Automatic high-density parking lot intelligent scheduling method and system

Abstract

The invention discloses an automatic high-density parking lot intelligent scheduling method and system. Comprising the following steps: S1, collecting real-time state information of a parking lot and correcting data based on a neural network model; s2, according to the obtained real-time state information of the parking lot, calculating a matching degree coefficient of the vehicle and the parking space; and S3, scheduling the vehicles through an automatic valet parking system according to the calculated matching degree coefficient of the vehicles. According to the invention, repeated operation possibly generated in an automatic high-density parking lot scheduling system is reduced, the scheduling efficiency of the automatic high-density parking lot is improved, and the scheduling cost and time are saved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent parking scheduling, and in particular to an automated high-density parking lot intelligent scheduling method and system. Background Art

[0002] As cities expand and their populations increase, the contradiction between limited urban space and the rapid growth of motor vehicle ownership gradually emerges. Major first-tier and second-tier cities are facing the common problem of "difficult parking". Increasing parking facilities by planning more parking demand land requires a large amount of land resources and government investment, which is difficult to achieve sustainable development.

[0003] High-density parking lot design provides a possibility to improve the land utilization rate of parking lots and alleviate the problem of "difficult parking". In traditional parking lots for manually driven cars, in order to ensure the accessibility of each car, parking spaces need to be arranged next to the lane. Therefore, about 1 / 3 of the space in the parking lot cannot be used as parking spaces. High-density parking lots do not need to ensure that vehicles are parked in the adjacent lanes, because the central control system and automatic valet parking technology can move the obstacle vehicles between the target vehicle and the lane so that the target vehicle can reach the destination.

[0004] Compared with traditional parking lots, a high-density parking lot with one end closed and one end for entry and exit can increase space utilization by about 30%, making full use of the site space. It can be arranged alone in an outdoor venue, or arranged in an underground garage of an existing building, and renovating the interior of an existing parking lot. Increasing parking density also requires a more reasonable parking scheduling method to store and retrieve vehicles. If the scheduling method of a high-density parking lot is not good, it may lead to frequent vehicle movement, long customer waiting time, and excessive energy consumption. Therefore, it is of great significance to design an intelligent scheduling method for high-density parking lots with high space utilization, low energy consumption, and short vehicle retrieval time. Summary of the invention

[0005] The purpose of the present invention is to provide an automated high-density parking lot intelligent scheduling method and system to address the deficiencies in the prior art, which can effectively improve the efficiency of parking and retrieval and the level of parking services.

[0006] To achieve the purpose of the present invention, the technical solution adopted by the present invention is as follows:

[0007] The method provided by the present invention comprises the following steps:

[0008] S1. Parking lot real-time status information collection and data correction based on neural network model;

[0009] S2. Calculate the matching coefficient between the vehicle and the parking space according to the corrected real-time parking status information;

[0010] S3. Dispatch the vehicles according to the calculated matching coefficients of the vehicles.

[0011] The system provided by the present invention includes a real-time state sensing unit, a data correction unit, a matching degree calculation unit and an automatic valet parking unit;

[0012] The real-time status sensing unit is used to collect vehicle information and parking status, update parking lot real-time status information, and obtain vehicle entry / exit requirements;

[0013] The data correction unit includes a pre-trained neural network model for obtaining a more accurate vehicle exit moment;

[0014] The matching degree calculation unit calculates the matching degree coefficient between the vehicle and the parking space according to the real-time parking lot status information after data correction, which is used in the scheduling process;

[0015] The automatic valet parking unit utilizes the automatic valet parking technology to dispatch the vehicle according to the calculated matching coefficient.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention senses the real-time status information of the parking lot and obtains the vehicle entry / exit requirements, calculates the matching coefficient between the vehicle and the parking space according to the real-time status information after data correction, allocates the parking space to the vehicle according to the size of the matching coefficient, and realizes vehicle dispatch by using automatic valet parking technology. The method reduces the repetitive operations that may be generated in the automated high-density parking dispatch system, improves the dispatch efficiency of the automated high-density parking lot, and saves dispatch costs and time. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of the method flow provided in an embodiment of the present application.

[0018] Figure 2 Schematic diagram of automated high-density parking lot.

[0019] Figure 3 For estimated value and correction value With the true value Cumulative distribution plot of relative error.

[0020] Figure 4 It is a fitting curve graph for the historical parking time data of vehicles entering and leaving the parking lot.

[0021] Figure 5 To enter the vehicle dispatch flow chart.

[0022] Figure 6 This is the flow chart for exit vehicle dispatch. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] like Figure 1 As shown, this embodiment provides an automated high-density parking lot intelligent scheduling method, comprising the following steps:

[0025] S1. Parking lot real-time status information collection and data correction based on neural network model;

[0026] S2. Calculate the matching coefficient between the vehicle and the parking space according to the corrected real-time parking status information;

[0027] S3. Dispatching the vehicle through the automatic valet parking system according to the calculated matching coefficient of the vehicle.

[0028] It should be further explained that, in the specific implementation process, the automated high-density parking lot stores vehicles like cargo in N parallel stacking lanes with a capacity of M, such as Figure 2 As shown, the automatic valet parking system is used to enable the vehicle to enter, leave the parking lot, and move within the parking lot.

[0029] Furthermore, the real-time parking lot status information is updated in one-second time steps, including the parking space occupancy status, parked vehicle information, and vehicle entry / exit requirements. The parking space occupancy status is recorded as 0 (occupied) or 1 (unoccupied, i.e., available); the parked vehicle information includes the vehicle's arrival time and exit time; the vehicle's exit time is an estimate provided by the user when arriving at the parking lot. After data correction, Calculate the matching coefficient.

[0030] In some embodiments, in step S1, sensing the real-time status information of the parking lot and performing data correction specifically includes the following:

[0031] Based on the acquired parking space occupancy status, parked vehicle information, and vehicle entry / exit requirements, a parking lot real-time status information table is generated, wherein the first column of the parking space real-time status information table is the parking space number, the second column is the parking space occupancy status, the third column is the arrival time of the vehicle currently parked in the parking space, and the fourth column is the exit time of the vehicle currently parked in the parking space.

[0032] The automatic identification system at the parking lot entrance obtains vehicle entry requirements and collects the estimated vehicle exit time provided by the user Data correction is performed based on the pre-trained neural network model to obtain the corrected exit time

[0033] The system's dispatching status is determined based on the acquired real-time parking lot status information. When there is a demand for a vehicle to enter and the parking lot is not full, and when there is a demand for a vehicle to exit, the subsequent dispatching steps are entered.

[0034] Furthermore, the training process of the neural network model specifically includes the following:

[0035] Collect parking lot historical data, including user ID, date, vehicle arrival time, and user-estimated vehicle exit time and the actual vehicle exit time

[0036] In this embodiment, the historical data collected is the vehicle entry and exit data of a parking lot for two weeks, with a total of 1238 records. After removing the unclear and cross-day data, there are a total of 973 samples; the date is expressed as a period, with Monday to Sunday counted as 1 to 7; the user provides an estimated vehicle exit time interval, and the average of the upper and lower bounds of the estimated interval is taken as the user's estimated vehicle exit time Input value. When the estimated parking time is too short (<1 hour) or too long (>12 hours), the sum of the lower limit (1 hour) or upper limit (12 hours) of the parking time and the vehicle arrival time is taken as the user's estimated vehicle exit time. Enter a value.

[0037] Construct and train a BP neural network model. The BP neural network model structure includes an input layer, a hidden layer and an output layer. The input layer determines the number of input nodes according to the number of features; the output layer is used to output the corrected exit time.

[0038] In this embodiment, 70% of the data is divided into a training set, 20% of the data is a validation set, and 10% of the data is a test set. In each iteration, the output is calculated by forward propagation, and the weight is updated by back propagation. The learning rate, number of hidden layers, number of nodes and other parameters are adjusted according to the validation set to optimize the model. A four-layer BP neural network is used, including an input layer, two hidden layers and an output layer. The number of neurons in the input layer is 3: date, vehicle arrival time, user-estimated vehicle exit time The number of neurons in the output layer is 1, corresponding to the corrected exit time The number of neurons in the two hidden layers is 4 and 3 respectively. The transfer function of the hidden layer is the Tan-Sigmoid function, and the transfer function of the output layer is the Purelin function.

[0039] Estimated value and correction value With the true value The relative error cumulative distribution is as follows Figure 3 As shown, compared with the user's estimated vehicle exit time Modified exit time With the true value The error is smaller. Only 35.1% of the data is relative to the true value The error is less than 3%, and after correction reached 80.4%,

[0040] In some embodiments, in step S2, the matching coefficient between the vehicle and the parking space is calculated according to the corrected real-time parking lot status information, as follows:

[0041] The matching coefficient P 匹配度 The calculation includes the ideal depth factor P i Calculation and accessibility coefficient P j calculate:

[0042]

[0043] Wherein, α1, α2 and β are the gradient and scaling coefficient of the coefficient term. In this embodiment, α1=2, a2=1, β=5.

[0044] The ideal depth factor P i The degree of matching between the vehicle and the depth x of the parking space is obtained based on the historical parking time distribution characteristics of the parking lot. The parking time distribution characteristics are obtained by fitting a curve to the parking time data of vehicles entering and leaving the parking lot in the past.

[0045] like Figure 4 As shown, in this embodiment, the least square method is used to perform quadratic polynomial linear fitting, that is, to estimate In Solving for f = 16.24x 2 -3.116x+0.8737. Correspondingly, the ideal depth x i The calculation is as follows:

[0046]

[0047] Vehicle parking time 停车时长 The shorter the ideal depth x i The shallower the ideal depth factor P i =|xx i |.

[0048] The accessibility coefficient P jThe heuristic cost represents the scheduling cost required to park the vehicle in the parking space. When there is no obstacle in the route of parking the vehicle in the parking space, the accessibility is the highest, P j = 0. When there are obstructed vehicles in the parking route, the accessibility coefficient is expressed as the negative of the ratio of the number of obstructed vehicles m to the total number of parking spaces M in the plank road:

[0049]

[0050] It should be further explained that, in the specific implementation process, in step S3, the vehicle is dispatched by the automatic valet parking system according to the calculated matching coefficient of the vehicle, which specifically includes the following:

[0051] In some embodiments, the entering vehicle scheduling includes allocating parking spaces for the entering vehicles and reallocating parking spaces for the obstacle vehicles, such as Figure 5 shown.

[0052] According to the calculated matching coefficient of the incoming vehicle, the parking space with the largest matching coefficient is allocated to the incoming vehicle. j = 0, there is no obstacle vehicle, and the vehicle entering the parking space is directly parked in the assigned parking space. j <0, there is an obstacle vehicle. The matching coefficient is calculated for the obstacle vehicle and the parking space with the largest matching coefficient is reallocated to it, so that the entering vehicle can park in the allocated parking space.

[0053] The exit vehicle dispatching includes reallocation of parking spaces for obstacle vehicles, such as Figure 6 As shown. When the accessibility coefficient P of the parking space where the exiting vehicle is located j = 0, there is no obstacle vehicle, and the exiting vehicle leaves the parking lot directly. j <0, there is an obstacle vehicle. The matching coefficient of the obstacle vehicle is calculated and the parking space with the largest matching coefficient is reallocated to it, so that the exiting vehicle leaves the parking lot.

[0054] The embodiment of the present application also provides an automated high-density parking lot intelligent dispatching system, including a real-time state sensing unit, a data correction unit, a matching degree calculation unit, and an automatic valet parking unit;

[0055] The real-time status sensing unit is used to collect vehicle information and parking status, update parking lot real-time status information, and obtain vehicle entry / exit requirements;

[0056] The data correction unit includes a pre-trained neural network model for obtaining a more accurate vehicle exit moment;

[0057] The matching degree calculation unit calculates the matching degree coefficient between the vehicle and the parking space according to the real-time parking lot status information after data correction, which is used in the scheduling process;

[0058] The automatic valet parking unit utilizes the automatic valet parking technology to dispatch the vehicle according to the calculated matching coefficient.

[0059] Finally, it should be noted that the above embodiments are only used to illustrate and explain the present invention, and are not intended to limit the present invention to the described embodiments. In addition, those skilled in the art can understand that the present invention is not limited to the above embodiments, and more variations and modifications can be made according to the present invention, and these variations and modifications are all within the scope of protection required by the present invention.

Claims

1. An automated high-density parking lot intelligent dispatching method, characterized in that: The following steps are involved: S1. Parking lot real-time status information collection and data correction based on neural network model; S2. Calculate the matching coefficient between the vehicle and the parking space according to the corrected real-time parking status information; S3. Dispatch the vehicles according to the calculated matching coefficients of the vehicles.

2. The automated high-density parking lot intelligent dispatching method according to claim 1 is characterized by: The vehicles in the automated high-density parking lot are stored in a plurality of parallel aisles, and an automatic valet parking system is used to enable vehicles to enter the parking lot, leave the parking lot, and move within the parking lot.

3. The method for intelligent dispatching of an automated high-density parking lot according to claim 1 or 2, characterized in that: The real-time parking lot status information includes the occupancy status of parking spaces, parked vehicle information, vehicle entry requirements, and vehicle exit requirements.

4. The automated high-density parking lot intelligent dispatching method according to claim 3 is characterized by: The parked vehicle information includes the arrival time and exit time of the vehicle; the exit time of the vehicle is an estimated value provided by the user when arriving at the parking lot.

5. The method for intelligent dispatching of automated high-density parking lots according to claim 4 is characterized by: Generate a parking lot real-time status information table based on the acquired parking lot real-time status information, wherein the first column of the parking lot real-time status information table is the number of the parking space, the second column is the occupancy status of the parking space, the third column is the arrival time of the vehicle currently parked in the parking space, and the fourth column is the exit time of the vehicle currently parked in the parking space; The vehicle entry requirements are obtained through the automatic recognition system at the parking lot entrance, and the estimated vehicle exit time provided by the user is collected. The data is corrected based on the pre-trained neural network model to obtain the corrected exit time.

6. The method for intelligent dispatching of automated high-density parking lots according to claim 1 is characterized by: The matching coefficient includes an ideal depth coefficient and an accessibility coefficient; The ideal depth coefficient represents the matching degree between the depth of the vehicle and the parking space, which is obtained according to the historical parking time distribution characteristics of the parking lot; the parking time distribution characteristics are obtained by collecting the parking time data of the vehicles entering and leaving the parking lot in the past and fitting the curve; The accessibility coefficient represents the heuristic cost of the scheduling cost required to park the vehicle in the parking space. When there is no obstacle vehicle in the route to the parking space, the accessibility is set to 0; when there is an obstacle vehicle in the route to the parking space, the accessibility coefficient is the negative of the ratio of the number of obstacle vehicles to the total number of parking spaces in the plank road.

7. The method for intelligent dispatching of automated high-density parking lots according to claim 6 is characterized by: The vehicle dispatching includes entering vehicle dispatching and exiting vehicle dispatching; The dispatching of incoming vehicles includes allocating parking spaces for incoming vehicles and reallocating parking spaces for vehicles with obstacles; allocating a parking space with the largest matching coefficient to the incoming vehicle according to the calculated matching coefficient of the incoming vehicle; When the accessibility coefficient is 0, there is no barrier vehicle and the entering vehicle is parked directly in the allocated parking space; When the accessibility coefficient is negative, there is an obstacle vehicle; Calculate the matching coefficient for the obstacle vehicle and reallocate the parking space with the largest matching coefficient to it, so that the entering vehicle can park in the allocated parking space; The exit vehicle dispatching includes reallocation of parking spaces for the obstacle vehicles; When the accessibility coefficient of the parking space where the exiting vehicle is located is 0, there is no obstacle vehicle, and the exiting vehicle leaves the parking lot directly; When the accessibility coefficient is negative, there is an obstacle vehicle. The matching coefficient is calculated for the obstacle vehicle and the parking space with the largest matching coefficient is reallocated to it, so that the exiting vehicle leaves the parking lot.

8. An automated high-density parking lot intelligent dispatching system, characterized in that: It includes a real-time state sensing unit, a data correction unit, a matching degree calculation unit and an automatic valet parking unit; The real-time status sensing unit is used to collect vehicle information and parking status, update parking lot real-time status information, and obtain vehicle entry requirements and vehicle exit requirements; The data correction unit includes a pre-trained neural network model for obtaining a more accurate vehicle exit moment; The matching degree calculation unit calculates the matching degree coefficient between the vehicle and the parking space according to the real-time parking lot status information after data correction, which is used in the scheduling process; The automatic valet parking unit utilizes the automatic valet parking technology to dispatch the vehicle according to the calculated matching coefficient.

Images