Charging and discharging control system for power battery of new energy tourism and sightseeing vehicle

Abstract

The invention discloses a charging and discharging control system for a power battery of a new energy tourism and sightseeing vehicle, and relates to the technical field of battery charging and discharging. Comprising a power generation prediction module, a demand analysis module, a decision management module and a dynamic supervision module, wherein the power generation prediction module is used for extracting future travel and future weather data from a sightseeing bus management system and a meteorological platform; the technical key points are as follows: a standardized API (Application Program Interface) cooperates with an existing sightseeing vehicle management system to collect travel data such as departure time, routes, loads and driving styles of the sightseeing vehicle, travel segmentation and weather data association are carried out by combining meteorological data, and a charging and discharging regulation and control strategy is generated after power generation prediction, demand analysis and decision management and is fed back to the sightseeing vehicle management system. The edge network is used for monitoring execution data in real time to form a closed loop, energy flow and operation scheduling dynamic matching is achieved, the energy utilization efficiency and the operation economy are improved, the use effect is good, and the method has good use prospects.

Description

Technical Field

[0001] The present invention relates to the technical field of battery charging and discharging, and in particular to a power battery charging and discharging control system for a new energy tourist sightseeing vehicle. Background Art

[0002] With the continuous advancement of solar photovoltaic power generation technology, the conversion efficiency of solar panels will continue to increase and the cost will be further reduced, making the application of photovoltaic power generation function in new energy tourist sightseeing vehicles more and more common.

[0003] As a clean energy means of transportation, new energy sightseeing vehicles have the characteristics of high energy efficiency and environmental protection, and have become the main development direction of future transportation.

[0004] New energy tourist buses, known for their environmentally friendly features, are a popular choice for scenic area tours. Their efficient operation is crucial to the safety of their charging and discharging control system. This system monitors battery status in real time, precisely regulating charging current and voltage to ensure battery safety and extend battery life. It also optimizes discharge strategies to improve energy efficiency and vehicle range.

[0005] During the off-season, the number of tourists in scenic spots drops sharply, and the frequency of use of new energy sightseeing vehicles decreases significantly, with a large number of vehicles being idle. At this time, the photovoltaic power generation system in the scenic area continues to work. If the generated electricity cannot be consumed in time, it may cause waste. At this time, the electricity on the new energy sightseeing vehicles can be sold through V2G.

[0006] When the scenic area's own power demand is low, these sightseeing vehicles can supply stored energy back to the grid. During this process, the charge-discharge control system continuously functions, monitoring the battery status in real time and precisely adjusting the charge and discharge parameters to ensure a safe, stable, and efficient bidirectional energy flow. This is why several electric vehicle charge-discharge control systems have been developed.

[0007] The existing patent application publication number is CN110171323A, and the patent name is a V2G-based electric vehicle charging and discharging control system and usage method. The invention patent records that it includes: a vehicle controller, an electric drive control system, a power battery and a battery management system, wherein: the electric drive system includes a DC / AC converter and a motor; the vehicle controller receives external information unidirectionally, and exchanges information with the battery management system and the DC / AC converter in both directions; the battery management system includes a power battery, which is electrically connected to the DC / AC converter; the DC / AC converter is electrically connected to the motor, and the motor is electrically connected to the power grid; when the motor is disconnected from the power grid, the DC / AC converter controls the output of the motor speed and torque. When the motor is connected to the power grid, the vehicle controller integrates information and issues charging and discharging instructions, and the DC / AC converter charges and discharges the power battery. The present invention has a simple structure, effectively improves the space utilization rate and charging and discharging power in the vehicle, and guarantees the interests of users.

[0008] The central idea of ​​the solution recorded in the above patent is how to realize the charging and discharging of electric vehicles, and further study the charging and discharging of electric vehicles to realize the adjustment of the charging and discharging speed. However, it is only a simple adjustment of the charging and discharging process. It does not take the overall consideration into account, and does not consider whether the sightseeing vehicle is in use, external electricity price fluctuations, etc. Therefore, the effect it plays is relatively simple. It is equivalent to an independent system with little connection with the outside world. It is completely unable to connect with the existing sightseeing vehicle management system to realize overall scheduling. Therefore, its applicability is relatively low, and it cannot be comprehensively analyzed. It does not meet the requirements of new energy tourist sightseeing vehicles. For this reason, we have developed a new energy tourist sightseeing vehicle power battery charging and discharging control system. Summary of the Invention

[0009] (1) Technical problems solved

[0010] In response to the shortcomings of the existing technology, the present invention provides a new energy tourist sightseeing vehicle power battery charging and discharging control system, which cooperates with the existing sightseeing vehicle management system through a standardized API interface to collect its departure time, route, load, driving style and other travel data, and combines meteorological data to segment the journey and associate the weather data. After power generation forecasting, demand analysis, and decision-making management, the present invention generates a charging and discharging control strategy and feeds it back to the sightseeing vehicle management system. The edge network is used to monitor the execution data in real time to form a closed loop, realize dynamic matching of energy flow and operation scheduling, improve energy utilization efficiency and operational economy, have good use effect, have good use prospects, and effectively solve the problems raised in the background technology.

[0011] (2) Technical solution

[0012] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0013] A new energy tourist sightseeing vehicle power battery charging and discharging control system includes a power generation prediction module, a demand analysis module, a decision management module, and a dynamic supervision module:

[0014] Power generation prediction module: This module extracts future itinerary and weather data from the sightseeing bus management system and meteorological platform, segments the future itinerary, and uses data fusion technology to label future weather data on future sightseeing bus itinerary sub-segments. The photovoltaic power generation prediction model is then used to predict the power generation of all future sightseeing bus itinerary sub-segments.

[0015] Demand analysis module: Build a random forest regression model, process the future sightseeing bus itinerary data and future weather data, and input them into the model to predict the energy consumption demand of the sightseeing bus's future itinerary;

[0016] Decision-making management module: obtains power generation, energy consumption demand and current power data of sightseeing vehicles, calculates the comprehensive control value, compares the comprehensive control value with the corresponding control range, and formulates the control strategy based on the comparison results;

[0017] Dynamic supervision module: Executes control strategies, obtains control strategy execution data in real time through the edge network, generates and applies optimization strategies through dynamic supervision models, and dynamically manages the charging and discharging of sightseeing vehicles.

[0018] Furthermore, standardized application programming interface technology is used to collect future travel and future weather data. Future travel includes departure time, route data, estimated vehicle passenger load and driver style data. Future weather data includes temperature, humidity, wind speed and weather conditions.

[0019] Furthermore, the segmentation of the future itinerary includes segmenting the stationary state of the sightseeing car by combining time series analysis and segmenting the driving state of the sightseeing car by using a spatial clustering algorithm to obtain the sightseeing car itinerary subsegments;

[0020] The steps for using data fusion technology to identify future weather data in future sightseeing bus itinerary sub-segments are as follows:

[0021] Deeply structure the future sightseeing bus itinerary segment data so that the future sightseeing bus itinerary segment contains a unique identifier, start and end timestamps, spatial coordinate sequence, state type, and associated segment ID;

[0022] Standardize and convert future weather data into a unified time granularity;

[0023] Based on the timestamps and spatial coordinates of the future sightseeing bus itinerary sub-segments, an association is established with future weather data;

[0024] Process spatial data and temporal data based on spatial dimension interpolation and temporal dimension interpolation;

[0025] Use the DTW algorithm to align the time series of travel and weather data caused by vehicle speed fluctuations;

[0026] Construct a semantic network, using named entity recognition technology from natural language processing and association rule algorithms from data mining to automatically extract and annotate semantic tags, thus forming semantic labels for each sub-segment of the future sightseeing bus itinerary;

[0027] Semantic tags are divided into basic tags and environmental tags.

[0028] Furthermore, the steps for predicting the power generation of the sightseeing bus trip sub-segment using the photovoltaic power generation prediction model are as follows:

[0029] Get the semantic labels of the future sightseeing bus itinerary sub-segments;

[0030] The data in the database are searched based on the basic tags of the future sightseeing bus itinerary sub-segments, and the data with the same basic tags are retained to obtain a preliminary screening data set;

[0031] Compare the environmental labels of the future sightseeing bus itinerary sub-segments with the environmental labels in the preliminary screening dataset and calculate the similarity. The specific formula is as follows:

[0032]

[0033] Where: Simx is the similarity, a is an adjustable parameter, 0<a<1, m is the number of dimensions of the environmental label, wi is the weight data of the i-th environmental label dimension, Lvi is the value of the i-th environmental label dimension in the environmental label of the future trip sub-segment, Lsi is the value of the i-th environmental label dimension in the environmental label of the historical trip sub-segment in the preliminary screening dataset, Gjsim is the intermediate similarity, exp is the exponential function with the natural constant e as the base, b is an adjustable parameter, 0<b<1.

[0034] Arrange the similarities from high to low, select the power generation corresponding to the first K groups of similarities, and perform weighted processing on the power generation to obtain the power generation of the sightseeing car journey sub-segment. The weighted processing formula is as follows:

[0035]

[0036] Where Pf is the predicted power generation of the sightseeing car trip sub-segment, Pfsj is the power generation of the jth group, e is a natural constant, c is the attenuation coefficient, c>0, c is fitted using the least squares objective function, and ΔTj is the interval between the time point corresponding to the i-th Pfsj and the current time.

[0037] Furthermore, the steps for building the random forest regression model are as follows:

[0038] Obtain historical data, clean the collected data, and then encode the categorical data;

[0039] Use the minimum-maximum normalization method to process historical data and divide the processed data into training set and test set;

[0040] From the training set, construct the training set of each decision tree by random sampling with replacement;

[0041] Randomly select a part of the features from all the features to construct the nodes of the decision tree, find the optimal split point according to the criterion of minimizing the mean square error, build a decision tree, and form a random forest regression model;

[0042] The random forest regression model is trained using the test set, and the mean square error evaluation indicator is used to output the random forest regression model whose test accuracy is greater than or equal to the preset threshold.

[0043] Furthermore, the comprehensive control value = (rated power of the sightseeing car + power generation of all future sightseeing car travel sub-segments - energy consumption demand - safety power) / rated power of the sightseeing car × 100%, and the control range includes the charging range (-100%, BZ1), the fluctuation range [BZ1, BZ2] and the discharging range (BZ2, 100%).

[0044] Furthermore, the comprehensive control value is compared with the corresponding control interval, and a control strategy is formulated based on the comparison results, including:

[0045] If the comprehensive control value is in the charging range, that is, the first charging adopts night valley charging, the sightseeing bus is charged to full power, and the second charging adopts dynamic charging, then the control strategy is first charging + dynamic charging;

[0046] If the comprehensive control value is not in the charging range, the secondary control value is calculated, which is: (current power of the sightseeing car + power generation of all future sightseeing car trip sub-segments - energy consumption demand - safety power) / rated power of the sightseeing car × 100%;

[0047] If the secondary control value is within the charging interval, night valley charging is adopted, and the charging capacity = |secondary control value × sightseeing car rated capacity|. In this case, the control strategy is single night charging.

[0048] If the secondary control value is within the fluctuation range, the control strategy is to not perform charging or discharging.

[0049] If the secondary control value is in the discharge range, discharge is performed, and the discharge capacity = secondary control value × rated capacity of the sightseeing car, and dynamic discharge is performed. At this time, the control strategy is dynamic discharge.

[0050] Furthermore, the dynamic charging and discharging schemes are optimized through a charging and discharging strategy model based on a deep Q network;

[0051] During dynamic charging, the input of the charging and discharging strategy optimization model based on the deep Q network is the charging power, dynamic electricity price data and the loss data corresponding to the charging speed. The output is the optimal charging power, charging schedule and charging amount at each moment.

[0052] The input of the charging and discharging strategy optimization model based on the deep Q network during dynamic discharge is the discharge power, dynamic electricity price data and loss data corresponding to the discharge speed, and the output is the optimal discharge power, discharge time schedule and discharge amount at each moment.

[0053] Furthermore, the edge network is used to obtain the control strategy execution data in real time. Sensors deployed on sightseeing vehicles and charging facilities on the edge network are used to collect the charging temperature, discharging temperature, current electricity price, charging power and discharging power in real time when the charging and discharging strategy is executed.

[0054] When the control strategy execution data is collected through the edge network, it is transmitted to the edge server in real time through the edge network. The edge server performs preliminary cleaning and preprocessing on the data, removes outliers and duplicate data, and stores the processed data in the local cache.

[0055] Furthermore, the dynamic supervision model includes:

[0056] Data input layer: Receives charging and discharging strategy execution data transmitted by the edge network and performs normalization processing;

[0057] Feature extraction layer: Uses spatiotemporal graph neural networks to extract spatiotemporal features from data and analyzes the correlation between charge and discharge status, electricity price, and temperature in time series and spatial distribution.

[0058] Decision Analysis Layer: Based on a deep Q-network decision-making mechanism, with the goal of minimizing charging and discharging costs and maximizing energy efficiency, it combines the results of the feature extraction layer to calculate Q values ​​under different states, evaluate the pros and cons of the current charging and discharging strategy, and adjust the strategy based on the dual factors of electricity price and charging temperature.

[0059] Strategy output layer: outputs optimization strategies, including discharge optimization strategies or charging optimization strategies;

[0060] Charging optimization strategy includes charging power and charging time;

[0061] The discharge optimization strategy includes discharge power and discharge time.

[0062] (3) Beneficial effects

[0063] The present invention provides a new energy tourist sightseeing vehicle power battery charge and discharge control system, which has the following beneficial effects:

[0064] 1. The present invention describes a new energy tourist sightseeing vehicle power battery charge and discharge control system. It cooperates with the existing sightseeing vehicle management system to collect its departure time, route, load, driving style and other travel data, combines meteorological data to segment the journey and correlate weather data, and generates a charge and discharge control strategy after power generation forecasting, demand analysis, and decision management. The strategy is fed back to the sightseeing vehicle management system, and the edge network is used to monitor the execution data in real time to form a closed loop, realize the dynamic matching of energy flow and operation scheduling, improve energy utilization efficiency and operational economy, and has wide adaptability, good use effect, and good application prospects.

[0065] 2. The present invention records a new energy tourist sightseeing vehicle power battery charging and discharging control system. When in use, it improves energy management accuracy through multi-dimensional data integration and spatiotemporal modeling: it collects sightseeing vehicle travel data and meteorological data, uses time series analysis and spatial clustering algorithms to segment the travel into stationary / driving states, and uses data fusion technology to achieve deep coupling of travel sub-segments and weather data. Combined with the similarity retrieval and exponential weighting algorithm of the photovoltaic power generation prediction model, it provides charging and discharging decision-making with finer spatiotemporal granularity and stronger environmental adaptability. Power generation prediction support, significantly improving the scientificity and reliability of energy flow management of new energy sightseeing vehicles, and the overall use effect is relatively good.

[0066] 3. The present invention records a new energy tourist sightseeing vehicle power battery charging and discharging control system, which realizes accurate estimation of power generation in spatiotemporal scenarios through multi-source data fusion and photovoltaic power generation prediction model, and trains the cleaned coded historical data through random forest regression model to improve the accuracy of energy consumption demand prediction. It realizes power balance and cost optimization based on comprehensive control value and multi-interval threshold combined with night valley power, dynamic charging and discharging and V2G discharge strategies, and further collects data in real time through edge network, generates optimization strategies through spatiotemporal graph neural network and deep Q network, realizes multi-dimensional dynamic optimization of charging and discharging of new energy sightseeing vehicles, significantly improves energy utilization efficiency, battery life and grid coordination capability, has good use effect and has good prospects for use. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 This is a flow chart of the power battery charging and discharging control system for a new energy tourist sightseeing vehicle according to the present invention. DETAILED DESCRIPTION

[0068] The following will clearly and completely describe the technical solutions in the embodiments of the present invention 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 making creative efforts are within the scope of protection of the present invention.

[0069] The central idea of ​​the solution recorded in the existing patent is to judge the state of the electric vehicle's power battery, realize the judgment of the power battery's charge and discharge state based on the corresponding mechanism, and further study and improve the charging and discharging of the electric vehicle to realize the adjustment of the charging and discharging speed.

[0070] However, in actual use, this solution has the following limitations:

[0071] Limited applicability: This method is suitable for daily charging and discharging (e.g., leaving home early and returning late to avoid peak and valley periods). However, for long-term use in scenic spots, where daily operation varies, it is not possible to achieve scheduled discharge during peak periods, nor to charge only during valley periods.

[0072] Low degree of coordination: This system mainly realizes the separate management of power batteries and cannot be combined with usage needs, and cannot be dynamically planned and adjusted based on usage needs and electricity prices.

[0073] The present invention breaks through the data uniformity, extensive prediction and static control of existing technologies through multiple innovations in multi-source data fusion, photovoltaic power generation prediction, energy consumption demand prediction and dynamic decision-making and optimization technologies, and constructs a full-process intelligent system of data collection-space-time modeling-intelligent decision-making-dynamic optimization, which significantly improves the refinement and efficiency of energy management of new energy sightseeing vehicles, provides reusable technical specifications for energy charging and discharging management of new energy sightseeing vehicles, and lays the technical foundation for charging and discharging management.

[0074] Example 1

[0075] See also Figure 1 This embodiment provides a new energy tourist sightseeing vehicle power battery charge and discharge control system, the specific scheme is as follows:

[0076] 1. System architecture;

[0077] 1.1 Data collection;

[0078] Data collection is responsible for real-time collection of various data related to the charging and discharging management of sightseeing vehicles, and various sensors are deployed on sightseeing vehicles and charging facilities.

[0079] Data collection mainly collects vehicle status, environment and charging and discharging parameters. The vehicle status depends on the vehicle status sensor. The vehicle status sensor collects battery status data such as battery remaining power, battery voltage, current, etc., and monitors the battery health status and charging and discharging capabilities in real time, providing a basic basis for the formulation of charging and discharging strategies. For example, by obtaining the remaining battery power, the current energy reserve status of the vehicle can be judged.

[0080] Environmental data relies on environmental sensors, including temperature sensors and light intensity sensors, which are used to collect environmental data such as charging temperature and solar radiation intensity. The layout of sensors in use depends on the characteristics of the scenic area. For example, the influence of wind needs to be considered in mountainous areas and plains, so additional corresponding sensors need to be arranged.

[0081] For scenic spots with lower requirements, environmental data can be obtained directly from the meteorological platform. The amount of data obtained by this method is relatively small and the accuracy is also low. For scenic spots with high requirements, this method is not applicable.

[0082] The charging and discharging parameters are obtained through sensors installed on the charging equipment. They can monitor parameters such as charging power, discharging power, actual charging and discharging amount in real time, so as to understand the real-time status of the charging and discharging process and provide data support for subsequent strategy optimization.

[0083] 1.2 Transmission network;

[0084] The network layer bears the heavy responsibility of data transmission and adopts edge network technology to achieve efficient and stable transmission of perception layer data. It mainly uses communication technologies such as 5G networks and low-power wide area networks to quickly transmit data collected by sensors to edge servers. With its high speed and low latency, the 5G network ensures that data with high real-time requirements (such as urgent battery anomaly data) can be transmitted in a timely manner; low-power wide area networks are suitable for sensor data transmission that is sensitive to power consumption and has a small data transmission volume, reducing the overall energy consumption of the system. At the same time, the edge network has certain local computing and data processing capabilities, which can perform preliminary cleaning and preprocessing of data during the data transmission process, reducing invalid data transmission and improving data transmission efficiency and quality.

[0085] For the data collected from the sightseeing bus management system and the meteorological platform, standardized application program interface (API) technology is used to establish a communication connection through the RESTful or SOAP protocol based on the Web API specifications provided by the system. This interface technology can accurately locate and obtain future itinerary data stored in the sightseeing bus management system, including structured data such as vehicle operation plans, driving routes, departure times, and station stop information. For data acquisition from the meteorological platform, it also relies on its open data interface to request future weather data in the prescribed data format (such as JSON, XML), covering multi-dimensional meteorological information such as temperature, humidity, light intensity, wind speed and direction, and precipitation probability. Through data interface technology, efficient and stable docking of data sources is achieved, laying the foundation for subsequent data collection.

[0086] When collecting data, use ETL technology to deeply process the connected data.

[0087] For example, in the data extraction stage, for the sightseeing bus management system, future travel data is extracted regularly or in real time according to the preset time period and data screening rules; for the meteorological platform data, accurate extraction is carried out according to the geographical area and time range. In the data conversion stage, the extracted heterogeneous data is formatted in a unified manner, such as standardizing the time format of different data sources and unifying the units of numerical data. At the same time, the data is cleaned to remove duplicate records, correct erroneous data, and fill in missing values. Finally, through data loading operations, the processed future travel data and weather data are stored in the system's data warehouse or designated database table, providing standardized and high-quality data support for subsequent data analysis and model training.

[0088] For the execution data of control strategies, sensor nodes and communication modules are deployed on each sightseeing vehicle with the help of Internet of Things (IoT) technology. Onboard sensors collect real-time data on the vehicle's operating status, such as current location, driving speed, battery power, etc. These data are transmitted to the sightseeing vehicle management system in real time via wireless networks (such as 4G / 5G, NB-IoT), thereby ensuring that future travel data in the system has the characteristics of real-time updates. At the same time, the meteorological platform also relies on widely distributed meteorological monitoring stations and IoT sensor networks, such as meteorological satellites, ground weather stations, meteorological radars and other equipment, to collect meteorological data in real time, aggregate and process it through the data center, and finally provide it to the system through the data interface, ensuring the timeliness and accuracy of future weather data.

[0089] 1.3 Data Analysis

[0090] Data analysis is the core of this system, which integrates multiple key models and data processing functions to achieve in-depth analysis of data and optimized strategy formulation.

[0091] Data analysis includes data preprocessing, namely data cleaning, integration, storage and standardization.

[0092] Data analysis also includes various models, such as power generation prediction, random forest regression model calculation, decision making in the decision management module, and dynamic management of the dynamic supervision model.

[0093] 1.4 Strategy Application

[0094] The strategy application mainly involves applying the data analysis results to actual scenarios and realizing the monitoring and management of the charging and discharging process of sightseeing vehicles.

[0095] For example, it receives optimized charging and discharging strategies and controls the operating parameters of the sightseeing vehicle charging and discharging equipment, such as adjusting the charging power, starting or stopping the charging and discharging operations, etc., to ensure that the sightseeing vehicle is charged and discharged according to the optimal strategy and realize the rational use of energy.

[0096] For example, the edge server can monitor various data during the charging and discharging process of the sightseeing vehicle in real time. Once abnormal data is found (such as the charging temperature is too high, the charging and discharging power fluctuations are too large, etc.), an alarm will be issued immediately and the abnormal information will be uploaded to the management platform. Management personnel can take timely measures based on the alarm information to ensure the safe operation of vehicles and equipment.

[0097] For example, it provides managers with a visual interface to display information such as the operating status of sightseeing vehicles, charging and discharging data, and strategy execution effects. At the same time, the management platform regularly evaluates and optimizes the dynamic supervision model based on real-time monitoring data and user feedback, adjusts model parameters and algorithms, and achieves continuous improvement and optimization of the system.

[0098] 2. System Solution

[0099] A new energy tourist sightseeing vehicle power battery charging and discharging control system, such as Figure 1 As shown, it includes power generation forecast module, demand analysis module, decision management module and dynamic supervision module:

[0100] This system is based on the data of the existing sightseeing bus management system. Only when the sightseeing bus management system has formulated the corresponding sightseeing bus plan can subsequent analysis and judgment be realized. Therefore, the basis of this system is data collection and processing.

[0101] In this system, sufficient data must be collected to ensure operation. The data collection and processing are based on the power generation prediction module. The power generation prediction module extracts future travel and weather data from the sightseeing car management system and meteorological platform, segments the future travel, and uses data fusion technology to mark the future weather data on the future sightseeing car travel sub-segments. The photovoltaic power generation prediction model is used to predict the power generation of all future sightseeing car travel sub-segments.

[0102] Standardized application programming interface technology is used to collect future travel and future weather data. Future travel includes departure time, route data, estimated vehicle passenger load and driver style data. Future weather data includes temperature, humidity, wind speed and weather conditions.

[0103] In order to ensure the accuracy and periodicity of the data, 1 day will be used in the future.

[0104] Therefore, the future itinerary collected is the corresponding data for the sightseeing bus on the next day, which generally includes information such as departure time, driving route, stop points, expected driving time, and number of people on board. In order to ensure the operation of the scenic area, traffic control is usually adopted to ensure that the sightseeing bus operates according to the settings under normal circumstances. Therefore, the accuracy of the system analysis can be guaranteed.

[0105] For example, vehicle numbered 003 departs at 9:00, has 4 stops in total, runs for 2 hours, stops at the 2nd stop for 3 minutes, stops at the 3rd stop for 5 minutes, stops at the last stop for 20 minutes before returning, with a total mileage of 50 kilometers. It departs at 14:00, has 4 stops in total, runs for 2 hours, stops at the 2nd stop for 3 minutes, stops at the 3rd stop for 5 minutes, stops at the last stop for 20 minutes before returning, with a total mileage of 50 kilometers.

[0106] Future weather data can be obtained through the weather API to obtain the future weather forecast along the route, including date, time, location, weather conditions (sunny / rainy / cloudy, etc.), solar radiation intensity (W / m 2 ), temperature, humidity, wind speed, etc.

[0107] For example, at 9 o'clock, the weather is sunny, the solar radiation intensity is 800, and the temperature is 25℃; at 14 o'clock, the weather is sunny, the solar radiation intensity is 1000, and the temperature is 33℃.

[0108] After collecting the relevant data, since the vehicle does not maintain a uniform state, it is necessary to further subdivide the vehicle's operation, that is, divide its daily formation into multiple stages, and then evaluate and analyze each stage separately.

[0109] Segmenting the future itinerary includes segmenting the stationary state of the sightseeing car by combining time series analysis and segmenting the driving state of the sightseeing car by using a spatial clustering algorithm to obtain the sightseeing car itinerary subsegments;

[0110] The steps for using data fusion technology to identify future weather data in future sightseeing bus itinerary sub-segments are as follows:

[0111] Deeply structure the data of future sightseeing bus itineraries so that they contain unique identifiers, start and end timestamps, spatial coordinate sequences, status types, and associated road segment IDs, facilitating subsequent segment processing.

[0112] After data consistency processing, the collected vehicle speed data is subjected to time series analysis. A speed threshold (e.g., 0.5 km / h) is set. When the vehicle speed is below this threshold and the stationary time exceeds 2 minutes, it is determined to be in a stationary state. Using a time series analysis algorithm, feature extraction is performed on the stationary time series data. Statistics such as the mean, variance, and autocorrelation coefficient of the data are calculated to analyze the stability and periodicity of the stationary state. When the characteristics of the time series show significant changes, the segment node of the stationary state is determined. The continuous stationary state is divided into different sub-segments, and the start timestamp, end timestamp, duration, and vehicle location coordinates of each sub-segment are recorded.

[0113] The geographic coordinate sequence of the vehicle during its travel is obtained based on GPS positioning data. A spatial clustering algorithm, such as the density peak clustering algorithm (DBSCAN), is used to perform cluster analysis on the geographic coordinate points. Adjacent geographic coordinate points with high density are grouped together according to a pre-set neighborhood radius and density threshold, forming different spatial clusters. Each cluster corresponds to a specific area on the sightseeing vehicle's travel path, such as a road within a scenic area or a road section around a scenic spot. The process of a vehicle entering from one cluster to another is regarded as a segmented node of the driving state, and the continuous driving process is divided into multiple sub-segments based on spatial regions. Each sub-segment contains information such as the starting cluster, the ending cluster, the driving path, and the driving time.

[0114] Standardize and convert future weather data into a unified time granularity (e.g., 10 minutes, adjusted based on accuracy) to ensure consistency in data format;

[0115] The divided sub-segment data for future sightseeing bus itineraries is deeply structured, with each segment assigned a unique identifier (UUID). The data also records the start and end timestamps, spatial coordinate sequence (including multiple geographic coordinate points), state type (stationary or moving), and associated segment ID (matched using the scenic area's electronic map). The structured segment data is stored in a relational database for easy query and processing.

[0116] Acquired future weather data is standardized and converted to a uniform time granularity, supplementing missing data and removing outliers. Data with discontinuous timestamps is supplemented using linear interpolation. Abnormal meteorological data (such as temperatures outside the normal range) is corrected or eliminated through comparative analysis with data from surrounding time points. Furthermore, a geographic coordinate field is added to each meteorological data record, and geocoding aligns the location of meteorological monitoring stations with the coordinate system of the scenic area map, forming a standardized meteorological dataset.

[0117] Based on the timestamps and spatial coordinates of the future sightseeing bus itinerary sub-segments, an association is established with future weather data;

[0118] Based on the timestamps and spatial coordinates of the future sightseeing bus itinerary sub-segments, an association is established with future weather data. For the stationary sub-segments, the coordinates of their stop locations are used as a benchmark to match the contemporaneous data of the nearest meteorological monitoring station. For the driving sub-segments, the meteorological data of the corresponding time are associated with the key nodes on their driving paths (such as the starting point, midpoint, and end point of the section). During the association process, the dual matching of time and space is considered to ensure the accuracy and consistency of the data.

[0119] Process spatial data and temporal data based on spatial dimension interpolation and temporal dimension interpolation;

[0120] Under normal circumstances, spatial dimension interpolation uses a combination of inverse distance weighted interpolation and Kriging interpolation. However, when there is complex terrain or areas with large fluctuations in meteorological data, the Kriging algorithm is switched to model spatial autocorrelation through the variation function to improve interpolation accuracy. Since this process is frequently used in existing technologies, it will not be described in detail.

[0121] Under normal circumstances, when interpolating the time dimension, for travel sub-segments across time periods, when the time interval of the meteorological data is inconsistent with the time interval of the travel sub-segment, the cubic spline interpolation method is used to fit a smooth curve based on the data of the previous and next moments, and calculate the estimated values ​​of the meteorological parameters at each time point in the sub-segment to ensure data continuity in the time dimension.

[0122] Use the DTW algorithm to align the time series of travel and weather data caused by vehicle speed fluctuations;

[0123] To address the time misalignment between travel and weather data caused by vehicle speed fluctuations, a dynamic time warping algorithm is used to align the time series. Using the time series of the travel sub-segments as a benchmark, the Euclidean distance matrix between them and the weather data time series is calculated. Dynamic programming is used to find the optimal curved path, elastically stretching or compressing the weather data time axis to ensure that each travel sub-segment precisely matches the weather conditions of the corresponding time period.

[0124] Construct a semantic network, using named entity recognition technology from natural language processing and association rule algorithms from data mining to automatically extract and annotate semantic tags, thus forming semantic labels for each sub-segment of the future sightseeing bus itinerary;

[0125] Semantic tags are divided into basic tags and environmental tags.

[0126] The generated semantic tags are divided into two categories: basic tags and environmental tags. Basic tags mainly include the segment ID, start and end timestamps, trip distance, driving direction, and status type, which are used to describe the basic attributes of the trip sub-segment. Environmental tags mainly include weather conditions, solar radiation intensity, temperature, humidity, wind speed, etc., which are used to describe the environmental conditions corresponding to the trip sub-segment. By classifying semantic tags, it is easier to perform subsequent tag-based data retrieval and analysis, which can reduce the workload of subsequent processing and improve the efficiency of analysis.

[0127] The steps for predicting the power generation of the sightseeing bus trip sub-segment using the photovoltaic power generation prediction model are as follows:

[0128] Obtain semantic labels for future sightseeing bus itinerary sub-segments. For historical data, use named entity recognition technology from natural language processing and association rule algorithms from data mining to automatically extract and annotate semantic labels.

[0129] Based on the basic tags of the future sightseeing bus itinerary sub-segments, the data in the database is retrieved, and SQL connection queries are performed in the database to retain the data with the same basic tags to obtain a preliminary screening data set, mainly to eliminate data and reduce workload;

[0130] For example, when searching for a road section, only the data of that road section is retained in the preliminary screening data set, which can significantly reduce the amount of data for subsequent comparison and analysis.

[0131] In order to facilitate the subsequent similarity calculation, the variables in the environmental labels need to be one-hot encoded and converted into binary vectors. For example, sunny days are represented as [1,0,0,0,0,0,0], and cloudy days are represented as [0,1,0,0,0,0,0]. This encoding method maps categorical variables to Euclidean space, which is convenient for subsequent similarity calculation.

[0132] The numerical labels in the environmental labels need to be normalized. The normalization process uses the minimum-maximum normalization method to map the data to the [0,1] interval. If you want to improve the accuracy, you can also use Z-score normalization for further processing.

[0133] Compare the environmental labels of the future sightseeing bus itinerary sub-segments with the environmental labels in the preliminary screening dataset and calculate the similarity. The specific formula is as follows:

[0134]

[0135] Where: Simx is the similarity, a is an adjustable parameter, 0<a<1, m is the number of dimensions of the environmental label, wi is the weight data of the i-th environmental label dimension, Lvi is the value of the i-th environmental label dimension in the environmental label of the future trip sub-segment, Lsi is the value of the i-th environmental label dimension in the environmental label of the historical trip sub-segment in the preliminary screening dataset, Gjsim is the intermediate similarity, exp is the exponential function with the natural constant e as the base, b is an adjustable parameter, 0<b<1.

[0136] Based on the weighted processing method combining the weighted cosine similarity algorithm and the improved distance similarity, the final similarity calculated by the improved similarity calculation more comprehensively reflects the similarity between the two environmental label sets, which can further improve the accuracy of the calculation, thereby making the calculated results more representative.

[0137] When a is close to 1, the final similarity depends more on the weighted cosine similarity; when a is close to 0, the final similarity depends more on the improved Euclidean distance similarity.

[0138] The larger the b value, the more significant the impact of distance on similarity. Even if the numerical difference between the two label sets is small, the similarity result will change significantly. The smaller the b value, the smaller the impact of distance on similarity.

[0139] a is used to balance the weights of cosine similarity and distance similarity, and b is used to adjust the influence of distance on similarity.

[0140] In application, when a is high, b can be appropriately reduced to avoid excessive interference of distance in the judgment of directional similarity. When a is low, b needs to be used to balance the rationality of distance scaling to avoid the similarity result being too close to 0 or 1.

[0141] Arrange the similarities from high to low, select the power generation corresponding to the first K groups of similarities, and perform weighted processing on the power generation to obtain the power generation of the sightseeing car journey sub-segment. The weighted processing formula is as follows:

[0142]

[0143] Where Pf is the predicted power generation of the sightseeing car trip sub-segment, Pfsj is the power generation of the jth group, e is a natural constant, c is the attenuation coefficient, c>0, c is fitted using the least squares objective function, and ΔTj is the interval between the time point corresponding to the i-th Pfsj and the current time.

[0144] The principle of designing this formula is that the closer to the current time point, the higher the credibility of the data. Because the longer the time, the greater the change in the external environment. Therefore, it is necessary to use data at different times to assign attenuation weights so that recent data contributes more to the prediction results. The weight tends to decrease with the distance from the current time.

[0145] In order to reflect the importance of recent data, the exponential decay method is adopted, and the least squares method is used for fitting the attenuation factor. Regarding the least squares method fitting, it uses a computer to process data, mainly converting the objective function into a linear regression problem, minimizing the fitting error of the logarithmic weight and time, and converting it through the linear least squares formula to obtain the attenuation factor, and then directly substituted the attenuation factor into the formula, so no excessive description is given.

[0146] The present invention records a new energy tourist sightseeing vehicle power battery charging and discharging control system, which cooperates with the existing sightseeing vehicle management system to collect its departure time, route, load, driving style and other travel data, combines meteorological data to segment the journey and associate weather data, and generates a charging and discharging control strategy after power generation forecasting, demand analysis, and decision-making management and feeds it back to the sightseeing vehicle management system. It uses the edge network to monitor the execution data in real time to form a closed loop, realizes dynamic matching of energy flow and operation scheduling, improves energy utilization efficiency and operational economy, has wide adaptability, good use effect, and has good prospects for use.

[0147] After collecting relevant data and pre-processing the data for basic analysis, in order to understand whether to discharge or charge, further demand analysis is required, and demand analysis depends on the demand analysis module.

[0148] The demand analysis module constructs a random forest regression model, processes the future sightseeing bus itinerary data and future weather data, and inputs them into the model to predict the energy consumption demand of the sightseeing bus's future itinerary;

[0149] The steps to build a random forest regression model are as follows:

[0150] Obtain historical data, clean the collected data, and then encode the categorical data;

[0151] For example, check whether there are missing values ​​or outliers in the data, fill missing values ​​with mean, median or estimate based on relevant data; remove or correct outliers.

[0152] For example, for categorical data, such as driving routes, weather conditions (sunny, rainy, cloudy, etc.), driver numbers, etc., they are converted into numerical data using one-hot encoding or label encoding.

[0153] The historical data were processed using the minimum-maximum normalization method, and the processed data were divided into a training set and a test set; the ratio of the training set to the test set was 7:3.

[0154] From the training set, construct the training set of each decision tree by random sampling with replacement;

[0155] Randomly select a part of the features from all the features to construct the nodes of the decision tree, the number is The optimal split point is found by minimizing the mean squared error. For example, if the number of features is 16, then when splitting, 4 features are randomly selected (one node is divided into two sub-nodes, and this process is repeated until a stopping condition is met, such as the number of node samples is less than a certain threshold or the depth of the tree reaches a preset value), and a decision tree is constructed to form a random forest regression model.

[0156] The random forest regression model is trained using the test set, and the mean square error evaluation indicator is used to output the random forest regression model whose test accuracy is greater than or equal to the preset threshold.

[0157] When making predictions, for each decision tree, the test sample starts from the root node and traverses downward according to the node splitting conditions until it reaches the leaf node. The value of the leaf node is the predicted value of the decision tree for the test sample. Then the average of all predicted values ​​is calculated, and the average value is the output value of the random forest regression model.

[0158] For example, there is a set of historical data for sightseeing buses, including driving routes, driving distance, temperature, driving score, number of passengers, and actual power consumption data:

[0159] First, process the data by one-hot encoding the driving routes and normalizing all other numerical data. Then, divide the data into a training set and a test set. Use the training set to train a random forest regression model. Assume that after hyperparameter tuning, the number of decision trees is set to 50, with a maximum depth of 10. After training, validate the model using the test set. If the mean squared error on the test set is 1.2 and the mean absolute error is 0.8, the model has good predictive performance.

[0160] When a future sightseeing bus itinerary is scheduled as "Route A, driving distance 20 kilometers", the future weather data is "temperature 27℃", the driver data is "driving style score 7 points", and the load data is "15 passengers", these data are input into the trained model, and the predicted power consumption output by the model is the predicted power demand value of the sightseeing bus in this situation.

[0161] When understanding the power generation and demand, it is necessary to conduct a comprehensive analysis of the data in order to make appropriate decisions, which depends on the decision management module.

[0162] Decision-making management module: obtains power generation, energy consumption demand and current power data of sightseeing vehicles, calculates the comprehensive control value, compares the comprehensive control value with the corresponding control range, and formulates the control strategy based on the comparison results;

[0163] Comprehensive control value = (rated power of sightseeing car + power generation of all future sightseeing car travel sub-segments - energy consumption demand - safety power) / rated power of sightseeing car × 100%. The safety power is generally preset to 20% of the rated power of the sightseeing car. The control range includes the charging range (-100%, BZ1), the fluctuation range [BZ1, BZ2] and the discharging range (BZ2, 100%). The fluctuation range is for example [-8%, 8%].

[0164] Compare the comprehensive control value with the corresponding control interval, and formulate the control strategy based on the comparison results, including:

[0165] If the comprehensive control value is in the charging range, that is, the first charging adopts night valley charging, the sightseeing bus is charged to full power, and the second charging adopts dynamic charging, then the control strategy is first charging + dynamic charging;

[0166] This type of vehicle is not enough to complete the journey on one charge and requires a second charge. When in use, the arrangement of tourist sightseeing buses is more reasonable. Under normal circumstances, the power demand of the sightseeing buses will not reach twice the rated power of the sightseeing buses. Because there are a large number of sightseeing buses, they can be arranged reasonably.

[0167] Moreover, this application formulates a strategy for tourist sightseeing buses in scenic spots during off-peak hours. During busy hours, discharge and electricity prices are not considered, and they are charged as much as possible to meet the operation of the scenic spot.

[0168] If the comprehensive control value is not in the charging range, the secondary control value is calculated, which is: (current power of the sightseeing car + power generation of all future sightseeing car trip sub-segments - energy consumption demand - safety power) / rated power of the sightseeing car × 100%;

[0169] This means that the vehicle is fully charged enough, but the vehicle will not remain in a sufficient state if it is driven every day. Further analysis is needed to determine the charging or discharging capacity.

[0170] If the secondary control value is within the charging interval, night valley charging is adopted, and the charging capacity = |secondary control value×rated capacity of the sightseeing vehicle|. In this case, the control strategy is single night charging.

[0171] In this case, it means that the current power is insufficient and needs to be recharged.

[0172] If the secondary control value is within the fluctuation range, the control strategy is to not perform charging or discharging.

[0173] This situation indicates that the power level is just right, so no action is taken.

[0174] If the secondary control value is in the discharge range, discharge is performed, and the discharge capacity = secondary control value × rated capacity of the sightseeing car, and dynamic discharge is performed. At this time, the control strategy is dynamic discharge.

[0175] This situation indicates that there is sufficient electricity. Generally, this situation occurs during the off-season and weekdays. From an economic perspective, selling excess electricity to the power grid can maximize economic benefits.

[0176] The dynamic charging and discharging scheme is optimized through a charging and discharging strategy model based on a deep Q network;

[0177] During dynamic charging, the input of the charging and discharging strategy optimization model based on the deep Q network is the charging power, dynamic electricity price data and the loss data corresponding to the charging speed. The output is the optimal charging power, charging schedule and charging amount at each moment.

[0178] The input of the charging and discharging strategy optimization model based on the deep Q network during dynamic discharge is the discharge power, dynamic electricity price data and loss data corresponding to the discharge speed, and the output is the optimal discharge power, discharge time schedule and discharge amount at each moment.

[0179] The model architecture used for dynamic charging and dynamic discharging is the same, with only the input data and output results being different.

[0180] The decision logic layer of the model learns the optimal charging strategy based on the reward mechanism and the optimal discharging strategy based on the reward mechanism through continuous interaction between the intelligent agent and the environment.

[0181] The core of the model is to plan to minimize the total charging cost and maximize the total discharging benefit.

[0182] Discharging supports the V2G mode and feeds excess power back to the grid.

[0183] That is, the objective function of the model is to minimize the charging cost during dynamic charging, and to maximize the discharge benefit during dynamic discharging. There are certain constraints in the model, namely the safe charging requirements, that is, the power range and temperature range.

[0184] By using optimization algorithms such as the stochastic gradient descent algorithm, the strategy parameters of the intelligent agent are continuously adjusted while satisfying the constraints to optimize the objective function, thereby determining the optimal charging and discharging power, time and power at each moment.

[0185] The present invention describes a new energy tourist sightseeing vehicle power battery charge and discharge control system, which, when in use, improves energy management accuracy through multi-dimensional data integration and spatiotemporal modeling: it collects sightseeing vehicle travel data and meteorological data, uses time series analysis and spatial clustering algorithms to segment the travel into stationary / driving states, and uses data fusion technology to achieve deep coupling of travel sub-segments and weather data. Combined with the similarity retrieval and exponential weighting algorithm of the photovoltaic power generation prediction model, it provides power generation prediction support with finer spatiotemporal granularity and stronger environmental adaptability for charging and discharging decisions, significantly improving the scientific nature and reliability of energy flow management for new energy sightseeing vehicles, and achieving relatively good overall use effect.

[0186] Once the corresponding charge and discharge control strategy is formulated, it will be strictly implemented in accordance with the charge and discharge control strategy. However, in actual use, it will be affected by human factors such as the external environment, resulting in certain changes in the implementation of the plan. Therefore, dynamic supervision of the charge and discharge process is required. This step relies on the dynamic supervision module.

[0187] The dynamic supervision module executes the control strategy, obtains the control strategy execution data in real time through the edge network, generates and applies the optimization strategy through the dynamic supervision model, and dynamically manages the charging and discharging of the sightseeing vehicle.

[0188] Acquiring control strategy execution data in real time through the edge network is to use sensors deployed on sightseeing vehicles and charging facilities on the edge network to collect real-time charging temperature, discharging temperature, current electricity price, charging power and discharging power when the charging and discharging strategy is executed;

[0189] When the control strategy execution data is collected through the edge network, it is transmitted to the edge server in real time through the edge network. The edge server performs preliminary cleaning and preprocessing on the data, removes outliers and duplicate data, and stores the processed data in the local cache.

[0190] As a core component of scenic area sightseeing vehicle charging and discharging management, the dynamic supervision model achieves real-time monitoring and dynamic optimization of the charging and discharging process through the coordinated operation of the data input layer, feature extraction layer, decision analysis layer, and strategy output layer. Based on edge network data acquisition, the model combines spatiotemporal graph neural network and deep Q network technology, comprehensively considering both electricity price and charging temperature. It aims to reduce charging and discharging costs, improve energy efficiency, and ensure the efficiency and stability of sightseeing vehicle energy management.

[0191] The dynamic regulatory model includes:

[0192] Data input layer: Receives charging and discharging strategy execution data transmitted by the edge network and performs normalization processing;

[0193] For example, at a certain moment, the sensor collects data such as the charging power is 30kW, the remaining battery power is 40kWh, the charging temperature is 25℃, and the current electricity price is 0.8 yuan / kWh.

[0194] For example, if the charging power is normalized, the historical minimum value is 5kW, the maximum value is 50kW, and the current charging power is 30kW, then the normalized charging power is 0.56. The normalized data will be used as the input of the feature extraction layer to provide a standardized data basis for subsequent analysis.

[0195] Feature extraction layer: Uses spatiotemporal graph neural networks to extract spatiotemporal features from data and analyzes the correlation between charge and discharge status, electricity price, and temperature in time series and spatial distribution.

[0196] A spatiotemporal graph neural network is used to extract spatiotemporal features from the data. In charging and discharging management scenarios, data not only exhibits temporal variations but also spatial correlations between different sightseeing vehicles and charging facilities. The spatiotemporal graph neural network uses graph structure modeling, treating each sightseeing vehicle or charging facility as a node in the graph. The connections between nodes represent their spatial correlations. For example, vehicles in close geographic locations may be affected by similar environmental factors. In the temporal dimension, a recurrent neural network processes the time series data of each node to capture the data's changing trends over time.

[0197] For example, by analyzing the correlation between charging and discharging status, electricity prices, and temperature in time series and spatial distribution, the spatiotemporal graph neural network can learn the changing patterns of charging power at different times of the day, as well as the correlation between such changes and electricity price fluctuations and temperature changes.

[0198] For example, by studying historical data, it was found that when electricity prices are low and the temperature is suitable, the vehicle charging power tends to be higher.

[0199] In terms of spatial distribution, the spatiotemporal graph neural network can analyze the differences in charging and discharging status of vehicles in different regions, as well as the relationship between these differences and local electricity price policies and ambient temperature.

[0200] For example, in different parking lots in a scenic area, due to different electricity prices and temperature distributions, there are obvious differences in the charging and discharging behaviors of vehicles. By extracting and analyzing such spatiotemporal features, more in-depth and valuable information can be provided to the decision-making analysis layer.

[0201] Decision Analysis Layer: Based on a deep Q-network decision-making mechanism, with the goal of minimizing charging and discharging costs and maximizing energy efficiency, it combines the results of the feature extraction layer to calculate Q values ​​under different states, evaluate the pros and cons of the current charging and discharging strategy, and adjust the strategy based on the dual factors of electricity price and charging temperature.

[0202] A two-factor strategy is used for adjustment. If the charging temperature is high, the charging speed is reduced. If the electricity price is low at this time, the charging speed is reduced slightly. If the electricity price is high at this time, the charging speed is greatly reduced.

[0203] Strategy output layer: outputs optimization strategies, including discharge optimization strategies or charging optimization strategies;

[0204] Charging optimization strategy includes charging power and charging time;

[0205] For example, the output is to charge at a power of 25kW in the next 2 hours.

[0206] The discharge optimization strategy includes discharge power and discharge time.

[0207] For example, in the next hour, the battery is discharged at a power of 15 kW.

[0208] These optimization strategies are transmitted via the edge network to the sightseeing vehicle's charge and discharge management system, guiding the vehicle's actual charging and discharging operations and enabling dynamic optimization management of the vehicle's charging and discharging process. Simultaneously, the system continuously monitors the effectiveness of the strategy execution and feeds relevant data back to the data input layer, forming a closed-loop management system that continuously optimizes the charging and discharging strategies and improves management effectiveness.

[0209] The present invention records a new energy tourist sightseeing vehicle power battery charging and discharging control system. Through multi-source data fusion and photovoltaic power generation prediction model, it realizes accurate estimation of power generation in spatiotemporal scenarios, and trains the cleaned coded historical data through random forest regression model to improve the accuracy of energy consumption demand prediction. Based on comprehensive control value and multi-interval threshold combined with night valley power, dynamic charging and discharging and V2G discharge strategies, it realizes power balance and cost optimization. Further, through real-time data collection through edge network, optimization strategy is generated through spatiotemporal graph neural network and deep Q network, which realizes multi-dimensional dynamic optimization of charging and discharging of new energy sightseeing vehicles, significantly improves energy utilization efficiency, battery life and grid coordination capability, has good use effect and has good prospects for use.

[0210] The weight coefficient involved in the above formula is determined by the coefficient of variation method. The coefficient of variation method is a method of weighting each indicator according to the degree of variation between the current value of each evaluation indicator and the target value. If the numerical difference of a certain indicator is large, it can clearly distinguish the evaluated objects, indicating that the indicator has rich discrimination information, and thus the indicator should be given a larger weight. On the contrary, if the numerical difference of each evaluated object on a certain indicator is small, then the ability of this indicator to distinguish the evaluation objects is weak, and thus the indicator should be given a smaller weight. This method directly uses the information contained in each indicator to obtain the weight of the indicator through calculation, and therefore is objective.

[0211] In this application, the several formulas involved are all calculated by taking their numerical values ​​after removing the dimensions, and the formulas are established by collecting a large amount of data and performing software simulation to obtain a formula for the most recent real situation. Some coefficients or weights in the formulas are set by technical personnel in this field according to actual conditions, so they will not be elaborated here.

[0212] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art will appreciate that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution.

[0213] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, and may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment as needed.

[0214] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. A new energy sightseeing vehicle power battery charge and discharge control system, characterized by: include: Power generation prediction module: This module extracts future itinerary and weather data from the sightseeing bus management system and meteorological platform, segments the future itinerary, and uses data fusion technology to label future weather data on future sightseeing bus itinerary sub-segments. The photovoltaic power generation prediction model is then used to predict the power generation of all future sightseeing bus itinerary sub-segments. Demand analysis module: Build a random forest regression model, process the future sightseeing bus itinerary data and future weather data, and input them into the model to predict the energy consumption demand of the sightseeing bus's future itinerary; Decision-making management module: obtains power generation, energy consumption demand and current power data of sightseeing vehicles, calculates the comprehensive control value, compares the comprehensive control value with the corresponding control range, and formulates the control strategy based on the comparison results; Dynamic supervision module: Executes control strategies, obtains control strategy execution data in real time through the edge network, generates and applies optimization strategies through dynamic supervision models, and dynamically manages the charging and discharging of sightseeing vehicles.

2. A new energy sightseeing vehicle power battery charge and discharge control system according to claim 1, characterized in that: Standardized application programming interface technology is used to collect future travel and future weather data. Future travel includes departure time, route data, estimated vehicle passenger load and driver style data. Future weather data includes temperature, humidity, wind speed and weather conditions.

3. A new energy sightseeing vehicle power battery charge and discharge control system according to claim 2, characterized in that: Segmenting the future itinerary includes segmenting the stationary state of the sightseeing car by combining time series analysis and segmenting the driving state of the sightseeing car by using a spatial clustering algorithm to obtain the sightseeing car itinerary subsegments; The steps for using data fusion technology to identify future weather data in future sightseeing bus itinerary sub-segments are as follows: Deeply structure the future sightseeing bus itinerary segment data so that the future sightseeing bus itinerary segment contains a unique identifier, start and end timestamps, spatial coordinate sequence, state type, and associated segment ID; Standardize and convert future weather data into a unified time granularity; Based on the timestamps and spatial coordinates of the future sightseeing bus itinerary sub-segments, an association is established with future weather data; Process spatial data and temporal data based on spatial dimension interpolation and temporal dimension interpolation; Use the DTW algorithm to align the time series of travel and weather data caused by vehicle speed fluctuations; Construct a semantic network, using named entity recognition technology from natural language processing and association rule algorithms from data mining to automatically extract and annotate semantic tags, thus forming semantic labels for each sub-segment of the future sightseeing bus itinerary; Semantic tags are divided into basic tags and environmental tags.

4. A new energy sightseeing vehicle power battery charge and discharge control system according to claim 3, characterized in that: The steps for predicting the power generation of the sightseeing bus trip sub-segment using the photovoltaic power generation prediction model are as follows: Get the semantic labels of the future sightseeing bus itinerary sub-segments; The data in the database are searched based on the basic tags of the future sightseeing bus itinerary sub-segments, and the data with the same basic tags are retained to obtain a preliminary screening data set; Compare the environmental labels of the future sightseeing bus itinerary sub-segments with the environmental labels in the preliminary screening dataset and calculate the similarity. The specific formula is as follows: Where: Simx is the similarity, a is an adjustable parameter, 0<a<1, m is the number of dimensions of the environment label, wi is the weight data of the i-th environment label dimension, Lvi is the value of the i-th environment label dimension in the future trip sub-segment environment label, Lsi is the value of the i-th environment label dimension in the historical trip sub-segment environment label in the preliminary screening dataset, Gjsim is the intermediate similarity, exp is the exponential function with the natural constant e as the base, b is an adjustable parameter, 0<b<1. Arrange the similarities from high to low, select the power generation corresponding to the first K groups of similarities, and perform weighted processing on the power generation to obtain the power generation of the sightseeing car journey sub-segment. The weighted processing formula is as follows: Where Pf is the predicted power generation of the sightseeing car trip sub-segment, Pfsj is the power generation of the jth group, e is a natural constant, c is the attenuation coefficient, c>0, c is fitted using the least squares objective function, and ΔTj is the interval between the time point corresponding to the i-th Pfsj and the current time.

5. A new energy sightseeing vehicle power battery charge and discharge control system according to claim 4, characterized in that: The steps to build a random forest regression model are as follows: Obtain historical data, clean the collected data, and then encode the categorical data; Use the minimum-maximum normalization method to process historical data and divide the processed data into training set and test set; From the training set, construct the training set of each decision tree by random sampling with replacement; Randomly select a part of the features from all the features to construct the nodes of the decision tree, find the optimal split point according to the criterion of minimizing the mean square error, build a decision tree, and form a random forest regression model; The random forest regression model is trained using the test set, and the mean square error evaluation indicator is used to output the random forest regression model whose test accuracy is greater than or equal to the preset threshold.

6. A new energy sightseeing vehicle power battery charge and discharge control system according to claim 5, characterized in that: Comprehensive control value = (rated power of sightseeing car + power generation of all future sightseeing car travel sub-segments - energy consumption demand - safety power) / rated power of sightseeing car × 100%. The control range includes the charging range (-100%, BZ1), the fluctuation range [BZ1, BZ2] and the discharging range (BZ2, 100%).

7. A new energy sightseeing vehicle power battery charge and discharge control system according to claim 6, characterized in that: Compare the comprehensive control value with the corresponding control interval, and formulate the control strategy based on the comparison results, including: If the comprehensive control value is in the charging range, that is, the first charging adopts night valley charging, the sightseeing bus is charged to full power, and the second charging adopts dynamic charging, then the control strategy is first charging + dynamic charging; If the comprehensive control value is not in the charging range, the secondary control value is calculated, which is: (current power of the sightseeing car + power generation of all future sightseeing car trip sub-segments - energy consumption demand - safety power) / rated power of the sightseeing car × 100%; If the secondary control value is within the charging interval, night valley charging is adopted, and the charging capacity = |secondary control value × sightseeing car rated capacity|. In this case, the control strategy is single night charging. If the secondary control value is within the fluctuation range, the control strategy is to not perform charging or discharging. If the secondary control value is in the discharge range, discharge is performed, and the discharge capacity = secondary control value × rated capacity of the sightseeing car, and dynamic discharge is performed. At this time, the control strategy is dynamic discharge.

8. A new energy sightseeing vehicle power battery charge and discharge control system according to claim 7, characterized in that: The dynamic charging and discharging scheme is optimized through a charging and discharging strategy model based on a deep Q network; During dynamic charging, the input of the charging and discharging strategy optimization model based on the deep Q network is the charging power, dynamic electricity price data and the loss data corresponding to the charging speed. The output is the optimal charging power, charging schedule and charging amount at each moment. The input of the charging and discharging strategy optimization model based on the deep Q network during dynamic discharge is the discharge power, dynamic electricity price data and loss data corresponding to the discharge speed, and the output is the optimal discharge power, discharge time schedule and discharge amount at each moment.

9. A new energy sightseeing vehicle power battery charge and discharge control system according to claim 8, characterized in that: Acquiring control strategy execution data in real time through the edge network is to use sensors deployed on sightseeing vehicles and charging facilities on the edge network to collect real-time charging temperature, discharging temperature, current electricity price, charging power and discharging power when the charging and discharging strategy is executed; When the control strategy execution data is collected through the edge network, it is transmitted to the edge server in real time through the edge network. The edge server performs preliminary cleaning and preprocessing on the data, removes outliers and duplicate data, and stores the processed data in the local cache.

10. A new energy sightseeing vehicle power battery charge and discharge control system according to claim 9, characterized in that: The dynamic regulatory model includes: Data input layer: Receives charging and discharging strategy execution data transmitted by the edge network and performs normalization processing; Feature extraction layer: Uses spatiotemporal graph neural networks to extract spatiotemporal features from data and analyzes the correlation between charge and discharge status, electricity price, and temperature in time series and spatial distribution. Decision Analysis Layer: Based on a deep Q-network decision-making mechanism, with the goal of minimizing charging and discharging costs and maximizing energy efficiency, it combines the results of the feature extraction layer to calculate Q values ​​under different states, evaluate the pros and cons of the current charging and discharging strategy, and adjust the strategy based on the dual factors of electricity price and charging temperature. Strategy output layer: outputs optimization strategies, including discharge optimization strategies or charging optimization strategies; Charging optimization strategy includes charging power and charging time; The discharge optimization strategy includes discharge power and discharge time.

Images