Electric vehicle charging demand prediction method based on traffic balance

Abstract

The invention discloses an electric vehicle charging demand prediction method based on traffic balance, and the method comprises the steps: firstly building an electric vehicle and a road network model in two dimensions of time and space, and constructing an effective path set for the travel of the electric vehicle through employing an A * algorithm; the finite rationality of a user is considered, time-of-use electricity price is introduced to formulate the electricity price of a charging station, and a semi-dynamic traffic balance model based on a random utility theory is established in combination with traffic flow constraints to solve traffic flow distribution. And then, based on a probability numerical calculation method of a combined state of charge (CSOC), the charging load of the electric vehicle is solved, the charging demand of the electric vehicle is predicted, and the influence of the travel characteristics of the electric vehicle on the traffic capacity condition of a traffic network is comprehensively considered. And the defects of the driving path selection of the vehicle individuals and the path selection under the interaction among the vehicle groups are made up.

Description

Technical Field

[0001] The invention relates to the technical fields of power systems and electric vehicles, and in particular to a method for predicting charging demand of electric vehicles based on traffic balance. Background Art

[0002] With the surge in the number of electric vehicles, their charging demand poses a severe challenge to the transportation network and the power grid's acceptance capacity. As a mobile power load, the charging behavior of electric vehicles is deeply affected by the owner's personal decision-making, so it shows a high degree of randomness in time and space. This randomness not only increases the complexity of grid scheduling, but may also have an adverse impact on traffic flow. Therefore, accurately describing the probabilistic characteristics of electric vehicle charging load has become an important basis for evaluating the impact of electric vehicles on the power grid and transportation network, and optimizing the layout and operation strategy of charging stations.

[0003] At present, domestic and foreign scholars have conducted extensive and in-depth research on electric vehicle load forecasting, mainly from the two dimensions of research methods and solution strategies. In terms of research methods, most scholars use the travel chain travel mode from a micro perspective to model single electric vehicles. In terms of solution methods, most existing studies use the Monte Carlo method to simulate electric vehicle load forecasting as a common method, but this method is an extractive method and has the problem of slow calculation speed.

[0004] The travel characteristics of electric vehicles will affect the charging characteristics of electric vehicles and the flow of the traffic network, so the reasonable and accurate description of the travel behavior of electric vehicles is the key to the load forecasting of electric vehicles. The above research considers the individual travel and charging behavior of electric vehicles from the perspective of micro-models. The choice of travel routes mostly considers the shortest path, and pays less attention to the traffic capacity of the traffic network. The congestion caused by excessive traffic flow is the result of the collective interaction of single vehicles. Some existing studies only consider the driving route selection of individual vehicles and do not consider the route selection under the interaction between vehicle groups. When considering group travel, the group vehicle route selection chooses the shortest path travel problem, and there is a problem of uneven traffic distribution on the road section.

[0005] Therefore, the prior art still needs to be improved and enhanced. Summary of the invention

[0006] In view of the above-mentioned deficiencies in the prior art, the purpose of the present invention is to provide a method for predicting the charging demand of electric vehicles based on traffic balance, aiming to solve the shortcomings of the prior art that the travel characteristics of electric vehicles pay less attention to the traffic capacity of the traffic network, only consider the driving path selection of individual vehicles but do not consider the path selection under the interaction between vehicle groups.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] A method for predicting electric vehicle charging demand based on traffic balance, comprising:

[0009] Step 1: Establish electric vehicle and road network models in both time and space dimensions, and use the A* algorithm to construct a set of effective paths for electric vehicle travel;

[0010] Step 2: Considering the limited rationality of users, time-of-use electricity prices are introduced to formulate charging station electricity prices. Combined with traffic flow constraints, a semi-dynamic traffic equilibrium model based on random utility theory is established and used to solve the distribution of traffic flow.

[0011] Step 3: Based on the probabilistic numerical calculation method of the combined state of charge CSOC, the electric vehicle charging load is solved to realize the prediction of electric vehicle charging demand.

[0012] Furthermore, the step 1 includes:

[0013] The continuous average algorithm is used for iteration. Through the path probability of vehicle travel, the traffic network reaches an equilibrium state, and the new semi-dynamic traffic model is solved through repeated iterations.

[0014] Furthermore, the establishment of electric vehicle and road network models in two dimensions of time and space includes:

[0015] When electric vehicles are driving on a road network, before they can travel on the path, they must first abstract the actual road network and establish a corresponding road network model.

[0016] The elements in the road network are described by edges and nodes to form a topological graph of the road network. Edges represent roads and nodes represent points connecting roads. Therefore, the road network can be described by G = (N, A), where N represents the node set and A represents the edge set.

[0017] The congestion effect of roads can be expressed as a function of time and traffic flow. The relationship function between road travel time and traffic flow is shown as follows:

[0018]

[0019] Where: x a,t is the road traffic volume; is the zero flow time, indicating the travel time when the road traffic volume is 0; c a is the road capacity.

[0020] Furthermore, the use of the A* algorithm to construct a set of effective paths for electric vehicle travel includes:

[0021] When choosing a trip on a road network, electric vehicles will choose a path with the shortest travel time, so the shortest path is generally used as the driving path for electric vehicles;

[0022] The A* algorithm is used to construct an effective path set for electric vehicle travel OD pairs. The A* algorithm is a heuristic path search algorithm that selects the shortest path from the starting point to the end point by designing a heuristic evaluation function; the function expression is shown in the following formula:

[0023] f(u)=g(u)+h(u)

[0024] Where: g(u) is the cost of OD from the starting node to any end point u; h(u) is the estimated consumption cost from end point u to the target end point; when h(u) = 0, the feasible path set is quickly constructed, which is more in line with the actual travel path selection situation.

[0025] Furthermore, the step 2 includes:

[0026] The user optimal allocation principle means that the path travel cost is the same and is smaller than any other path cost, achieving the optimal equilibrium state for users; the time scale of the semi-dynamic traffic model is 15-90 minutes, which can approximate multiple static traffic problems and is less difficult to solve than the dynamic model. The semi-dynamic simulation is as follows:

[0027]

[0028] Where: N od is the set of all paths in (o,d); is the travel flow of the nth path (o, d) in time period t; is the travel demand at (o, d) in time period t;

[0029]

[0030] Where: The flow rate of each path in each period x a,t The relationship between the flow of section a and path n can be expressed by the association matrix express. Indicates on path n;

[0031]

[0032] Where: is the travel time on path n in period t;

[0033] Considering that not all vehicles depart at the same time, there is residual traffic. The residual traffic of the previous period needs to be taken into account in the next period and corrected within the period to establish a semi-dynamic traffic model with multiple period coupling relationships:

[0034]

[0035] Where: is the remaining flow, τ is the time length, is the modified time requirement of (o,d) in time period t.

[0036] Furthermore, the step three includes:

[0037] The continuous average algorithm MSA is used for iteration. When the probability of electric vehicles choosing a travel path is The traffic network reaches a balanced state, and the new semi-dynamic traffic model is solved by repeated iterations. The convergence condition is shown in the formula:

[0038]

[0039] Where: ε is the preset error, k is the number of iterations, and when the equilibrium state is reached, no individual can change the travel decision; the specific solution steps are as follows:

[0040] Initialize the road network impedance and set the initial parameters;

[0041] Divide a day into 24 time periods, set the number of iterations k and time period t;

[0042] Entering the tth period, starting k iterations, using the effective path method to obtain the feasible path set of each OD pair, and updating the impedance;

[0043] Solve the traffic residual flow phenomenon and obtain the corrected

[0044] Determine whether t is in the peak electricity price period, perform flow analysis on the adjacent paths of the charging station, and determine whether the flow constraint conditions are met. If not, reallocate the flow of the preceding path;

[0045] Calculating the probability of path selection based on the random utility model Correct the travel demand, update the path flow of the current period, and iterate. The formula is as follows;

[0046]

[0047] Determine whether the convergence condition is met. If not, return to step 3;

[0048] Determine whether all moments have been solved. If not, return to step 3 at t=t+1. If yes, output the path travel time and path flow.

[0049] The technical solution adopted by the present invention has the following beneficial effects:

[0050] In this application, the travel flow of electric vehicles is solved through macro traffic flow distribution. The traditional method only considers the driving path selection of individual vehicles and does not consider the path selection under the interaction between vehicle groups. This application considers the path selection of group vehicles when traveling in groups and chooses the shortest path to reduce the problem of uneven traffic distribution on road sections; avoids congestion caused by excessive traffic flow;

[0051] The travel and charging characteristics of electric vehicles and the road network are modeled separately, and an effective path set based on the A* algorithm is constructed. Then, the random utility theory is introduced to constrain the travel path selection of car owners to achieve more accurate modeling;

[0052] A charging price for public charging stations combined with the traffic flow of road network nodes is formulated through time-of-use electricity prices. A semi-dynamic traffic equilibrium model is established in combination with random utility theory to solve the distribution of traffic flow. The charging state of electric vehicles in each OD pair of each time period is described by the numerical calculation method of the combined state of charge (CSOC) charging load probability of electric vehicles. The final charging load probability prediction result is obtained to realize the prediction of electric vehicle charging demand, comprehensively consider the impact of the travel characteristics of electric vehicles on the traffic capacity of the traffic network, and make up for the shortcomings of the driving path selection of individual vehicles and the path selection under the interaction between vehicle groups. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 A schematic diagram of a framework of a method for predicting electric vehicle charging demand based on traffic balance provided by the present invention;

[0054] Figure 2 A schematic diagram of a method for generating an effective path for a method for predicting electric vehicle charging demand based on traffic balance provided by the present invention;

[0055] Figure 3 A CSOC charging load calculation model diagram of a method for predicting electric vehicle charging demand based on traffic balance provided by the present invention. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solution and effect of the present invention clearer and more specific, the present invention is further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0057] The present invention proposes a method for predicting electric vehicle charging load based on traffic balance. Figure 1As shown in the figure, firstly, electric vehicles and road networks are modeled in the two dimensions of time and space, and the A* algorithm is used to construct an effective path set for electric vehicle travel OD pairs; taking into account the limited rationality of users, a charging station electricity price combined with traffic flow constraints is proposed, and a semi-dynamic traffic equilibrium model is established by combining random utility theory with time-of-use electricity prices to solve the distribution of traffic flow; then the electric vehicle charging load is solved based on the probabilistic numerical calculation method of the combined state of charge (CSOC).

[0058] When electric vehicles are driving on a road network, before they can travel on a path, they must first abstract the actual road network and establish a corresponding road network model. This paper uses the concept of "graph" in graph theory in computer data structures to represent the road network, describing the elements in the road network with edges and nodes, and constructing a topological graph of the road network. Edges represent roads, and nodes represent points connecting roads. Therefore, the road network can be described by G = (N, A), where N represents the node set and A represents the edge set.

[0059] The congestion effect of roads can be expressed as a function of time and traffic volume. According to research, the relationship between road travel time and traffic volume is as follows:

[0060]

[0061] Where: x a,t is the road traffic volume; is the zero flow time, indicating the travel time when the road traffic volume is 0; c a is the road capacity.

[0062] like Figure 2 As shown in the figure, when choosing a travel path on a road network, electric vehicles will choose a path with the shortest travel time, so the shortest path is generally used for electric vehicle travel. However, in the equilibrium state, there can be many shortest travel paths for an OD pair. The enumeration method is a path search method, but this method will include many invalid paths into the feasible path set, which does not conform to the actual user's travel choice. Therefore, the A* algorithm is used to construct an effective path set for electric vehicle travel OD pairs. The A* algorithm is a heuristic path search algorithm that selects the shortest path from the starting point to the end point by designing a heuristic evaluation function. The function expression is shown in the following formula:

[0063] f(u)=g(u)+h(u)

[0064] Where: g(u) is the cost of OD from the starting node to any end point u; h(u) is the estimated consumption cost from end point u to the target end point. Let h(u) = 0 to quickly build a feasible path set, which is more in line with the actual travel path selection situation.

[0065] Traffic flow allocation is a method of describing the traffic network. It expresses the macroscopic allocation of traffic flow in a known traffic network under fixed travel demand and finds the flow of each road. The user optimal allocation principle means that the path travel cost is the same and is smaller than any other path cost, achieving the user's optimal equilibrium state. The time scale of the semi-dynamic traffic model is (15-90min), which can approximate multiple static traffic problems and is less difficult to solve than the dynamic model. However, the semi-dynamic model needs to consider a problem that in a certain time period, some vehicles cannot reach the destination in this time period, and reach the destination in the next time period, resulting in residual flow. Generally, semi-dynamic simulation can solve this kind of problem.

[0066]

[0067] Where: N od is the set of all paths in (o,d); is the travel flow of the nth path (o, d) in time period t; is the travel demand at (o, d) in time period t.

[0068]

[0069] Where: The flow rate of each path in each period x a,t The relationship between the flow of section a and path n can be expressed by the association matrix express. Indicates on path n.

[0070]

[0071] Where: is the travel time on path n in period t.

[0072] Taking into account that not all vehicles depart at the same time, there is residual flow. The residual flow of the previous period needs to be taken into account in the next period and corrected within the period to establish a semi-dynamic traffic model containing multiple period coupling relationships.

[0073]

[0074]

[0075] Where: is the remaining flow, τ is the time length, is the modified time requirement of (o,d) in time period t.

[0076] The continuous averaging algorithm (MSA) is used for iteration. When the probability of an electric vehicle choosing a travel path is The traffic network reaches a balanced state, and the new semi-dynamic traffic model is solved by repeated iterations. The convergence condition is shown in the formula:

[0077]

[0078] Where: ε is the preset error, k is the number of iterations. When the equilibrium state is reached, no individual can change the travel decision. The specific solution steps are as follows:

[0079] (1) Initialize the road network impedance and set the initial parameters.

[0080] (2) Divide a day into 24 time periods and set the number of iteration steps k and time period t.

[0081] (3) Entering the tth time period, starting k iterations, using the effective path method to obtain the feasible path set of each OD pair, and updating the impedance.

[0082] (4) Solve the traffic residual flow phenomenon and obtain the corrected

[0083] (5) Determine whether time t is in the peak electricity price period, perform flow analysis on the adjacent paths of the charging station, and determine whether the flow constraint conditions are met. If not, reallocate the flow on the preceding path.

[0084] (6) Calculate the path selection probability based on the random utility model Correct the travel demand, update the path flow of the current period, and iterate. The formula is as follows.

[0085]

[0086] (7) Determine whether the convergence condition is met. If not, return to step 3.

[0087] (8) Determine whether all time points have been solved. If not, return to step 3 at t = t + 1. If yes, output the path travel time and path flow.

[0088] The CSOC calculation method is used to describe the charge state changes of the electric vehicle group. When the electric vehicle group reaches the destination or the charging station before the destination, if the SOC is lower than the minimum threshold, charging is carried out. Charging is determined by the number of electric vehicles being charged, the charging probability, the charging power, and the charging efficiency. The charging time is affected by the CSOC of the starting period. By superimposing the charging demand of the OD pairs in all time periods, the daily charging load curve of the charging stations in the area can be obtained, and the impact of a series of OD pairs on each charging station can be analyzed separately. The solution process is as follows: Figure 3 shown.

[0089] Figure 2A method for generating a valid path is described, the specific steps are as follows:

[0090] Step 1: Initialize the impedance of the road network.

[0091] Step 2: Enter the start-end (OD) pairs for a time period and start numbering from 1.

[0092] Step 3: Select the current OD pair for processing.

[0093] Step 4: Use the A* algorithm to generate the first shortest path for the current OD pair.

[0094] Step 5: Add the generated shortest path to the feasible path set of the current OD pair.

[0095] Step 6: Consider the travel demand during the current period and combine all feasible paths together.

[0096] Step 7: Calculate the traffic volume of each road segment.

[0097] Step 8: Update the impedance of the road network based on the calculated traffic flow.

[0098] Step 9: Use the A* algorithm again to generate the next shortest path for the current OD pair.

[0099] Step 10: Check whether the newly generated path already exists in the path set. If yes (Y), jump to step 13; if no (N), continue to step 12.

[0100] Step 11: Process the next OD pair and increase the number by 1.

[0101] Step 12: Check whether all OD pairs have been processed. If yes (Y), execute step 14. If not (N), return to step 4 to continue processing the next OD pair.

[0102] Step 13: Generate the final set of feasible paths and the process ends.

[0103] Figure 2 and Figure 3 An iterative process is described in which a set of feasible paths is finally obtained by continuously updating the impedance of the road network and generating new shortest paths. This process may be used in traffic planning or optimization to ensure that traffic flow can be effectively managed within a specific period of time.

[0104] After the vehicle travels from the starting point to the end point, the charging station is established at the end point node and can only be charged at the end point. The analysis is not considered in terms of the time-sharing electricity price of charging. As time changes, the spatial distribution of traffic on each section of the road presents a semi-dynamic traffic equilibrium state. Users have irrational driving situations. The spatial distribution of traffic on some sections of the road is quite different. Some sections of the road have almost no traffic. The traffic on sections with the same driving direction and shorter distances is greater than that on sections with longer paths. Car owners generally choose to travel on shorter paths. When the traffic volume exceeds the road's capacity and the shortest path cannot meet the optimal allocation, car owners will choose other longer paths to travel, and the road network reaches a new equilibrium state.

[0105] Most electric vehicles can reach their destination within a certain period of time, and the remaining travel demand after correction is very small compared to that before correction. The higher the travel demand, the greater the corresponding power consumption and the higher the charging demand. After the electric vehicle completes this journey and goes to the charging station to charge, the peak value of the daily charging load will be reached after the peak period of electric vehicle travel, which reflects a positive proportional relationship between the charging load and the travel demand. The higher the travel demand of electric vehicles, the more electric vehicles will reach the corresponding charging threshold as the electric vehicle's own power is consumed.

[0106] In summary, the CSOC method used in this patent solves the charging load. The Monte Carlo method is compared with the method of this patent. The charging load results obtained by multiple Monte Carlo are roughly the same as those of the method of this patent, indicating that the prediction method of the electric vehicle cluster in the macro state has the effect and advantages that individual sampling cannot achieve, and Monte Carlo is repeated sampling, which is relatively time-consuming in calculation. The probability calculation method based on CSOC (overall state of charge) used in this patent has significant advantages in the numerical calculation of the probability distribution of electric vehicle charging load. First, this method effectively eliminates the variable coupling error in the modeling of multiple trips by disassembling the multiple trips of the vehicle into independent single trips for start-end point (OD) analysis. Secondly, it establishes a CSOC dynamic probability model considering the travel characteristics of the vehicle, which can determine the probability density function of the starting state of charge of each trip. Combined with the law of large numbers, the spatiotemporal probability distribution function of the charging load can be obtained. This method greatly improves the calculation efficiency while ensuring the calculation accuracy. Compared with the traditional Monte Carlo simulation method, it eliminates the coupling error problem and significantly improves the calculation efficiency. In addition, the CSOC probability model can also analyze the CSOC of the vehicle at each road network node to determine the probability density function of the state of charge (SOC) at the beginning of a single trip of the vehicle, and build a probability density model of the electric vehicle charging load. These advantages make the probability calculation method based on CSOC have high practical value and efficiency when dealing with the probability distribution problem of electric vehicle charging load.

[0107] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the schemes disclosed herein. The present invention is intended to cover any variations, uses or adaptations of the present invention that follow the general principles of the present invention and include common knowledge or customary techniques in the art that are not disclosed in this disclosure. The description and examples are to be considered exemplary only, and the true scope and spirit of the present invention are indicated by the claims.

Claims

1. A method for predicting electric vehicle charging demand based on traffic balance, characterized in that: include: Step 1: Establish electric vehicle and road network models in both time and space dimensions, and use the A* algorithm to construct a set of effective paths for electric vehicle travel; Step 2: Considering the limited rationality of users, time-of-use electricity prices are introduced to formulate charging station electricity prices. Combined with traffic flow constraints, a semi-dynamic traffic equilibrium model based on random utility theory is established and used to solve the distribution of traffic flow. Step 3: Based on the probabilistic numerical calculation method of the combined state of charge CSOC, the electric vehicle charging load is solved to realize the prediction of electric vehicle charging demand.

2. The method for predicting electric vehicle charging demand based on traffic balance according to claim 1 is characterized in that: The step one includes: The continuous average algorithm is used for iteration. Through the path probability of vehicle travel, the traffic network reaches an equilibrium state, and the new semi-dynamic traffic model is solved through repeated iterations.

3. The method for predicting electric vehicle charging demand based on traffic balance according to claim 1 is characterized in that: The electric vehicle and road network model is established in two dimensions of time and space, including: When electric vehicles are driving on a road network, before they can travel on the path, they must first abstract the actual road network and establish a corresponding road network model. The elements in the road network are described by edges and nodes to form a topological graph of the road network. Edges represent roads and nodes represent points connecting roads. Therefore, the road network can be described by G = (N, A), where N represents the node set and A represents the edge set. The congestion effect of roads can be expressed as a function of time and traffic flow. The relationship function between road travel time and traffic flow is shown as follows: Where: x a,t is the road traffic volume; is the zero flow time, indicating the travel time when the road traffic volume is 0; c a is the road capacity.

4. The method for predicting electric vehicle charging demand based on traffic balance according to claim 1, characterized in that: The effective path set for electric vehicle travel constructed by using the A* algorithm includes: When choosing a trip on a road network, electric vehicles will choose a path with the shortest travel time, so the shortest path is generally used as the driving path for electric vehicles; The A* algorithm is used to construct an effective path set for electric vehicle travel OD pairs. The A* algorithm is a heuristic path search algorithm that selects the shortest path from the starting point to the end point by designing a heuristic evaluation function; the function expression is shown in the following formula: f(u)=g(u)+h(u) Where: g(u) is the cost of OD from the starting node to any end point u; h(u) is the estimated consumption cost from end point u to the target end point; when h(u) = 0, the feasible path set is quickly constructed, which is more in line with the actual travel path selection situation.

5. The method for predicting electric vehicle charging demand based on traffic balance according to claim 1, characterized in that: The step 2 includes: The user optimal allocation principle means that the path travel cost is the same and is smaller than any other path cost, achieving the optimal equilibrium state for users; the time scale of the semi-dynamic traffic model is 15-90 minutes, which can approximate multiple static traffic problems and is less difficult to solve than the dynamic model. The semi-dynamic simulation is as follows: Where: N od is the set of all paths in (o,d); is the travel flow of the nth path (o, d) in time period t; is the travel demand at (o, d) in time period t; Where: The flow rate of each path in each period x a,t The relationship between the flow of section a and path n can be expressed by the association matrix express; Indicates on path n; Where: is the travel time on path n in period t; Considering that not all vehicles depart at the same time, there is residual traffic. The residual traffic of the previous period needs to be taken into account in the next period and corrected within the period to establish a semi-dynamic traffic model with multiple period coupling relationships: Where: is the remaining flow, τ is the time length, is the modified time requirement of (o,d) in time period t.

6. The method for predicting electric vehicle charging demand based on traffic balance according to claim 1, characterized in that: The step three includes: The continuous average algorithm MSA is used for iteration. When the probability of electric vehicles choosing a travel path is The traffic network reaches a balanced state, and the new semi-dynamic traffic model is solved by repeated iterations. The convergence condition is shown in the formula: Where: ε is the preset error, k is the number of iterations, and when the equilibrium state is reached, no individual can change the travel decision; the specific solution steps are as follows: Initialize the road network impedance and set the initial parameters; Divide a day into 24 time periods, set the number of iterations k and time period t; Entering the tth period, starting k iterations, using the effective path method to obtain the feasible path set of each OD pair, and updating the impedance; Solve the traffic residual flow phenomenon and obtain the corrected Determine whether t is in the peak electricity price period, perform flow analysis on the adjacent paths of the charging station, and determine whether the flow constraint conditions are met. If not, reallocate the flow of the preceding path; Calculating the probability of path selection based on the random utility model Correct the travel demand, update the path flow of the current period, and iterate. The formula is as follows; Determine whether the convergence condition is met. If not, return to step 3; Determine whether all moments have been solved. If not, return to step 3 at t=t+1. If yes, output the path travel time and path flow.

Images