Intelligent parking garage scheduling method and system based on storage space energy efficiency and duration prediction

Abstract

The present invention discloses a scheduling method and system for an intelligent three-dimensional parking garage based on parking space energy efficiency and duration prediction. Based on the time it takes for a vehicle to reach a parking space from the entrance, all parking spaces are divided into three energy efficiency zones: high efficiency, medium efficiency, and low efficiency. Vehicle storage durations are sorted from smallest to largest and then divided into three intervals, corresponding to vehicles stored in different energy efficiency zones. When a vehicle to be parked arrives at the entrance of the intelligent three-dimensional parking garage, the predicted duration of the vehicle's stay is calculated using a trained XGBoost algorithm based on the vehicle's historical parking data, and the vehicle is then automatically parked in the assigned parking space. This invention provides an effective scheduling solution that balances parking capacity, avoids extreme parking capacity, and makes scheduling more scientific and reasonable.

Description

Technical Field

[0001] The present invention relates to intelligent stereo garage scheduling, and in particular to an intelligent stereo garage scheduling method and system based on storage space energy efficiency and duration prediction. Background Art

[0002] With the acceleration of urbanization and the growth of car ownership, parking challenges are becoming increasingly prominent in city centers, schools, commercial complexes, tourist attractions, and large hospitals. This not only inconveniences people's lives but also exacerbates urban traffic pressures. For example, large hospitals, such as those in my country's top three hospitals, generally face parking difficulties. As a modern parking solution, multi-story parking garages are gradually gaining widespread adoption. However, traditional manual parking methods are no longer able to meet the growing demand. The development of multi-story parking garages is beginning to shift from traditional manual parking to a new stage of automation and intelligent development.

[0003] Traditional parking garages face a range of operational and management challenges, particularly in areas such as intelligence and information technology, which have resulted in operational efficiency and service quality failing to meet expectations. Currently, most parking garages still utilize traditional management models, relying on manual operation or simple proximity-based dispatching for vehicle access. This approach is not only inefficient but also suffers from significant drawbacks in terms of user experience, energy consumption, and safety. Summary of the Invention

[0004] Purpose of the invention: The purpose of the present invention is to provide an efficient intelligent parking garage scheduling method and system based on storage space energy efficiency and duration prediction.

[0005] Technical Solution: The intelligent parking garage scheduling method based on storage space energy efficiency and duration prediction described in the present invention includes the following steps:

[0006] Calculate the storage time of vehicles from the entrance to each storage berth, and divide all storage berths into three energy efficiency zones according to the storage time from the smallest to the largest, namely the high efficiency zone, the medium efficiency zone and the low efficiency zone;

[0007] Obtain the historical dwell time of several vehicles and sort them in ascending order. Standardize the historical dwell time and divide it into three equal intervals, namely, the first interval, the second interval, and the third interval, which correspond to vehicles stored in the high-efficiency zone, the medium-efficiency zone, and the low-efficiency zone, respectively.

[0008] Re-acquire the historical parking data features of several vehicles, including vehicle arrival date, arrival time, departure date and departure time, and use this historical parking data to train the XGBoost algorithm to predict vehicle parking duration;

[0009] When a vehicle arrives at the entrance of the smart parking garage, the trained XGBoost algorithm is used to predict the vehicle's parking duration based on the vehicle's historical parking data.

[0010] The predicted vehicle dwell time is standardized, and according to the range of the standardized predicted vehicle dwell time, the vehicle is allocated to a storage berth corresponding to different energy efficiency areas for storage.

[0011] Furthermore, during the off-peak period of parking, vehicles in the medium-efficiency and low-efficiency areas are moved to the high-efficiency or medium-efficiency areas. The remaining parking time of the vehicles in the medium-efficiency and low-efficiency areas is calculated, the remaining parking time is normalized, and the vehicles are moved to the high-efficiency or medium-efficiency areas according to the range of the normalized remaining parking time.

[0012] The low-peak period for inbound vehicles is a period when the average inbound vehicle flow rate is not greater than a threshold value.

[0013] Furthermore, the warehousing time period classification method includes: It is the peak period for storage. When It is the low peak period for storage;

[0014] in The first Average inbound vehicle flow in the hourly period, is the average value of the average inbound vehicle flow within the statistical time range, is the constant term coefficient, is the standard deviation of the average inbound vehicle flow within the statistical time range.

[0015] Furthermore, the features of the historical parking data of the vehicles that are retrieved also include: whether the arrival date and departure date of the vehicle are weekdays or holidays, the number of times the vehicle enters the parking lot, and the province and city mentioned in the vehicle license plate number.

[0016] Furthermore, before using the historical parking data to train the XGBoost algorithm to predict the vehicle stay time, the method also includes: preprocessing the historical parking data to eliminate abnormal data.

[0017] Furthermore, the method for removing abnormal data includes: calculating the interquartile range (IQR) of the box plot of historical parking data, and when the historical parking data is less than or greater than When , it is determined to be abnormal data, where Q1 is the lower quartile and Q3 is the upper quartile.

[0018] Furthermore, the storage time of the vehicle from the entrance to each storage berth includes the time for the elevator to pick up the vehicle from the entrance, the time for the elevator to move vertically to the designated floor, and the time for the horizontal transporter to move the vehicle to the designated storage berth;

[0019] The step of normalizing the predicted vehicle dwell time and allocating the vehicle to a storage space corresponding to a different energy efficiency zone for storage according to the range of the normalized predicted vehicle dwell time includes:

[0020] First, it is determined whether there are any vacant parking spaces in the intelligent stereoscopic parking garage. If there are vacant parking spaces, the driver parks the vehicle at the waiting point corresponding to the elevator and leaves. Otherwise, the driver drives the vehicle away. The elevator moves the vehicle to the floor where the storage parking spaces in the energy efficiency area are located, and the transverse transport removes the vehicle from the elevator and moves it to the storage parking spaces in the energy efficiency area. During this period, if there are no vacant elevators and / or transverse transports, the vehicle waits in place until the elevators and / or transverse transports are vacant.

[0021] The intelligent parking garage scheduling system based on storage space energy efficiency and duration prediction of the present invention comprises: after a vehicle to be parked arrives at the entrance of the intelligent parking garage, an elevator delivers the vehicle to a designated floor, and a transverse transporter moves the vehicle to a designated storage berth. The system comprises:

[0022] The energy efficiency zone division unit is used to calculate the storage time of vehicles from the entrance to each storage berth, and divide all storage berths into three energy efficiency zones according to the storage time from the smallest to the largest, namely the high efficiency zone, the medium efficiency zone and the low efficiency zone;

[0023] An energy efficiency zone matching unit is configured to obtain the historical stay times of several vehicles and sort them in ascending order, normalize the historical stay times and divide them into three equal intervals, namely, a first interval, a second interval, and a third interval, which correspond to vehicles stored in a high-efficiency zone, a medium-efficiency zone, and a low-efficiency zone, respectively;

[0024] The vehicle dwell time prediction unit is used to retrieve the historical parking data features of several vehicles, including the vehicle arrival date, arrival time, departure date and departure time, and use this historical parking data to train the XGBoost algorithm to predict the vehicle dwell time. When a parked vehicle arrives at the entrance of the intelligent parking garage, the trained XGBoost algorithm is used to obtain the predicted vehicle dwell time based on the vehicle's historical parking data.

[0025] The vehicle dispatching unit is used to standardize the predicted vehicle dwell time and allocate the vehicle to a storage berth corresponding to different energy efficiency zones for storage according to the interval of the standardized predicted vehicle dwell time.

[0026] The electronic device described in the present invention includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the computer program is loaded into the processor, it implements the intelligent stereoscopic parking garage scheduling method based on storage location energy efficiency and duration prediction.

[0027] The computer-readable storage medium of the present invention stores a computer program, and when the computer program is executed by a processor, the intelligent stereoscopic parking garage scheduling method based on storage location energy efficiency and duration prediction is implemented.

[0028] Beneficial effect: Compared with the existing technology, the advantages of the present invention are: the present invention combines warehouse management theory and machine learning to design an intelligent stereoscopic parking garage scheduling method based on storage space energy efficiency and duration prediction, dynamically optimizes parking resource allocation, and provides an effective scheduling solution for existing intelligent stereoscopic parking garage scheduling. It also provides a scientific solution for high-demand parking lots and provides new theoretical and practical references for research in related fields. Specifically: the present invention applies the ABC classification method of cargo storage to the division of stereoscopic parking garage berths, and divides the stereoscopic parking garage parking spaces into three categories based on the length of storage time: high efficiency area, medium efficiency area, and low efficiency area; in duration prediction, the three aspects of parking time, vehicle source, and historical warehousing situation are comprehensively considered, and the vehicle parking time is predicted based on the XGBoost algorithm, and then the corresponding energy efficiency area is matched according to the parking time, so that the fullness rate of each energy efficiency area is relatively balanced, avoiding extreme fullness rate, and at the same time reducing the average waiting time of customers, and the scheduling is more scientific and reasonable. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 Schematic diagram of an intelligent three-dimensional parking garage in an embodiment of the present invention.

[0030] Figure 2 This is a flow chart of the warehousing scheduling method according to an embodiment of the present invention.

[0031] Figure 3 Schematic diagram of energy efficiency areas according to an embodiment of the present invention.

[0032] Figure 4 This is a flow chart of the warehouse transfer scheduling method in an embodiment of the present invention.

[0033] Figure 5 This is a flow chart of the outbound method in an embodiment of the present invention.

[0034] Figure 6 This is a simulation diagram of the full inventory rate using the traditional nearest allocation method in an embodiment of the present invention.

[0035] Figure 7 This is a simulation diagram of the full inventory rate using the method described in the embodiment of the present invention.

[0036] Figure 8 This is a comparison chart of the impact of different numbers of elevators and transverse conveyors on average operation time in an embodiment of the present invention.

[0037] Figure 9 4 is a comparison chart showing the effect of different numbers of elevators and transverse conveyors on the average waiting time in an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0039] In the field of operations research, the management and scheduling of multi-story parking garages is considered a complex resource optimization problem, involving multiple aspects such as vehicle access methods, space allocation, and energy management. As garages expand in size and user needs diversify, the complexity of solving this problem increases. However, advances in machine learning and simulation are providing new insights and approaches to address these issues. How to effectively utilize these technologies to improve multi-story parking garage throughput efficiency and customer satisfaction has become a common challenge facing multi-story parking garage managers.

[0040] In-depth research on multiple parking garages revealed an imbalance between the rate of data accumulation and the depth of data application. Some garages fail to fully utilize accumulated operational data to optimize management and service processes. Furthermore, parking garage managers rarely prioritize the energy efficiency of parking spaces, resulting in unbalanced and inadequate utilization of parking resources. Based on this finding, we can consider leveraging big data processing technologies to analyze and mine large amounts of accumulated operational data, identify key and vulnerable links in the management and service processes, and prioritize and strengthen these links in the design of vehicle scheduling methods. Therefore, it is necessary to integrate warehouse management theory and the application of big data technologies with the management and scheduling issues of parking garages, leveraging the former to improve and enhance the latter's scheduling methods, thereby achieving more scientific and efficient management and scheduling solutions.

[0041] Taking the F automated parking garage as an example, the F garage has six floors, including a lobby and five parking garages. The lobby is located on the underground level, with the vehicle entrance located there. Each parking floor has 30 parking spaces, for a total of 150 parking spaces. Currently, the garage uses a proximity-based scheduling method. Research has revealed that the garage fails to fully utilize data resources. The current scheduling method results in uneven resource allocation, with some garage floors experiencing overcrowding or idle resources. Furthermore, this scheduling method results in long wait times for parking, low customer satisfaction, and a tendency to cause customer anxiety.

[0042] Therefore, the present invention provides an intelligent scheduling method for a stereo garage system based on the warehouse management theory and the historical data accumulated from the operation of the stereo garage. Figure 1 As shown in the figure, taking the intelligent stereo garage F as an example, a simulation model is established based on the actual operation process of the stereo garage, and the operation data provided by the intelligent stereo garage F is deeply analyzed and studied, thereby designing a vehicle scheduling method that comprehensively considers the energy efficiency of the storage space and the duration prediction. The intelligent stereo garage F includes an elevator 1, a horizontal conveyor 2, an elevator channel 3, a transport channel 4, a storage berth 5, a first screen 6, an automatic garage door 7, a second screen 8, a waiting point 9, a third screen 10, a camera 11, a detection platform 12, a fourth screen 13, and a waiting area 14. Figure 2 As shown, the method of the present invention includes the following steps.

[0043] (1) Conduct field research and analysis on the operation process of the stereo garage and collect the operating parameters of hardware facilities, such as elevators and horizontal conveyors;

[0044] (2) Calculate the total storage time of a vehicle from the garage entrance to each berth and divide the energy efficiency zones according to the ABC classification of cargo storage;

[0045] (3) Collect and organize the time characteristic data of customers’ parking, including the length of stay, vehicle source, historical parking situation, etc.;

[0046] (4) Establish a prediction model for vehicle parking time, including data preprocessing, data feature construction, machine learning model construction, etc.

[0047] (5) Allocate vehicles to different energy efficiency zones based on the predicted duration results;

[0048] (6) During the off-peak period of parking, vehicles in the medium and low efficiency areas are moved to the storage area, that is, the remaining parking time is re-predicted and the energy efficiency area is allocated.

[0049] The specific contents of each step are as follows.

[0050] Step 1: Energy efficiency division of storage spaces. Based on the ABC classification method for cargo storage, the F multi-story parking garage berths are divided into different energy efficiency zones.

[0051] Step 1.1: Field research and analysis of the F parking garage operation process (taking customer parking as an example), which is mainly divided into five stages: ① vehicle arrival at the intelligent parking garage, ② parking confirmation, ③ elevator handover, ④ horizontal vehicle loading and unloading, and ⑤ storage in the designated storage berth.

[0052] Step 1.2: Calculate the relationship between the time consumption and motion parameters during the vehicle access process.

[0053] The entire parking process can be divided into the following three stages:

[0054] 1) Lift Go to the ground floor car hall to pick up a car;

[0055] ;

[0056] 2) Lift Vertical movement to parking level, horizontal conveyor Move horizontally to the connection point of the corresponding lift, the lift and the cross conveyor transfer the vehicle;

[0057] ;

[0058] 3) The horizontal conveyor moves the vehicle to the designated storage berth;

[0059] ;

[0060] Sum: .

[0061] The entire car collection process can also be divided into three stages:

[0062] 1) The horizontal conveyor moves to the vehicle to be picked up storage berths;

[0063] ;

[0064] 2) Transport aircraft Horizontal movement to the lift Connection point, lift Move vertically to the corresponding parking level, horizontal transporter and lifts Transfer the vehicle;

[0065] ;

[0066] 3) Machine Go to the ground lobby floor to release the car;

[0067] ;

[0068] Sum: .

[0069] The meanings of the relevant parameters are shown in the following table.

[0070] Table 1 Parameter meaning

[0071]

[0072] Step 1.3: Calculate the total storage time of the vehicle from the entrance of the intelligent parking garage to each berth, divide the energy efficiency area based on the ABC classification method of warehouse management, and put the vehicle with the shortest storage time in the front. The storage berths are divided into high-efficiency areas, with the longest storage time in the front The first 50 storage berths with the shortest storage time are divided into high efficiency areas, the first 50 storage berths with the longest storage time are divided into low efficiency areas, and the remaining 50 storage berths are divided into medium efficiency areas. Figure 3 The following is a schematic diagram of energy efficiency area. Figure 3 The middle coordinate system represents the length, width and height of the intelligent stereo garage. The yellow points are high-efficiency areas, the green points are medium-efficiency areas, and the blue points are low-efficiency areas.

[0073] Step 2: Parking duration prediction. This embodiment mainly analyzes some factors that affect vehicle parking duration, including dwell time, vehicle source, and historical parking conditions. Based on the historical parking data of these three aspects, a machine learning algorithm is used to make predictions.

[0074] Step 2.1: Data Analysis: Analyze the historical parking data of the garage to identify factors that affect vehicle dwell time, including dwell time, parking reason, vehicle source, and medical stay status.

[0075] Step 2.2: Data preprocessing. By calculating the interquartile range (IQR) of the box plot of historical parking time, outliers are eliminated. or greater than , it is determined to be an outlier, where Q1 is the lower quartile and Q3 is the upper quartile. Removing outlier data can improve the accuracy of the prediction.

[0076] Step 2.3: Construct data features. Time data features include the car's arrival date, arrival time, departure date, and departure time. Factors such as the day of the week, weekday, and holiday also affect the number of times a customer parks in the garage and the length of time they park. Other data features include the city of the license plate, whether it is the local province or city, and the number of times a car has parked in the garage, which also lead to different parking times.

[0077] Step 2.4: Use the preprocessed data to predict parking duration based on the XGBoost algorithm.

[0078] 1) The original data set composed of data features is divided into training set and test set according to the ratio of 7:3. Then, 20% of the training set data is randomly sampled to form a validation set for cross-validation of the model.

[0079] 2) The parameters of the XGBoost algorithm are set to be based on tree model iteration, with a maximum search depth of 7, a learning rate of 0.01, an alpha of 0.01, a gamma of 0.1, and the RMSE loss function for model convergence;

[0080] 3) Calculate the gradient of the loss function with respect to the current prediction value to guide the construction of the next tree, add the newly constructed decision tree to the model, and update the overall prediction result;

[0081] 4) Repeat step 3 above, constructing a new decision tree each time until the predetermined number of trees or the predetermined number of iterations is reached;

[0082] 5) Use five-fold cross-validation to evaluate model performance and select appropriate parameters. After training, perform duration prediction and output the prediction results.

[0083] Step 3, combine Figure 1 The intelligent stereoscopic parking garage shown in the figure takes into account the storage method design of the storage space energy efficiency and parking time prediction, and performs scheduling based on the above storage space energy efficiency division and parking time prediction.

[0084] Step 3.1: The vehicle to be parked is located at the entrance of the intelligent stereo garage. The customer uses the fourth screen 13 to determine whether there is an available parking space in the garage. If so, the vehicle will enter the garage; if not, the vehicle will leave.

[0085] Step 3.2: After the vehicle enters the intelligent stereo garage, it is tested on the detection platform 12. If the weight and size of the vehicle meet the requirements, the vehicle enters the waiting area 14.

[0086] Step 3.3: Identify the vehicle through the camera 11, analyze the historical vehicle stay data, predict the vehicle stay time, and obtain the predicted stay time .

[0087] Step 3.4: Predict the length of stay Standardize and determine the range in which the vehicle is located, assign the vehicle to different energy efficiency zones, and then assign it to the nearest parking space in the energy efficiency zone based on the principle of proximity. The specific method is as follows.

[0088] 1) Collect the time T1 when M vehicles enter the smart stereo garage and the time T2 when they leave the smart stereo garage, calculate their parking time TT = T2-T1, and sort them in ascending order of parking time TT;

[0089] 2) Calculate the average parking time of M vehicles and standard deviation , standardize TT, ;

[0090] 3) Delete Abnormal data;

[0091] 4) Use percentile method to sort the remaining Divide into three equal parts and find the two quantiles of 33% and 67%. and , the standardized parking time is divided into three intervals. The first interval , corresponding to the vehicle storage in the high efficiency area; the second interval is , corresponding to the vehicle storage in the medium efficiency zone; the third interval is , corresponding vehicles are stored in low-efficiency areas.

[0092] 5) Predicting the length of stay Standardize to get , according to its corresponding interval, it is allocated to the corresponding energy efficiency area.

[0093] 6) If there are no free parking spaces in the energy efficiency zone to which the vehicle is assigned, the vehicle will be assigned to another energy efficiency zone with free parking spaces. The specific allocation plan is as follows: For vehicles assigned to the high-efficiency zone, if there are no free parking spaces in the high-efficiency zone, they will be assigned to the medium-efficiency zone first. If there are no free parking spaces in the medium-efficiency zone, they will be assigned to the low-efficiency zone; For vehicles assigned to the low-efficiency zone, if there are no free parking spaces in the low-efficiency zone, they will be assigned to the medium-efficiency zone first. If there are no free parking spaces in the medium-efficiency zone, they will be assigned to the high-efficiency zone; For vehicles assigned to the medium-efficiency zone, if there are no free parking spaces in the medium-efficiency zone, it depends on the vehicle's Value and 、 The distance between the values, if , it is allocated to the low-efficiency area first, otherwise it is allocated to the high-efficiency area.

[0094] Step 3.5: If there is an idle elevator 1, the driver will drive to the waiting point 9 of the corresponding idle elevator according to the instructions on the third screen 10.

[0095] Step 3.6: When the lift 1 reaches the car hall floor, the garage automatic door 7 opens, and the driver drives the car into the garage automatic door 7, parks the car, and exits the garage. After the driver clicks the first screen 6 to confirm the entry operation, the vehicle officially enters the entry process.

[0096] Step 3.7: The vehicle moves vertically with the elevator 1 through the elevator channel 3 to the assigned parking floor and connects with the idle horizontal conveyor 2 on that floor.

[0097] Step 3.8: After the transverse conveyor 2 moves the vehicle out of the elevator 1 by its own device, it moves horizontally on the handling channel 4 to the designated storage berth 5. If there is no idle transverse conveyor 2 on this floor, the vehicle waits for the transverse conveyor in the elevator 1.

[0098] Step 4, such as Figure 4 As shown, during the low-peak period of warehousing, vehicles in the medium-efficiency area and the low-efficiency area are transferred to the warehouse.

[0099] Step 4.1: Count the traffic in and out of the warehouse every hour of every day over a period of time, and calculate the average traffic flow in each hour. , ,For example The average inbound vehicle flow rate between 2:00 AM and 3:00 AM. In this embodiment, the hourly average inbound vehicle flow rate for 30 days is calculated.

[0100] Calculate the mean and standard deviation:

[0101] ;

[0102] ;

[0103] when It is the peak period for storage. When It is the low peak period for storage; for Average inbound vehicle flow in the hourly period, , is the average of the average inbound vehicle flow in 24 hours a day, is the constant term coefficient, is the standard deviation of the average inbound vehicle flow rate over 24 hours a day.

[0104] During actual calculations, the statistical time range can be determined based on historical traffic conditions, decision-maker needs and / or the opening hours of the intelligent multi-story parking garage, such as 6:00 a.m. to 6:00 p.m., 5:00 a.m. to 8:00 p.m., etc.

[0105] Step 4.2: Sort the remaining parking time of vehicles in the current medium and low efficiency areas from short to long, and determine whether the vehicle is assigned to a more efficient area. The remaining parking time of a vehicle is its predicted parking time minus the actual parking time. Judgment condition: Assume that the remaining parking time of the vehicle is ,Will To standardize the process, ,when , and the current vehicle is in the low efficiency area, it is allocated to the medium efficiency area. Similarly, when If the vehicle is not in the high-efficiency area, it will be allocated to the high-efficiency area.

[0106] Step 4.3: If the vehicle is assigned to a higher efficiency area, the garage system will determine whether there is an available parking space in the reallocated efficiency area; if there is an available parking space, the vehicle will be moved to the garage; otherwise, the vehicle will wait in place.

[0107] Step 4.4: Repeat steps 4.2 to 4.3 until all vehicles that meet the requirements are moved to the high-efficiency area, or the high-efficiency area is full and no vehicles in the low-efficiency area meet the requirements for transfer.

[0108] Step 5, such as Figure 5 As shown, the driver picks up the vehicle and the vehicle leaves the warehouse.

[0109] Step 5.1: The driver clicks the "Get Car" button on the second screen 8, scans the QR code and pays, and the vehicle officially enters the outbound process.

[0110] Step 5.2: Transverse conveyor 2 moves horizontally along the transport aisle 4 to the vehicle's currently parked storage space 5, then uses its own equipment to move the vehicle out of the current storage space 5. Transverse conveyor 2 moves horizontally along the transport aisle 4 and connects to elevator 1. If there is no idle elevator 1, the vehicle waits on transverse conveyor 2.

[0111] Step 5.3: Lift 1 moves vertically through liftway 3 to the parking floor where the vehicle to be unloaded is located. Transverse transporter 2 uses its own device to move the vehicle onto lift 1. Lift 1 moves vertically through liftway 3 to the parking hall floor.

[0112] Step 5.4: When the elevator rises to the car lobby floor, the garage automatic door 7 opens. The driver enters the garage automatic door 7 and drives the vehicle out, and the vehicle leaves the garage.

[0113] The method of the present invention is verified by specific experiments below.

[0114] The improved XGBoost algorithm used in this paper was compared with the support vector regression (SVR) algorithm and the long short-term memory (LSTM) algorithm for predicting parking duration. The results are shown in Table 2. It can be seen that the improved XGBoost algorithm of this invention achieved the best prediction results. Compared with the SVR algorithm, the improved XGBoost algorithm reduced the RMSE by 84.48% and the MAE by 83.80%. Compared with the LSTM algorithm, the improved XGBoost algorithm reduced the RMSE by 88.09% and the MAE by 46.23%.

[0115] Table 2 Comparison of prediction results of different algorithms

[0116]

[0117] In order to observe the differences in vehicle dispatching results under different methods, this embodiment focuses on Figure 1 The intelligent parking garage F shown in the figure has 624 vehicle entry and exit tasks, of which 312 vehicles are parked and 312 are retrieved. The analysis is conducted from 6:00 AM to 6:00 PM. Peak entry times are 8:00 AM to 9:00 AM and 2:00 PM to 3:00 PM, while 12:00 PM to 1:00 PM is off-peak. The remaining hours are off-peak. Figure 6 and Figure 7 3 and 4 are schematic diagrams of simulation results according to the traditional allocation method and the simulation results according to the method of the present invention. Table 3 shows the comparison of simulation results of the traditional allocation method and the simulation results of the method of the present invention.

[0118] Table 3 Comparison of simulation results

[0119]

[0120] The differences between the two methods in terms of average fullness, maximum fullness, average operation time, and average waiting time are mainly concentrated in high-efficiency areas and low-efficiency areas. Compared with the current proximity allocation method, the average fullness and maximum fullness of the high-efficiency area under the method of the present invention decreased by 22.6% and 22.84% respectively, while the average fullness and maximum fullness of the low-efficiency area increased by 353.82% and 280.17% respectively. Overall, the standard deviations of the average fullness and maximum fullness of each energy efficiency area under the method of the present invention were 7.38 and 9.62 respectively, while the standard deviations of the average fullness and maximum fullness of each energy efficiency area under the proximity allocation method were 13.29 and 34.43 respectively. In contrast, the proximity allocation scheduling method will lead to uneven resource utilization in each energy efficiency area, and high-efficiency areas may be full. However, the fullness of each layer under the method of the present invention is more balanced, avoiding extreme fullness. In the long run, the vehicle scheduling of the present invention is more scientific and reasonable.

[0121] Furthermore, the method proposed in this paper also reduced average customer waiting time. Increasing the number of incoming vehicles to 454 vehicles reduced the average customer waiting time from 462.33 seconds to 427.33 seconds, a decrease of 7.57%. Therefore, considering the combined effects of fullness and average customer waiting time, the scheduling method proposed in this paper, based on storage space energy efficiency and duration prediction, is more suitable for stereo parking systems.

[0122] To explore the universality of the method proposed in this invention, simulation experiments were conducted for different vehicle flows and numbers of hardware facilities, and the service performance of the garage under different numbers of hardware facilities was explored when the inbound vehicle flows were 312, 454, and 618.

[0123] Figure 8 The figure shows the comparison of the effect of different numbers of elevators and transverse conveyors on the average operation time under the same traffic volume. Figure 8 As can be seen, under the same traffic volume, the number of elevators and transverse conveyors has little effect on average operation time. However, when traffic volume increases, the average operation time of the garage also increases. Since the movement of elevators and transverse conveyors within the garage is almost unaffected by traffic volume, this indicates that increased traffic volume will increase the waiting time for coordination between elevators and transverse conveyors.

[0124] Figure 9 The figure shows the comparison of the effect of different numbers of elevators and transverse conveyors on the average waiting time under the same traffic volume. Figure 9 As can be seen, both 3 elevators and 10 transverse conveyors and 4 elevators and 10 transverse conveyors can meet the parking needs of customers with vehicle volumes of 312 and 454, respectively. At a vehicle volume of 618, 4 elevators and 10 transverse conveyors reduced the average waiting time by 4031.59 seconds compared to 4 elevators and 5 transverse conveyors, a reduction of 86.85%. There was no significant difference in the average waiting time between 3 elevators and 10 transverse conveyors and 4 elevators and 10 transverse conveyors, indicating that increasing the number of transverse conveyors can significantly reduce customer waiting time.

[0125] The intelligent parking garage scheduling system based on storage space energy efficiency and duration prediction of the present invention comprises: after a vehicle to be parked arrives at the entrance of the intelligent parking garage, an elevator delivers the vehicle to a designated floor, and a transverse transporter moves the vehicle to a designated storage berth. The system comprises:

[0126] The energy efficiency zone division unit is used to calculate the storage time of vehicles from the entrance to each storage berth, and divide all storage berths into three energy efficiency zones according to the storage time from the smallest to the largest, namely the high efficiency zone, the medium efficiency zone and the low efficiency zone;

[0127] An energy efficiency zone matching unit is configured to obtain the historical stay times of several vehicles and sort them in ascending order, normalize the historical stay times and divide them into three equal intervals, namely, a first interval, a second interval, and a third interval, which correspond to vehicles stored in a high-efficiency zone, a medium-efficiency zone, and a low-efficiency zone, respectively;

[0128] The vehicle dwell time prediction unit is used to retrieve the historical parking data features of several vehicles, including the vehicle arrival date, arrival time, departure date and departure time, and use this historical parking data to train the XGBoost algorithm to predict the vehicle dwell time. When a parked vehicle arrives at the entrance of the intelligent parking garage, the trained XGBoost algorithm is used to obtain the predicted vehicle dwell time based on the vehicle's historical parking data.

[0129] The vehicle dispatching unit is used to standardize the predicted vehicle dwell time and allocate the vehicle to a storage berth corresponding to different energy efficiency zones for storage according to the interval of the standardized predicted vehicle dwell time.

[0130] The electronic device described in the present invention includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the computer program is loaded into the processor, it implements the intelligent stereoscopic parking garage scheduling method based on storage location energy efficiency and duration prediction.

[0131] The computer-readable storage medium of the present invention stores a computer program, and when the computer program is executed by a processor, the intelligent stereoscopic parking garage scheduling method based on storage location energy efficiency and duration prediction is implemented.

[0132] The computer-readable storage media may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, flash memory, or any other medium that can be used to store program code in the form of instructions or data structures and that can be accessed by a computer.

[0133] The processor is configured to execute the computer program stored in the memory to implement the various steps in the method involved in the above embodiment.

Claims

1. An intelligent parking garage scheduling method based on storage space energy efficiency and duration prediction, characterized in that: After the vehicle to be parked arrives at the entrance of the intelligent stereo garage, the elevator sends the vehicle to the designated floor, and the horizontal transporter moves the vehicle to the designated storage berth. The intelligent stereo garage scheduling method includes the following steps: The storage time from the entrance to each storage space is calculated, and all storage spaces are divided into three energy efficiency zones, namely, high efficiency zone, medium efficiency zone, and low efficiency zone, based on the storage time from the entrance to each storage space. The storage time from the entrance to each storage space includes the time it takes for the elevator to pick up the vehicle from the entrance, the time it takes for the elevator to move vertically to the designated floor, and the time it takes for the horizontal conveyor to move the vehicle to the designated storage space. Obtain the historical dwell time of several vehicles and sort them in ascending order. Standardize the historical dwell time and divide it into three equal intervals, namely, the first interval, the second interval, and the third interval, which correspond to vehicles stored in the high-efficiency zone, the medium-efficiency zone, and the low-efficiency zone, respectively. Re-acquire the historical parking data features of several vehicles, including vehicle arrival date, arrival time, departure date and departure time, and use this historical parking data to train the XGBoost algorithm to predict vehicle parking duration; When a vehicle arrives at the entrance of the smart parking garage, the trained XGBoost algorithm is used to predict the vehicle's parking duration based on the vehicle's historical parking data. The predicted vehicle dwell time is standardized, and according to the interval of the standardized predicted vehicle dwell time, the vehicle is allocated to the storage berths corresponding to different energy efficiency areas for storage. The specific method includes: firstly judging whether there are vacant berths in the intelligent stereoscopic parking garage; if there are vacant berths, the driver parks the vehicle at the waiting point corresponding to the elevator and leaves; otherwise, the driver drives the vehicle away; the elevator moves the vehicle to the floor where the storage berths in the energy efficiency area are located, and the transverse transporter removes the vehicle from the elevator and moves it to the storage berths in the energy efficiency area; during this period, if there are no vacant elevators and / or transverse transporters, the vehicle waits in place until the elevators and / or transverse transporters are vacant; if there are no vacant storage berths in the energy efficiency area to which it is allocated, the vehicle is allocated to the remaining energy efficiency areas with vacant storage berths; During the low-peak period of parking, vehicles in the medium-efficiency zone and the low-efficiency zone are moved to the parking lot. The method is as follows: the remaining parking time of the vehicles in the medium-efficiency zone and the low-efficiency zone is calculated, the remaining parking time is standardized, and the vehicles are moved to the high-efficiency zone or the medium-efficiency zone according to the range of the standardized remaining parking time. The low-peak period of parking is the period when the average parking volume is not greater than the threshold. The parking time period classification method includes: when It is the peak period for storage. When It is the low peak period for storage; The first Average inbound vehicle flow in the hourly period, is the average value of the average inbound vehicle flow within the statistical time range, is the constant term coefficient, is the standard deviation of the average inbound vehicle flow within the statistical time range.

2. The intelligent parking garage scheduling method based on storage space energy efficiency and duration prediction according to claim 1 is characterized in that: The features of the historical parking data of the vehicles that are retrieved also include: whether the arrival date and departure date of the vehicle are weekdays or holidays, the number of times the vehicle enters the parking lot, and the province and city to which the vehicle license plate number belongs.

3. The intelligent parking garage scheduling method based on storage space energy efficiency and duration prediction according to claim 1 is characterized in that: Before using the historical parking data to train the XGBoost algorithm to predict the vehicle parking time, the method also includes: preprocessing the historical parking data to eliminate abnormal data.

4. The intelligent parking garage scheduling method based on storage space energy efficiency and duration prediction according to claim 3 is characterized in that: The method for eliminating abnormal data includes: calculating the interquartile range (IQR) of the box plot of historical parking data; when the historical parking data is less than or greater than When , it is determined to be abnormal data, where Q1 is the lower quartile and Q3 is the upper quartile.

5. An intelligent parking garage scheduling system based on storage space energy efficiency and duration prediction, characterized by: After the vehicle to be parked arrives at the entrance of the intelligent stereo garage, the elevator will take the vehicle to the designated floor, and the horizontal transporter will move the vehicle to the designated storage berth. The system includes: An energy efficiency zone division unit is used to calculate the storage time of a vehicle from the entrance to each storage berth, and divide all storage berths into three energy efficiency zones, i.e., a high efficiency zone, a medium efficiency zone, and a low efficiency zone, based on the storage time from the entrance to each storage berth. The storage time of a vehicle from the entrance to each storage berth includes the time it takes for the elevator to pick up the vehicle from the entrance, the time it takes for the elevator to move vertically to the designated floor, and the time it takes for the horizontal conveyor to move the vehicle to the designated storage berth. An energy efficiency zone matching unit is configured to obtain the historical stay times of several vehicles and sort them in ascending order, normalize the historical stay times and divide them into three equal intervals, namely, a first interval, a second interval, and a third interval, which correspond to vehicles stored in a high-efficiency zone, a medium-efficiency zone, and a low-efficiency zone, respectively; The vehicle dwell time prediction unit is used to retrieve the historical parking data features of several vehicles, including the vehicle arrival date, arrival time, departure date and departure time, and use this historical parking data to train the XGBoost algorithm to predict the vehicle dwell time. When a parked vehicle arrives at the entrance of the intelligent parking garage, the trained XGBoost algorithm is used to obtain the predicted vehicle dwell time based on the vehicle's historical parking data. The vehicle dispatching unit is used to standardize the predicted vehicle dwell time, and allocate the vehicle to storage berths corresponding to different energy efficiency areas for storage according to the interval of the standardized predicted vehicle dwell time. The specific method includes: firstly judging whether there are vacant berths in the intelligent stereo garage; if there are vacant berths, the driver parks the vehicle at the waiting point corresponding to the elevator and leaves; otherwise, the driver drives the vehicle away; the elevator moves the vehicle to the floor where the storage berths of the energy efficiency area are located, and the transverse transport removes the vehicle from the elevator and moves it to the storage berths of the energy efficiency area; during this period, if there are no vacant elevators and / or transverse transports, the vehicle waits in place until the elevators and / or transverse transports are vacant; if there are no vacant storage berths in the energy efficiency area to which it is allocated, the vehicle is allocated to the remaining energy efficiency areas with vacant storage berths; During the low-peak period of parking, vehicles in the medium-efficiency zone and the low-efficiency zone are moved to the parking lot. The method is as follows: the remaining parking time of the vehicles in the medium-efficiency zone and the low-efficiency zone is calculated, the remaining parking time is standardized, and the vehicles are moved to the high-efficiency zone or the medium-efficiency zone according to the range of the standardized remaining parking time. The low-peak period of parking is the period when the average parking volume is not greater than the threshold. The parking time period classification method includes: when It is the peak period for storage. When It is the low peak period for storage; The first Average inbound vehicle flow in the hourly period, is the average value of the average inbound vehicle flow within the statistical time range, is the constant term coefficient, is the standard deviation of the average inbound vehicle flow within the statistical time range.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the computer program is loaded into the processor, the intelligent stereoscopic parking garage scheduling method based on storage space energy efficiency and duration prediction according to any one of claims 1 to 4 is implemented.

7. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the intelligent stereoscopic parking garage scheduling method based on storage space energy efficiency and duration prediction according to any one of claims 1 to 4 is implemented.

Images