A Big Data-Based Intelligent Operation Method for New Energy Public Buses

Abstract

A big data-based intelligent operation method for new energy buses fully utilizes real-world vehicle big data to obtain the segmented SOC changes of bus routes under the influence of time, road conditions, and other factors, as well as the comprehensive equivalent SOC changes of each segment. This guides the rational route allocation for buses awaiting departure. By clustering vehicle operation data tags, the method accurately identifies the specific conditions and energy consumption of each bus route in different time periods, seasons, months, traffic conditions, etc. This allows for the intelligent overall planning and arrangement of all vehicles at various bus depots across multiple routes. Compared to the traditional fixed vehicle and timetable operation mode of bus routes, this provides greater flexibility, making it possible to realize intelligent transportation across the entire region and effectively avoiding accidents and safety hazards.

Description

Technical Field

[0001] This invention belongs to the field of new energy bus operation planning technology, specifically involving a smart operation method for new energy buses based on big data. Background Technology

[0002] Currently, pure electric new energy vehicles are being increasingly adopted in public transportation, gradually showing a trend of completely replacing traditional fuel-powered buses and providing possibilities for intelligent bus operation. However, the contradiction between the high charging demand of electric vehicles and the shortage of charging infrastructure has been difficult to eliminate for a long time. Furthermore, as the performance and capacity of onboard power batteries continue to decline during use, the driving range and charging intervals are gradually shortening, further exacerbating charging shortages and range anxiety. This is especially true for electric buses, which are characterized by long-term high-load operation. To ensure that vehicles can return to the depot after multiple runs, the charging cycle and interval cannot be guaranteed to be scientifically reasonable, making overcharging and over-discharging difficult to avoid. This leads to faster battery performance and lifespan degradation, making the aforementioned contradiction particularly severe and hindering the realization of intelligent operation to some extent. Therefore, how to precisely analyze vehicle energy consumption and driving range based on the characteristics of bus route operation and optimize route scheduling to ultimately achieve comprehensive optimization of operational efficiency and the charging process is a pressing technical problem that needs to be solved in this field. Summary of the Invention

[0003] In view of this, and to address the technical problems existing in this field, the present invention provides a method for intelligent operation of new energy buses based on big data, specifically including the following steps:

[0004] Step 1: Collect data from the actual operation of the new energy bus, including battery voltage, battery current, SOC, vehicle latitude and longitude coordinates, vehicle speed, motor speed, time, ambient temperature, air conditioning working status, current route, etc., and upload it to the big data platform;

[0005] Step 2: After receiving the data uploaded by each new energy bus, the big data platform performs corresponding preprocessing, and establishes corresponding data tags for each frame, such as battery voltage, battery current, SOC, vehicle latitude and longitude coordinates, vehicle speed, motor speed, ambient temperature, air conditioning working status, and current operating route. Based on the time data, it also establishes data tags for daily time periods, months, and seasons. The established data tags are used to generate a set of operation data tags for all new energy buses.

[0006] Step 3: Grid the map of the city, town or other specific area to which the bus belongs, and divide each bus route in the area into several route segments based on the same grid length; calculate the SOC change of each new energy bus when it passes through different segments, and generate a time-segment-SOC change data table for each vehicle.

[0007] Step 4: For a specific route segment, use the vehicle's latitude and longitude coordinates and the current route data label to filter out the corresponding data labels of buses that have passed through this segment in the historical period from the operation data label set. Combine these with the corresponding SOC change values ​​from the data table to form multiple arrays corresponding to each frame. Perform a clustering algorithm on the arrays and use the SOC change value corresponding to the calculated cluster center as the equivalent SOC change value of this route segment under different operating conditions. Combine the time period label corresponding to the cluster center with the equivalent SOC change value of each consecutive segment to reconstruct the complete equivalent SOC change value of the route under different operating conditions.

[0008] Step 5: Extract the current SOC of each new energy bus currently at the bus depot. By comparing the change in the complete equivalent SOC of different routes during the corresponding time period, determine the specific route to be run by each bus, the departure time, and arrange for charging of vehicles that are insufficient to complete any route.

[0009] Furthermore, in step four, before filtering the data labels, several specific operating conditions are first divided according to some operating data characteristics. Then, a cost function is established using the remaining operating data characteristics as input and the complete line energy consumption corresponding to the several specific operating conditions as output. The neural network is trained using historical real vehicle operating data to obtain the weights of the remaining motion data characteristics under different specific operating conditions. The weights are used in conjunction with the weight thresholds to remove some operating data labels, and the remaining operating data labels are used in subsequent steps.

[0010] Furthermore, in step four, after performing a clustering algorithm on the data group, clustering is performed again using the corresponding vehicle speed, motor speed, ambient temperature, and air conditioning operating status data labels in the neighborhood of each cluster center, so as to further classify the obtained operating conditions, and determine the equivalent SOC change of the continuous section and the complete equivalent SOC change of the reconstructed line based on each cluster center of the re-clustering.

[0011] Furthermore, the big data platform will also extract the corresponding SOC, vehicle latitude and longitude coordinates, vehicle speed, motor speed, daily time period, month, and season data labels from the arrays in the neighborhood of each cluster center, and provide them to the relevant big data platforms and units as auxiliary information for optimizing the public transport network and traffic management in specific areas.

[0012] Furthermore, the data labels corresponding to outliers in the clustering process are extracted synchronously to analyze the abnormal conditions of the vehicle itself or its operating conditions on each vehicle.

[0013] The method provided by the present invention fully considers that most of the new energy buses operating on public transport routes in urban and rural areas of my country use the same model, and the specifications of the on-board power batteries are basically consistent. Moreover, within a small area such as a single city, the environmental conditions such as temperature and weather of all vehicles can be regarded as completely the same. Therefore, by taking into account the driving data of all new energy buses in this urban area, it can ensure that the relationship between battery SOC and driving range is basically consistent. The same remaining power corresponds to the same driving range, avoiding the problem of inaccurate clustering results or failure to converge due to the existence of multiple models and battery types.

[0014] The intelligent operation method for new energy buses based on big data provided by this invention fully utilizes real-world big data from new energy buses to obtain the corresponding segmental SOC changes of bus routes under the influence of time, road conditions, and other factors, as well as the comprehensive equivalent SOC changes of each segment. This guides the rational route allocation for waiting buses. By clustering vehicle operation data tags, the specific conditions and energy consumption of each bus route in different time periods, seasons, months, traffic conditions, etc., can be accurately identified. This allows for the intelligent overall planning and arrangement of all vehicles at each bus depot across multiple routes, providing greater flexibility compared to the traditional fixed vehicle and timetable operation method. Utilizing clustered neighborhoods and related operational data corresponding to specific operating conditions helps to promptly identify problems in bus route planning and traffic management, and also enables timely detection of abnormal vehicle conditions or operating conditions. This provides the possibility for the realization of intelligent transportation across the entire region and can effectively prevent accidents and safety hazards. Detailed Implementation

[0015] The technical solution of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] For urban and rural traffic conditions, the SOC (State of Charge) and energy consumption of vehicles can vary significantly depending on specific daily times, seasons, months, and traffic events. For example, for the same route, the SOC of a vehicle will obviously decrease faster during morning and evening rush hours compared to other times of the day. In summer and winter, vehicles often use air conditioning, resulting in different SOC trends compared to spring and autumn. Of course, considering factors such as battery capacity degradation in winter, the SOC changes in summer and winter need to be calculated separately. Some commercial areas experience traffic congestion around them on holidays, while weekday morning and evening rush hours are not significantly different from other times. For primary and secondary schools, road conditions around schools differ drastically between the school term and summer / winter breaks, and the start times of morning and evening rush hours also differ. Some key areas also have regular traffic control measures. Transportation hubs such as train stations and long-distance bus stations experience higher passenger loads and traffic congestion compared to other areas. These special circumstances indicate different operating conditions for public transport routes. To achieve intelligent public transport operation, a detailed analysis of the energy consumption and vehicle range resulting from these operating conditions is essential.

[0017] On the other hand, for any complete bus route, the actual operating conditions of buses continuously change as they pass through different sections due to traffic or geographical reasons. For example, in the special cases mentioned above, the energy consumption of vehicles passing through commercial areas, schools, controlled road sections, and transportation hubs may be very different from other sections. Even if congestion, load, or onboard electrical appliances such as air conditioning cause similar power consumption in the same section, the differences between operating conditions can still be reflected by operating data such as vehicle speed, motor speed, and changes in latitude and longitude. Therefore, dividing the entire bus route into multiple segments by using a gridded map can make the equivalent energy consumption calculation results of the complete route under different operating conditions more accurate. Combined with the premise of uniform bus type and battery type in the same area, it helps to assign a suitable route to the vehicle based on its real-time power level.

[0018] Based on the above considerations, the present invention provides the following intelligent operation method for new energy buses based on big data, specifically including the following steps:

[0019] Step 1: Collect data from the actual operation of the new energy bus, including battery voltage, battery current, SOC, vehicle latitude and longitude coordinates, vehicle speed, motor speed, time, ambient temperature, air conditioning working status, current route, etc., and upload it to the big data platform;

[0020] Step 2: After receiving the data uploaded by each new energy bus, the big data platform performs corresponding preprocessing, and establishes corresponding data tags for each frame, such as battery voltage, battery current, SOC, vehicle latitude and longitude coordinates, vehicle speed, motor speed, ambient temperature, air conditioning working status, and current operating route. Based on the time data, it also establishes data tags for daily time periods, months, and seasons. The established data tags are used to generate a set of operation data tags for all new energy buses.

[0021] Step 3: Grid the map of the specific area, such as the city or town where the bus belongs, and divide each bus route in the area into several route segments based on the same grid length; calculate the SOC change of each new energy bus when it passes through different segments, and generate a corresponding SOC change data table.

[0022] Step 4: For a specific route segment, use the vehicle's latitude and longitude coordinates and the current route data label to filter out the corresponding data labels of buses that have passed through this segment in the historical period from the operation data label set. Combine these with the corresponding SOC change values ​​from the data table to form multiple arrays corresponding to each frame. Perform a clustering algorithm on the arrays and use the SOC change value corresponding to the calculated cluster center as the equivalent SOC change value of this route segment under different operating conditions. Combine the time period label corresponding to the cluster center with the equivalent SOC change value of each consecutive segment to reconstruct the complete equivalent SOC change value of the route under different operating conditions.

[0023] If the State of Charge (SOC) of new energy buses changes systematically across different times of day, months, quarters, and traffic conditions, the operational data labels serve as array coordinates in clustering to differentiate the specific operating conditions of different sections. For example, the energy consumption of a certain section may be similar during peak hours in spring and autumn (when congestion is high) and during off-peak hours in summer and winter (when air conditioning is on), but differences in operating conditions can still be reflected through data labels such as vehicle speed, motor speed, and air conditioning status. Increasing the types of operational data further improves the accuracy of identifying operating conditions. Therefore, the complete equivalent SOC change of the route obtained through the above clustering process can be effectively used to guide the scheduling of vehicles awaiting departure. For instance, in the moderate temperatures of spring and autumn, the traffic conditions and energy consumption changes of certain urban sections are relatively similar at the same time. In clustering, the data for these two seasons use the same "spring and autumn" seasonal data label. Differences in route energy consumption within the same season are mostly related to traffic conditions at different times; therefore, the data for morning and evening peak hours can be labeled with the same "peak congestion" time period label, while other times use the "normal smooth" time period label. As can be seen from the above examples, clustering ultimately yields only a limited number of operating condition types, thus helping to ensure controllable computational overhead and avoid generating too many classification results. Of course, the number of classifications also depends on the method and quantity of motion data label settings, the fineness of filtering, and the setting of cluster neighborhoods. For routes where environmental or traffic conditions cause complex variations in operating conditions for the same route, more labels or smaller neighborhoods can be appropriately used.

[0024] Step 5: Extract the current SOC of each new energy bus currently at the bus depot. By comparing it with the complete equivalent SOC change of different routes, determine the specific route each bus will run, departure time, and arrange for charging of vehicles insufficient to complete any route. Vehicles to be dispatched can also be set with a certain margin, such as 10% to 15%, of the corresponding complete equivalent SOC change to cope with various emergencies. For example, for situations where the operating conditions and energy consumption of the same route during the same period in spring and autumn are relatively similar, the complete equivalent SOC change of the route obtained from historical operating data of several months in spring can guide the route assignment of each bus in autumn. For routes during the morning and evening peak hours of the current spring and summer semesters, the complete equivalent SOC change of the same period can be calculated using historical operating data from the first half of the school year as the basis for scheduling the same bus route in the second half of the school year; or historical operating data from the same period in previous years can be used to guide the operation of the current year. Those skilled in the art can also flexibly choose how to effectively utilize the complete equivalent SOC change obtained from historical operating data based on the teachings of this invention. When applied to driverless buses to eliminate driver rest time, this invention can further significantly shorten the departure interval of routes and comprehensively improve the intelligent operation efficiency of the entire public transportation network.

[0025] In a preferred embodiment of the present invention, before filtering data labels in step four, several specific operating conditions are first divided based on some operating data characteristics. Then, a cost function is established using the remaining operating data characteristics as input and the complete line energy consumption corresponding to the several specific operating conditions as output. A neural network is trained using historical real-vehicle operating data to obtain the weights of the remaining motion data characteristics under different specific operating conditions. Using the weights and combining them with each weight threshold, some operating data labels are removed, and the retained operating data labels are used in subsequent steps. In this way, the degree of influence of different operating data characteristics on line energy consumption can be calculated in advance for certain specific operating conditions, such as the morning and evening peak hours during the autumn and winter semesters, using the idea of ​​correlation analysis. This allows for the retention of more important motion data labels and the removal of some less influential labels, thereby further reducing the number of elements in the clustering array and the overall computational load.

[0026] In a preferred embodiment of the present invention, after performing a clustering algorithm on the data group in step four, clustering is performed again using the corresponding vehicle speed, motor speed, ambient temperature, and air conditioning operating status data labels in the neighborhood of each cluster center. This is to further classify the obtained operating conditions, and based on each cluster center from the second clustering, the equivalent SOC change of the continuous segment and the complete equivalent SOC change of the reconstructed route are determined sequentially. This secondary clustering allows for further identification of the subdivided operating conditions corresponding to individual motion data features, which helps to achieve more accurate route planning by combining the remaining battery power of the actual vehicle. However, the computational load inevitably increases to some extent.

[0027] In a preferred embodiment of the present invention, the big data platform also extracts the corresponding SOC, vehicle latitude and longitude coordinates, vehicle speed, motor speed, daily time period, monthly, and seasonal data labels from the arrays within the neighborhood of each cluster center. These data are then provided to the relevant big data platform and relevant units as auxiliary information for optimizing the public transport network and managing traffic within a specific area. For example, for sections and time periods prone to congestion due to problems with road design and traffic control, vehicle operation data passing through these areas will show lower vehicle speeds and motor speeds. This real-world vehicle big data can provide important optimization basis for relevant departments, facilitating the realization of intelligent traffic operation within the region.

[0028] In a preferred embodiment of the present invention, when some bicycles have abnormalities such as severe battery degradation or thermal runaway, their corresponding operating data are likely to be outliers in clustering. Therefore, the data labels corresponding to the outliers in the clustering process are extracted simultaneously to analyze the abnormal conditions of the vehicle itself or its operating conditions on each bicycle, which can effectively identify safety hazards and prevent accidents from occurring.

[0029] It should be understood that the sequence number of each step in the embodiments of the present invention does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0030] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent operation of new energy buses based on big data, characterized in that: Specifically, the following steps are included: Step 1: Collect data from the actual operation of the new energy bus, including battery voltage, battery current, SOC, vehicle latitude and longitude coordinates, vehicle speed, motor speed, time, ambient temperature, air conditioning working status, and current route, and upload it to the big data platform; Step 2: After receiving the data uploaded by each new energy bus, the big data platform performs corresponding preprocessing, and establishes corresponding data tags for each frame, including battery voltage, battery current, SOC, vehicle latitude and longitude coordinates, vehicle speed, motor speed, ambient temperature, air conditioning working status, and current operating route. Based on the time data, it also establishes data tags for daily time periods, months, and seasons. The established data tags are used to generate a set of operation data tags for all new energy buses. Step 3: Grid the map of the city and town areas to which the buses belong, and divide each bus route in the area into several route segments based on the same grid length; calculate the SOC change of each new energy bus when it passes through different segments, and generate a corresponding SOC change data table. Step 4: For a specific route segment, use the vehicle's latitude and longitude coordinates and the current route data label to filter out the corresponding data labels of buses that have passed through this segment in the historical period from the operation data label set. Combine the data labels with the SOC change data table generated in Step 3 to form multiple arrays corresponding to each frame. Perform a clustering algorithm on the arrays and use the SOC change corresponding to the calculated cluster center as the equivalent SOC change of this route segment under different operating conditions. Combine the time period label corresponding to the cluster center with the equivalent SOC change of each continuous segment to reconstruct the complete equivalent SOC change of the route under different operating conditions. Step 5: Extract the current SOC of each new energy bus currently at the bus depot. By comparing the change in the complete equivalent SOC of different routes during the corresponding time period, determine the specific route and departure time of each bus, and arrange for vehicles that are insufficient to complete any route to be charged.

2. The method as described in claim 1, characterized in that: In step four, before filtering the data labels, several specific operating conditions are first divided according to some operating data characteristics. Then, a cost function is established using the remaining operating data characteristics as input and the complete line energy consumption corresponding to the several specific operating conditions as output. The neural network is trained using historical real vehicle operating data to obtain the weights of the remaining operating data characteristics under different specific operating conditions. The weights are used in combination with the weight thresholds to remove some operating data labels, and the remaining operating data labels are used in subsequent steps.

3. The method as described in claim 1, characterized in that: In step four, after performing a clustering algorithm on the data group, clustering is performed again using the corresponding data labels of vehicle speed, motor speed, ambient temperature, and air conditioning operating status in the neighborhood of each cluster center. This is to further classify the obtained operating conditions, and based on each cluster center of the re-clustering, the equivalent SOC change of the continuous section and the complete equivalent SOC change of the reconstructed line are determined sequentially.

4. The method as described in claim 1, characterized in that: The big data platform will also extract the corresponding SOC, vehicle latitude and longitude coordinates, vehicle speed, motor speed, daily time period, month, and season data labels from the arrays in the neighborhood of each cluster center. These data will be used as auxiliary information for optimizing the public transport network and traffic management within the city and town areas and provided to the relevant big data platforms and units.

5. The method as described in claim 1, characterized in that: During the clustering process, the data labels corresponding to outliers are extracted simultaneously to analyze the abnormal conditions of the vehicles themselves or their operating conditions on each vehicle.