An improved markov chain-based electric vehicle charging and discharging probability prediction method

Abstract

This invention discloses a method for predicting the probability of electric vehicle charging and discharging based on an improved Markov chain. Taking charging and discharging piles as the research object, firstly, the states of the Markov chain are abstracted. The current total power consumption of the charging and discharging piles at the beginning of the cycle is obtained, and the state intervals are divided according to the adjustable range of the power consumption of the charging and discharging piles. Secondly, the decision-making behaviors of the charging and discharging piles are divided into four types: charging behavior, fast discharging behavior, slow discharging behavior, and stationary behavior. Then, the TPC method is used to establish a control strategy, determine the number of charging and discharging piles allocated for charging and discharging behaviors at a certain moment, and calculate the transition probability to establish a state transition matrix. This method is beneficial for leveraging the role of charging and discharging piles in the control of electric vehicle charging and discharging under the consideration of randomness, and for providing an intuitive and quantitative analysis of the charging and discharging amount of electric vehicles under control.

Description

Technical Field

[0001] This invention belongs to the field of power dispatching technology, specifically relating to a method for predicting the probability of charging and discharging of electric vehicles based on an improved Markov chain. Background Technology

[0002] With the development of the electric vehicle industry, the randomness of electric vehicle charging and discharging has led to a large influx of electric vehicles into the grid, resulting in a widening peak-valley difference in the power grid. Against this backdrop, electric vehicle charging and discharging stations play a crucial role as executors of dispatch instructions issued by aggregators or virtual power plants. Existing research has considered the randomness of electric vehicle charging and discharging, applying Markov chains for solutions. However, these studies primarily consider the user's perspective, focusing on electric vehicles as the research object and travel distance as a reference, describing the changes in the state of charge of the power battery during a single electric vehicle user's daily journey. The drawback is that, from the aggregator's perspective, the high spatial randomness of electric vehicle users makes unified control difficult. Focusing on charging and discharging stations facilitates the aggregation and control of electric vehicle clusters; however, research on charging and discharging stations is still relatively limited. Furthermore, current applications of Markov chains are limited to probabilistic predictions of given electricity consumption behavior, without providing control strategies to optimize the results. Summary of the Invention

[0003] To address the shortcomings mentioned in the background art, the present invention aims to provide a method for predicting the probability of electric vehicle charging and discharging based on an improved Markov chain. This method collects historical charging and discharging pile usage rates, the proportion of charging and discharging behaviors, and the total electricity consumption of the current charging and discharging pile cluster. Based on the Markov chain model, it enables the prediction of the total electricity consumption after adjustment of the charging and discharging piles under the uncertainty characteristics of electric vehicle charging and discharging.

[0004] The objective of this invention can be achieved through the following technical solution: a method for predicting the charging and discharging probability of electric vehicles based on an improved Markov chain, comprising the following steps:

[0005] Obtain the current total power consumption of the charging and discharging pile cluster at the start of the prediction period, and divide the state intervals according to the adjustable range of the total power consumption of the charging and discharging pile cluster, which are all possible states of the Markov chain.

[0006] Classify the decision-making behavior of individual charging and discharging piles, and calculate the state transition of the charging and discharging pile cluster after the decision-making behavior is selected based on the decision-making behavior of each charging and discharging pile.

[0007] Obtain the historical proportion of charging and discharging behavior of the charging and discharging pile cluster at the same time, and use it as the probability of charging behavior and discharging behavior of a single charging and discharging pile.

[0008] The TPC method is used to establish a control strategy, calculate the state transition probability of the charging and discharging pile cluster under the control strategy, list the Markov state transition matrix, and predict the charging and discharging probability of electric vehicles based on the matrix.

[0009] Preferably, the state interval partitioning process includes the following steps:

[0010] First, establish the state model of the Markov chain:

[0011] Using time T as a period, the total electricity consumption W of the charging and discharging pile cluster from the start of the statistics to any given time is considered as state S. At the beginning of the prediction period, the total electricity consumption W0 of the current charging and discharging pile cluster is obtained, and its current state interval is taken as the initial state S0. After a time interval Δt, the current state interval of the total electricity consumption of the charging and discharging pile cluster is taken as the next state S1. After a time interval i·Δt, the current state interval of the total electricity consumption of the charging and discharging pile cluster is taken as state S. i The final state is S T / Δt ;

[0012] Secondly, divide the state intervals to obtain all possible states:

[0013] The adjustable power consumption range is [W0-W]. max ,W0+W max ], W max To determine the maximum capacity of the charging pile cluster within a prediction period, the range is divided into M state intervals, each with a size of [missing value]. Then the i-th state interval is

[0014] Establish the benchmark interval as

[0015] Preferably, the decision-making behavior of a single charging / discharging pile and the state transition process of the charging pile cluster include the following steps:

[0016] First, the decision-making behavior a of a single charging / discharging pile. i There are four types:

[0017] Charging behavior: Electric vehicles charging at charging stations, a i =1+;

[0018] Fast-discharge behavior: Charging and discharging stations quickly charge electric vehicles, a i =1--;

[0019] Slow-release behavior: Charging stations slowly charge electric vehicles, a i =1-;

[0020] Static behavior: The charging / discharging station neither charges nor discharges, a i =0;

[0021] Secondly, determine the state transition of the charging and discharging pile cluster after the decision-making behavior is selected:

[0022] At time t, the charging and discharging pile cluster is in the i-th state interval, where a and b are the lower and upper limits of this state interval, respectively. The state interval at time t+Δt is...

[0023]

[0024] Where N is the total number of charging and discharging stations. Let P be the current charging / discharging power of the i-th charging / discharging pile. -- For the fast discharge power of the charging and discharging pile, P - For the slow discharge power of the charging and discharging pile, P + For the charging and discharging pile's charging power, P -- <0, P - <0, P + >0.

[0025] Preferably, the process of establishing the Markov state transition matrix using the TPC method includes the following steps:

[0026] First, to maintain the total electricity consumption of the charging and discharging pile cluster within a benchmark range, a control strategy model is established:

[0027] Suppose that the total electricity consumption of the current charging and discharging pile cluster is in the i-th state interval, and its upper and lower limits differ from the baseline interval. The number of charging / discharging piles allocated for discharging is: The number of units used for charging is Where α is the linear coefficient and N is the total number of charging and discharging piles;

[0028] Secondly, calculate the probability of charging and discharging behavior of a single charging and discharging pile:

[0029] Collect the number of charging / discharging stations for charging behavior, fast discharge behavior, and slow discharge behavior over m historical time periods:

[0030]

[0031] Where, ρ + ρ -- ρ - These represent the probabilities of charging behavior, fast discharge behavior, and slow discharge behavior for a single charging / discharging station, respectively, where N is the probability of charging behavior, fast discharge behavior, and slow discharge behavior. a5ai+,i N a5ai-,i N represents the number of charging / discharging charging piles allocated for charging and discharging after adjustment at the i-th time period. +,i N --,i N -,iThese represent the number of charging / discharging stations in charging, fast discharging, and slow discharging states at the i-th time period, respectively.

[0032] Constraints: N +,i ≤N a5ai+,i N --,i +N -,i ≤N a5ai-,i ;

[0033] Finally, calculate the state transition probability:

[0034] The total electricity consumption of the charging and discharging pile cluster changes as it transitions from the i-th state interval to the j-th state interval. Suppose that satisfying the state transition requires N ；eed— The charging / discharging pile is in slow discharge behavior, N ；eed-- One charging / discharging pile is in a fast discharge behavior, N need+ One charging / discharging station is in the charging activity, N need0 If a charging / discharging pile is in a static state, then the probability ρ of transitioning from the i-th state interval to the j-th state interval is... i,j for:

[0035]

[0036] in (a,b)∈{(N a5ai- N need- ),(N a5ai- N need-- ),(N a5ai+ N need+ The constraints are as follows:

[0037] ΔW=(P - ·N ；eed— +P -- ·N need-- +P + ·N need+ )·Δt

[0038] N ；eed— +N need-- +N need1 +N need0 =N

[0039] N need— +N need-- ≤N a5ai-

[0040] N need1 ≤N a5ai1

[0041] Finally, the product of all state transitions of the charging pile cluster within a cycle is calculated to obtain the probability prediction for each charging and discharging scenario, and the state transition matrix is ​​listed:

[0042]

[0043] In the formula, ρ 1,M and ρ M,1 Let ρ be the probability of transitioning from the 1st state interval to the Mth state interval and the probability of transitioning from the Mth state interval to the 1st state interval, respectively. 1,1 and ρ M,M These represent the probabilities of keeping the 1st and Mth state intervals unchanged, respectively.

[0044] The beneficial effects of this invention are:

[0045] (1) This invention takes charging and discharging piles as the research object and controls the charging and discharging power by allocating the number of charging piles and discharging piles, thus avoiding the uncontrollability of user behavior choices in studies targeting electric vehicle users.

[0046] (2) This invention provides a TPC method, and under this method, a Markov state transition probability matrix is ​​established to calculate the electric vehicle power consumption after one cycle under the control strategy. By comparing the electric vehicle power consumption under no control, the effect of the control strategy on stabilizing the power grid can be displayed intuitively and quantitatively.

[0047] (3) The present invention takes into account the uncertainty of electric vehicle charging and discharging. Based on historical data, it intuitively derives the stabilizing effect of a control strategy. On the one hand, it helps to reduce the fluctuation of electricity consumption caused by electric vehicles entering the grid. On the other hand, it helps aggregators or virtual power plants to evaluate the effectiveness of the control strategy. Attached Figure Description

[0048] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] like Figure 1 As shown, the electric vehicle charging and discharging probability prediction method based on improved Markov chains according to the present invention includes the following steps:

[0051] Step 1: Divide the state intervals.

[0052] Using time T as a period, the total electricity consumption W of the charging and discharging pile cluster from the start of the statistics to any given time is considered as state S. The state interval where the total electricity consumption W0 of the charging and discharging pile cluster is located at the beginning of the prediction period is the initial state S0. The state interval where the total electricity consumption of the charging and discharging pile cluster is located after an interval of time Δt is the next state S1, and so on, until the state S is reached after an interval of time i·Δt. i The final state is S T / Δt .

[0053] Obtain the total power consumption W0 of the charging and discharging pile cluster at the start of the prediction period, and based on this, derive the adjustable power range [W0-W]. max ,W0+W max The range is divided into M state intervals, each with a size of [missing value]. Then the i-th state interval is To reduce peak-valley fluctuations, a benchmark interval is established as follows:

[0054] Step 2: Classify the decision-making behavior of charging and discharging piles and determine the state transition process.

[0055] First, the decision-making behavior a of a single charging / discharging pile. i There are four types:

[0056] Charging behavior, electric vehicles charging at charging stations: a i =1+;

[0057] Fast charging behavior, charging and discharging stations quickly charge electric vehicles: a i =1--;

[0058] Slow-release behavior, where charging stations slowly charge electric vehicles: a i =1-;

[0059] At rest, the charging / discharging station neither charges nor discharges: a i =0.

[0060] Secondly, consider the transition of the state interval of the charging and discharging pile cluster after the decision-making behavior is selected: the state interval of the charging and discharging pile cluster at time t is [a,b], and the state interval at time t+Δt is...

[0061]

[0062] Where N is the total number of charging and discharging stations. Let P be the current charging / discharging power of the i-th charging / discharging pile. -- For the fast discharge power of the charging and discharging pile, P - P1 represents the slow discharge power of the charging and discharging pile, and P2 represents the charging power of the charging and discharging pile. -- <0, P -<0, P1>0.

[0063] Step 3: Calculate the probability of charging and discharging behavior of a single charging and discharging pile.

[0064] Based on the individual charging / discharging pile decision-making behavior classification in step 2, the number of charging / discharging piles exhibiting charging behavior, fast discharge behavior, and slow discharge behavior at m historical time points is collected. The selection probabilities for charging behavior, fast discharge behavior, and slow discharge behavior of a single charging / discharging pile are calculated:

[0065]

[0066] Wherein, ρ1, ρ -- ρ - These represent the probabilities of charging behavior, fast discharge behavior, and slow discharge behavior for a single charging / discharging station, respectively, where N is the probability of charging behavior, fast discharge behavior, and slow discharge behavior. a5ai1,i N a5ai-,i N represents the number of charging / discharging charging piles allocated for charging and discharging after adjustment at the i-th time period. 1,i N --,i N -,i These represent the number of charging / discharging stations in charging, fast-discharging, and slow-discharging states at the i-th time period, respectively.

[0067] Constraints: N 1,i ≤N a5ai1,i N --,i +N -,i ≤N a5ai-,i .

[0068] Step 4: Employing the Transfer Probability Calculation Considering State Control of Charging Pile (TPC) method, a control strategy is first established. Based on the difference between the current state interval and the baseline interval, the number of charging / discharging piles is allocated for discharging and charging respectively. By adjusting the number of available discharging and charging piles, the charging / discharging power of the user cluster is limited. The state transition probability under this control strategy is calculated, and a Markov state transition matrix is ​​generated. Based on this matrix, the charging / discharging probability of electric vehicles is predicted. The charging pile cluster state undergoes multiple transitions within one cycle. The product of all transition processes within one cycle is calculated to obtain the probability prediction for each charging / discharging scenario.

[0069] Based on the state intervals defined in step 1, calculate the change in total electricity consumption of the charging and discharging pile when transitioning from the i-th state to the j-th state.

[0070] Based on the state intervals obtained in step 1 and step 3, the probability of charging and discharging behavior of a single charging and discharging pile is calculated to determine the number of charging and discharging piles N required to transition from the i-th state interval to the j-th state interval. need1 Number N of fast-discharge charging / discharging stations need-- Number N of slow-release charging / discharging stations ；eed— Number N of static charging / discharging piles ；eed0 Calculate the state transition probability from the i-th state interval to the j-th state interval.

[0071]

[0072] in (a,b)∈{(N a5ai- N ；eed- ),(N a5ai- N ；eed-- ),(N a5ai1 N need1 )}

[0073] Constraints:

[0074] ΔW=(P - ·N need— +P -- ·N need-- +P1·N need1 )·Δt

[0075] N need— +N need-- +N need1 +N need0 =N

[0076] N need— +N need-- ≤N a5ai-

[0077] N need1 ≤N a5ai1

[0078] Finally, based on the Markov state transition matrix, the charging pile cluster undergoes multiple state transitions within a cycle. By calculating the product of the probabilities of all transition processes within a cycle, the probability prediction for each charging and discharging process can be obtained. Simultaneously, the probability prediction for the state interval where the total electricity consumption ends in the cycle can be calculated. The state transition matrix is ​​listed below:

[0079]

[0080] In the formula, ρ 1,M and ρ M,1 Let ρ be the probability of transitioning from the 1st state interval to the Mth state interval and the probability of transitioning from the Mth state interval to the 1st state interval, respectively.1,1 and ρ M,M These represent the probabilities of keeping the 1st and Mth state intervals unchanged, respectively.

Claims

1. An improved Markov chain-based electric vehicle charging and discharging probability prediction method, characterized in that, Includes the following steps: Obtain the current total power consumption of the charging and discharging pile cluster at the start of the prediction period, and divide the state interval according to the adjustable range of the total power consumption of the charging and discharging pile cluster, which serves as all possible states of the Markov chain. Classify the decision-making behavior of individual charging and discharging piles, and calculate the state transition of the charging and discharging pile cluster after the decision-making behavior is selected based on the decision-making behavior of each charging and discharging pile. Obtain the historical proportion of charging and discharging behavior of the charging and discharging pile cluster at the same time, and use it as the probability of charging behavior and discharging behavior of a single charging and discharging pile. A control strategy is established using the TPC method. The state transition probability of the charging and discharging pile cluster under the control strategy is calculated, and the Markov state transition matrix is ​​listed. Based on this matrix, the charging and discharging probability of electric vehicles is predicted. The process of establishing the Markov state transition matrix using the TPC method includes the following steps: First, to maintain the total electricity consumption of the charging and discharging pile cluster within a benchmark range, a control strategy model is established: Assume the total electricity consumption of the current charging and discharging pile cluster is at the [number]th [level]. Each state interval has its upper and lower limits differing from the baseline interval. Then the number of charging / discharging piles allocated for discharging is: The number used for charging is ,in The coefficients are linear. This represents the total number of charging and discharging stations. Secondly, calculate the probability of charging and discharging behavior of a single charging and discharging pile: Collection history The number of charging / discharging stations during charging behavior, fast discharge behavior, and slow discharge behavior at each time period: ； in, , , These represent the probabilities of charging behavior, fast discharge behavior, and slow discharge behavior at a single charging / discharging station, respectively. , The first The number of charging and discharging stations allocated for charging and discharging after adjustment at a certain time period. , , The first The number of charging / discharging stations that are in charging, fast discharging, and slow discharging behavior at any given time; Constraints: ; Finally, calculate the state transition probability: From the The state interval transitions to the first... The total electricity consumption of the charging and discharging pile cluster changes within each state interval. Assume that the state transition requirements are met. The charging / discharging station is in a slow-release behavior. The charging and discharging pile is in a fast-discharge behavior. Each charging and discharging station is in the process of charging. If the charging / discharging pile is stationary, then from the first... The state interval transitions to the first... The probability of each state interval for: ， in , The constraints are as follows: ， ， ， ， Finally, the product of all state transitions of the charging pile cluster within a cycle is calculated to obtain the probability prediction for each charging and discharging scenario, and the state transition matrix is ​​listed: ， In the formula, and They transition from the first state interval to the second state interval, respectively. The probability of the nth state interval and from the nth state interval The probability of transitioning from one state interval to the first state interval. and Keep the first and the second respectively The probability that a state interval remains unchanged.

2. The electric vehicle charging and discharging probability prediction method based on improved Markov chains according to claim 1, characterized in that, State interval partitioning process Includes the following steps: First, establish the state model of the Markov chain: In time As one cycle, the electricity consumption of the charging and discharging pile cluster is counted from the start of the cycle to any given time. Considered a state At the start of the prediction period, the total electricity consumption of the current charging and discharging pile cluster is obtained. Take its current state interval as the initial state. The subsequent interval The state range in which the total power consumption of the charging and discharging pile cluster is located is the next state. ,interval The state range in which the total electricity consumption of the charging and discharging pile cluster is after a certain time is called state. The final state is ; Secondly, divide the state intervals to obtain all possible states: The adjustable range of power consumption is: , To determine the maximum capacity of the charging pile cluster within a forecast period, the range is divided into: There are state intervals, each state interval having a size of . Then the first The number of state intervals is ; Establish the benchmark interval as .

3. The electric vehicle charge / discharge probability prediction method based on improved Markov chains according to claim 1, characterized in that, The decision-making behavior of a single charging / discharging pile and the state transition process of a charging pile cluster include the following steps: First, the decision-making behavior of a single charging / discharging pile. There are four types: Charging behavior: Electric vehicles charging at charging stations. ; Fast charging behavior: Charging and discharging stations quickly charge electric vehicles. ; Slow-release behavior: Charging stations slowly charge electric vehicles. ; Inactive behavior: The charging / discharging station neither charges nor discharges. ; Secondly, determine the state transition of the charging and discharging pile cluster after the decision-making behavior is selected: The charging and discharging pile cluster is located at the first Each state interval , These are the lower and upper limits of the state range, respectively. The state interval at time is ; ， in, This represents the total number of charging and discharging stations. For the first Current charging and discharging power of each charging and discharging station To achieve fast discharge power for charging and discharging piles, For slow discharge power of the charging and discharging pile, The charging power of the charging and discharging pile, , , .

Images