Bus dispatching arrival prediction system and prediction method thereof
By setting up electronic fences and signal modules on buses and stations, collecting and processing driving data, and using self-learning models and weighted moving average method, high-precision prediction of bus departure time is achieved, and prediction accuracy in the existing technology is solved Low problems improve operational efficiency and passenger experience.
Patent Information
- Application Number
- CN202411024216.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-29
- Publication Date
- 2024-11-19
AI Technical Summary
The existing bus departure time prediction technology is not high, resulting in low operational efficiency and poor passenger experience. It is difficult to accurately predict the arrival time in cities with high uncertainty in traffic.
The electronic fence and signal module are used to collect bus driving data, preprocess and store it through the signal receiving module, the classification training module classifies the data, and builds a self-learning model. The weighted moving average method is used to calculate the average time of bus departure to the station. , achieve more accurate predictions.
It improves the efficiency of bus operation and passenger travel experience, provides more accurate bus arrival time prediction information, and helps bus companies optimize scheduling.
Smart Images

Figure CN118982913A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of bus departure and arrival prediction, in particular to a bus departure and arrival prediction system and a prediction method thereof. Background Art
[0002] With the development of smart public transportation and the continuous improvement of public transportation service levels, requirements have been put forward for the prediction of bus departure and arrival times. By predicting the departure and arrival of buses, a scheduling basis can be provided for operations, improving operational efficiency. At the same time, it can provide passengers with bus arrival time prediction information to facilitate passengers' travel.
[0003] Currently, bus companies estimate how long it takes to run a bus route based on experience, such as the mileage of the bus route and the number of stops required. This results in low prediction accuracy, which in turn leads to low accuracy in bus departure from the starting station, making the waiting time uncertain for passengers and providing a poor experience. This is not conducive to passenger travel and cannot effectively improve bus dispatch efficiency.
[0004] A Chinese patent discloses a bus arrival time prediction method based on a quantile convolutional network (authorization announcement number CN112990536B). This patented technology uses two prediction modes for bus arrival time: point prediction and quantile prediction. It extracts richer feature information from historical data and better captures future uncertainties. At the same time, it can overcome the problem of sparse or sporadic unavailability of historical data for certain bus routes. By using a small amount of historical bus GPS data and known change characteristics at future times (such as month, week, hour, whether bus lanes are open, etc.), it is possible to learn the complex mode of bus operation and obtain multi-step prediction results for a period of time in the future. However, there is uncertainty in urban transportation. After the bus is dispatched, the arrival time cannot be accurately predicted, which leads to poor operating efficiency of the bus company. Summary of the invention
[0005] The purpose of the present invention is to provide a bus departure and arrival prediction system and a prediction method thereof to solve the problems raised in the above-mentioned background technology.
[0006] To achieve the above object, the present invention provides the following technical solutions:
[0007] A bus departure and arrival prediction system includes an electronic fence set within a certain range of each bus stop, a signal transmission module installed on each bus, and a signal receiving module, a database, a classification training module and a data prediction module located in the background. The electronic fence is used for the travel data of the bus within the travel route;
[0008] The signal transmitting module is used to receive the travel data of the bus departure and arrival every day, and send it to the signal receiving module;
[0009] The signal receiving module is used to receive the driving data, pre-process and clean the driving data, remove abnormal data and invalid data, and then store them in the database to form historical data;
[0010] The classification training module is used to classify the types of historical data and output the classified historical data;
[0011] The data prediction module is used to extract various historical data classified in the database within a recent period of time and construct a self-learning model; a weight is assigned to each historical data through the self-learning model, and then the average time for bus departure and arrival is calculated based on the assigned weight, so as to predict the departure and arrival time of a bus line throughout the day.
[0012] As a further solution of the present invention: the specific method of electronic fence collection is as follows:
[0013] S11, first the electronic fence identifies the time point when each bus enters and exits the electronic fence, and obtains the stop time of each bus;
[0014] S12, by identifying the time point when each bus leaves the current electronic fence and the time point when it enters the electronic fence of the next bus stop,
[0015] S13. Calculate the average travel speed and travel time between stations based on the distance between stations to form travel data.
[0016] As a further solution of the present invention: the historical data stored in the database includes the mileage between stations, the duration of stop, the average travel speed between stations and the duration of travel between stations.
[0017] As a further solution of the present invention: in the classification training module, the types of the classified historical data include weather type, date type, time period type and road section mileage type.
[0018] As a further solution of the present invention: the average time of the bus departure and arrival is calculated using the weighted moving average method, and the calculation formula is as follows:
[0019]
[0020] In the above formula (1), MWA(i) represents the average time between bus departure and arrival; n represents the number of calendar days from the prediction day to the extraction of historical data; x i represents the historical data from the farthest calendar day from the forecast day to the i-th day; w i represents the weight of the historical data from the farthest calendar day from the forecast day to the i-th day, where w i Meet w1+
[0021] w2+...+w n =1.
[0022] As a further solution of the present invention: the weight assigned to each item of the historical data is determined according to the date of data collection; wherein, the data closer to the prediction date is assigned a greater weight, and the data farther from the prediction date is assigned a smaller weight; the weight of each item is determined according to a decreasing arithmetic progression, and the calculation formula is as follows:
[0023]
[0024] In the above formula (2), w i It represents the weight of historical data from the farthest time to the i-th day, and j represents the number of calendar days from the prediction day to extract historical data.
[0025] A prediction method for a bus departure and arrival prediction system comprises the following steps:
[0026] S1. First, collect the driving data of the bus within the driving route through the electronic fence; and send the driving data to the signal receiving module through the signal transmitting module;
[0027] S2, pre-processing and cleaning the driving data through the signal receiving module, removing abnormal data and invalid data, and storing them in the database to form historical data;
[0028] S3. Then, the types of historical data are classified through the classification training module, and the classified historical data are output;
[0029] S4. Then, the data prediction module is used to extract the classified historical data in the database in the recent period, and a self-learning model is constructed. A weight is assigned to each historical data through the self-learning model, and then the average time of bus departure and arrival is calculated according to the assigned weight, and a departure and arrival prediction model is obtained. Then, a complete departure and arrival model of a bus route throughout the day is established based on the prediction model.
[0030] Compared with the prior art, the present invention has the following beneficial effects:
[0031] The present invention first collects the driving data of bus departure and arrival, and stores the processed driving data in a database to form historical data; then classifies the types of historical data and constructs a self-learning model; assigns weights to each historical data through the self-learning model, and then calculates the average time of bus departure and arrival according to the assigned weights to obtain a departure and arrival prediction model, and then establishes a full-day departure and arrival complete model of a bus line according to the prediction model; thereby providing a scheduling basis for operations and improving operational efficiency, and at the same time providing passengers with bus arrival time prediction information to facilitate passengers' travel. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] Figure 1 It is a structural diagram of a bus departure and arrival prediction system;
[0033] Figure 2 A schematic diagram of electronic fence data collection in a bus departure and arrival prediction system;
[0034] Figure 3 A schematic diagram of historical data classification in a bus departure and arrival prediction system;
[0035] Figure 4 A schematic diagram of weight distribution of historical data in a bus departure and arrival prediction system;
[0036] Figure 5 A curve diagram of the time consumption between stations in a prediction method of a bus departure and arrival prediction system;
[0037] Figure 6 for Figure 5 The curve chart of the time taken by the three bus shifts to go between stops;
[0038] Figure 7 A flowchart of a prediction method of a bus departure and arrival prediction system;
[0039] Figure 8 A curve diagram of a prediction model for bus departure and arrival in a prediction method of a bus departure and arrival prediction system;
[0040] Fig. 9 The curve diagram is a complete model of the departure and arrival of a bus line in a prediction method of a bus departure and arrival prediction system. DETAILED DESCRIPTION
[0041] See also Figures 1 to 6 In an embodiment of the present invention, a bus departure and arrival prediction system includes an electronic fence set within a certain range of each bus stop, a signal transmission module installed on each bus, and a signal receiving module, a database, a classification training module and a data prediction module located in the background, and the electronic fence is used for the travel data of the bus within the travel route;
[0042] The signal transmitting module is used to receive the travel data of the bus every day, that is, the travel data from the first stop to the last stop, and send it to the signal receiving module;
[0043] The signal receiving module is used to receive driving data, pre-process and clean the driving data, remove abnormal data and invalid data, and then store it in the database to form historical data;
[0044] The classification training module is used to classify the types of historical data and output the classified historical data;
[0045] The data prediction module is used to extract the classified historical data in the database in the recent period and build a self-learning model; the self-learning model assigns weights to each historical data, and then calculates the average time for bus departure and arrival based on the assigned weights, so as to predict the departure and arrival times of a bus route throughout the day.
[0046] Preferably, Figure 2 As shown, the specific method of electronic fence collection is as follows:
[0047] S11, first the electronic fence identifies the time point when each bus enters and exits the electronic fence, and obtains the stop time of each bus;
[0048] S12, by identifying the time point when each bus leaves the current electronic fence and the time point when it enters the electronic fence of the next bus stop,
[0049] S13. Calculate the average travel speed and travel time between stations based on the distance between stations to form travel data.
[0050] Preferably, the historical data stored in the database includes mileage between stations, stop duration, average travel speed between stations and travel duration between stations.
[0051] Preferably, in the classification training module, the types of the classified historical data include weather type, date type, time period type and road section mileage type; Figure 3 As shown, weather type classification can be classified into sunny days, cloudy days, rainy days, and snowy days. Rainy days can be further classified into light rain, moderate rain, heavy rain, and rainstorms. The acquisition of weather conditions is not limited to official weather forecasts; date type classification can be classified from Monday to Sunday, and can be classified according to weekdays, special dates, and statutory holidays; due to the different traffic volumes in different time periods, time period types can be classified according to the morning peak from 7 to 9 a.m., the evening peak from 5 to 7 p.m., the night from 10 to 6 p.m., and the rest of the time periods; due to the different distances between stations, when the distance between stations is long, the number of stations within the same mileage will become less, and even if other conditions of the vehicle are the same, the time required may not be the same, so the mileage type between stations can be classified according to the different mileages between stations.
[0052] Preferably, the average time of bus departure and arrival is calculated using the weighted moving average method, and the calculation formula is as follows:
[0053]
[0054] In the above formula (1), MWA(i) represents the average time between bus departure and arrival; n represents the number of calendar days from the prediction day to the extraction of historical data. For example, if the historical data within the last 120 days is extracted from the prediction day, n is 120; x i represents the historical data from the farthest calendar day from the forecast day to the i-th day. For example, if the historical data of the last 120 days is collected, then x i is the historical data from 120 days ago to the i-th day; w i represents the weight of the historical data from the farthest calendar day from the forecast day to the i-th day, where w i Satisfy w1+w2+...+w n =1.
[0055] Preferably, Figure 4 As shown in the figure, the weight assigned to each historical data is determined according to the date of data collection; the closer the data is to the prediction date, the greater the weight is assigned, and the farther the data is from the prediction date, the smaller the weight is assigned; the size of each weight is determined according to a decreasing arithmetic progression, and the calculation formula is as follows:
[0056]
[0057] In the above formula (2), w i It represents the weight of historical data from the farthest time to the i-th day, and j represents the number of calendar days from the prediction day to extract historical data.
[0058] See also Figures 7 to 9 In an embodiment of the present invention, a prediction method of a bus departure and arrival prediction system includes the following steps:
[0059] S1, first collect the driving data of the bus departure and arrival through the electronic fence; and send the driving data to the signal receiving module through the signal transmitting module;
[0060] S2, pre-processing and cleaning the driving data through the signal receiving module, removing abnormal data and invalid data, and storing them in the database to form historical data;
[0061] S3. Then, the types of historical data are classified through the classification training module, and the classified historical data are output;
[0062] S4. Then, the data prediction module is used to extract the classified historical data in the database in the recent period, and a self-learning model is constructed. A weight is assigned to each historical data through the self-learning model, and then the average time for the bus to depart and arrive is calculated according to the assigned weight, and a prediction model for the departure and arrival is obtained. Then, a complete model for the departure and arrival of a bus line throughout the day is established based on the prediction model. For example, the historical data of the last 120 days is taken, with 288 data points per day and one data point every 5 minutes.
[0063] In order to better illustrate the technical effects of the present invention, the following embodiments are described:
[0064] For example, a bus route has 13 stations, named the first station, stations A to K, and the last station. Take the historical data between 7:00 and 7:05 every Monday morning in the past month. It is found that the departure data is unbalanced, such as Figure 5 As shown;
[0065] According to the statistical interval of 5 minutes, the weight value is neutralized to calculate a neutral value. Assuming that there are 3 shifts in the historical data obtained in 5 minutes, the dates of these 3 shifts from recent to far are as follows: Figure 6 As shown; the time between the first station and station A is x i ; After allocating the weights, we get the median time between stations; then we use the weighted moving average method to calculate the average time for bus departure and arrival, and get a prediction model for departure and arrival, such as Figure 8 As shown; then, a complete model of the departure and arrival of a bus route throughout the day is established based on the prediction model Fig. 9 shown.
[0066] What is described above is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.
Claims
1. A bus departure and arrival prediction system, characterized in that: It includes an electronic fence set up within a certain range of each bus stop, a signal transmission module installed on each bus, and a signal receiving module, a database, a classification training module and a data prediction module located in the background. The electronic fence is used for the driving data of the bus within the driving route; The signal transmitting module is used to receive the travel data of the bus departure and arrival every day, and send it to the signal receiving module; The signal receiving module is used to receive the driving data, pre-process and clean the driving data, remove abnormal data and invalid data, and then store them in the database to form historical data; The classification training module is used to classify the types of historical data and output the classified historical data; The data prediction module is used to extract various historical data classified in the database within a recent period of time and construct a self-learning model; a weight is assigned to each historical data through the self-learning model, and then the average time for bus departure and arrival is calculated based on the assigned weight, so as to predict the departure and arrival time of a bus line throughout the day.
2. A bus departure and arrival prediction system according to claim 1, characterized in that: The specific method of electronic fence collection is as follows: S11, first the electronic fence identifies the time point when each bus enters and exits the electronic fence, and obtains the stop time of each bus; S12, identifying the time point when each bus leaves the current electronic fence and enters the electronic fence of the next bus stop; S13. Calculate the average travel speed and travel time between stations based on the distance between stations to form travel data.
3. A bus departure and arrival prediction system according to claim 1, characterized in that: The historical data stored in the database include mileage between stations, stop duration, average travel speed between stations and travel duration between stations.
4. A bus departure and arrival prediction system according to claim 1, characterized in that: In the classification training module, the types of the classified historical data include weather type, date type, time period type and road section mileage type.
5. A bus departure and arrival prediction system according to claim 1, characterized in that: The average time of bus departure and arrival is calculated by weighted moving average method, and the calculation formula is as follows: In the above formula (1), MWA(i) represents the average time between bus departure and arrival; n represents the number of calendar days from the prediction day to the extraction of historical data; x i represents the historical data from the farthest calendar day from the forecast day to the i-th day; w i represents the weight of the historical data from the farthest calendar day from the forecast day to the i-th day, where w i Satisfy w1+w2+...+w n =1.
6. A bus departure and arrival prediction system according to claim 5, characterized in that: The weight assigned to each item of historical data is determined according to the date of data collection; the closer the data is to the forecast date, the greater the weight is assigned, and the farther the data is from the forecast date, the smaller the weight is assigned; the weight of each item is determined according to a decreasing arithmetic progression, and the calculation formula is as follows: In the above formula (2), w i represents the weight of historical data from the farthest time to the i-th day, and j represents the number of calendar days from the prediction day to extract historical data.
7. A prediction method for implementing the bus departure and arrival prediction system according to claim 1, characterized in that: The following steps are involved: S1, first collect the driving data of the bus departure and arrival through the electronic fence; and send the driving data to the signal receiving module through the signal transmitting module; S2, pre-processing and cleaning the driving data through the signal receiving module, removing abnormal data and invalid data, and storing them in the database to form historical data; S3. Then, the types of historical data are classified through the classification training module, and the classified historical data are output; S4. Then, the data prediction module is used to extract the classified historical data in the database in the recent period, and a self-learning model is constructed. A weight is assigned to each historical data through the self-learning model, and then the average time of bus departure and arrival is calculated according to the assigned weight, and a departure and arrival prediction model is obtained. Then, a complete departure and arrival model of a bus route throughout the day is established based on the prediction model.
Citation Information
Cited By
Garbage clearance method and system
CN119494449A
A garbage collection method and system
CN119494449B
