WO2023109099A1 - 基于非侵入式检测的充电负荷概率预测系统及方法 - Google Patents

基于非侵入式检测的充电负荷概率预测系统及方法 Download PDF

Info

Publication number
WO2023109099A1
WO2023109099A1 PCT/CN2022/103801 CN2022103801W WO2023109099A1 WO 2023109099 A1 WO2023109099 A1 WO 2023109099A1 CN 2022103801 W CN2022103801 W CN 2022103801W WO 2023109099 A1 WO2023109099 A1 WO 2023109099A1
Authority
WO
WIPO (PCT)
Prior art keywords
charging
vehicle
image
charging load
historical
Prior art date
Application number
PCT/CN2022/103801
Other languages
English (en)
French (fr)
Inventor
龙虹毓
詹小胜
徐洋
尹霄
黄昭成
赵胤豪
何宇强
刘上华
Original Assignee
重庆邮电大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 重庆邮电大学 filed Critical 重庆邮电大学
Publication of WO2023109099A1 publication Critical patent/WO2023109099A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01K—MEASURING TEMPERATURE; MEASURING QUANTITY OF HEAT; THERMALLY-SENSITIVE ELEMENTS NOT OTHERWISE PROVIDED FOR
    • G01K13/00—Thermometers specially adapted for specific purposes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/067—Enterprise or organisation modelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/10—Protocols in which an application is distributed across nodes in the network
    • H04L67/1097—Protocols in which an application is distributed across nodes in the network for distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/02—Services making use of location information
    • H04W4/029—Location-based management or tracking services
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/30—Services specially adapted for particular environments, situations or purposes
    • H04W4/38—Services specially adapted for particular environments, situations or purposes for collecting sensor information
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/30—Services specially adapted for particular environments, situations or purposes
    • H04W4/40—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P]
    • H04W4/44—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P] for communication between vehicles and infrastructures, e.g. vehicle-to-cloud [V2C] or vehicle-to-home [V2H]

Definitions

  • the invention relates to the technical field of charging load forecasting, in particular to a charging load probabilistic forecasting system and method based on non-invasive detection.
  • the charging load of electric vehicles has strong randomness in both time and space. With the popularization of electric vehicles in the future, its charging load will have an increasing impact on the operation of urban distribution networks, especially for central Small city grid.
  • the electric vehicle charging load is connected to the power grid on a large scale, on the one hand, it will affect the power quality of the distribution network, it will bring voltage offset, three-phase imbalance and harmonic pollution to the distribution network, and further increase the peak-to-valley difference It directly affects the reliability of the power grid; on the other hand, after large-scale access, the network loss of the distribution network and the service life of the transformer will change, which will affect the economic operation of the power grid.
  • Electric vehicle charging load forecasting is the basis for the analysis of the impact of electric vehicle access on the grid, the planning and control operation of distribution network, the two-way interaction between electric vehicles and the grid, and the coordination research between electric vehicles and other energy and transportation systems. Due to the temporal and spatial randomness of the charging behavior of electric vehicles, the prediction of charging load involves very complex influencing factors, and different consideration angles will form different load forecasting models and results. For the normal and reliable operation of the urban power grid, the charging load of electric vehicles in the future should be accurately predicted. It is urgent to investigate and analyze the impact of large-scale access of electric vehicles on the grid structure, power quality, load curve, dispatching control, etc. An adaptive solution for the coordinated development of automobiles, so as to more effectively promote the promotion and application of new energy electric vehicles.
  • the current load forecasting methods for electric vehicles are mainly based on the factors affecting the load of electric vehicles, which are divided into short-term load forecasting methods for power systems, Monte Carlo simulation methods, and other new electric vehicle load forecasting methods.
  • short-term load forecasting methods for power systems Monte Carlo simulation methods, and other new electric vehicle load forecasting methods.
  • the impact of temperature control load is not considered, that is, in colder or hotter seasons, the opening of the air conditioner in the car will increase the power consumption of electric vehicles and aggravate the driver's "mileage anxiety". ", making the charging needs of electric vehicles more frequent.
  • the object of the present invention is to provide a charging load probabilistic prediction system and method based on non-intrusive detection, adding environmental temperature factors to predict the charging demand of electric vehicles, so as to reasonably select the layout planning of charging facilities.
  • One aspect of the present invention is to provide a charging load probability prediction system based on non-invasive detection, including:
  • the image acquisition unit captures the pictures of passing vehicles through the traffic cameras deployed on each road, and processes the picture data
  • the temperature detection unit obtains the ambient temperature through a temperature sensor installed on the side of the traffic camera;
  • a positioning unit installed on the traffic camera, is used to obtain the position of the traffic camera, and then obtain the position of the electric vehicle at the time of shooting;
  • the Internet of Vehicles service system based on the geographical location of the traffic camera, obtains the charging station information around at this time;
  • the cloud storage computing platform is connected in communication with the image acquisition unit, the temperature detection unit, the Internet of Vehicles service system, and the positioning unit, and is used for data processing and storage of the historical vehicle pictures collected by the image acquisition unit, and by Identify license plate information, vehicle owner's head deflection action, and vehicle owner's facial expression characteristics from historical vehicle picture data, combine electric vehicle historical air conditioner on status data and historical charging records to train the charging load prediction model, and use it to receive the traffic camera collection and processing in real time
  • the vehicle picture and the real-time temperature data of the temperature sensor and classify the data obtained by vehicle picture recognition according to the temperature data range, and input it into the corresponding charging load prediction model to obtain the probability of when and where the target will be charged and the charging load. Time accumulation is used for the planning and construction of charging facilities in certain areas.
  • the image acquisition unit includes:
  • Traffic cameras are deployed on various traffic roads to capture pictures of passing vehicles
  • An image preprocessing module connected in communication with the traffic camera, for performing data enhancement, normalization, and grayscale preprocessing on the passing vehicle pictures;
  • the first communication module uses a 5G network to connect to the cloud storage computing platform, and uploads the preprocessed vehicle image to the cloud storage computing platform;
  • the temperature detection unit includes:
  • a temperature sensor installed beside the traffic camera, is used to monitor the ambient temperature
  • the third communication module uses a 5G network to connect to the cloud storage computing platform, and uploads the ambient temperature to the cloud storage computing platform;
  • the positioning unit includes:
  • a positioning module configured to obtain the position of the traffic camera, and then obtain the position of the electric vehicle at the moment of shooting;
  • the fourth communication module uses a 5G network to connect to the cloud storage computing platform, and uploads location information data to the cloud storage computing platform;
  • the car networking service system includes:
  • the database module is used to store the charging station information in various places
  • Charging recording module used to store historical charging data of various electric vehicles
  • the air conditioner use status recording module is used to store the historical air conditioner use status data of various electric vehicles
  • the fifth communication module uses a 5G network to connect to the cloud storage computing platform, and uploads charging station information, historical charging data of electric vehicles, and historical air conditioner usage status data to the cloud storage computing platform.
  • the cloud storage computing platform includes:
  • the historical image database module is used to store preprocessed historical vehicle pictures to form a historical database
  • the image recognition module is used to carry out the license plate information, the head deflection action of the owner, and the facial expression characteristics of the owner of the electric vehicle in the processed vehicle picture;
  • the model training module by identifying the license plate information, the head deflection action of the car owner, and the facial expression characteristics of the car owner from the historical vehicle picture data, combined with the historical data of the electric vehicle's air conditioner on state and historical charging records to train the charging load prediction model;
  • the model prediction module is used to receive the vehicle picture collected and processed by the traffic camera in real time and the real-time temperature data of the temperature sensor, classify the data obtained by vehicle picture identification according to the temperature data range, input it into the corresponding charging load prediction model, and obtain The probability of when and where the target will be charged and the charging load are accumulated over time to obtain the planning and construction of charging facilities in certain areas;
  • the second communication module uses a 5G network to connect the image acquisition unit, the temperature detection unit, the Internet of Vehicles service system, and the positioning unit.
  • Another aspect of the present invention is to provide a charging load probability prediction method based on non-intrusive detection, comprising the following steps:
  • the temperature sensor next to the traffic camera uses the temperature sensor next to the traffic camera to detect the ambient temperature, determine the corresponding temperature of the photo taken at a certain time, and judge whether the air conditioner in the car is turned on. If the air conditioner in the car is turned on, input the first load demand forecasting model to predict the charging load. If the air conditioner is not turned on, input the second load demand forecasting model for charging load forecasting;
  • the picture database is composed of historical picture data of m cameras, t time points, and n targets, with a total of mnt basic pictures.
  • the identification and acquisition of the license plate information includes the following steps:
  • License plate image preprocessing converting the vehicle image from the RGB channel to the HSV channel, converting the HSV channel image into a grayscale image, and then binarizing and morphologically processing the grayscale image;
  • License plate location using the function cv2.findContours() to perform rectangle detection on the grayscale image after morphological processing, locate the license plate area, and segment the license plate area;
  • Segmenting license plate characters performing horizontal correction of the license plate, removal of the license plate frame and rivets, and character segmentation operations in sequence on the segmented license plate area, wherein the horizontal correction of the license plate includes tilt angle detection and tilt correction;
  • License plate character recognition convert the characters on the image into feature vectors through HOG feature extraction, and classify and distinguish through the SVM classification algorithm;
  • Carry out data processing on the pictures of passing vehicles obtain vehicle type information and vehicle owner's driving position information, and judge the vehicle owner's head deflection direction and the corresponding direction of the instrument panel through the vehicle type information and the vehicle owner's driving position information;
  • the RNN main loop and LSTM information memory obtains the convolution sampling feature vector from the feature sampling layer as input; then, enters the loop network according to the time series, and the loop unit LSTM extracts information to generate state information; finally, outputs the feature vector for classification;
  • ⁇ (t) represents the output path ⁇ at time t
  • y t represents the network output of RNN at time t
  • the predicted label conditional probability is expressed as the sum of the conditional probabilities of the corresponding Path,
  • V denotes the operator that transforms the output path ⁇ into the target L
  • the expression of the input image at the current moment is the label with the highest conditional probability:
  • the random forest model gives the predicted value of the sample by averaging the predicted values of all regression trees for:
  • the ambient temperature detected by the temperature sensor as 14° C. to 28° C. under the condition of not driving the in-vehicle air conditioner, otherwise it is under the condition of turning on the in-vehicle air conditioner.
  • the charging load probability prediction method based on non-intrusive detection of the present invention adopts the method of non-invasive detection of charging probability and charging load, and everything after obtaining the load demand prediction model does not need to be associated with the car owner, so as to avoid affecting the car owner emotions, resulting in large errors in forecast results.
  • the charging load probability prediction method based on non-intrusive detection of the present invention receives the image data collected and processed by the traffic camera and the real-time temperature data of the temperature sensor in real time, and classifies the image data according to the temperature results, and inputs the corresponding load demand prediction model , so as to predict the probability and charging load of when and where the target electric vehicle will be charged.
  • the invention can predict the electric vehicle charging demand in the prediction area, so as to reasonably select the layout planning of the charging facilities.
  • FIG. 1 is a schematic diagram of the system architecture of the charging load probability prediction system based on non-invasive detection in the present invention
  • Fig. 2 is a schematic flow chart of the steps of the charging load probability prediction method based on non-invasive detection in the present invention
  • Figure 3 is a simplified diagram of the driver's head rotation model.
  • An embodiment of the present invention provides a charging load probability prediction system based on non-intrusive detection, as shown in Figure 1, including an image acquisition unit, a temperature detection unit, a positioning unit, a car networking service system, and a cloud storage computing platform.
  • the image acquisition unit acquires pictures of passing vehicles through the traffic cameras deployed on each road, and performs image data processing; the temperature detection unit acquires the ambient temperature through the temperature sensor installed next to the traffic camera; the positioning unit is installed on the traffic camera It is used to obtain the position of the traffic camera, and then obtain the location of the electric vehicle at the time of shooting; the Internet of Vehicles service system, based on the geographical location of the traffic camera, obtains the information of the surrounding charging stations at this time.
  • the cloud storage computing platform communicates with the image acquisition unit, temperature detection unit, Internet of Vehicles service system, and positioning unit, and is used for data processing and storage of historical vehicle pictures collected by the image acquisition unit, and through historical vehicle picture data
  • the license plate information, the head deflection action of the car owner, and the facial expression characteristics of the car owner are recognized, and the charging load prediction model is trained in combination with the historical air-conditioning status data and historical charging records of electric vehicles, as well as the vehicle pictures and temperature sensors used for real-time reception of traffic camera collection and processing.
  • the real-time temperature data of the vehicle is classified according to the range of the temperature data, and the data obtained by vehicle picture recognition is input into the corresponding charging load prediction model to obtain the probability of when and where the target will be charged and the charging load. Through time accumulation, what can be calculated?
  • the probability of charging when and where and the charging load are used in the planning and construction of charging facilities in certain areas.
  • the image acquisition unit includes a traffic camera, an image preprocessing module and a first communication module.
  • Traffic cameras are deployed on various traffic roads to capture pictures of passing vehicles;
  • the image preprocessing module communicates with the traffic cameras and is used to perform data enhancement, normalization, and grayscale processing on the pictures of passing vehicles;
  • a communication module which uses a 5G network to connect to the cloud storage computing platform, and uploads the preprocessed vehicle image to the cloud storage computing platform;
  • the temperature detection unit includes a temperature sensor and a third communication module.
  • the temperature sensor is installed next to the traffic camera to monitor the ambient temperature; the third communication module uses a 5G network to connect to the cloud storage computing platform, and uploads the ambient temperature to the cloud storage computing platform. platform.
  • the positioning unit includes a positioning module and a fourth communication module.
  • the positioning module is used to obtain the position of the traffic camera, and then obtain the position of the electric vehicle at the time of shooting;
  • the data is uploaded to the cloud storage computing platform.
  • the Internet of Vehicles service system includes a database module and a charging record module.
  • the database module is used to store charging station information in various places;
  • the charging record module is used to store historical charging data of various The historical air conditioner usage status data of electric vehicles;
  • the fifth communication module using 5G network to connect to the cloud storage computing platform, uploads charging station information, historical charging data of electric vehicles and historical air conditioner usage status data to the cloud storage computing platform.
  • the cloud storage computing platform includes a historical image database module, an image recognition module, a model training module, a model prediction module and a second communication module.
  • the historical image database module is connected in communication with the image pre-processing module, and is used to store the pre-processed historical vehicle pictures to form a historical database;
  • the image recognition module is connected in communication with the image pre-processing module, and is used for processing the vehicle pictures. Carry out license plate information, owner's head deflection movements, and vehicle owner's facial expression characteristics for electric vehicles;
  • the model training module recognizes license plate information, vehicle owner's head deflection movements, and vehicle owner's facial expression characteristics from historical vehicle image data, combined with the history of electric vehicles.
  • State data and historical charging records train the charging load prediction model;
  • the model prediction module is used to receive real-time vehicle pictures collected and processed by traffic cameras and real-time temperature data from temperature sensors, and classify the data obtained by vehicle picture recognition according to the range of temperature data. Input the corresponding charging load prediction model to get the probability of when and where the target will be charged and the charging load. Through time accumulation, the planning and construction of charging facilities in certain areas are obtained; the second communication module uses 5G network to connect the image acquisition unit, Temperature detection unit, car networking service system, positioning unit.
  • S20 Perform data processing on historical vehicle pictures to obtain license plate information, vehicle owner's head deflection action, and vehicle owner's facial expression features;
  • S60 Use the temperature sensor on the side of the traffic camera to detect the ambient temperature, determine the corresponding temperature of the photos taken at a certain time, determine whether the air conditioner in the car is turned on, and set the ambient temperature detected by the temperature sensor to 14°C to 28°C to avoid driving Under the condition of the air conditioner in the car, otherwise it is the condition of turning on the air conditioner in the car. If the air conditioner in the car is turned on, input the first load demand forecasting model for charging load forecasting; if the air conditioner is not turned on, input the second load demand forecasting model for charging load forecasting;
  • the picture database is composed of historical picture data of m cameras, t time points, and n targets, with a total of mnt basic pictures.
  • the identification and acquisition of license plate information includes the following steps:
  • License plate image preprocessing converting the vehicle image from the RGB channel to the HSV channel, converting the HSV channel image into a grayscale image, and then performing binarization and morphological processing on the grayscale image;
  • License plate location use the function cv2.findContours() to perform rectangle detection on the grayscale image after morphological processing, locate the license plate area, and segment the license plate area;
  • License plate horizontal correction includes tilt angle detection and tilt correction
  • the vehicle owner's head deflection action is obtained as follows:
  • Carry out data processing on the pictures of passing vehicles obtain vehicle model information and vehicle owner's driving position information, and judge the vehicle owner's head deflection direction and the corresponding direction of the dashboard through the vehicle model information and vehicle owner's driving position information;
  • the driver's head rotation model can be simplified to the model shown in Figure 3.
  • the line segment formed by connecting the two eyes of the driver with his head facing straight ahead is represented as a line segment
  • the spatial connecting line segment between the two eyes is represented as a line segment
  • the line segment is numerically and are equal
  • the projection of the distance between the eyes on the forward plane after head rotation is expressed as a line segment That is, the distance between the two eyes shown in the two-dimensional image after the head is rotated, the line segment can be obtained through geometric analysis with line segment
  • the acute angle between is equal to the angle of the head rotation, then the calculation formula of the head rotation angle is expressed as
  • ⁇ value is negative, it is judged that the head is rotating to the left; if the value of ⁇ is positive, it is judged that the head is rotating right; if the right rotation angle of the head is greater than 30 degrees, it is judged that the driver is looking at the dashboard .
  • the acquisition of the owner's facial expression feature is as follows:
  • the RNN main loop and LSTM information memory obtains the convolution sampling feature vector from the feature sampling layer as input; then, enters the loop network according to the time series, and the loop unit LSTM extracts information to generate state information; finally, outputs the feature vector for classification;
  • ⁇ (t) represents the output path ⁇ at time t
  • y t represents the network output of RNN at time t
  • the predicted label conditional probability is expressed as the sum of the conditional probabilities of the corresponding Path,
  • V denotes the operator that transforms the output path ⁇ into the target L
  • the expression of the input image at the current moment is the label with the highest conditional probability:
  • the charging load demand forecasting model is established as follows:
  • the random forest model gives the predicted value of the sample by averaging the predicted values of all regression trees for:

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Signal Processing (AREA)
  • General Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Health & Medical Sciences (AREA)
  • General Business, Economics & Management (AREA)
  • Tourism & Hospitality (AREA)
  • Marketing (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Evolutionary Biology (AREA)
  • Computational Linguistics (AREA)
  • Development Economics (AREA)
  • Biophysics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Mathematical Physics (AREA)
  • Game Theory and Decision Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Educational Administration (AREA)
  • Public Health (AREA)
  • Water Supply & Treatment (AREA)
  • Primary Health Care (AREA)

Abstract

涉及充电负荷预测技术领域，尤其涉及基于非侵入式检测的充电负荷概率预测系统及方法，用来解决充电负荷预测以及规划问题。基于非侵入式检测的充电负荷概率预测系统包括图像采集单元、温度检测单元、定位单元、车联网服务系统、云存储计算平台，在云存储计算平台中能够实时接收交通摄像头采集的车辆图片和实时温度数据，并按温度数据范围对车辆图片识别获取的数据进行分类，输入对应的充电负荷预测模型中，得到目标将要何时何地充电的概率以及充电负荷，通过时间积累，可以将何时何地充电的概率以及充电负荷用于某些地区的充电设施规划建设。可以对预测区域内的电动汽车充电需求进行预测，并合理选择充电设施的布局规划。

Description

基于非侵入式检测的充电负荷概率预测系统及方法 技术领域
本发明涉及充电负荷预测技术领域，尤其涉及基于非侵入式检测的充电负荷概率预测系统及方法。
背景技术
电动汽车的充电负荷在时间与空间上都存在很强的随机性，随着今后电动汽车的普及，其充电负荷会对城市配电网的运行带来越来越大的影响，尤其是对中小型城市电网。电动汽车充电负荷大规模接入电网后，一方面对配电网电能质量的影响，它会给配电网带来电压偏移、三相不平衡和谐波污染，使峰谷差进一步拉大直接影响电网的可靠性；另一方面，大规模接入后使得配电网的网损和变压器使用年限发生变化，影响电网的经济运行。
电动汽车充电负荷预测是开展电动汽车接入对电网的影响分析、配电网规划与控制运行、电动汽车与电网双向互动及电动汽车与其他能源、交通等系统协调研究的基础。由于电动汽车的充电行为具有时间上和空间上的随机性，充电负荷的预测涉及十分复杂的影响因素，不同的考虑角度将形成不同的负荷预测模型和结果。为了城市电网能正常可靠运行，应准确预测出未来电动汽车的充电负荷，迫切需要调研分析电动汽车大规模接入对电网结构、电能质量、负荷曲线、调度控制等方面的影响，形成电网与电动汽车协调发展的适应性方案，从而更有效促进新能源电动汽车的推广和应用。目前对电动汽车的负荷预测方法，主要是基于电动汽车负荷影响因素，分为电力系统短期负荷预测方法、蒙特卡洛模拟法以及其他新型电动汽车负荷预测方法。但从目前的研究方法来看，其中并没有考虑温控负荷的影响，即在较冷或较热的季节时，车内空调的开启会增加电动汽车耗电量，加剧驾驶员的“里程焦虑”，使电动汽车的充电需求更加频繁。
环境温度引起的电动汽车用户频繁充电行为，加剧了典型季节的电网负载，温度成为电动汽车充电负荷预测不可忽略的影响因素之一。目前将环境温度作为充电需求预测影响因素之一的研究甚少。
发明内容
有鉴于此，本发明的目的是提供基于非侵入式检测的充电负荷概率预测系统及方法，加入环境温度影响因素对电动汽车充电需求进行预测，以便合理选择充电设施的布局规划。
本发明通过以下技术手段解决上述技术问题：本发明的一方面在于提供了一种基于非侵入式检测的充电负荷概率预测系统，包括：
图像采集单元，通过部署在各道路上的交通摄像头拍摄获取过往车辆图片，并进行图片数据处理；
温度检测单元，通过安装在所述交通摄像头旁侧的温度传感器获取环境温度；
定位单元，安装在所述交通摄像头上，用于获取所述交通摄像头的位置，进而获取拍摄时刻电动汽车所处的位置；
车联网服务系统，基于所述交通摄像头的地理位置，获得此时周围的充电站信息；
云存储计算平台，与所述图像采集单元、温度检测单元、车联网服务系统、定位单元均通信连接，用于对所述图像采集单元采集到的历史车辆图片进行数据处理和存储，以及通过从历史车辆图片数据中识别出车牌信息、车主头部偏转动作、车主面部表情特征，结合电动汽车历史空调开启状态数据和历史充电记录训练充电负荷预测模型，以及用于实时接收所述交通摄像头采集处理的车辆图片和温度传感器的实时温度数据，并按温度数据范围对车辆图片识别获取的数据进行分类，输入对应的充电负荷预测模型中，得到目标将要何时何地充电的概率以及充电负荷，通过时间积累，用于某些地区的充电设施规划建设。
进一步，所述图像采集单元包括：
交通摄像头，部署在各交通道路上用于拍摄获取过往车辆图片；
图像预处理模块，与所述交通摄像头通信连接，用于对所述过往车辆图片进行数据增强、归一化、灰度化预处理；
第一通信模块，采用5G网络连接云存储计算平台，将预处理后的车辆图像上传至云存储计算平台；
所述温度检测单元包括：
温度传感器，安装在所述交通摄像头旁侧，用于监测环境温度；
第三通信模块，采用5G网络连接云存储计算平台，将所述环境温度上传至云存储计算平台；
所述定位单元包括：
定位模块，用于获取所述交通摄像头的位置，进而获取拍摄时刻电动汽车所处的位置；
第四通信模块，采用5G网络连接云存储计算平台，将位置信息数据上传至云存储计算平台；
所述车联网服务系统包括：
数据库模块，用于存储各地的充电站信息；
充电记录模块，用于存储各种电动汽车的历史充电数据；
空调使用状态记录模块，用于存储各种电动汽车历史空调使用状态数据；
第五通信模块，采用5G网络连接云存储计算平台，将各地的充电站信息、电动汽车的历史充电数据和历史空调使用状态数据上传至云存储计算平台。
进一步，所述云存储计算平台包括：
历史图像数据库模块，用于存储经过预处理的历史车辆图片，形成历史数据库；
图像识别模块，用于对处理后的车辆图片中的电动汽车进行车牌信息、车主头部偏转动作、车主面部表情特征；
模型训练模块，通过从历史车辆图片数据中识别出车牌信息、车主头部偏转动作、车主面部表情特征，结合电动汽车历史空调开启状态数据和历史充电记录训练充电负荷预测模型；
模型预测模块，用于实时接收所述交通摄像头采集处理的车辆图片和温度传感器的实时温度数据，并按温度数据范围对车辆图片识别获取的数据进行分类，输入对应的充电负荷预测模型中，得到目标将要何时何地充电的概率以及充电负荷，通过时间积累，得到某些地区的充电设施规划建设；
第二通信模块，采用5G网络连接所述图像采集单元、温度检测单元、车联网服务系统、定位单元。
本发明的另一方面在于提供了基于非侵入式检测的充电负荷概率预测方法，包括以下步骤：
通过部署在道路上的交通摄像头拍摄采集过往的历史车辆图片，建立图片数据库；
对所述历史车辆图片进行数据处理，获取车牌信息、车主头部偏转动作、车主面部表情特征；
基于所述车牌信息，通过车联网系统获取该电动汽车充电记录，包括充电地点和充电时间；
以所述车主头部偏转动作和所述车主面部表情特征作为自变量，以充电时间和充电地点作为因变量，分别建立每个车主的充电负荷需求预测模型；
通过车联网系统获取电动汽车历史空调使用状态，通过空调开启信息训练优化充电负荷需求预测模型，获得空调开启时的第一负荷需求预测模型和空调关闭时的第二负荷需求预测模型；
通过交通摄像头旁侧的温度传感器检测环境温度，明确某时刻拍摄的照片的对应温度，判断是否有开启车内空调，若是有开启车内空调则输入第一负荷需求预测模型进行充电负荷预测，若是没有开启空调则输入第二负荷需求预测模型进行充电负荷预测；
再结合交通摄像头的定位单元，能够获取某一时刻某处的电动车将要何时何地充电的概率和充电负荷预测。
进一步，所述图片数据库是由m个摄像头、t个时刻、n个目标的历史图片数据组成，共mnt个基础图片。
进一步，所述车牌信息的识别获取包括以下步骤：
车牌图像预处理，将车辆图片从RGB通道转换到HSV通道，将HSV通道图像转化为灰度图像，再将灰度图像进行二值化和形态学处理；
车牌定位，使用函数cv2.findContours()对形态学处理后的灰度图像进行矩形检测，定位车牌区域，将所述车牌区域分割出来；
车牌字符分割，对分割出来的所述车牌区域依次进行车牌水平矫正、车牌边框和铆钉的去除、字符分割操作，所述车牌水平矫正包括倾斜角度检测、倾斜矫正；
车牌字符识别，通过HOG特征提取将图像上的字符转化成特征向量，并通过SVM分类算法分类判别；
得到结果，通过分类判别识别为是否是新能源汽车，以及车牌号码。
进一步，所述车主头部偏转动作的获取如下：
对过往车辆图片进行数据处理，获取车型信息和车主驾驶位置信息，并通过所述车型信息和所述车主驾驶位置信息判断车主的头部偏转方向与仪表盘的对应方向；
根据人体头部运动的规律，在假设人体头部按照头部中心轴线转动而不发生垂直方向上左右摆动的条件下，将驾驶人头部朝向正前方情况下的两眼连成的线段表示为线段
Figure PCTCN2022103801-appb-000001
头部旋转过一个θ角度后两眼之间的空间连接线段表示为线段
Figure PCTCN2022103801-appb-000002
在数值上线段
Figure PCTCN2022103801-appb-000003
和
Figure PCTCN2022103801-appb-000004
是相等的，头部旋转后两眼之间距离在正向平面上的投影表示为线段
Figure PCTCN2022103801-appb-000005
通过几何分析可以得到线段
Figure PCTCN2022103801-appb-000006
与线段
Figure PCTCN2022103801-appb-000007
之间所夹锐角与头部旋转过的角度相等，则头部旋转角度计算公式表示为
Figure PCTCN2022103801-appb-000008
计算得到的θ值为负值，则判断为头部左旋转，θ值为正值，则判断为头部右旋转。
进一步，所述车主面部表情特征的获取如下：
在特征采样层中，对车辆图片进行图像数据预处理、人脸检测、人脸定位、卷积特征学习和特征采样，输入原始车辆图像序列x＝{x 1,x 2,…,x T}，对图像进行预处理，消除光照影响；在时刻t＝1，采用Fast-CNN方法快速定位人脸图像，并进行分割；在后续时刻，进行快速跟踪人脸图像并分割；然后，人脸图像进入卷积神经网络学习，通过交叉卷积和池化，生成图像抽象特征；最后，输入到K-平均采样层，该层对x T-K+1,x T-K+2,…,x T共K个连续图像特征进行平均采样，作为输入特征进入循环网络学习；
RNN主循环和LSTM信息记忆，从特征采样层获取卷积采样特征向量作为输入；然后，根据时间序列进入循环网络，循环单元LSTM提取信息生成状态信息；最后，输出特征向量进行分类；
对循环网络层学习到的特征向量输出序列进行分类，设T为序列长度，L为标签的长度，则当序列T出现时标签路径为L的T次方种可能，称其中每一种可能称为一条“Path”，其条件概率公式
Figure PCTCN2022103801-appb-000009
其中，π (t)表示在t时刻的输出路径π，y t表示在时刻t的RNN的网络输出，
预测标签条件概率表示为相应Path的条件概率之和，
Figure PCTCN2022103801-appb-000010
其中，V表示将输出路径π转换为目标L的运算符，
当输入序列进行训练时，当前时刻输入图像表情为条件概率最大的标签：
Figure PCTCN2022103801-appb-000011
识别出车主愤怒、厌恶、恐惧、高兴、悲伤、惊讶、中性各种表情。
进一步，所述充电负荷需求预测模型建立如下：
假设用N组数据进行模型训练，进行随机森林算法分析：
a.将电动汽车的充电行为作为预测变量X o，把驾驶员在充电前一小时内面部表现出的愤怒、厌恶、恐惧、高兴、悲伤、惊讶、中性和头部偏转这八个因素作为影响变量X i；
b.令抽样次数b＝1,2,…,B s，重复步骤c、d；
c.通过自主重复抽样技术从X o、X i中随机选择一个子样本集作为训练集；
d.通过训练集得到一个回归树模型r fb；
训练结束后，对一个新的样本x，随机森林模型通过平均所有回归树的预测值给出该样本的预测值
Figure PCTCN2022103801-appb-000012
为：
Figure PCTCN2022103801-appb-000013
进一步，所述判断是否有开启车内空调步骤中，设定所述温度传感器检测到的环境温度为14℃～28℃为不开车内空调条件下，否则为开启车内空调条件。
本发明的基于非侵入式检测的充电负荷概率预测方法，采用非侵入式检测充电概率和充电负荷的方法，在获取到负荷需求预测模型之后的一切事情都不需要与车主发生关联，避免影响车主情绪，造成预测结果较大误差。本发明的基于非侵入式检测的充电负荷概率预测方法，通过实时接收交通摄像头采集处理的图像数据，以及温度传感器的实时温度数据，并按温度结果对图像数据分类，输入对应的负荷需求预测模型，从而预测出目标电动汽车将要何时何地充电的概率和充电负荷。本发明可以对预测区域内的电动汽车充电需求进行预测，以便合理选择充电设施的布局规划。
附图说明
图1是本发明基于非侵入式检测的充电负荷概率预测系统的系统架构示意图；
图2是本发明基于非侵入式检测的充电负荷概率预测方法的步骤流程示意图；
图3是驾驶员头部旋转模型简化图。
具体实施方式
下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本发明保护的范围。
本发明的说明书和权利要求书及上述附图中的术语“第一”“第二”等是用于区别不同的对象，而不是用于描述特定的顺序。
本发明实施例中提供了一种基于非侵入式检测的充电负荷概率预测系统，如图1所示，包括图像采集单元、温度检测单元、定位单元、车联网服务系统、云存储计算平台。
图像采集单元，通过部署在各道路上的交通摄像头拍摄获取过往车辆图片，并进行图片数据处理；温度检测单元，通过安装在交通摄像头旁侧的温度传感器获取环境温度；定位单元，安装在交通摄像头上，用于获取交通摄像头的位置，进而获取拍摄时刻电动汽车所处的位置；车联网服务系统，基于交通摄像头的地理位置，获得此时周围的充电站信息。
云存储计算平台，与图像采集单元、温度检测单元、车联网服务系统、定位单元均通信连接，用于对图像采集单元采集到的历史车辆图片进行数据处理和存储，以及通过从历史车辆图片数据中识别出车牌信息、车主头部偏转动作、车主面部表情特征，结合电动汽车历史空调开启状态数据和历史充电记录训练充电负荷预测模型，以及用于实时接收交通摄像头采集处理的车辆图片和温度传感器的实时温度数据，并按 温度数据范围对车辆图片识别获取的数据进行分类，输入对应的充电负荷预测模型中，得到目标将要何时何地充电的概率以及充电负荷，通过时间积累，可以将何时何地充电的概率以及充电负荷用于某些地区的充电设施规划建设。
在本发明的一个实施例中，图像采集单元包括交通摄像头、图像预处理模块和第一通信模块。交通摄像头，部署在各交通道路上用于拍摄获取过往车辆图片；图像预处理模块，与交通摄像头通信连接，用于对过往车辆图片进行数据增强、归一化、灰度化处理预处理；第一通信模块，采用5G网络连接云存储计算平台，将预处理后的车辆图像上传至云存储计算平台；
温度检测单元包括温度传感器和第三通信模块，温度传感器，安装在交通摄像头旁侧，用于监测环境温度；第三通信模块，采用5G网络连接云存储计算平台，将环境温度上传至云存储计算平台。
定位单元包括定位模块和第四通信模块，定位模块，用于获取交通摄像头的位置，进而获取拍摄时刻电动汽车所处的位置；第四通信模块，采用5G网络连接云存储计算平台，将位置信息数据上传至云存储计算平台。
车联网服务系统包括数据库模块和充电记录模块，数据库模块，用于存储各地的充电站信息；充电记录模块，用于存储各种电动汽车的历史充电数据；空调使用状态记录模块，用于存储各种电动汽车历史空调使用状态数据；第五通信模块，采用5G网络连接云存储计算平台，将各地的充电站信息、电动汽车的历史充电数据和历史空调使用状态数据上传至云存储计算平台。
在本发明的一个实施例中，云存储计算平台包括历史图像数据库模块、图像识别模块、模型训练模块、模型预测模块和第二通信模块。历史图像数据库模块，与图像预处理模块通信连接，用于存储经过预处理的历史车辆图片，形成历史数据库；图像识别模块，与图像预处理模块通信连接，用于对处理后的车辆图片中的电动汽车进行车牌信息、车主头部偏转动作、车主面部表情特征；模型训练模块，通过从历史车辆图片数据中识别出车牌信息、车主头部偏转动作、车主面部表情特征，结合电动汽车历史空调开启状态数据和历史充电记录训练充电负荷预测模型；模型预测模块，用于实时接收交通摄像头采集处理的车辆图片和温度传感器的实时温度数据，并按温度数据范围对车辆图片识别获取的数据进行分类，输入对应的充电负荷预测模型中，得到目标将要何时何地充电的概率以及充电负荷，通过时间积累，得到某些地区的充电设施规划建设；第二通信模块，采用5G网络连接图像采集单元、温度检测单元、车联网服务系统、定位单元。
本发明的另一个实施例中提供了基于非侵入式检测的充电负荷概率预测方法，如图2所示，包括以下步骤：
S10.通过部署在道路上的交通摄像头拍摄采集过往的历史车辆图片，建立图片数据库；
S20.对历史车辆图片进行数据处理，获取车牌信息、车主头部偏转动作、车主面部表情特征；
S30.基于车牌信息，通过车联网系统获取该电动汽车充电记录，包括充电地点和充电时间；
S40.以车主头部偏转动作和车主面部表情特征作为自变量，以充电时间和充电地点作为因变量，分别建立每个车主的充电负荷需求预测模型；
S50.通过车联网系统获取电动汽车历史空调使用状态，通过空调开启信息训练优化充电负荷需求预测模型，获得空调开启时的第一负荷需求预测模型和空调关闭时的第二负荷需求预测模型；
S60.通过交通摄像头旁侧的温度传感器检测环境温度，明确某时刻拍摄的照片的对应温度，判断是否有开启车内空调，设定温度传感器检测到的环境温度为14℃～28℃为不开车内空调条件下，否则为开启车内空调条件。若是有开启车内空调则输入第一负荷需求预测模型进行充电负荷预测，若是没有开启空调则输入第二负荷需求预测模型进行充电负荷预测；
S70.再结合交通摄像头的定位单元，能够获取某一时刻某处的电动车将要何时何地充电的概率和充电负荷预测。
在本发明的一个实施例中，图片数据库是由m个摄像头、t个时刻、n个目标的历史图片数据组成，共mnt个基础图片。
在本发明的一个实施例中，车牌信息的识别获取包括以下步骤：
B10.车牌图像预处理，将车辆图片从RGB通道转换到HSV通道，将HSV通道图像转化为灰度图像，再将灰度图像进行二值化和形态学处理；
B20.车牌定位，使用函数cv2.findContours()对形态学处理后的灰度图像进行矩形检测，定位车牌区域，将车牌区域分割出来；
B30.车牌字符分割，对分割出来的车牌区域依次进行车牌水平矫正、车牌边框和铆钉的去除、字符分割操作，车牌水平矫正包括倾斜角度检测、倾斜矫正；
B40.车牌字符识别，通过HOG特征提取将图像上的字符转化成特征向量，并通过SVM分类算法分类判别；
B50.得到结果，通过分类判别识别为是否是新能源汽车，以及车牌号码。
在本发明的一个实施例中，车主头部偏转动作的获取如下：
对过往车辆图片进行数据处理，获取车型信息和车主驾驶位置信息，并通过车型信息和车主驾驶位置信息判断车主的头部偏转方向与仪表盘的对应方向；
根据人体头部运动的规律，在假设人体头部按照头部中心轴线转动而不发生垂直方向上左右摆动的条件下，驾驶员的头部旋转模型可以简化为如图3所示的模型。将驾驶人头部朝向正前方情况下的两眼连成的线段表示为线段
Figure PCTCN2022103801-appb-000014
头部旋转过一个θ角度后两眼之间的空间连接线段表示为线段
Figure PCTCN2022103801-appb-000015
由于人的两眼之间的距离不会发生改变，因此在数值上线段
Figure PCTCN2022103801-appb-000016
和
Figure PCTCN2022103801-appb-000017
是相等的，头部旋转后两眼之间距离在正向平面上的投影表示为线段
Figure PCTCN2022103801-appb-000018
也就是头部发生旋转后在二维图像中表现出来的两眼之间的距离，通过几何分析可以得到线段
Figure PCTCN2022103801-appb-000019
与线段
Figure PCTCN2022103801-appb-000020
之间所夹锐角与头部旋转过的角度相等，则头部旋转角度计算公式表示为
Figure PCTCN2022103801-appb-000021
计算得到的θ值为负值，则判断为头部左旋转，θ值为正值，则判断为头部右旋转，头部右旋转角度大于30度则判断为驾驶员偏头看向仪表盘。
在本发明的一个实施例中，车主面部表情特征的获取如下：
在特征采样层中，对车辆图片进行图像数据预处理、人脸检测、人脸定位、卷积特征学习和特征采样，输入原始车辆图像序列x＝{x 1,x 2,…,x T}，对图像进行预处理，消除光照影响；在时刻t＝1，采用Fast-CNN方法快速定位人脸图像，并进行分割；在后续时刻，进行快速跟踪人脸图像并分割；然后，人脸图像进入卷积神经网络学习，通过交叉卷积和池化，生成图像抽象特征；最后，输入到K-平均采样层，该层对x T-K+1,x T-K+2,…,x T共K个连续图像特征进行平均采样，作为输入特征进入循环网络学习；
RNN主循环和LSTM信息记忆，从特征采样层获取卷积采样特征向量作为输入；然后，根据时间序列进入循环网络，循环单元LSTM提取信息生成状态信息；最后，输出特征向量进行分类；
对循环网络层学习到的特征向量输出序列进行分类，设T为序列长度，L为标签的长度，则当序列T出现时标签路径为L的T次方种可能，称其中每一种可能称为一条“Path”，其条件概率公式
Figure PCTCN2022103801-appb-000022
其中，π (t)表示在t时刻的输出路径π，y t表示在时刻t的RNN的网络输出，
预测标签条件概率表示为相应Path的条件概率之和，
Figure PCTCN2022103801-appb-000023
其中，V表示将输出路径π转换为目标L的运算符，
当输入序列进行训练时，当前时刻输入图像表情为条件概率最大的标签：
Figure PCTCN2022103801-appb-000024
识别出车主愤怒、厌恶、恐惧、高兴、悲伤、惊讶、中性各种表情。
在本发明的一个实施例中，充电负荷需求预测模型建立如下：
假设用N组数据进行模型训练，进行灰色关联分析：
假设用N组数据进行模型训练，进行随机森林算法分析：
a.将电动汽车的充电行为作为预测变量X o，把驾驶员在充电前一小时内面部表现出的愤怒、厌恶、恐惧、高兴、悲伤、惊讶、中性和头部偏转这八个因素作为影响变量X i；
b.令抽样次数b＝1,2,…,B s，重复步骤c、d；
c.通过自主重复抽样技术从X o、X i中随机选择一个子样本集作为训练集；
d.通过训练集得到一个回归树模型r fb；
训练结束后，对一个新的样本x，随机森林模型通过平均所有回归树的预测值给出该样本的预测值
Figure PCTCN2022103801-appb-000025
为：
Figure PCTCN2022103801-appb-000026
以上实施例仅用以说明本发明的技术方案而非限制，尽管参照较佳实施例对本发明进行了详细说明，本领域的普通技术人员应当理解，可以对本发明的技术方案进行修改或者等同替换，而不脱离本发明技术方案的宗旨和范围，其均应涵盖在本发明的权利要求范围当中。本发明未详细描述的技术、形状、构造部分均为公知技术。

Claims (10)

  1. 基于非侵入式检测的充电负荷概率预测系统，其特征在于，包括：
    图像采集单元，通过部署在各道路上的交通摄像头拍摄获取过往车辆图片，并进行图片数据处理；
    温度检测单元，通过安装在所述交通摄像头旁侧的温度传感器获取环境温度；
    定位单元，安装在所述交通摄像头上，用于获取所述交通摄像头的位置，进而获取拍摄时刻电动汽车所处的位置；
    车联网服务系统，基于所述交通摄像头的地理位置，获得此时周围的充电站信息；
    云存储计算平台，与所述图像采集单元、温度检测单元、车联网服务系统、定位单元均通信连接，用于对所述图像采集单元采集到的历史车辆图片进行数据处理和存储，以及通过从历史车辆图片数据中识别出车牌信息、车主头部偏转动作、车主面部表情特征，结合电动汽车历史空调开启状态数据和历史充电记录训练充电负荷预测模型，以及用于实时接收所述交通摄像头采集处理的车辆图片和温度传感器的实时温度数据，并按温度数据范围对车辆图片识别获取的数据进行分类，输入对应的充电负荷预测模型中，得到目标将要何时何地充电的概率以及充电负荷，通过时间积累，用于某些地区的充电设施规划建设。
  2. 根据权利要求1所述的基于非侵入式检测的充电负荷概率预测系统，其特征在于，所述图像采集单元包括：
    交通摄像头，部署在各交通道路上用于拍摄获取过往车辆图片；
    图像预处理模块，与所述交通摄像头通信连接，用于对所述过往车辆图片进行数据增强、归一化、灰度化预处理；
    第一通信模块，采用5G网络连接云存储计算平台，将预处理后的车辆图像上传至云存储计算平台；
    所述温度检测单元包括：
    温度传感器，安装在所述交通摄像头旁侧，用于监测环境温度；
    第三通信模块，采用5G网络连接云存储计算平台，将所述环境温度上传至云存储计算平台；
    所述定位单元包括：
    定位模块，用于获取所述交通摄像头的位置，进而获取拍摄时刻电动汽车所处的位置；
    第四通信模块，采用5G网络连接云存储计算平台，将位置信息数据上传至云存储计算平台；
    所述车联网服务系统包括：
    数据库模块，用于存储各地的充电站信息；
    充电记录模块，用于存储各种电动汽车的历史充电数据；
    空调使用状态记录模块，用于存储各种电动汽车历史空调使用状态数据；
    第五通信模块，采用5G网络连接云存储计算平台，将各地的充电站信息、电动汽车的历史充电数据和历史空调使用状态数据上传至云存储计算平台。
  3. 根据权利要求2所述的基于非侵入式检测的充电负荷概率预测系统，其特征在于，所述云存储计算平台包括：
    历史图像数据库模块，用于存储经过预处理的历史车辆图片，形成历史数据库；
    图像识别模块，用于对处理后的车辆图片中的电动汽车进行车牌信息、车主头部偏转动作、车主面部表情特征；
    模型训练模块，通过从历史车辆图片数据中识别出车牌信息、车主头部偏转动作、车主面部表情特征，结合电动汽车历史空调开启状态数据和历史充电记录训练充电负荷预测模型；
    模型预测模块，用于实时接收所述交通摄像头采集处理的车辆图片和温度传感器的实时温度数据，并按温度数据范围对车辆图片识别获取的数据进行分类，输入对应的充电负荷预测模型中，得到目标将要何时何地充电的概率以及充电负荷，通过时间积累，得到某些地区的充电设施规划建设；
    第二通信模块，采用5G网络连接所述图像采集单元、温度检测单元、车联网服务系统、定位单元。
  4. 基于非侵入式检测的充电负荷概率预测方法，其特征在于，包括以下步骤：
    通过部署在道路上的交通摄像头拍摄采集过往的历史车辆图片，建立图片数据库；
    对所述历史车辆图片进行数据处理，获取车牌信息、车主头部偏转动作、车主面部表情特征；
    基于所述车牌信息，通过车联网系统获取该电动汽车充电记录，包括充电地点和充电时间；
    以所述车主头部偏转动作和所述车主面部表情特征作为自变量，以充电时间和充电地点作为因变量，分别建立每个车主的充电负荷需求预测模型；
    通过车联网系统获取电动汽车历史空调使用状态，通过空调开启信息训练优化充电负荷需求预测模型，获得空调开启时的第一负荷需求预测模型和空调关闭时的第二负荷需求预测模型；
    通过交通摄像头旁侧的温度传感器检测环境温度，明确某时刻拍摄的照片的对应温度，判断是否有开启车内空调，若是有开启车内空调则输入第一负荷需求预测模型进行充电负荷预测，若是没有开启空调则输入第二负荷需求预测模型进行充电负荷预测；
    再结合交通摄像头的定位单元，能够获取某一时刻某处的电动车将要何时何地充电的概率和充电负荷预测。
  5. 根据权利要求4所述的基于非侵入式检测的充电负荷概率预测方法，其特征在于，所述图片数据库是由m个摄像头、t个时刻、n个目标的历史图片数据组成，共mnt个基础图片。
  6. 根据权利要求5所述的基于非侵入式检测的充电负荷概率预测方法，其特征在于，所述车牌信息的识别获取包括以下步骤：
    车牌图像预处理，将车辆图片从RGB通道转换到HSV通道，将HSV通道图像转化为灰度图像，再将灰度图像进行二值化和形态学处理；
    车牌定位，使用函数cv2.findContours()对形态学处理后的灰度图像进行矩形检测，定位车牌区域，将所述车牌区域分割出来；
    车牌字符分割，对分割出来的所述车牌区域依次进行车牌水平矫正、车牌边框和铆钉的去除、字符分割操作，所述车牌水平矫正包括倾斜角度检测、倾斜矫正；
    车牌字符识别，通过HOG特征提取将图像上的字符转化成特征向量，并通过SVM分类算法分类判别；
    得到结果，通过分类判别识别为是否是新能源汽车，以及车牌号码。
  7. 根据权利要求6所述的基于非侵入式检测的充电负荷概率预测方法，其特征在于，所述车主头部偏转动作的获取如下：
    对过往车辆图片进行数据处理，获取车型信息和车主驾驶位置信息，并通过所述车型信息和所述车主驾驶位置信息判断车主的头部偏转方向与仪表盘的对应方向；
    根据人体头部运动的规律，在假设人体头部按照头部中心轴线转动而不发生垂直方向上左右摆动的条件下，将驾驶人头部朝向正前方情况下的两眼连成的线段表示为线段
    Figure PCTCN2022103801-appb-100001
    头部旋转过一个θ角度后两眼之间的空间连接线段表示为线段
    Figure PCTCN2022103801-appb-100002
    在数值上线段
    Figure PCTCN2022103801-appb-100003
    和
    Figure PCTCN2022103801-appb-100004
    是相等的，头部旋转后两眼之间距离在正向平面上的投影表示为线段
    Figure PCTCN2022103801-appb-100005
    通过几何分析可以得到线段
    Figure PCTCN2022103801-appb-100006
    与线段
    Figure PCTCN2022103801-appb-100007
    之间所夹锐角与头部旋转过的角度相等，则头部旋转角度计算公式表示为
    Figure PCTCN2022103801-appb-100008
    计算得到的θ值为负值，则判断为头部左旋转，θ值为正值，则判断为头部右旋转。
  8. 根据权利要求7所述的基于非侵入式检测的充电负荷概率预测方法，其特征在于，所述车主面部表情特征的获取如下：
    在特征采样层中，对车辆图片进行图像数据预处理、人脸检测、人脸定位、卷积特征学习和特征采样，输入原始车辆图像序列x＝{x 1,x 2,…,x T}，对图像进行预处理，消除光照影响；在时刻t＝1，采用Fast-CNN方法快速定位人脸图像，并进行分割；在后续时刻，进行快速跟踪人脸图像并分割；然后，人脸图像进入 卷积神经网络学习，通过交叉卷积和池化，生成图像抽象特征；最后，输入到K-平均采样层，该层对x T-K+1,x T-K+2,…,x T共K个连续图像特征进行平均采样，作为输入特征进入循环网络学习；
    RNN主循环和LSTM信息记忆，从特征采样层获取卷积采样特征向量作为输入；然后，根据时间序列进入循环网络，循环单元LSTM提取信息生成状态信息；最后，输出特征向量进行分类；
    对循环网络层学习到的特征向量输出序列进行分类，设T为序列长度，L为标签的长度，则当序列T出现时标签路径为L的T次方种可能，称其中每一种可能称为一条“Path”，其条件概率公式
    Figure PCTCN2022103801-appb-100009
    其中，π (t)表示在t时刻的输出路径π，y t表示在时刻t的RNN的网络输出，
    预测标签条件概率表示为相应Path的条件概率之和，
    Figure PCTCN2022103801-appb-100010
    其中，V表示将输出路径π转换为目标L的运算符，
    当输入序列进行训练时，当前时刻输入图像表情为条件概率最大的标签：
    Figure PCTCN2022103801-appb-100011
    识别出车主愤怒、厌恶、恐惧、高兴、悲伤、惊讶、中性各种表情。
  9. 根据权利要求8所述的基于非侵入式检测的充电负荷概率预测方法，其特征在于，所述充电负荷需求预测模型建立如下：
    假设用N组数据进行模型训练，进行随机森林算法分析：
    a.将电动汽车的充电行为作为预测变量X o，把驾驶员在充电前一小时内面部表现出的愤怒、厌恶、恐惧、高兴、悲伤、惊讶、中性和头部偏转这八个因素作为影响变量X i；
    b.令抽样次数b＝1,2,…,B s，重复步骤c、d；
    c.通过自主重复抽样技术从X o、X i中随机选择一个子样本集作为训练集；
    d.通过训练集得到一个回归树模型r fb；
    训练结束后，对一个新的样本x，随机森林模型通过平均所有回归树的预测值给出该样本的预测值
    Figure PCTCN2022103801-appb-100012
    为：
    Figure PCTCN2022103801-appb-100013
  10. 根据权利要求9所述的基于非侵入式检测的充电负荷概率预测方法，其特征在于，所述判断是否有开启车内空调步骤中，设定所述温度传感器检测到的环境温度为14℃～28℃为不开车内空调条件下，否则为开启车内空调条件。
PCT/CN2022/103801 2021-12-15 2022-07-05 基于非侵入式检测的充电负荷概率预测系统及方法 WO2023109099A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202111535061.4A CN114418298A (zh) 2021-12-15 2021-12-15 基于非侵入式检测的充电负荷概率预测系统及方法
CN202111535061.4 2021-12-15

Publications (1)

Publication Number Publication Date
WO2023109099A1 true WO2023109099A1 (zh) 2023-06-22

Family

ID=81268255

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/103801 WO2023109099A1 (zh) 2021-12-15 2022-07-05 基于非侵入式检测的充电负荷概率预测系统及方法

Country Status (2)

Country Link
CN (1) CN114418298A (zh)
WO (1) WO2023109099A1 (zh)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116359602A (zh) * 2023-03-07 2023-06-30 北京智芯微电子科技有限公司 非侵入式电动车充电识别方法、装置、介质和智能电表
CN116552299A (zh) * 2023-07-11 2023-08-08 深圳市南霸科技有限公司 一种可移动式电动汽车应急充电系统及方法
CN116872780A (zh) * 2023-09-08 2023-10-13 国网浙江省电力有限公司杭州供电公司 一种电动汽车充电供应控制方法、装置、终端及介质
CN117058468A (zh) * 2023-10-11 2023-11-14 青岛金诺德科技有限公司 用于新能源汽车锂电池回收的图像识别与分类系统
CN117520979A (zh) * 2023-11-03 2024-02-06 长沙云软信息技术有限公司 一种基于ocr识别的地磅测量设备及其使用方法
CN117885593A (zh) * 2024-03-14 2024-04-16 江苏智融能源科技有限公司 一种基于大数据的充电站数据管控方法及系统
CN117969946A (zh) * 2024-04-01 2024-05-03 山东普源电力工程有限公司 一种照明系统能耗检测方法及系统

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114418298A (zh) * 2021-12-15 2022-04-29 重庆邮电大学 基于非侵入式检测的充电负荷概率预测系统及方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106908075A (zh) * 2017-03-21 2017-06-30 福州大学 大数据采集与处理系统及基于其电动汽车续航估计方法
CN107392400A (zh) * 2017-09-04 2017-11-24 重庆大学 计及实时交通与温度的ev充电负荷时空分布预测方法
US20180211116A1 (en) * 2017-01-25 2018-07-26 Echelon Corporation Distributed system for mining, correlating, and analyzing locally obtained traffic data including video
CN110363331A (zh) * 2019-06-21 2019-10-22 国网天津市电力公司电力科学研究院 一种应用于电动汽车安全的云平台服务系统与方法
CN114418298A (zh) * 2021-12-15 2022-04-29 重庆邮电大学 基于非侵入式检测的充电负荷概率预测系统及方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20180211116A1 (en) * 2017-01-25 2018-07-26 Echelon Corporation Distributed system for mining, correlating, and analyzing locally obtained traffic data including video
CN106908075A (zh) * 2017-03-21 2017-06-30 福州大学 大数据采集与处理系统及基于其电动汽车续航估计方法
CN107392400A (zh) * 2017-09-04 2017-11-24 重庆大学 计及实时交通与温度的ev充电负荷时空分布预测方法
CN110363331A (zh) * 2019-06-21 2019-10-22 国网天津市电力公司电力科学研究院 一种应用于电动汽车安全的云平台服务系统与方法
CN114418298A (zh) * 2021-12-15 2022-04-29 重庆邮电大学 基于非侵入式检测的充电负荷概率预测系统及方法

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116359602A (zh) * 2023-03-07 2023-06-30 北京智芯微电子科技有限公司 非侵入式电动车充电识别方法、装置、介质和智能电表
CN116552299A (zh) * 2023-07-11 2023-08-08 深圳市南霸科技有限公司 一种可移动式电动汽车应急充电系统及方法
CN116552299B (zh) * 2023-07-11 2023-09-15 深圳市南霸科技有限公司 一种可移动式电动汽车应急充电系统及方法
CN116872780A (zh) * 2023-09-08 2023-10-13 国网浙江省电力有限公司杭州供电公司 一种电动汽车充电供应控制方法、装置、终端及介质
CN116872780B (zh) * 2023-09-08 2023-12-15 国网浙江省电力有限公司杭州供电公司 一种电动汽车充电供应控制方法、装置、终端及介质
CN117058468A (zh) * 2023-10-11 2023-11-14 青岛金诺德科技有限公司 用于新能源汽车锂电池回收的图像识别与分类系统
CN117058468B (zh) * 2023-10-11 2023-12-19 青岛金诺德科技有限公司 用于新能源汽车锂电池回收的图像识别与分类系统
CN117520979A (zh) * 2023-11-03 2024-02-06 长沙云软信息技术有限公司 一种基于ocr识别的地磅测量设备及其使用方法
CN117520979B (zh) * 2023-11-03 2024-05-31 长沙云软信息技术有限公司 一种基于ocr识别的地磅测量设备及其使用方法
CN117885593A (zh) * 2024-03-14 2024-04-16 江苏智融能源科技有限公司 一种基于大数据的充电站数据管控方法及系统
CN117885593B (zh) * 2024-03-14 2024-05-24 江苏智融能源科技有限公司 一种基于大数据的充电站数据管控方法及系统
CN117969946A (zh) * 2024-04-01 2024-05-03 山东普源电力工程有限公司 一种照明系统能耗检测方法及系统

Also Published As

Publication number Publication date
CN114418298A (zh) 2022-04-29

Similar Documents

Publication Publication Date Title
WO2023109099A1 (zh) 基于非侵入式检测的充电负荷概率预测系统及方法
Arya et al. Transfer learning-based road damage detection for multiple countries
WO2017156772A1 (zh) 一种乘客拥挤度的计算方法及其系统
Lin et al. A Real‐Time Vehicle Counting, Speed Estimation, and Classification System Based on Virtual Detection Zone and YOLO
CN106710228B (zh) 一种客货分道交通参数监测系统的实现方法
He et al. Deep learning based geometric features for effective truck selection and classification from highway videos
CN112836683A (zh) 用于便携式摄像设备的车牌识别方法、装置、设备和介质
KR102186974B1 (ko) 차량의 주차면 분석을 위한 지능형 cctv 시스템
Hoang et al. Optimizing YOLO performance for traffic light detection and end-to-end steering control for autonomous vehicles in Gazebo-ROS2
Lee et al. An intelligent driving assistance system based on lightweight deep learning models
CN113723273A (zh) 一种车辆轨迹信息确定方法、装置及计算机设备
Liu et al. Research on security of key algorithms in intelligent driving system
Rouf et al. Real-time vehicle detection, tracking and counting system based on YOLOv7
Jawed et al. Data-driven vehicle trajectory forecasting
Shahane et al. A self-driving car platform using raspberry Pi and Arduino
TWM618607U (zh) 自動化交通轉向量化調查報告產出系統
Van Cuong et al. AI-Driven Vehicle Recognition for Enhanced Traffic Management: Implications and Strategies
Bezawada et al. Modeling lane and road characteristics for drive quality rating
Zhou et al. Design of a commercial vehicle monitoring system based on multiple sensors
Muhammad Detection parking Spaces by using the ResNet50 Algorithm
Prawinsankar et al. Traffic Congession Detection through Modified Resnet50 and Prediction of Traffic using Clustering
Xu et al. Research on parking Space Detection and Prediction Model based on CNN-LSTM
CN117612140B (zh) 道路场景的识别方法、装置和存储介质及电子设备
CN113721255B (zh) 基于激光雷达与视觉融合的列车站台停车点精准检测方法
Aldakkhelallah et al. Development of new technologies for real-time traffic monitoring

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22905864

Country of ref document: EP

Kind code of ref document: A1