Intelligent decision analysis system and method based on multi-source data

Abstract

The invention discloses an intelligent decision analysis system and method based on multi-source data, and belongs to the technical field of artificial intelligence. According to the invention, multi-source data related to bus operation is acquired, real-time acquisition and uploading of bus operation data are realized, data preprocessing is carried out, and the multi-source data are integrated; feature engineering is carried out, and a passenger flow prediction model is established; based on a passenger flow prediction model, obtaining predicted passenger flows in different time periods of the day, and predicting a driver state; establishing a bus scheduling model by using a deep learning model, and inputting predicted passenger flow, predicted driver state, historical data of bus operation and urban traffic condition data to obtain a bus scheduling scheme on that day; when the dispatched buses in a certain period of time of the bus route are not enough, a route from a bus station closest to a starting station of the bus route to the starting station is obtained, and idle buses are dispatched; and if no idle bus exists in the bus station, the buses of the rescue bus route allowed to be dispatched are dispatched.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an intelligent decision analysis system and method based on multi-source data. Background Art

[0002] In modern urban transportation systems, buses, as important public transportation tools, undertake a large number of passenger transport tasks. However, the complexity and variability of bus operations pose many challenges to scheduling and management. Traditional bus scheduling systems often rely on manual experience and static planning and are unable to cope with dynamically changing passenger demand and traffic conditions, which leads to the following major problems:

[0003] Traditional passenger flow forecasting methods lack the comprehensive use of multi-source data, and the forecast results are often not accurate enough to effectively guide dispatching decisions; traditional dispatching systems usually adopt preset fixed plans and cannot be flexibly adjusted according to real-time passenger demand and traffic conditions, resulting in unreasonable vehicle dispatching and long waiting times for passengers; when faced with emergencies or temporary needs, traditional dispatching systems are often slow to respond and unable to quickly dispatch idle vehicles for support, affecting the continuity of bus operations and service quality. Summary of the invention

[0004] The purpose of the present invention is to provide an intelligent decision analysis system and method based on multi-source data to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] An intelligent decision analysis method based on multi-source data comprises the following steps:

[0007] S100, obtain multi-source data related to bus operation, use the Internet of Vehicles to realize real-time collection and upload of bus operation data, pre-process the collected data through the big data platform, and integrate multi-source data;

[0008] S200. Obtain historical passenger flow data and related data affecting passenger flow in the big data platform, perform feature engineering, use the ARIMA time series model to establish a passenger flow prediction model, train the model, adjust model parameters, and use cross-validation to verify the model effect;

[0009] S300, based on the passenger flow prediction model, the predicted passenger flow in different time periods of the day is obtained, and the driver status is predicted by using the random forest algorithm based on the predicted passenger flow, time data and weather data in combination with the driver status database; the historical data of bus operation and urban traffic status data are obtained in the big data platform, and the bus scheduling model is established using the deep learning model LSTM, and the predicted passenger flow, predicted driver status, historical data of bus operation and urban traffic status data are input to obtain the bus scheduling plan for the day;

[0010] S400. Based on the daily bus dispatch plan, when there are not enough vehicles to dispatch on a bus line during a certain period of time, the number of idle buses in different bus stations is obtained, and the route from the bus station closest to the bus line's starting station to the bus line's starting station is obtained through the A* algorithm combined with weather data and urban traffic conditions data, and the idle buses are dispatched; if there are no idle buses in the bus stations, rescue bus lines are obtained by sorting them in descending order according to the size of the overlapping part of the lines, and the vehicles allowed to be dispatched are obtained according to the daily bus dispatch plan of the rescue bus lines, and the vehicles allowed to be dispatched are dispatched.

[0011] According to step S100, the multi-source data related to the operation of the bus includes: GPS positioning data, on-board sensor data, passenger flow data, time data, weather data, population density data, bus line user rating data, bus transfer data and major event data; the GPS positioning data records the latitude and longitude coordinates, speed and driving direction of the bus through the GPS device on each bus, the on-board sensor data collects information inside and outside the bus, the passenger flow data obtains the time, location and number of passengers getting on and off the bus through the passenger card swiping device, the electronic ticketing system and the in-vehicle monitoring video analysis, the time data includes a specific timestamp, the weather data is real-time weather data provided by the meteorological department, the population density data is population density data of each region, provided by the statistical department, and reflects the distribution of permanent population in a specific area, the bus line user rating data is user ratings and evaluations obtained from the user feedback system of the bus company, reflecting the passenger satisfaction with the bus line service, the bus transfer data records the information of passengers transferring between different bus lines, and the major event data records include information on major events, holidays and emergencies.

[0012] According to step S100, a GPS device, an on-board sensor and a 5G communication module are installed on each bus; the GPS device collects the location information of the bus in real time, the on-board sensor monitors the running status of the vehicle and the environmental data inside and outside the vehicle in real time, the passenger card swiping device and the monitoring system record the passenger flow data in real time, and the collected data is uploaded to the big data platform in real time through the 5G communication module, and the big data platform receives the real-time data uploaded by each bus;

[0013] The collected data is preprocessed through the big data platform to remove noise data, process missing values ​​and outliers, unify the data formats from different sources, aggregate the data in time and space, and summarize them by hour and by station; the preprocessed data is integrated with other data sources to form a unified data set.

[0014] According to step S200, the relevant data affecting passenger flow includes time data, weather data, population density data, bus line user rating data, bus transfer data and major event data; feature engineering is performed through correlation analysis to extract features of historical passenger flow data and relevant data affecting passenger flow; the correlation analysis uses the Pearson correlation coefficient ρX,Y, and the formula is as follows:

[0015]

[0016] Among them, X is the historical passenger flow data, Y is the relevant data affecting passenger flow, Cov(X,Y) is the covariance between X and Y, which measures the common change trend between the two, and σX and σY are the standard deviations of X and Y respectively.

[0017] Differential processing, model identification and model fitting are performed on the features of historical passenger flow data and related data that affect passenger flow extracted by feature engineering. The model identification is used to determine the parameters of the ARIMA model. The model fitting uses historical data to fit the ARIMA model and establish an ARIMA time series model. The mathematical formula is as follows:

[0018]

[0019] in, is the predicted passenger flow at time t, is the predicted value of passenger flow at time ti, is the predicted value of passenger flow at time tj, c is a constant term, representing the baseline value of the model, φ i is the autoregressive coefficient, which indicates the impact of passenger flow at time ti on time t, p is the order of the autoregressive part, which indicates the number of time lags of the historical passenger flow data used, and φ j is the moving average coefficient, which indicates the effect of the error term at time tj on time t, q is the order of the moving average part, which indicates the number of time lags of the historical error term used, ∈ t is the error term at time t, representing the difference between the predicted value and the actual value.

[0020] The data is divided into a training set and a validation set. The training set data is used to train the ARIMA model, and the model parameters are adjusted. The least squares method is used to optimize the parameters of the ARIMA model so that the difference between the predicted value and the actual value is minimized. The formula is as follows:

[0021]

[0022] Among them, f t is the actual passenger flow value at time t, n is the number of time periods, indicating the total number of data points used;

[0023] The cross-validation method is used to verify the model effect, and the evaluation indicators are used to evaluate the prediction performance of the model. The evaluation indicators include root mean square error RMSE and mean absolute error MAE. The formula is as follows:

[0024]

[0025] According to step S300, the driver status database includes historical driver status data, and the historical driver status data includes the driver's fatigue level, the driver's health status, the driver's working hours and the driver's technical level; extract the predicted passenger flow, time data, weather data and historical driver status data in the driver status database, use the random forest algorithm to predict the driver's status in different time periods, perform model training and model verification to obtain a trained driver status prediction model, input the predicted passenger flow, time data and weather data of the day, and predict the driver's status in different time periods of the day;

[0026] Extract time features, traffic features, and bus operation features from historical bus operation data and urban traffic status data, select important features through correlation analysis, define the number of layers of the LSTM network and the number of neurons in each layer, divide the important features into training sets and test sets, use the training set data to train the LSTM model, and use the test set data to verify the model performance;

[0027] For the extraction of time features, timestamps are extracted from the original data and converted into useful time features, such as date, hour, day of the week, season, etc. Based on historical data, morning and evening peak hours are identified and marked; combined with holiday data, holidays in historical data are marked;

[0028] For the extraction of traffic characteristics, we can extract the traffic flow information of specific road sections and time periods from the urban traffic status data, extract the real-time congestion index and historical congestion data of specific areas, identify and record traffic accidents, road construction and other events and their impact scope and time;

[0029] To extract the operation characteristics of buses, the operation trajectory, station stop time, driving speed and other information of each line are extracted from the bus GPS data; the vehicle's fuel consumption, acceleration and deceleration times, braking frequency and other data are analyzed to evaluate the vehicle's operating status; the number of people getting on and off at each station and its changing trend are extracted from historical data.

[0030] The predicted passenger flow, predicted driver status, historical bus operation data and urban traffic condition data for the day are input into the trained LSTM model, and the LSTM model outputs the bus dispatch plan for different time periods of the day, which includes the departure time, departure frequency, vehicle allocation and driver scheduling for each line.

[0031] According to step S400, the vehicle dispatching and operation status of each line are monitored in real time through the big data platform, and a line is marked when it is found that the vehicle dispatching is insufficient in a certain period of time and cannot meet the passenger demand; the number and distribution of idle buses are queried from different bus depots and stations, and the idle vehicle data of each bus depot is obtained; the weather data and urban traffic status data are integrated, and the starting point and end point of the A* algorithm are initialized according to the positions of different bus depots and stations and the starting stations of the bus lines, respectively, the starting point is a different bus depot, and the end point is the starting station of the bus line; the A* algorithm searches for the optimal path from the starting point to the end point according to the weather data and the urban traffic status data, outputs the bus depot closest to the starting station of the bus line and the corresponding optimal path, and formulates a dispatch plan;

[0032] Among them, weather data includes temperature, rainfall, wind speed, etc., which affect the vehicle's driving speed and safety; obtaining traffic condition data requires real-time monitoring of urban traffic flow, congestion and traffic incidents, and traffic condition data affects driving time and route selection.

[0033] When performing the A* algorithm, the urban road network is defined as nodes and edges, where nodes represent intersections and edges represent roads. The estimated distance from the current node to the end point is used to adjust the valuation in combination with weather and traffic conditions. The search process is as follows: the starting point is placed in an open list, the node with the lowest valuation is selected from the open list, its adjacent nodes are expanded, the actual travel time and path safety are calculated in combination with weather and traffic conditions, the optimal path is found, and the nearest bus station and the corresponding path are output.

[0034] According to the dispatch plan, idle buses are directed to depart along the optimal route to provide support to the target route. The driving conditions of dispatched vehicles are monitored in real time, and feedback is provided based on the actual situation and the A* algorithm is used for real-time adjustments.

[0035] In the dispatch plan, determine the optimal path for idle buses from the starting point to the target line, plan the departure time and estimated arrival time of the vehicle, describe the route in detail, and mark the key nodes and turning points. Command idle buses to depart along the optimal path according to the plan, use the Internet of Vehicles and GPS technology to monitor the driving conditions of the vehicles in real time, obtain information such as location, speed, and driving route, provide feedback based on real-time traffic conditions and vehicle driving conditions, use the A* algorithm to re-evaluate the path, and make adjustments when necessary.

[0036] According to step S400, if it is confirmed that there are no idle vehicles in all bus stations, the overlapping parts of different bus routes are calculated according to multi-source data, and the lines are sorted in descending order according to the size of the overlapping parts, and the lines with the largest overlapping parts are preferentially selected as rescue routes; the daily dispatch plan of the sorted rescue routes is obtained from the big data platform, and according to the dispatch plan of the rescue routes and the current demand, it is determined that a specific bus is dispatched without affecting normal operation;

[0037] Considering the current location and operating status of a specific bus, the most suitable vehicle is selected from the vehicles allowed for dispatch, and the A* algorithm is used to plan the optimal driving path from the current route to the target route for the dispatched vehicle. A dispatch plan is formulated. During the execution of the dispatch plan, feedback is provided based on the actual situation and the A* algorithm is used for real-time adjustments.

[0038] Extract the dispatch plan for the day of the sorted rescue route from the big data platform, including the vehicle location, operating status and dispatch plan, and evaluate whether it is necessary to dispatch vehicles from the rescue route in combination with current needs. Ensure that the dispatched vehicles will not affect the normal operation of the rescue route, evaluate the remaining tasks and operating status of the dispatched vehicles, and select the most suitable dispatched vehicles based on the current location, operating status and remaining tasks of the vehicles. Develop a detailed rescue dispatch plan, including departure time, driving route, estimated arrival time, etc., command the selected vehicles for dispatch according to the plan, monitor the vehicle driving conditions in real time, and ensure the effective implementation of the dispatch plan.

[0039] Intelligent decision-making analysis system based on multi-source data, including:

[0040] Multi-source data acquisition module: including: data acquisition unit and data preprocessing unit; the data acquisition unit is responsible for acquiring multi-source data related to bus operation, using the Internet of Vehicles to realize real-time collection and uploading of bus operation data, and the data preprocessing unit is responsible for preprocessing the collected data through the big data platform and integrating multi-source data;

[0041] Passenger flow prediction module: including: feature engineering unit and time series model building unit; the feature engineering unit is responsible for obtaining historical passenger flow data and related data affecting passenger flow in the big data platform, and performing feature engineering; the time series model building unit uses the ARIMA time series model to establish a passenger flow prediction model, train the model, adjust the model parameters, and use the cross-validation method to verify the model effect;

[0042] The dispatching scheme generation module includes: a passenger flow prediction unit, a driver status prediction unit, a dispatching model building unit and a dispatching scheme generation unit; wherein the passenger flow prediction unit obtains the predicted passenger flow in different time periods of the day based on the passenger flow prediction model; the driver status prediction unit uses the random forest algorithm to predict the driver status based on the predicted passenger flow, time data and weather data in combination with the driver status database; the dispatching model building unit is responsible for obtaining the historical data of bus operation and urban traffic status data in the big data platform, and using the deep learning model LSTM to establish a bus dispatching model; the dispatching scheme generation unit is responsible for inputting the predicted passenger flow, predicted driver status, historical data of bus operation and urban traffic status data to obtain the bus dispatching scheme for the day;

[0043] Emergency dispatch module: including: idle vehicle dispatch unit and rescue vehicle dispatch unit; wherein, the idle vehicle dispatch unit is based on the daily dispatch plan of the bus. When the bus line does not have enough dispatch vehicles in a certain period of time, the number of idle buses in different bus depots is obtained, and the route from the bus depot closest to the bus line starting station to the bus line starting station is obtained through the A* algorithm combined with weather data and urban traffic conditions data, and the idle buses are dispatched. The rescue vehicle dispatch unit is responsible for obtaining the rescue bus line in descending order according to the size of the overlapping part of the line when there are no idle buses in the bus depot, and obtaining the vehicles allowed to be dispatched according to the daily dispatch plan of the rescue bus line, and dispatching the vehicles allowed to be dispatched.

[0044] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0045] The present invention adopts ARIMA time series model to predict passenger flow, combines feature engineering and cross-validation methods, improves prediction accuracy, and can more accurately reflect passenger demand in different time periods.

[0046] The present invention is based on the LSTM deep learning model, combines historical data and real-time traffic condition data, and generates a daily bus dispatch plan, which can be flexibly adjusted according to dynamically changing passenger flow and traffic conditions, and optimize vehicle dispatch and driver scheduling.

[0047] When there is a shortage of vehicles, the present invention uses the A* algorithm and the method for sorting the size of the overlapping parts of the lines to quickly find the optimal scheduling plan, ensure the timeliness and efficiency of the emergency response, and improve the continuity of public transportation operations and service quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0049] Figure 1 It is a schematic diagram of the steps of the intelligent decision analysis method based on multi-source data of the present invention;

[0050] Figure 2 It is a system structure diagram of the intelligent decision analysis system based on multi-source data of the present invention. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] See also Figure 1 and Figure 2 , the present invention provides a technical solution:

[0053] According to one embodiment of the present invention, Figure 1 As shown in the schematic diagram of the steps of the intelligent decision analysis method based on multi-source data, the intelligent decision analysis method based on multi-source data includes the following steps:

[0054] S100, obtain multi-source data related to bus operation, use the Internet of Vehicles to realize real-time collection and upload of bus operation data, pre-process the collected data through the big data platform, and integrate multi-source data;

[0055] S200. Obtain historical passenger flow data and related data affecting passenger flow in the big data platform, perform feature engineering, use the ARIMA time series model to establish a passenger flow prediction model, train the model, adjust model parameters, and use cross-validation to verify the model effect;

[0056] S300, based on the passenger flow prediction model, the predicted passenger flow in different time periods of the day is obtained, and the driver status is predicted by using the random forest algorithm based on the predicted passenger flow, time data and weather data in combination with the driver status database; the historical data of bus operation and urban traffic status data are obtained in the big data platform, and the bus scheduling model is established using the deep learning model LSTM, and the predicted passenger flow, predicted driver status, historical data of bus operation and urban traffic status data are input to obtain the bus scheduling plan for the day;

[0057] S400. Based on the daily bus dispatch plan, when there are not enough vehicles to dispatch on a bus line during a certain period of time, the number of idle buses in different bus stations is obtained, and the route from the bus station closest to the bus line's starting station to the bus line's starting station is obtained through the A* algorithm combined with weather data and urban traffic conditions data, and the idle buses are dispatched; if there are no idle buses in the bus stations, rescue bus lines are obtained by sorting them in descending order according to the size of the overlapping part of the lines, and the vehicles allowed to be dispatched are obtained according to the daily bus dispatch plan of the rescue bus lines, and the vehicles allowed to be dispatched are dispatched.

[0058] According to step S100, the multi-source data related to the operation of the bus includes: GPS positioning data, on-board sensor data, passenger flow data, time data, weather data, population density data, bus line user rating data, bus transfer data and major event data; the GPS positioning data records the latitude and longitude coordinates, speed and driving direction of the bus through the GPS device on each bus, the on-board sensor data collects information inside and outside the bus, the passenger flow data obtains the time, location and number of passengers getting on and off the bus through the passenger card swiping device, the electronic ticketing system and the in-vehicle monitoring video analysis, the time data includes a specific timestamp, the weather data is real-time weather data provided by the meteorological department, the population density data is population density data of each region, provided by the statistical department, and reflects the distribution of permanent population in a specific area, the bus line user rating data is user ratings and evaluations obtained from the user feedback system of the bus company, reflecting the passenger satisfaction with the bus line service, the bus transfer data records the information of passengers transferring between different bus lines, and the major event data records include information on major events, holidays and emergencies.

[0059] According to step S100, a GPS device, an on-board sensor and a 5G communication module are installed on each bus; the GPS device collects the location information of the bus in real time, the on-board sensor monitors the running status of the vehicle and the environmental data inside and outside the vehicle in real time, the passenger card swiping device and the monitoring system record the passenger flow data in real time, and the collected data is uploaded to the big data platform in real time through the 5G communication module, and the big data platform receives the real-time data uploaded by each bus;

[0060] The collected data is preprocessed through the big data platform to remove noise data, process missing values ​​and outliers, unify the data formats from different sources, aggregate the data in time and space, and summarize them by hour and by station; the preprocessed data is integrated with other data sources to form a unified data set.

[0061] According to step S200, the relevant data affecting passenger flow includes time data, weather data, population density data, bus line user rating data, bus transfer data and major event data; feature engineering is performed through correlation analysis to extract features of historical passenger flow data and relevant data affecting passenger flow; the correlation analysis uses the Pearson correlation coefficient ρX,Y, and the formula is as follows:

[0062]

[0063] Among them, X is the historical passenger flow data, Y is the relevant data affecting the passenger flow, Cov(X,Y) is the covariance between X and Y, which measures the common change trend between the two, and σX and σY are the standard deviations of X and Y, respectively. In this embodiment, the historical passenger flow data is taken from January 1, 2019 to December 31, 2020, a total of two years of data. Time data is marked by date, hour, day of the week, and holidays. Weather data includes temperature, rainfall, wind speed, etc. Population density data includes population density in various areas of the city. Bus line user rating data is the passenger satisfaction rating of the bus line. Bus transfer data includes the number of transfers and transfer station data. Major event data is the time and place of major events.

[0064] Differential processing, model identification and model fitting are performed on the features of historical passenger flow data and related data that affect passenger flow extracted by feature engineering. The model identification is used to determine the parameters of the ARIMA model. The model fitting uses historical data to fit the ARIMA model and establish an ARIMA time series model. The mathematical formula is as follows:

[0065]

[0066] in, is the predicted passenger flow at time t, is the predicted value of passenger flow at time ti, is the predicted value of passenger flow at time tj, c is a constant term, representing the baseline value of the model, φ i is the autoregressive coefficient, which indicates the impact of passenger flow at time ti on time t, p is the order of the autoregressive part, which indicates the number of time lags of the historical passenger flow data used, and φ j is the moving average coefficient, which indicates the effect of the error term at time tj on time t, q is the order of the moving average part, which indicates the number of time lags of the historical error term used, ∈ t is the error term at time t, representing the difference between the predicted value and the actual value. Through model identification, the parameters of the ARIMA model are determined to be p=2, q=2.

[0067] The data is divided into a training set and a validation set. The training set is the data from January 1, 2019 to June 30, 2020, and the validation set is the data from July 1, 2020 to December 31, 2020. Use the training set data to train the ARIMA model, adjust the model parameters, and use the least squares method to optimize the parameters of the ARIMA model so that the difference between the predicted value and the actual value is minimized. The formula is as follows:

[0068]

[0069] Among them, f t is the actual passenger flow value at time t, n is the number of time periods, indicating the total number of data points used;

[0070] The cross-validation method is used to verify the model effect, and the evaluation indicators are used to evaluate the prediction performance of the model. The evaluation indicators include root mean square error RMSE and mean absolute error MAE. The formula is as follows:

[0071]

[0072] The RMSE on the training set is 25.7, MAE is 19.4, and the RMSE on the validation set is 28.3, MAE is 21.1. The passenger flow prediction effect on the validation set is good and can accurately reflect the changing trend of passenger flow.

[0073] According to step S300, the driver status database includes historical driver status data, and the historical driver status data includes the driver's fatigue level, the driver's health status, the driver's working hours and the driver's technical level; extract the predicted passenger flow, time data, weather data and historical driver status data in the driver status database, use the random forest algorithm to predict the driver's status in different time periods, perform model training and model verification to obtain a trained driver status prediction model, input the predicted passenger flow, time data and weather data of the day, and predict the driver's status in different time periods of the day;

[0074] Extract time features, traffic features, and bus operation features from historical bus operation data and urban traffic status data, select important features through correlation analysis, define the number of layers of the LSTM network and the number of neurons in each layer, divide the important features into training sets and test sets, use the training set data to train the LSTM model, and use the test set data to verify the model performance;

[0075] For the extraction of time features, timestamps are extracted from the original data and converted into useful time features, such as date, hour, day of the week, season, etc. Based on historical data, morning and evening peak hours are identified and marked; combined with holiday data, holidays in historical data are marked;

[0076] For the extraction of traffic characteristics, we can extract the traffic flow information of specific road sections and time periods from the urban traffic status data, extract the real-time congestion index and historical congestion data of specific areas, identify and record traffic accidents, road construction and other events and their impact scope and time;

[0077] To extract the operation characteristics of buses, the operation trajectory, station stop time, driving speed and other information of each line are extracted from the bus GPS data; the vehicle's fuel consumption, acceleration and deceleration times, braking frequency and other data are analyzed to evaluate the vehicle's operating status; the number of people getting on and off at each station and its changing trend are extracted from historical data.

[0078] The predicted passenger flow, predicted driver status, historical bus operation data and urban traffic condition data for the day are input into the trained LSTM model, and the LSTM model outputs the bus dispatch plan for different time periods of the day, which includes the departure time, departure frequency, vehicle allocation and driver scheduling for each line.

[0079] In this embodiment, the input data are: predicted passenger flow for the day: the passenger flow during the morning peak period of a certain route is predicted to be 500 people / hour; predicted driver status for the day: the driver fatigue index is high during the morning peak period; urban traffic conditions for the day: the congestion index of a specific road section is high during the morning peak period;

[0080] Some of the outputs of the LSTM model are: Scheduling plan: Route A: Departure time: every 10 minutes from 7:00 to 9:00; Departure frequency: every 10 minutes; Vehicle allocation: add 2 buses during the morning rush hour; Driver scheduling: adjust the driver shift time according to the fatigue index;

[0081] The verification results are: the average waiting time of passengers during the morning peak period is reduced by 30%; the punctuality rate of bus operation is increased by 20%.

[0082] According to step S400, the vehicle dispatching and operation status of each line are monitored in real time through the big data platform, and a line is marked when it is found that the vehicle dispatching is insufficient in a certain period of time and cannot meet the passenger demand; the number and distribution of idle buses are queried from different bus depots and stations, and the idle vehicle data of each bus depot is obtained; the weather data and urban traffic status data are integrated, and the starting point and end point of the A* algorithm are initialized according to the positions of different bus depots and stations and the starting stations of the bus lines, respectively, the starting point is a different bus depot, and the end point is the starting station of the bus line; the A* algorithm searches for the optimal path from the starting point to the end point according to the weather data and the urban traffic status data, outputs the bus depot closest to the starting station of the bus line and the corresponding optimal path, and formulates a dispatch plan;

[0083] Among them, weather data includes temperature, rainfall, wind speed, etc., which affect the vehicle's driving speed and safety; obtaining traffic condition data requires real-time monitoring of urban traffic flow, congestion and traffic incidents, and traffic condition data affects driving time and route selection.

[0084] When performing the A* algorithm, the urban road network is defined as nodes and edges, where nodes represent intersections and edges represent roads. The estimated distance from the current node to the end point is used to adjust the valuation in combination with weather and traffic conditions. The search process is as follows: the starting point is placed in an open list, the node with the lowest valuation is selected from the open list, its adjacent nodes are expanded, the actual travel time and path safety are calculated in combination with weather and traffic conditions, the optimal path is found, and the nearest bus station and the corresponding path are output.

[0085] According to the dispatch plan, idle buses are directed to depart along the optimal route to provide support to the target route. The driving conditions of dispatched vehicles are monitored in real time, and feedback is provided based on the actual situation and the A* algorithm is used for real-time adjustments.

[0086] In the dispatch plan, determine the optimal path for idle buses from the starting point to the target line, plan the departure time and estimated arrival time of the vehicle, describe the route in detail, and mark the key nodes and turning points. Command idle buses to depart along the optimal path according to the plan, use the Internet of Vehicles and GPS technology to monitor the driving conditions of the vehicles in real time, obtain information such as location, speed, and driving route, provide feedback based on real-time traffic conditions and vehicle driving conditions, use the A* algorithm to re-evaluate the path, and make adjustments when necessary.

[0087] According to step S400, if it is confirmed that there are no idle vehicles in all bus stations, the overlapping parts of different bus routes are calculated according to multi-source data, and the lines are sorted in descending order according to the size of the overlapping parts, and the lines with the largest overlapping parts are preferentially selected as rescue routes; the daily dispatch plan of the sorted rescue routes is obtained from the big data platform, and according to the dispatch plan of the rescue routes and the current demand, it is determined that a specific bus is dispatched without affecting normal operation;

[0088] Considering the current location and operating status of a specific bus, the most suitable vehicle is selected from the vehicles allowed for dispatch, and the A* algorithm is used to plan the optimal driving path from the current route to the target route for the dispatched vehicle. A dispatch plan is formulated. During the execution of the dispatch plan, feedback is provided based on the actual situation and the A* algorithm is used for real-time adjustments.

[0089] Extract the dispatch plan for the day of the sorted rescue route from the big data platform, including the vehicle location, operating status and dispatch plan, and evaluate whether it is necessary to dispatch vehicles from the rescue route in combination with current needs. Ensure that the dispatched vehicles will not affect the normal operation of the rescue route, evaluate the remaining tasks and operating status of the dispatched vehicles, and select the most suitable dispatched vehicles based on the current location, operating status and remaining tasks of the vehicles. Develop a detailed rescue dispatch plan, including departure time, driving route, estimated arrival time, etc., command the selected vehicles for dispatch according to the plan, monitor the vehicle driving conditions in real time, and ensure the effective implementation of the dispatch plan.

[0090] According to another embodiment of the present invention, Figure 2 As shown in the system structure diagram of the intelligent decision analysis system based on multi-source data, the intelligent decision analysis system based on multi-source data includes:

[0091] Multi-source data acquisition module: including: data acquisition unit and data preprocessing unit; the data acquisition unit is responsible for acquiring multi-source data related to bus operation, using the Internet of Vehicles to realize real-time collection and uploading of bus operation data, and the data preprocessing unit is responsible for preprocessing the collected data through the big data platform and integrating multi-source data;

[0092] Passenger flow prediction module: including: feature engineering unit and time series model building unit; the feature engineering unit is responsible for obtaining historical passenger flow data and related data affecting passenger flow in the big data platform, and performing feature engineering; the time series model building unit uses the ARIMA time series model to establish a passenger flow prediction model, train the model, adjust the model parameters, and use the cross-validation method to verify the model effect;

[0093] The dispatching scheme generation module includes: a passenger flow prediction unit, a driver status prediction unit, a dispatching model building unit and a dispatching scheme generation unit; wherein the passenger flow prediction unit obtains the predicted passenger flow in different time periods of the day based on the passenger flow prediction model; the driver status prediction unit uses the random forest algorithm to predict the driver status based on the predicted passenger flow, time data and weather data in combination with the driver status database; the dispatching model building unit is responsible for obtaining the historical data of bus operation and urban traffic status data in the big data platform, and using the deep learning model LSTM to establish a bus dispatching model; the dispatching scheme generation unit is responsible for inputting the predicted passenger flow, predicted driver status, historical data of bus operation and urban traffic status data to obtain the bus dispatching scheme for the day;

[0094] Emergency dispatch module: including: idle vehicle dispatch unit and rescue vehicle dispatch unit; wherein, the idle vehicle dispatch unit is based on the daily dispatch plan of the bus. When the bus line does not have enough dispatch vehicles in a certain period of time, the number of idle buses in different bus depots is obtained, and the route from the bus depot closest to the bus line starting station to the bus line starting station is obtained through the A* algorithm combined with weather data and urban traffic conditions data, and the idle buses are dispatched. The rescue vehicle dispatch unit is responsible for obtaining the rescue bus line in descending order according to the size of the overlapping part of the line when there are no idle buses in the bus depot, and obtaining the vehicles allowed to be dispatched according to the daily dispatch plan of the rescue bus line, and dispatching the vehicles allowed to be dispatched.

[0095] In this embodiment, for the emergency dispatch module, the bus dispatch plan for the day is as follows: total number of routes: 20, dispatch time of each route: 6:00-22:00, one bus every 10 minutes, number of vehicles on each route: 30;

[0096] The idle bus data are as follows: Number of bus depots: 5, idle vehicles at each bus depot: 10;

[0097] Weather data and traffic condition data are as follows: Weather data: temperature (30°C), rainfall (0mm), wind speed (5km / h); Traffic condition data: real-time traffic flow, congestion index, traffic accident data;

[0098] The data of bus route overlap are as follows: Length of overlap (unit: km): routes A and B (10 km), routes A and C (5 km), routes A and D (15 km);

[0099] In the idle vehicle dispatch unit, route A has insufficient vehicle dispatch during the morning rush hour (7:00-9:00), and 2 buses need to be dispatched; data input: the day's dispatch plan, the number of idle vehicles at each bus station, weather data and traffic conditions data; A* algorithm implementation: number of urban road network nodes: 100; average number of connection edges for each node: 4; weather and traffic conditions are considered during the search process to adjust the path valuation.

[0100] In the rescue vehicle dispatching unit, all bus depots and stations have no idle buses, so vehicles need to be dispatched from the lines with the largest overlap. Data input: bus line overlap data, the rescue bus line's daily dispatch plan, and the number of vehicles allowed to be dispatched. Sorting and dispatching implementation: sorting in descending order by the length of the overlap, and dispatching the line with the largest overlap to line D.

[0101] The dispatch results are as follows: In the idle vehicle dispatch unit, the initial data is: idle vehicle station distribution: station 1 (2 vehicles), station 2 (4 vehicles), station 3 (1 vehicle), station 4 (2 vehicles), station 5 (1 vehicle); the A* algorithm result is: the optimal path: station 2 to the starting station of line A (path length: 8 kilometers); dispatch time: 20 minutes; the dispatch plan is: dispatch 2 buses from station 2, and it is expected to arrive at the starting station of line A in 20 minutes; the effect evaluation is: the timeliness of vehicle dispatch during the morning peak period: improved by 80%; the waiting time of passengers is reduced by 25%;

[0102] In the rescue vehicle dispatch unit, the initial data is: overlapping part of the bus lines: Line D (15 kilometers), Line B (10 kilometers), Line C (5 kilometers); the sorting results are: overlapping part sorting: Line D>Line B>Line C; the dispatch plan is: dispatch 2 buses from Line D; estimated dispatch time: 30 minutes; effect evaluation is: impact of dispatched vehicles on the original line: reduction of 10%; passenger demand satisfaction rate: increase of 15%.

[0103] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0104] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent decision analysis method based on multi-source data, characterized in that: The method comprises the following steps: S100, obtain multi-source data related to bus operation, use the Internet of Vehicles to realize real-time collection and upload of bus operation data, pre-process the collected data through the big data platform, and integrate multi-source data; S200. Obtain historical passenger flow data and related data affecting passenger flow in the big data platform, perform feature engineering, use the ARIMA time series model to establish a passenger flow prediction model, train the model, adjust model parameters, and use cross-validation to verify the model effect; S300, based on the passenger flow prediction model, the predicted passenger flow in different time periods of the day is obtained, and the driver status is predicted by using the random forest algorithm based on the predicted passenger flow, time data and weather data in combination with the driver status database; the historical data of bus operation and urban traffic status data are obtained in the big data platform, and the bus scheduling model is established using the deep learning model LSTM, and the predicted passenger flow, predicted driver status, historical data of bus operation and urban traffic status data are input to obtain the bus scheduling plan for the day; S400. Based on the daily bus dispatch plan, when there are not enough vehicles to dispatch on a bus line during a certain period of time, the number of idle buses in different bus stations is obtained, and the route from the bus station closest to the bus line's starting station to the bus line's starting station is obtained through the A* algorithm combined with weather data and urban traffic conditions data, and the idle buses are dispatched; if there are no idle buses in the bus stations, rescue bus lines are obtained by sorting them in descending order according to the size of the overlapping part of the lines, and the vehicles allowed to be dispatched are obtained according to the daily bus dispatch plan of the rescue bus lines, and the vehicles allowed to be dispatched are dispatched.

2. The intelligent decision analysis method based on multi-source data according to claim 1 is characterized by: According to step S100, the multi-source data related to the operation of the bus includes: GPS positioning data, on-board sensor data, passenger flow data, time data, weather data, population density data, bus line user rating data, bus transfer data and major event data; the GPS positioning data records the latitude and longitude coordinates, speed and driving direction of the bus through the GPS device on each bus, the on-board sensor data collects information inside and outside the bus, the passenger flow data obtains the time, location and number of passengers getting on and off the bus through the passenger card swiping device, the electronic ticketing system and the in-vehicle monitoring video analysis, the time data includes a specific timestamp, the weather data is real-time weather data provided by the meteorological department, the population density data is population density data of each region, provided by the statistical department, and reflects the distribution of permanent population in a specific area, the bus line user rating data is user ratings and evaluations obtained from the user feedback system of the bus company, reflecting the passenger satisfaction with the bus line service, the bus transfer data records the information of passengers transferring between different bus lines, and the major event data records include information on major events, holidays and emergencies.

3. The intelligent decision analysis method based on multi-source data according to claim 2 is characterized in that: According to step S100, a GPS device, an on-board sensor and a 5G communication module are installed on each bus; The GPS device collects the location information of the bus in real time, the on-board sensor monitors the running status of the vehicle and the environmental data inside and outside the vehicle in real time, the passenger card swiping device and the monitoring system record the passenger flow data in real time, and the collected data is uploaded to the big data platform in real time through the 5G communication module. The big data platform receives the real-time data uploaded by each bus; The collected data is preprocessed through the big data platform to remove noise data, process missing values ​​and outliers, unify the data formats from different sources, aggregate the data in time and space, and summarize by hour and by station; Integrate the preprocessed data with other data sources to form a unified data set.

4. The intelligent decision analysis method based on multi-source data according to claim 1 is characterized in that: According to step S200, the relevant data affecting passenger flow includes time data, weather data, population density data, bus line user rating data, bus transfer data and major event data; feature engineering is performed through correlation analysis to extract features of historical passenger flow data and relevant data affecting passenger flow; the correlation analysis uses the Pearson correlation coefficient ρX,Y, and the formula is as follows: Among them, X is the historical passenger flow data, Y is the relevant data affecting passenger flow, Cov(X,Y) is the covariance between X and Y, which measures the common change trend between the two, and σX and σY are the standard deviations of X and Y respectively.

5. The intelligent decision analysis method based on multi-source data according to claim 4 is characterized in that: Differential processing, model identification and model fitting are performed on the features of historical passenger flow data and related data that affect passenger flow extracted by feature engineering. The model identification is used to determine the parameters of the ARIMA model. The model fitting uses historical data to fit the ARIMA model and establish an ARIMA time series model. The mathematical formula is as follows: in, is the predicted passenger flow at time t, is the predicted passenger flow at time ti, is the predicted value of passenger flow at time tj, c is a constant term, representing the baseline value of the model, φ i is the autoregressive coefficient, which indicates the impact of passenger flow at time ti on time t, p is the order of the autoregressive part, which indicates the number of time lags of the historical passenger flow data used, and φ j is the moving average coefficient, which indicates the effect of the error term at time tj on time t, q is the order of the moving average part, which indicates the number of time lags of the historical error term used, ∈ t is the error term at time t, representing the difference between the predicted value and the actual value.

6. The intelligent decision analysis method based on multi-source data according to claim 5 is characterized by: The data is divided into a training set and a validation set. The training set data is used to train the ARIMA model, and the model parameters are adjusted. The least squares method is used to optimize the parameters of the ARIMA model so that the difference between the predicted value and the actual value is minimized. The formula is as follows: Among them, f t is the actual passenger flow value at time t, n is the number of time periods, indicating the total number of data points used; The cross-validation method is used to verify the model effect, and the evaluation indicators are used to evaluate the prediction performance of the model. The evaluation indicators include root mean square error RMSE and mean absolute error MAE. The formula is as follows:

7. The intelligent decision analysis method based on multi-source data according to claim 1 is characterized in that: According to step S300, the driver status database includes historical driver status data, and the historical driver status data includes the driver's fatigue level, the driver's health status, the driver's working hours and the driver's technical level; extract the predicted passenger flow, time data, weather data and historical driver status data in the driver status database, use the random forest algorithm to predict the driver's status in different time periods, perform model training and model verification to obtain a trained driver status prediction model, input the predicted passenger flow, time data and weather data of the day, and predict the driver's status in different time periods of the day; Extract time features, traffic features, and bus operation features from historical bus operation data and urban traffic status data, select important features through correlation analysis, define the number of layers of the LSTM network and the number of neurons in each layer, divide the important features into training sets and test sets, use the training set data to train the LSTM model, and use the test set data to verify the model performance; The predicted passenger flow, predicted driver status, historical bus operation data and urban traffic condition data for the day are input into the trained LSTM model, and the LSTM model outputs the bus dispatch plan for different time periods of the day, which includes the departure time, departure frequency, vehicle allocation and driver scheduling for each line.

8. The intelligent decision analysis method based on multi-source data according to claim 1 is characterized by: According to step S400, the vehicle dispatching and operation status of each route are monitored in real time through the big data platform, and a route is marked when it is found that the vehicle dispatching of a certain route is insufficient during a certain period and cannot meet the passenger demand; Query the number and distribution of idle buses in different bus stations to obtain idle vehicle data for each bus station; integrate weather data and urban traffic data, and initialize the starting point and end point of the A* algorithm according to the locations of different bus stations and bus line departure stations, where the starting point is different bus stations and the end point is the bus line departure station; The A* algorithm searches for the optimal path from the starting point to the end point based on weather data and urban traffic conditions, outputs the bus station closest to the starting station of the bus line and the corresponding optimal path, and formulates a scheduling plan; According to the dispatch plan, idle buses are directed to depart along the optimal route to provide support to the target route. The driving conditions of dispatched vehicles are monitored in real time, and feedback is provided based on the actual situation and the A* algorithm is used for real-time adjustments.

9. The intelligent decision analysis method based on multi-source data according to claim 8 is characterized in that: According to step S400, if it is confirmed that there are no idle vehicles in all bus stations, the overlapped parts of different bus routes are calculated according to the multi-source data, and the routes are sorted in descending order according to the size of the overlapped parts, and the routes with the largest overlapped parts are preferentially selected as rescue routes; Obtain the dispatch plan of the day for the sorted rescue routes from the big data platform, and determine the dispatch of specific buses without affecting normal operations based on the dispatch plan of the rescue routes and current demand; Considering the current location and operating status of a specific bus, the most suitable vehicle is selected from the vehicles allowed for dispatch, and the A* algorithm is used to plan the optimal driving path from the current route to the target route for the dispatched vehicle. A dispatch plan is formulated. During the execution of the dispatch plan, feedback is provided based on the actual situation and the A* algorithm is used for real-time adjustments.

10. An intelligent decision analysis system based on multi-source data, using the intelligent decision analysis method based on multi-source data according to any one of claims 1 to 9, characterized in that: include: Multi-source data acquisition module: including: data acquisition unit and data preprocessing unit; the data acquisition unit is responsible for acquiring multi-source data related to bus operation, using the Internet of Vehicles to realize real-time collection and uploading of bus operation data, and the data preprocessing unit is responsible for preprocessing the collected data through the big data platform and integrating multi-source data; Passenger flow prediction module: including: feature engineering unit and time series model building unit; the feature engineering unit is responsible for obtaining historical passenger flow data and related data affecting passenger flow in the big data platform, and performing feature engineering; the time series model building unit uses the ARIMA time series model to establish a passenger flow prediction model, train the model, adjust the model parameters, and use the cross-validation method to verify the model effect; The dispatching scheme generation module includes: a passenger flow prediction unit, a driver status prediction unit, a dispatching model building unit and a dispatching scheme generation unit; wherein the passenger flow prediction unit obtains the predicted passenger flow in different time periods of the day based on the passenger flow prediction model; the driver status prediction unit uses the random forest algorithm to predict the driver status based on the predicted passenger flow, time data and weather data in combination with the driver status database; the dispatching model building unit is responsible for obtaining the historical data of bus operation and urban traffic status data in the big data platform, and using the deep learning model LSTM to establish a bus dispatching model; the dispatching scheme generation unit is responsible for inputting the predicted passenger flow, predicted driver status, historical data of bus operation and urban traffic status data to obtain the bus dispatching scheme for the day; Emergency dispatch module: including: idle vehicle dispatch unit and rescue vehicle dispatch unit; wherein, the idle vehicle dispatch unit is based on the daily dispatch plan of the bus. When the bus line does not have enough dispatch vehicles in a certain period of time, the number of idle buses in different bus depots is obtained, and the route from the bus depot closest to the bus line starting station to the bus line starting station is obtained through the A* algorithm combined with weather data and urban traffic conditions data, and the idle buses are dispatched. The rescue vehicle dispatch unit is responsible for obtaining the rescue bus line in descending order according to the size of the overlapping part of the line when there are no idle buses in the bus depot, and obtaining the vehicles allowed to be dispatched according to the daily dispatch plan of the rescue bus line, and dispatching the vehicles allowed to be dispatched.

Images