A method for dynamically calculating inspection time of an unmanned aerial vehicle in-lane parking charge management system

Abstract

The application discloses an unmanned aerial vehicle (UAV) in-road parking charging management system inspection time dynamic calculation method, comprising the following steps: step one, fitting the probability density function of vehicle arrival and parking time length of the in-road parking area based on historical parking data; step two, calculating the parking time length of the vehicle based on real-time vehicle information of the UAV inspection; step three, calculating the change probability of all parking space states in the inspection time interval based on the Bayes theorem; step four, establishing a parking revenue maximization model based on the in-road charging rules; step five, solving the parking revenue maximization model based on the double simulated annealing algorithm to obtain the take-off time of the UAV dynamic inspection. The application can be used with the existing in-road parking charging management system based on the autonomous inspection UAV, effectively, real-timely and dynamically calculates the inspection time of the UAV, improves the economy and scientificity of the system, helps the system to be applied in the actual commercialization scene, and effectively implements the landing.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of traffic engineering, and particularly relates to a UAV (Unmanned Aerial Vehicle) inspection time dynamic calculation method for a UAV in-street parking charging management system. BACKGROUND

[0002] Under the condition of continuous growth of private car ownership in China, the problem of parking difficulty in urban residential areas is becoming more and more prominent, and the parking resource shortage situation has become an important problem to be solved in urban development. The in-street parking area becomes one of the key strategies to alleviate the parking problem due to its short construction period and low cost. At present, the in-street parking management mainly relies on the management modes of police patrol, meter charging and high-position video charging, which are not suitable for the management of residential in-street parking. With the active promotion of the state to the "low-altitude economy", some researchers have begun to carry out related research on the application of automatic inspection UAV in parking management, and have proposed an automatic inspection parking management system based on UAV, which has a broad application prospect due to its advantages of rapid and efficient coverage of a large area. Compared with the existing method, the system has the advantages of rapidness, efficiency and low carbon. However, the UAV inspection frequency is closely related to the economy and scientificity of the system, and there is very little research on it at present, which leads to the difficulty of the system in practical commercial application scenarios and makes it difficult to effectively implement. Therefore, it is particularly important to determine the UAV take-off time in real time, dynamically and effectively in the in-street parking management system, and to realize the dynamic calculation of the inspection time based on the UAV in-street parking charging management system. SUMMARY

[0003] In order to solve the problems in the prior art, the application provides a UAV in-street parking charging management system inspection time dynamic calculation method, which comprises the following steps:

[0004] Step one, fitting the probability density function of vehicle arrival and parking duration of in-street parking area based on historical parking data;

[0005] Step two, calculating the parked duration of the vehicle based on the real-time vehicle information of UAV inspection;

[0006] Step three, calculating the change probability of all parking space states in the inspection time interval based on Bayes theorem;

[0007] Step four, establishing a parking revenue maximization model based on in-street charging rules;

[0008] Step five, solving the parking revenue maximization model based on double simulated annealing algorithm to obtain the UAV dynamic inspection take-off time.

[0009] Further, in step one, the specific steps include:

[0010] Collecting vehicle parking information data in the on-street parking area, including: license plate information, vehicle entering time, leaving time, latitude and longitude of parking location, vehicle photo;

[0011] Analyzing the quality of vehicle parking information data and performing data preprocessing to remove false information and unreasonable information;

[0012] Statistical analysis is performed on the preprocessed vehicle parking information data to obtain the parking duration regularity of vehicles in the area and the vehicle arrival regularity, and the probability density function of vehicle parking duration and the probability density function of vehicle arrival are calculated.

[0013] Further, in step two, through real-time vehicle information detected by unmanned aerial vehicle inspection, the same vehicle is detected by unmanned aerial vehicle inspection for two consecutive times in two different existence states of "in" or "not in", the vehicle entering time and leaving time The interval is between the time, so as to calculate the parking duration Time of the vehicle.

[0014] Further, the entering time leaving time And the parking duration Time calculation method is:

[0015] When a vehicle T i The vehicle state is "not in" at the inspection time, but T i+1 The vehicle state is "in" at the inspection time, and the expected value of the entering time of the vehicle is calculated:

[0016]

[0017] In the formula, T i Is the time of the i-th inspection of the unmanned aerial vehicle;

[0018] When a vehicle T i The vehicle state is "in" at the inspection time, but T i+1 The vehicle state is "not in" at the inspection time, and the expected value of the leaving time of the vehicle is calculated:

[0019]

[0020] The parking duration of the vehicle is:

[0021]

[0022] For one unmanned aerial vehicle inspection, it is found that the state of the vehicle in the system does not change, that is, the vehicle is always parked there, and the parking duration of the vehicle is:

[0023]

[0024] Bayesian methods are very suitable for small sample situations. By using prior knowledge and prior distribution, Bayesian methods can provide more stable and reliable predictions in the case of insufficient data.

[0025] Further, in step three, after the completion of a certain inspection by the unmanned aerial vehicle, the next flight will be carried out in the unmanned aerial vehicle inspection time interval [T min ,T max ] based on the vehicle parking duration and the probability density function of the parking duration, the probability value of the next time in the parked vehicle leaving is calculated using the Bayesian method:

[0026]

[0027] In the formula, P leave Time ij is the probability of the jth vehicle parking duration being Time in the ith inspection, x ij is the xth hour of the jth vehicle parking in the ith inspection;

[0028] According to the probability density function of the vehicle arrival, the probability value P come of the vehicle arrival at any future time can be calculated.

[0029] Further, in step four, based on the parking management policy of the area, the parking fee rule is determined, and the in-lane parking fee of the area is set as a function of the parking duration, denoted as:

[0030] fee=Gt

[0031] In the formula, G t is the parking duration function;

[0032] Combined with the parking space state change probability value of all vehicles in the unmanned aerial vehicle inspection time interval obtained in step three, a parking revenue maximization model is established:

[0033]

[0034] In the formula, j is the vehicle number of the ith inspection, N is the parking number; λ is the vehicle arrival probability cost coefficient, yuan.

[0035] Due to the capacity limitation of the parking lot, it is assumed that the capacity of the parking lot is M, then:

[0036] N≤M

[0037] The time interval of the unmanned aerial vehicle inspection is limited:

[0038] T min ≤x ij <T max .

[0039] Further, in step five, the above parking revenue maximization model is solved based on a double simulated annealing algorithm. The algorithm introduces a second "simulated annealing" process on the basis of the classical simulated annealing algorithm, increases the depth and breadth of the search, and thus finds the optimal solution in a larger range. It can adapt to different types of optimization problems, including complex nonlinear and multi-peak problems, and can reduce the influence of local optimum, and has strong advantages in finding global optimal solution.

[0040] The update of the solution of the algorithm needs to rely on the Metropolis sampling criterion, and the change of the objective function from state i to state j can be described as:

[0041]

[0042] Beneficial effects:

[0043] The present application mainly proposes a UAV inspection time dynamic calculation method based on on-street parking management, collects real-time information of on-street parked vehicles through a UAV, extracts vehicle entry and exit time, calculates parking duration, then calculates the charging condition corresponding to the parking duration based on the probability value of all vehicles parking in the next UAV inspection time interval, and constructs a UAV inspection time dynamic calculation method based on on-street parking management.

[0044] The UAV inspection time dynamic method proposed by the present application can effectively, real-timely and dynamically calculate the UAV inspection time based on the on-street parking charging management system of the UAV, avoid resource waste caused by too frequent UAV take-off of the system, reduce the too large vehicle charging error caused by too low frequency, and help commercialization of the existing UAV on-street parking charging management system. In addition, compared with the fixed time UAV flight method (UAV take-off time interval is 30 min), the charging efficiency can be significantly improved, the charging accuracy rate reaches more than 95%, the problem of lacking a theoretical basis for determining the inspection cycle in the on-street parking automation inspection is solved, and specific methods and ideas are provided for future on-street low-carbon energy-saving parking charging. BRIEF DESCRIPTION OF DRAWINGS

[0045] Figure 1 The flow chart of the UAV dynamic inspection time calculation method proposed by the present application;

[0046] Figure 2 The on-street parking duration probability density function graph of the specific embodiment of the present application;

[0047] Figure 3 The on-street vehicle arrival probability density function graph of the specific embodiment of the present application. DETAILED DESCRIPTION

[0048] Please refer to the attachedFigure 1 The specific implementation method of the present application is specifically described below.

[0049] The implementation object is a two-way eight-lane on-street parking lot in Harbin.

[0050] Step 1: Use the method of manual recording to observe and count the on-street parking lot, record parking data, including license plate information, vehicle entry time, departure time, latitude and longitude of parking location, vehicle photo, etc., and record historical parking data multiple times.

[0051] Then upload the recorded data to the computer, visualize the data quality and perform data preprocessing, and through PyCharm software programming, eliminate vehicle data information with only entry time, only departure time, no license plate information, vehicle photo, error information and unreasonable information.

[0052] Then, based on the processed data, use Origin software to perform fitting analysis on the data to obtain the vehicle arrival distribution and parking duration distribution of the on-street parking area, as shown in Figure 2 and Figure 3 .

[0053] Among them, the probability density function of vehicle arrival is:

[0054] Among them, the probability density function of vehicle arrival is:

[0055]

[0056] The probability density function of parking duration is:

[0057]

[0058] Step 2: The UAV will start flying at the beginning of the daily charging time interval (i.e. 8:00), and complete the last flight before the end point (i.e. 20:00). Any flight after the first flight of the UAV each day has a "last flight". At the first flight, the vehicles on the parking spaces will be identified and recorded, and the time when the vehicle is photographed is recorded as the entry time; at the end of the "last flight" of the i-th flight, the vehicle information photographed and identified is compared with the information library, the newly appeared vehicle information is added to the database of entering vehicles, the disappeared vehicle information is added to the database of leaving vehicles, and the remaining vehicles are added to the database for parking information update.

[0059] Through the real-time vehicle information detected by the UAV inspection, the time interval of vehicle entry and exit of the area can be determined, which is between the time when the same vehicle is detected by the UAV for two consecutive inspections, such as for a vehicle, Ti The vehicle state is "in" at the inspection time, but T i+1 The vehicle state is "not in" at the inspection time, and vice versa. Then, the entering time of the entering vehicle is calculated based on the probability density function of vehicle arrival:

[0060]

[0061] The leaving time of the leaving vehicle can also be calculated based on the probability density function of vehicle parking duration:

[0062]

[0063] The parking duration of the vehicle is:

[0064]

[0065] For vehicles whose state does not change in the system, i.e., the vehicle is always parked here, the entering time is calculated as above, and the parking duration of such vehicles is:

[0066]

[0067] Step 3: After completing a flight, the updated vehicle parking data, including license plate number, vehicle entering time, and parking duration information, can be obtained from Step 2. The UAV will perform the next flight in the UAV inspection time interval [T min , T max ]. Where T min is the minimum time interval for UAV takeoff, set to 6 minutes in this example; T max is the maximum time interval for UAV takeoff, set to 30 minutes in this example.

[0068] Based on the probability density functions of vehicle parking duration and vehicle arrival, the Bayesian method is used to calculate the probability value of the next time interval for the parked vehicle to leave. Bayesian method is very suitable for small sample situation. By using prior knowledge and prior distribution, Bayesian method can provide more stable and reliable prediction in the case of insufficient data.

[0069] According to the probability density function of vehicle parking duration, the probability value of having a vehicle still parked at any future time can be calculated:

[0070]

[0071] And according to the probability density function of vehicle arrival, the probability value P come of having a vehicle arrive at any future time can be calculated.

[0072] Step 4: Referencing the on-street parking management regulations issued by the Harbin government, determine the charging rules for the studied road section. Parking time for this section is divided into free parking and paid parking periods. Free parking is from 8:00 PM to 8:00 AM the next day, and paid parking is from 8:00 AM to 8:00 PM. The specific charging rules for this section are: free parking for the first 30 minutes; a charge of 5 yuan for parking from 30 minutes to 2 hours; for parking exceeding 2 hours, an additional 1 yuan is charged per hour beyond 2 hours. These charging rules can be expressed using a function, i.e.:

[0073]

[0074] The DJI Mavic 3 drone flies and records video within a paid parking area. Through these steps, the probability that a vehicle will remain parked at any future time after a given inspection by the drone can be obtained. Based on this, the on-street parking fee management rules can be determined, and a parking revenue maximization model can be constructed.

[0075]

[0076] Step 5: After establishing the model, the parking revenue maximization model is solved using the double simulated annealing algorithm. This algorithm introduces a second "simulated annealing" process based on the classic simulated annealing algorithm, increasing the depth and breadth of the search, thus finding the optimal solution over a wider range. It can adapt to different types of optimization problems, including complex nonlinear and multimodal problems, reducing the influence of local optima and possessing a strong advantage in finding the global optimum.

[0077] The algorithm's solution update relies on the Metropolis sampling criterion, and the change in the objective function from state i to state j can be described as follows:

[0078]

[0079] This paper describes the intelligent computation process of the double simulated annealing algorithm using a Python program. The initial temperature is 5230℃, the restart temperature ratio is 0.00002, the access parameter is 2.62, the acceptance probability parameter is -5, and the initial solution is randomly generated. This method achieves a charging accuracy of up to 95%, demonstrating excellent optimization performance and indicating that the model possesses high optimization capabilities.

Claims

1. A method for dynamically calculating the inspection time of an unmanned aerial vehicle (UAV) in-lane parking fee management system, comprising the following steps: Step one: fitting the probability density function of vehicle arrival and parking duration in the in-lane parking area based on historical parking data; Step two: calculating the parking duration of a vehicle based on real-time vehicle information of UAV inspection; Step three: calculating the change probability of all parking space states in the inspection time interval based on Bayes' theorem; Step four: establishing a parking revenue maximization model based on in-lane charging rules; Step five: solving the parking revenue maximization model based on a double simulated annealing algorithm to obtain the take-off time of UAV dynamic inspection; In step three, after the UAV completes a certain inspection, the next flight will be carried out in the UAV inspection time interval [T min ,T max ] interval, based on the vehicle parking time and the parking time probability density function, the probability value of the next time when the parked vehicle leaves is calculated using the Bayesian method: In the formula, P leave Time ij is the probability of the i-th inspection of the j-th vehicle parking for Time, x ij is the i-th inspection of the j-th vehicle parked for x hours; The probability value P of having a vehicle arriving at any future time can be calculated from the probability density function of the vehicle arrival come ; In step four, based on the parking management policy of the area, the parking charging rules are determined, and the in-lane parking fee is a function of the parking duration, denoted as: fee=Gt In the formula, G t is the charging function related to the parking duration; Based on the change probability value of all vehicle parking space states in the UAV inspection time interval obtained in step three, a parking revenue maximization model is established: In the formula, j is the vehicle number of the i-th inspection, N is the number of parking spaces, and λ is the vehicle arrival probability cost coefficient, yuan. Due to the capacity limit of the parking lot, it is assumed that the capacity of the parking lot is M, then: N≤M The time interval of UAV inspection is limited: T min ≤x ij <T max . 2.The method of claim 1, wherein, In step one, the specific steps include: Collecting vehicle parking information data in the in-lane parking area, including: license plate information, vehicle entry time, departure time, latitude and longitude of parking location, vehicle photo; Analyzing the quality of vehicle parking information data and performing data preprocessing to remove incorrect information and unreasonable information; Statistical analysis is performed on the preprocessed vehicle parking information data to obtain the parking duration rule and vehicle arrival rule of the area, and the probability density function of vehicle parking duration and the probability density function of vehicle arrival are calculated. 3.The method of claim 1, wherein, In step two, the real-time vehicle information detected by the drone inspection is used to calculate the vehicle entry time by detecting the same vehicle in two consecutive inspections, showing either "present" or "absent". and departure time The time interval is between which the vehicle has been stopped for a period of time (Time).

4. The method of claim 3, wherein the method further comprises: determining a time interval between two consecutive inspection times; and determining the inspection time based on the time interval and the current time. Entry time Exit time The Time calculation method is: When a certain vehicle T i The vehicle state is "not in" at the inspection time, but T i+1 The vehicle state is "in" at the inspection time, and the expected value of the entering time of the vehicle is calculated as In the formula, T i is the time of the i-th inspection of the UAV; When a certain vehicle T i The vehicle state is "in" at the inspection time, but T i+1 The vehicle state is "not in" at the inspection time, and the expected value of the vehicle departure time when the vehicle departs is calculated as The parking duration of a vehicle is: For a UAV inspection, it is found that the state of the vehicle in the system does not change, i.e. the vehicle is parked there, then the parking duration of the vehicle is:

5. The method of claim 1, wherein the method further comprises: determining a time of day when the UAV is to be parked; and determining a time of day when the UAV is to be retrieved. In step five, the parking revenue maximization model is solved based on the double simulated annealing algorithm. On the basis of the classical simulated annealing algorithm, a second "simulated annealing" process is introduced to increase the depth and breadth of the search. 6.The method of claim 5, wherein, The update of the algorithm solution of the double simulated annealing algorithm relies on the Metropolis sampling criterion, and the change of the objective function from state i to state j is described as:

Images