A personalized prediction method for available parking resources

Abstract

The present invention discloses a personalized available parking resource prediction method, which comprises the following steps: S1, establishing a parking behavior probability model for each vehicle; S2, establishing a set of parking space docking states; S3, for a remote reservation or on-site entry application of a specific vehicle, matching and predicting the parking behavior probability model of the vehicle with each parking space state one by one, and calculating the available probability, conflict duration, and usage duration ratio in the case of the vehicle being paired with each parking space; S4, for the target vehicle, predicting and calculating the available probability, conflict duration, and usage duration ratio of each parking space. Based on the original prediction method, the prediction parameters of the present invention are refined, and then a model for predicting the entry and exit time and docking duration of each vehicle is established. By aggregating and calculating the models of each vehicle's behavior, the available parking space resources of the overall parking lot / parking building at the target time period can be predicted more accurately.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a personalized available parking resource prediction method. Background Art

[0002] In existing parking management, generally the overall available parking space quantity at a certain moment or time period is predicted, and vehicle entry applications are responded based on the prediction results. For road cruising vehicles (referring to vehicles that temporarily arrive and apply for docking), if there are remaining cruising parking spaces available at that time, the vehicle is allowed to dock; for reserved vehicles (referring to vehicles that reserve docking in advance and then arrive later), the remaining parking space quantity within the target docking time period of the vehicle is estimated, and if there are still remaining parking spaces, the reservation of this vehicle is accepted.

[0003] The existing technology has the following disadvantages:

[0004] (1) The existing solutions emphasize common features and do not depict the individual parking characteristics of single vehicles, and the prediction granularity is relatively coarse. The existing solutions estimate the available parking space quantity at a certain moment or time period based on the overall vehicle entry and exit situation of the parking lot / building, and decide whether to respond to the vehicle entry application based on this. The prediction granularity of the overall remaining parking space quantity is relatively coarse, and it is impossible to predict the available resources for a single vehicle, often resulting in inaccurate predictions leading to idle parking spaces or occupancy conflicts.

[0005] (2) The existing solutions do not fully consider the uncertainty of parking behaviors. In reality, the entry and exit times of a single vehicle do not strictly conform to the reserved time period, so after allocating a parking space for it according to its reserved time period, occupancy conflicts with other vehicles often occur due to the uncertainty of its own entry and exit.

[0006] The information disclosed in this background art section is only intended to deepen the understanding of the overall background art of the present invention, and should not be regarded as an admission or any form of suggestion that this information constitutes the prior art known to those skilled in the art. Summary of the Invention

[0007] The purpose of the present invention is to provide a personalized available parking resource prediction method to solve the technical problems existing in the prior art.

[0008] To achieve the above purpose, the present invention adopts the following technical solutions:

[0009] The present invention provides a personalized available parking resource prediction method, including the following steps:

[0010] S1. Establish a parking behavior probability model for each vehicle, where the probability model includes an entry time distribution vector and a parking duration distribution matrix;

[0011] S2. Establish a set A of parking space docking states A = {p1, p2,..., p x}, where x is the current total number of parking spaces, and p x = {car1, car2,...}, car1 = {V, L}, V is the entry time distribution vector of car1, and L is the parking duration distribution matrix of car1; that is, there may be multiple existing or reserved vehicles on each parking space, and each vehicle has its entry time distribution vector and parking duration distribution matrix;

[0012] S3. For the remote reservation or on-site entry application of a specific vehicle, match and predict the parking behavior probability model of this vehicle with each parking space state one by one, and calculate the available probability, conflict duration, and usage duration ratio in the case of this vehicle being paired with each parking space;

[0013] S4. For the target vehicle, predict and calculate the available probability, conflict duration, and usage duration ratio of each parking space.

[0014] Preferably, the specific method for establishing the parking behavior probability model of each vehicle in step S1 is as follows:

[0015] For the possible entry time periods, they can be evenly divided into N equal parts, from t1, t2,..., to t n , corresponding to the entry time distribution vector V = [v1, v2,..., v i ,..., v n-1 , v n , which is used to represent the discrete probability distribution of the vehicle entering in this time period, and v i represents the specific probability value of the vehicle entering at the i-th moment;

[0016] The parking duration distribution matrix L is an N * M matrix, where N is the number of N discrete time segments in the above time period, and M represents the M possible parking durations of this vehicle, M = {s1, s2,..., s m}. For example, the first row l1 = [l 1,1 , l 1,2 ,..., l 1,j ,..., l 1,m-1 , l 1,m of the parking duration distribution matrix, where l 1,2 represents the probability value of this vehicle leaving after staying for s2 duration after entering at t1.

[0017] Preferably, the specific calculation method of step S3 is as follows:

[0018] (1) Conduct a random sampling once according to the parking behavior probability models of all occupied vehicles in the parking space docking state p x to obtain a set of specific time intervals for the docking times of the occupied vehicles;

[0019] (2) Perform a random sampling on the parking behavior probability model of the vehicle to be admitted to obtain the specific interval of the parking time of the target vehicle.

[0020] (3) Add the parking time interval of the target vehicle to the set of parking time intervals of the occupied vehicles, check whether there is an overlap between any two time intervals in the set, and calculate the cumulative overlapping duration and the proportion of the parking space usage duration in this case.

[0021] (4) Repeat steps (1)-(3) several times according to the set number of times, and perform an average statistics on whether there is an overlap, the overlapping duration, and the parking space utilization rate obtained several times to obtain the expected overlapping probability (i.e., the available probability), the expected overlapping duration (i.e., the conflict duration), and the proportion of the usage duration under the vehicle-parking space matching.

[0022] Preferably, the result usage method after predicting and calculating the available probability, conflict duration, and proportion of the usage duration of each parking space in step S4 is as follows: (1) Conditional screening: The operator sets thresholds for the three parameters. When the three parameters of a certain parking space do not meet the threshold conditions, the parking space is not allocated to the target vehicle; (2) Optimal decision-making: The operator sets the weights of the three parameters in advance, and the parking space with the highest value obtained by weighted averaging is allocated to the target vehicle.

[0023] Adopting the above technical solution, the present invention has the following beneficial effects:

[0024] 1) Simple implementation: It can be achieved through software and model upgrades without adding additional facilities and equipment on site.

[0025] 2) Strong versatility: A probability model can be established through historical data, which is applicable to all types of parking scenarios, including on-street parking, multi-storey parking buildings, parking lots, etc.

[0026] 3) It realizes the prediction of available parking spaces for individual vehicles, supports the refined decision-making of the operator, can reduce occupancy conflicts, improve the overall utilization rate, and enhance economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0028] Figure 1 It is a flowchart of the personalized available parking resource prediction method provided by the embodiment of the present invention.

[0029] Figure 2 This is the specific calculation flowchart of step S3 provided by the embodiment of the present invention. Detailed implementation manner

[0030] The following further describes in detail the implementation manner of the present invention in conjunction with the drawings and embodiments. The detailed description and drawings of the following embodiments are used to exemplarily illustrate the principle of the present invention, but cannot be used to limit the scope of the present invention, that is, the present invention is not limited to the described preferred embodiments, and the scope of the present invention is defined by the claims.

[0031] During the parking management process, it is necessary to estimate the available parking space resources to decide whether to serve a certain vehicle. Intelligent parking includes two methods: cruising stop and reservation stop. Vehicles for cruising stop do not need to reserve parking spaces and randomly arrive at the parking area and apply for entry; vehicles for reservation stop need to reserve parking spaces in advance from the operator, and the operator predicts the remaining available parking space quantity and probability during the target time period, and decides whether to allocate parking space resources to cruising or reserved vehicles accordingly. In order to predict the remaining available parking space quantity and probability during the target time period for the target vehicle (including vehicles applying for entry during cruising and vehicles reserving parking spaces), the present invention adopts the following method:

[0032] First, establish a probability model of the parking behavior (entry time, exit time, parking duration) of each vehicle based on historical big data; then, based on the behavior of the vehicles that have already or will park at each parking space currently (there may be multiple vehicles), aggregate and calculate the probability model of the available time period of each parking space during the target time period; finally, perform one-by-one matching calculation between the probability model of the target vehicle's parking behavior during the target time period and the probability model of the available time period of each parking space during the target time period to obtain the available probability and available duration of each "vehicle-parking space" pairing situation, which supports the operator to decide whether to provide service for this vehicle. Specifically as follows:

[0033] Combined with Figure 1 shown, this embodiment provides a personalized available parking resource prediction method, including the following steps:

[0034] S101. Establish a probability model of the parking behavior of each vehicle, where the probability model includes an entry time distribution vector and a parking duration distribution matrix;

[0035] S102. Establish a set A of parking space occupancy states A = {p1, p2,..., p x}, where x is the current total number of parking spaces, and p x = {car1, car2,...}, car1 = {V, L}, where V is the entry time distribution vector of car1 and L is the parking duration distribution matrix of car1; that is, there may be multiple existing or reserved vehicles on each parking space, and each vehicle has its own entry time distribution vector and parking duration distribution matrix;

[0036] S103. For a remote reservation or on-site entry application for a specific vehicle, match and predict the parking behavior probability model of this vehicle with each parking space status one by one, and calculate the available probability, conflict duration, and usage duration ratio in the case of pairing this vehicle with each parking space;

[0037] S104. For the target vehicle, predict and calculate the available probability, conflict duration, and usage duration ratio of each parking space.

[0038] Specifically, the specific method for establishing the parking behavior probability model of each vehicle in step S101 is as follows:

[0039] For possible entry time periods, it can be evenly divided into N equal parts, from t1, t2,..., to t n , corresponding to the entry time distribution vector V = [v1, v2,..., v i ,..., v n-1 , v n , which is used to represent the discrete probability distribution of the vehicle entering in this time period, and v i represents the specific probability value of the vehicle entering at the i-th moment;

[0040] The parking duration distribution matrix L is an N*M matrix, where N is the number of N discrete time segments of the above time period, and M represents M possible durations for this vehicle to park, M = {s1, s2,..., s m}. For example, the first row l1 = [l 1,1 , l 1,2 ,..., l 1,j ,..., l 1,m-1 , l 1,m of the parking duration distribution matrix, l 1,2 represents the probability value that this vehicle leaves after staying for s2 duration after entering at t1.

[0041] Specifically, as shown in Figure 2 , the specific calculation method of step S103 is as follows:

[0042] (1) Conduct a random sampling once according to the parking behavior probability models of all occupied vehicles in the occupied state p x of this parking space to obtain a set of specific time intervals for the parking times of the occupied vehicles;

[0043] (2) Conduct a random sampling once on the parking behavior probability model of this vehicle to be entered to obtain the specific time interval for the parking time of the target vehicle;

[0044] (3) Add the parking time interval of the target vehicle to the set of parking time intervals of the occupied vehicles, check whether any two time intervals in the set overlap, calculate the cumulative overlapping duration in this case, and the proportion of the parking space usage duration;

[0045] (4) Repeat steps (1)-(3) a certain number of times according to the set number of times, and perform average statistics on whether there is overlap, the overlapping duration, and the parking space utilization rate obtained several times to obtain the expected overlapping probability (i.e., the available probability), the expected overlapping duration (i.e., the conflict duration), and the proportion of the usage duration under the vehicle-parking space matching.

[0046] Specifically, the result usage method after predicting and calculating the available probability, conflict duration, and proportion of the usage duration of each parking space in step S104 is as follows: (1) Condition screening: The operator sets thresholds for the three parameters. When the three parameters of a certain parking space do not meet the threshold conditions, the parking space is not allocated to the target vehicle; (2) Optimal decision-making: The operator sets the weights of the three parameters in advance, and the parking space with the highest numerical value obtained after weighted averaging is allocated to the target vehicle.

[0047] To better understand the present invention, the present invention will be described in detail below in conjunction with specific embodiments:

[0048] 1. Divide the time period from 8:00 to 9:00 into 6 time segments, each segment being 10 minutes; for a target vehicle to enter the venue, its entry time distribution vector is V = [0, 0.1, 0.4, 0.3, 0.2, 0]. For example, the probability of entering between 8:20 and 8:30 is 0.4; its parking duration distribution matrix is:

[0049]

[0050] 2. There are currently two parking spaces 1 and 2, where:

[0051] Vehicle A is parked on parking space 1, and vehicle B has reserved parking space 1;

[0052] Vehicle C is parked on parking space 2, and vehicles D and E have reserved parking space 2;

[0053] Vehicles A, B, C, D, and E all have their own entry time distribution vectors and parking duration distribution matrices.

[0054] 3. Match and predict the target vehicle with parking space 1:

[0055] First sampling: According to the random sampling of the probability model contained in the entry vector and the parking duration distribution matrix, generate the parking interval of the target vehicle [8:25, 9:12], generate the A parking interval [7:30 - 8:30], and generate the B parking interval [9:20 - 11:00]. There is an overlap; the overlap duration is 5 minutes; the usage duration ratio between 7 - 10 o'clock is (180 - 38) / 180 = 78.9%.

[0056] Second sampling: Generate the parking interval of the target vehicle [8:17, 9:20], generate the A parking interval [7:20 - 8:13], and generate the B parking interval [9:10 - 11:10]. There is an overlap; the overlap duration is 10 minutes; the usage duration ratio between 7 - 10 o'clock is (180 - 20 - 4) / 180 = 86.7%.

[0057] Third sampling:...

[0058] After averaging the results of several samplings, the non - overlapping incidence rate (available probability) is 87%, the conflict duration is 8.7 minutes, and the usage duration ratio is 85.5%.

[0059] 4. Match and predict the target vehicle with parking space 2:

[0060] The principle is the same as above. After averaging the results of several samplings, the non - overlapping incidence rate (available probability) is 67%, the conflict duration is 25.0 minutes, and the usage duration occupancy is 95.5%.

[0061] Based on the prediction results, decide whether to allocate a parking space for the target vehicle and which parking space to allocate. Since the operator plans to control the available probability above 80% and the conflict duration within 10 minutes, the target vehicle and parking space 2 do not meet the threshold conditions and are not considered; allocate parking space 1 for the target vehicle.

[0062] In summary, based on the original prediction method, the technical solution of this patent application refines the prediction parameters, turns to predict the model of the entry and exit time and parking duration of each vehicle, and then aggregates and calculates the model of each vehicle's behavior to more accurately predict the available parking space resources of the overall parking lot / parking building during the target period; considering the micro - uncertainty of parking behavior, uses a probability model to depict the single - vehicle behavior model, and specifically depicts the probability model of single - vehicle behavior through historical big data, further improving the accuracy of predicting the number of available parking spaces.

[0063] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A personalized prediction method for available parking resources, characterized in that, It includes the following steps: S1. Establish a parking behavior probability model for each vehicle. The probability model includes an arrival time distribution vector and a parking duration distribution matrix; S2. Establish a set A of parking space docking states A = {p1, p2,..., p x}, where x is the current total number of parking spaces, and p x = {car1, car2,...}, car1 = {V, L}, V is the arrival time distribution vector of car1, and L is the parking duration distribution matrix of car1; that is, there may be multiple existing or reserved vehicles on each parking space, and each vehicle has its arrival time distribution vector and parking duration distribution matrix; S3. For a remote reservation or on-site entry application of a specific vehicle, match and predict the parking behavior probability model of this vehicle with each parking space status one by one, and calculate the available probability, conflict duration, and usage duration ratio in the case of this vehicle being paired with each parking space; S4. For the target vehicle, predict and calculate the available probability, conflict duration, and usage duration ratio of each parking space.

2. The personalized available parking resource prediction method according to claim 1, wherein The specific method for establishing the parking behavior probability model for each vehicle in step S1 is as follows: For a possible entry time period, it can be evenly divided into N equal parts, from t1, t2,..., to t n , corresponding to the entry time distribution vector V = [v1, v2,..., v i ,..., v n-1 , v n , which is used to represent the discrete probability distribution of the vehicle's entry during this time period. v i represents the specific probability value of the vehicle entering at the i-th moment; The parking duration distribution matrix L is an N*M matrix, where N is the number of N discrete time segments in the above time period, and M represents M possible parking durations for the vehicle, M = {s1, s2,..., s m}.

3. The personalized available parking resource prediction method according to claim 1, wherein The specific calculation method of step S3 is as follows: (1) Perform a random sampling on the probability models of the parking behaviors of all occupied vehicles according to the parking space occupancy status p x to obtain a set of specific intervals of the parking times of the occupied vehicles; (2) Conduct a random sampling on the parking behavior probability model of the vehicle to be entered to obtain the specific interval of the parking time of the target vehicle; (3) Add the parking time interval of the target vehicle to the set of parking time intervals of the occupied vehicles, check whether there is an overlap between any two time intervals in the set, and calculate the cumulative overlapping duration and the usage duration ratio of this parking space in this case; (4) Repeat steps (1)-(3) a certain number of times according to the set number of times, and conduct an average statistics on whether there is an overlap, the overlapping duration, and the parking space utilization rate obtained several times to obtain the available probability, conflict duration, and usage duration ratio under the vehicle-parking space matching.

4. The personalized available parking resource prediction method according to claim 1, wherein The result usage method after predicting and calculating the available probability, conflict duration, and usage duration ratio of each parking space in step S4 is as follows: (1) Conditional screening: The operator sets thresholds for the three parameters. When the three parameters of a certain parking space do not meet the threshold conditions, the parking space is not allocated to the target vehicle; (2) Optimal decision-making: The operator sets the weights of the three parameters in advance, and the parking space with the highest value obtained after weighted averaging is allocated to the target vehicle.

Images