Station traffic long-term prediction method and device and computer equipment

Abstract

The invention provides a station flow long-term prediction method and device and computer equipment, and the method comprises the steps: firstly calculating the year-on-year trend ratio of the product of the current passenger flow of a reference day and the average passenger flow before a prediction day relative to the average passenger flow before the reference day after obtaining necessary data, and then carrying out the prediction according to historical weather data and historical passenger flow data, a regression analysis method is adopted, flow influence coefficients under different weathers are obtained through fitting, then a first station flow prediction model is constructed by adopting a multi-condition summation method according to a year-on-year trend ratio and the flow influence coefficients, and then a passenger flow increment ratio generated by activities of a prediction day is calculated by taking the current day passenger flow of a reference day as a benchmark according to activity characteristics of the prediction day. And finally, according to the first station flow prediction model and the standardized passenger flow increment ratio, integrating to form a second station flow prediction model, and calculating the current-day passenger flow of the prediction day by adopting the second station flow prediction model. Compared with the prior art, the method is more suitable for long-term prediction of predictable events.

Description

Technical Field

[0001] The present invention belongs to the technical field of traffic data processing, and in particular relates to a method, a device and a computer device for long-term prediction of site flow. Background Art

[0002] As a basic task in the operation of subway rail systems, vehicle flow prediction is mainly used to assist in the design of train operation diagrams and the planning of subway lines and schedules. In addition, through vehicle flow prediction related methods and supporting APPs, travelers can also be informed of the flow of people in advance so that they can make travel plans in advance.

[0003] Most of the existing technologies for predicting passenger flow in stations are dedicated to short-term prediction, mainly collecting continuous passenger flow data in the previous time period to predict future short-term flow. This short-term prediction method will make the adjustment time of operators and management strategies (adjusting train schedules, train routes, and flow control arrangements, etc.) too short, which is not conducive to timely response. When there are some foreseeable events in urban activities (holidays, weather or activities), the accuracy of the prediction will be greatly reduced because the collected data and the data to be predicted are not continuous in time. Summary of the invention

[0004] The present invention provides a method, device and computer equipment for long-term prediction of site traffic, which are used to solve the problems existing in the prior art.

[0005] In a first aspect, in order to achieve the above-mentioned object of the invention, the present invention provides a method for long-term prediction of site traffic, comprising: S10: obtaining a reference day, the passenger flow on the reference day, the average passenger flow before the reference day, the average passenger flow before the forecast day, the activity characteristics of the forecast day, and the weather characteristics of the forecast day according to the date characteristics of the forecast day; The average passenger flow before the reference date refers to the average passenger flow in the scheduled time period before the reference date, and the average passenger flow before the forecast date refers to the average passenger flow in the scheduled time period before the forecast date; S20: Calculate the year-on-year trend ratio of the product of the passenger flow on the reference day and the average passenger flow on the forecast day relative to the average passenger flow on the reference day; S30: Based on the historical weather data of any day before the forecast date and the corresponding historical passenger flow, a regression analysis method is used to fit the flow influence coefficient under different weather conditions; S40: Based on the year-on-year trend ratio, flow impact coefficient and weather characteristics of the forecast day, a multi-condition summation method is used to construct a flow prediction model for the first station; S50: according to the activity characteristics of the forecast day, taking the passenger flow of the reference day as a benchmark, calculating the passenger flow increment ratio caused by the activity on the forecast day, and performing standardization on the passenger flow increment ratio and its time label; S60: A second site traffic prediction model is formed based on the first site traffic prediction model and the normalized passenger flow increment ratio, and the second site traffic prediction model is used to calculate the passenger flow on the prediction day.

[0006] Through the above technical scheme, the method of the present invention fully considers the impact of foreseeable events (holidays, weather, activities) on passenger flow, does not rely on real-time traffic data, can support long-term predictions many days in advance, and solves the problem of reduced prediction accuracy in existing short-term predictions due to the discontinuity between the short-term period and the prediction period.

[0007] Preferably, the calculation formula of step S20 is:

[0008] in, Representative forecast day i Year-on-year trend ratio; Represents the average passenger flow predicted on the previous day; Represents the average passenger flow on the previous day. Representative reference date r The daily passenger flow.

[0009] The first site traffic prediction model in step S40 is specifically: in, The coefficient obtained by fitting in step S30 is the flow influence coefficient; Forecast day i The first item of passenger flow on the day; Represents sunny days during the holiday season; Represents rainy days during the holiday season. Indicates the day when the weather is predicted to be rainy j The year-on-year trend ratio.

[0010] Preferably, the calculation model of step S50 is:

[0011] in, For the number of people participating in the activity, is the normalized passenger flow increment ratio over time, s The start time of the activity. The time difference between the activity end time and the activity start time.

[0012] Preferably, the second site traffic prediction model in step S60 is specifically:

[0013] in, Representative forecast day iThe daily passenger flow.

[0014] Preferably, the reference date includes working days, public holidays and statutory holidays, and the predicted date includes working days, public holidays and statutory holidays. If the predicted date is a statutory holiday, the time interval between the reference date and the predicted date is one year.

[0015] Preferably, the predicted daily activity characteristics include at least the activity type, activity venue, number of participants, activity start time and activity end time.

[0016] In a second aspect, in order to achieve the above-mentioned object of the invention, the present invention further provides a long-term prediction device for site traffic to implement the above-mentioned long-term prediction method for site traffic, the device comprising: A data acquisition module is used to acquire a reference day, as well as passenger flow on the reference day, average passenger flow before the reference day, average passenger flow before the forecast day, activity characteristics on the forecast day, and weather characteristics on the forecast day according to date characteristics of the forecast day; A trend ratio calculation module, used to calculate the year-on-year trend ratio of the product of the passenger flow on the reference day and the average passenger flow on the forecast day relative to the average passenger flow on the reference day; The influence coefficient calculation module is used to fit the flow influence coefficient under different weather conditions by using regression analysis method based on the historical weather data of any day before the forecast date and the corresponding historical passenger flow; A model building module is used to build a flow prediction model for the first station by using a multi-condition summation method based on the year-on-year trend ratio, flow impact coefficient and weather characteristics on the forecast day; The standardization module is used to calculate the passenger flow increment ratio generated by the activity on the forecast day based on the activity characteristics of the forecast day and the passenger flow on the reference day as the benchmark, and to standardize the passenger flow increment ratio and its time label; The prediction module is used to integrate the first site traffic prediction model and the standardized passenger flow increment ratio to form a second site traffic prediction model, and use the second site traffic prediction model to calculate the daily passenger flow on the prediction day.

[0017] In a third aspect, in order to achieve the above-mentioned purpose of the invention, the present invention also provides a computer device, including an internal memory, a processor, a non-volatile storage medium, and a computer program stored in the non-volatile storage medium and executable on the processor, wherein the processor implements the above-mentioned site traffic long-term prediction method when executing the computer program.

[0018] The present invention has at least the following advantages: 1) Compared with the prior art, the method of the present invention does not rely on real-time traffic data, can support long-term predictions for many days in advance, solves the problem of reduced prediction accuracy in existing short-term predictions due to the discontinuity between the short-term period and the prediction period, and has higher prediction accuracy.

[0019] 2) Compared with the existing technology, the method of the present invention fully considers the impact of foreseeable events (holidays, weather, activities) on passenger flow, solves the prediction problem of multiple events occurring simultaneously, is more in line with the actual situation, and has stronger universality. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0021] Figure 1 This is a flow chart of the long-term prediction method for site traffic in Embodiment 1 of the present invention.

[0022] Figure 2 It is a schematic diagram of the structure of the long-term prediction device for site traffic in the second embodiment of the present invention.

[0023] Figure 3 It is a schematic diagram of the structure of the computer device in the third embodiment of the present invention. DETAILED DESCRIPTION

[0024] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] This embodiment discloses a method, device and computer equipment for long-term prediction of site traffic, which are used to solve the problems existing in the prior art.

[0026] Embodiment 1 like Figure 1 As shown, this embodiment discloses a method for long-term prediction of site traffic, including: S10: Obtain a reference day according to the date characteristics of the forecast day, as well as the passenger flow on the reference day, the average passenger flow before the reference day, the average passenger flow before the forecast day, the activity characteristics of the forecast day, and the weather characteristics of the forecast day.

[0027] In this embodiment, the reference day includes working days, public holidays and statutory holidays, and the predicted day includes working days, public holidays and statutory holidays. When the predicted day happens to be a statutory holiday, the reference day corresponding to the predicted day takes the corresponding statutory holiday date of the previous year. When the predicted day happens to be a public holiday, the reference day corresponding to the predicted day can take the corresponding public holiday date of the previous year or the public holiday of the previous week. In addition, if the predicted day is a statutory holiday, it needs to meet the following three conditions to match the corresponding reference day, namely: holiday name, total number of holiday days and the number of days of holiday. The predicted day activity features include activity type (such as concerts, sports events, large-scale trade activities, large-scale promotional activities, etc.), activity venue (such as gymnasiums, supermarkets, squares, exhibition halls, etc.), number of people in the event (i.e., number of registrations, such as number of online ticket bookings, number of scheduled exhibition participants, etc.), activity start time and activity end time. The passenger flow on the reference day is the passenger flow of the target site on the reference day, the average passenger flow on the reference day is the overall average passenger flow of the target site a few days before the reference day, and the average passenger flow on the prediction day is the overall average passenger flow of the target site a few days before the prediction day.

[0028] Specifically, the data related to the above reference days are all historical data, while the activity characteristic data of the predicted day can be obtained through the Internet.

[0029] Compared with the prior art, the present invention fully considers the impact of statutory holidays on the prediction. If the prediction date is a statutory holiday, the time interval between the reference date and the prediction date is one year, which is more in line with the actual situation and has stronger universality.

[0030] S20: Calculate the year-on-year trend ratio of the product of the passenger flow on the reference day and the average passenger flow on the predicted day relative to the average passenger flow on the reference day.

[0031] Specifically, the specific calculation formula of the above step S20 is: (1) in, Representative forecast day i Year-on-year trend ratio; Representative forecast day i The predicted average passenger flow on the previous day; Representative reference date r The average passenger flow on the reference day, Representative reference date r The passenger flow on the reference day.

[0032] S30: Based on the historical weather data of any day before the forecast date and the corresponding historical passenger flow, a regression analysis method is used to fit the flow influence coefficient under different weather conditions.

[0033] Specifically, the fitting regression formula for sunny days in the above step S30 is: (2) in, is the flow influence coefficient, is the historical passenger flow on any sunny day. Correspondingly, the formula for rainy days is the same as formula (2), with only the coefficient value being different, which will not be repeated here.

[0034] It should be explained that the method of the present invention proposes two assumptions based on practical knowledge. First, rainy days will lead to a decrease in travel demand and a decrease in station passenger flow. Second, the reduced travel demand on rainy days will be transferred to the next sunny day, so that the station passenger flow on sunny days will greatly increase. Therefore, based on formula (2), step S40 can be obtained.

[0035] S40: Based on the year-on-year trend ratio, flow impact coefficient and weather characteristics on the forecast day, a multi-condition summation method is used to construct a flow prediction model for the first station.

[0036] Specifically, the first site traffic prediction model is as follows: (3) in, Represents sunny days during the holiday season; Represents rainy days during the holiday season. Indicates the day when the weather is predicted to be rainy j The year-on-year trend ratio.

[0037] After the above step S40, the method of the present invention further includes: S50: According to the activity characteristics of the forecast day, taking the passenger flow of the reference day as a benchmark, the passenger flow increment ratio generated by the activity on the forecast day is calculated, and the passenger flow increment ratio and its time label are standardized.

[0038] In this embodiment, there is no relevant activity held on the reference day relative to the predicted day.

[0039] Specifically, a conventional polynomial regression model can be used to predict the peak activity period and peak passenger flow within the target station based on the activity type, activity venue, number of registered activity participants, activity start time, and activity end time. Based on the passenger flow on a reference day without an activity, the passenger flow increment ratio generated by the activity on the predicted day is calculated, and the passenger flow increment ratio and its time label are standardized to predict the passenger flow status under another activity.

[0040] It should be explained that the incremental traffic ratio is the ratio of the incremental traffic generated by the activity to the number of participants. Since different activities have different start and end times, the time tags need to be standardized.

[0041] Specifically, the calculation model of the above step S50 is: (4) in, For the number of people participating in the activity, is the normalized passenger flow increment ratio over time, s The start time of the activity. The time difference between the activity end time and the activity start time.

[0042] S60: A second site traffic prediction model is formed based on the first site traffic prediction model and the normalized passenger flow increment ratio, and the second site traffic prediction model is used to calculate the passenger flow on the prediction day.

[0043] Specifically, the second site traffic prediction model in step S60 is specifically:

[0044] in, Representative forecast day i The daily passenger flow.

[0045] It should be noted that the flow impact coefficient of the above-mentioned second site flow prediction model can be optimized by an iterative method. After the second site flow prediction model is formed, it is only necessary to substitute the relevant data of the corresponding reference day and the known data of the prediction day to obtain the passenger flow on the prediction day. The statistical method proposed by the method of the present invention includes three steps, integrating three factors respectively. First, the annual trend statistics are evaluated by determining the reference day of the previous year and deriving trend parameters from the site flow on non-holidays. Secondly, the weather parameters are calibrated according to historical data, and the flow transfer method is applied to establish the relationship between the same holiday day and another day, so that travel during the holiday becomes a whole. Third, activity parameters are learned from historical activities to quantify the incremental site flow caused by the activity and its distribution along time.

[0046] Through the above steps S10-S60, compared with the prior art, the method of the present invention has the following beneficial effects: 1. This technical solution is applicable to site traffic prediction under foreseeable events (holidays, weather, activities).

[0047] 2. The method of the present invention is more applicable under foreseeable events. Without relying on real-time data, in the scenario where holidays, rainstorms, and concerts occur simultaneously, the hourly flow prediction accuracy of 57.65% of the stations exceeds 80%, and the hourly flow prediction accuracy of 84.36% of the stations exceeds 70%.

[0048] 3. The method of the present invention can realize the traffic volume prediction of holidays by time and station, and provide fine-grained holiday travel status prediction.

[0049] Embodiment 2 like Figure 2 As shown, this embodiment discloses a long-term prediction device for site traffic to implement the long-term prediction method for site traffic in embodiment 1, and the device includes: A data acquisition module is used to acquire a reference day, as well as passenger flow on the reference day, average passenger flow before the reference day, average passenger flow before the forecast day, activity characteristics on the forecast day, and weather characteristics on the forecast day according to date characteristics of the forecast day; A trend ratio calculation module, used to calculate the year-on-year trend ratio of the product of the passenger flow on the reference day and the average passenger flow on the forecast day relative to the average passenger flow on the reference day; The influence coefficient calculation module is used to fit the flow influence coefficient under different weather conditions by using regression analysis method based on the historical weather data of any day before the forecast date and the corresponding historical passenger flow; A model building module is used to build a flow prediction model for the first station by using a multi-condition summation method based on the year-on-year trend ratio, flow impact coefficient and weather characteristics on the forecast day; The standardization module is used to calculate the passenger flow increment ratio generated by the activity on the forecast day based on the activity characteristics of the forecast day and the passenger flow on the reference day as the benchmark, and to standardize the passenger flow increment ratio and its time label; The prediction module is used to integrate the first site traffic prediction model and the standardized passenger flow increment ratio to form a second site traffic prediction model, and use the second site traffic prediction model to calculate the daily passenger flow on the prediction day.

[0050] Embodiment 3 like Figure 3 As shown, this embodiment discloses a computer device, including an internal memory, a processor, a non-volatile storage medium, and a computer program stored in the non-volatile storage medium and executable on the processor. When the processor executes the computer program, the above-mentioned site traffic long-term prediction method is implemented.

[0051] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above-mentioned method embodiment; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk, etc. Various media that can store program codes.

[0052] For those skilled in the art, various other corresponding changes and deformations can be made according to the technical solutions and concepts described above, and all of these changes and deformations should fall within the protection scope of the claims of the present invention.

Claims

1. A method for long-term prediction of site traffic, characterized in that: include: S10: obtaining a reference day, the passenger flow on the reference day, the average passenger flow before the reference day, the average passenger flow before the forecast day, the activity characteristics of the forecast day, and the weather characteristics of the forecast day according to the date characteristics of the forecast day; The average passenger flow before the reference date refers to the average passenger flow in a predetermined time period before the reference date, and the average passenger flow before the forecast date refers to the average passenger flow in a predetermined time period before the forecast date; S20: Calculate the year-on-year trend ratio of the product of the passenger flow on the reference day and the average passenger flow before the predicted day relative to the average passenger flow before the reference day; S30: Based on the historical weather data of any day before the forecast date and the corresponding historical passenger flow, a regression analysis method is used to fit the flow influence coefficient under different weather conditions; S40: constructing a first site flow prediction model by using a multi-condition summation method according to the year-on-year trend ratio, the flow impact coefficient and the weather characteristics of the forecast day; S50: According to the activity characteristics of the forecast day, taking the passenger flow of the reference day as a benchmark, calculating the passenger flow increment ratio generated by the activity on the forecast day, and normalizing the passenger flow increment ratio and its time label; S60: A second site traffic prediction model is formed based on the first site traffic prediction model and the normalized passenger flow increment ratio, and the second site traffic prediction model is used to calculate the passenger flow on the prediction day.

2. The method for long-term prediction of site traffic according to claim 1, characterized in that: The calculation formula of the year-on-year trend ratio in step S20 is: in, Representative forecast day i Year-on-year trend ratio; Represents the average passenger flow predicted on the previous day; Represents the average passenger flow on the previous day. Representative reference date r The daily passenger flow.

3. The method for long-term prediction of site traffic according to claim 1, characterized in that: The first site traffic prediction model in step S40 is specifically: in, is obtained by fitting in step S30 and is the flow influence coefficient; Forecast day i The first item of passenger flow on the day; Represents sunny days during the holiday season; Represents rainy days during the holiday season. Indicates the forecast day for rainy weather j The year-on-year trend ratio.

4. The method for long-term prediction of site traffic according to claim 3, characterized in that: The predicted daily activity characteristics include at least one or more of the activity type, activity venue, number of participants, activity start time, and activity end time.

5. The method for long-term prediction of site traffic according to any one of claim 4, characterized in that: The calculation model of step S50 is: in, For the number of people participating in the activity, is the normalized passenger flow increment ratio over time, s is the start time of the activity, is the time difference between the activity start time and the activity end time.

6. The method for long-term prediction of site traffic according to claim 5, characterized in that: The second site traffic prediction model in step S60 is specifically: in, Representative forecast day i The daily passenger flow.

7. The method for long-term prediction of site traffic according to claim 1, characterized in that: The reference date includes working days, public holidays and statutory holidays; the predicted date includes working days, public holidays and statutory holidays; if the predicted date is a statutory holiday, the time interval between the reference date and the predicted date is one year.

8. A long-term prediction device for site traffic, the long-term prediction device for site traffic being used to implement the long-term prediction method for site traffic as claimed in any one of claims 1 to 7, the device comprising: A data acquisition module is used to acquire a reference day, as well as passenger flow on the reference day, average passenger flow before the reference day, average passenger flow before the forecast day, activity characteristics on the forecast day, and weather characteristics on the forecast day according to date characteristics of the forecast day; A trend ratio calculation module, used to calculate the year-on-year trend ratio of the product of the passenger flow on the reference day and the average passenger flow before the predicted day relative to the average passenger flow before the reference day; The influence coefficient calculation module is used to fit the flow influence coefficient under different weather conditions by using regression analysis method based on the historical weather data of any day before the forecast date and the corresponding historical passenger flow; A model building module, used to build a first site flow prediction model by using a multi-condition summation method according to the year-on-year trend ratio, the flow impact coefficient and the weather characteristics of the forecast day; A standardization module, for calculating the passenger flow increment ratio generated by the activity on the predicted day based on the activity characteristics of the predicted day and taking the passenger flow on the reference day as a benchmark, and performing standardization processing on the passenger flow increment ratio and its time label; The prediction module is used to integrate the first site traffic prediction model and the standardized passenger flow increment ratio to form a second site traffic prediction model, and use the second site traffic prediction model to calculate the daily passenger flow on the prediction day.

9. A computer device, characterized in that: The method comprises an internal memory, a processor, a non-volatile storage medium, and a computer program stored in the non-volatile storage medium and executable on the processor. When the processor executes the computer program, the method for long-term prediction of site traffic as described in any one of claims 1 to 7 is implemented.

Images