {s1, s2, ..., sm}。

The state space S: the set of all possible states, for example, S = {s1, s2, ..., sm}.

初始状态概率分布 π0：在时间 n=0 时，系统处于各个状态的概率分布。

The initial state probability distribution π0: the probability distribution of the system being in each state at time n=0.

转移概率 P：状态之间的转移概率，通常表示为一个矩阵，其中 Pij = P(Xn+1 = sj | Xn = si) 是在时间 n 处于状态 si 的条件下，在时间 n+1 转移到状态 sj 的概率。

The transition probability P: the probability of transitioning between states, usually represented as a matrix, where Pij = P(Xn+1 = sj | Xn = si) is the probability of transitioning from state si at time n to state sj at time n+1.

2. 马尔可夫性质

2.Markov Property

马尔可夫性质表明未来状态的概率分布仅依赖于当前状态，而与之前的状态历史无关。数学上，这可以表示为：

P(Xn+1 = sj | Xn = si, Xn-1 = sin-1, ..., X0 = si0) = P(Xn+1 = sj | Xn = si)

The Markov property states that the probability distribution of future states depends only on the current state and not on the history of previous states. Mathematically, this can be expressed as:

P(Xn+1 = sj | Xn = si, Xn-1 = sin-1, ..., X0 = si0) = P(Xn+1 = sj | Xn = si)

3. 状态的分类

3.State Classification

吸收状态：一旦进入该状态，系统将永远停留在该状态。

Absorbing states: Once the system enters this state, it will stay in it forever.

瞬时状态：系统在这些状态中停留的概率为零。

Transient states: The probability of the system staying in these states is zero.

周期性状态：系统只能在特定时间回到状态 i。

Periodic states: The system can only return to state i at specific times.

不可约状态：从任何状态 i 都可以到达任何状态 j。

Reducible states: It is possible to go from any state i to any state j.

4. 长期行为和平稳分布

4.Long-term Behavior and Stationary Distribution

马尔可夫链的长期行为可以通过其平稳分布来描述。如果存在一个概率分布 π，使得对于所有的 n： πP = π

The long-term behavior of a Markov chain can be described by its stationary distribution. If there exists a probability distribution π such that for all n:πP = π

则 π 是马尔可夫链的平稳分布。这意味着，如果系统在足够长的时间后达到平稳分布，那么未来状态的概率分布将不再随时间变化。

Then π is the stationary distribution of the Markov chain. This means that if the system reaches a stationary distribution after a sufficiently long time, the probability distribution of future states will no longer change over time.

5. 案例分析

5.Case Analysis

案例1：地铁拥堵系数分析

Case 1: Subway Congestion Coefficient Analysis

需求：预测未来一个小时的地铁拥堵情况，设置地铁拥堵系数为三个可选参数（1,2,3）。

Requirements: Predict the subway congestion situation in the next hour, setting the subway congestion coefficient to three optional parameters (1, 2, 3).

依赖条件：

Dependent conditions:

1. 根据上一个站的情况，统计并计算后一个站拥堵系数情况的状态发生概率，构成状态分布矩阵 S。

1. Based on the situation of the previous station, statistics and calculations of the congestion coefficient situation of the next station are made to form the state distribution matrix S.

2. 站台之间具有强关联性质，即路过了上一个站台那么下一个站台发生拥堵的可能性是具有一定的规律性的，有一定的概率可以统计得出。

2. There is a strong correlation between platforms, that is, the possibility of congestion at the next platform after passing the previous platform has a certain regularity, and a certain probability can be statistically derived.

分析：通过构建马尔可夫链模型，我们可以预测未来地铁站的拥堵情况。这个案例展示了马尔可夫链在处理具有时间序列特性的问题时的实用性。

Analysis: By constructing a Markov chain model, we can predict the congestion situation of future subway stations. This case demonstrates the practicality of Markov chains in dealing with problems with time series characteristics.

案例2：天气预报模型

Case 2: Weather Forecasting Model

应用：使用马尔可夫链建立天气预测模型，只需要最近或现在的动态资料则可按转移概率预测将来，这样就可以很方便地达到预测天气变化的目的。

Applications: Using Markov chains to establish a weather forecasting model, only the most recent or current dynamic data is needed to predict the future according to transition probabilities, which can conveniently achieve the purpose of predicting weather changes.

今天的分享就到这里了。

如果您对今天的文章有独特的看法，

欢迎给我们留言，

让我们相约明天，

祝您今天过得开心！

That's all for today's share.

If you have a unique view of today's article,

Please leave us a message,

Let's meet tomorrow,

Have a nice day!

本文由LearningYard新学苑原创，如有侵权，请联系删除。

参考资料：哔哩哔哩