技术领域technical field
本发明涉及大数据技术领域，尤其涉及与新能源汽车充电站选址因素有关的研究和分析，具体涉及基于深度学习的多特征充电选址影响因素分析方法及系统。The present invention relates to the field of big data technology, in particular to research and analysis related to site selection factors for new energy vehicle charging stations, and in particular to a deep learning-based analysis method and system for influencing factors of multi-feature charging site selection.
背景技术Background technique
随着能源技术的进一步发展，新能源汽车受到越来越多人的青睐。而新能源汽车由于其本身的结构和动力特点，对充电站的需求极其强烈。理想的充电站布设应尽可能满足此类车辆的出行需求，同时又能够为运营商带来良好的经济效益。With the further development of energy technology, new energy vehicles are favored by more and more people. Due to its own structure and power characteristics, new energy vehicles have an extremely strong demand for charging stations. The ideal charging station layout should meet the travel needs of such vehicles as much as possible, and at the same time bring good economic benefits to operators.
因此，充电站的选址问题便显得十分重要。衡量一个充电站是否运营良好的标准，往往是看其站点内充电桩的利用率情况。如果充电桩利用率较高，则说明该站点的选址较为合理。通过分析影响充电桩利用率的因素，可以得到与充电站选址相关的重要因素。Therefore, the location of the charging station is very important. The standard to measure whether a charging station is operating well is often to look at the utilization rate of charging piles in its site. If the utilization rate of the charging pile is high, it means that the location of the station is more reasonable. By analyzing the factors affecting the utilization rate of charging piles, the important factors related to the location of charging stations can be obtained.
深度学习的发展为此类问题的分析提供了极大的帮助。由于与选址有关的因素种类繁多，且各因素之间可能存在相互关联，因此，使用传统的方法不利于对多特征多联系的场景进行建模，而深度学习凭借独特的多层网络结构，适用于对多特征和多联系进行建模和分析。The development of deep learning has provided great help for the analysis of such problems. Due to the wide variety of factors related to site selection, and there may be interrelationships among the factors, the use of traditional methods is not conducive to modeling multi-feature and multi-connection scenarios, and deep learning relies on the unique multi-layer network structure. Suitable for modeling and analysis of multiple features and multiple relationships.
现有的研究更多关注的是共享单车的配置问题。他们通过分析车辆的时序特征，例如轨迹数据(车辆的行驶路线和情况等)，来更好地规划共享单车的资源配置。但其忽略了站点本身的一些重要特征，如充电桩类型，充电费用和周围环境等。Existing research focuses more on the allocation of shared bicycles. By analyzing the time series characteristics of vehicles, such as trajectory data (vehicles' driving routes and conditions, etc.), they can better plan the resource allocation of shared bicycles. But it ignores some important characteristics of the station itself, such as charging pile type, charging cost and surrounding environment.
发明内容Contents of the invention
为解决现有技术中存在的上述问题，本发明提供了一种基于深度学习的多特征充电选址影响因素分析方法及系统。In order to solve the above-mentioned problems existing in the prior art, the present invention provides a method and system for analyzing factors affecting multi-feature charging location selection based on deep learning.
本发明所述的分析方法，基于站点本身的特征，在诸多种类中，先经过数据筛选挑选出一批具有代表性的特征，通常情况下，不可避免的，其中会包含一部分噪声数据，而对这部分噪声的处理也是极为重要的一步。数据清洗完成后，模型的选择和调整优化也极大影响着最终影响因素的确定。The analysis method described in the present invention is based on the characteristics of the site itself. Among many types, a batch of representative features are first selected through data screening. Usually, it is inevitable that some noise data will be included, and for The processing of this part of the noise is also an extremely important step. After the data cleaning is completed, the selection and adjustment and optimization of the model also greatly affect the determination of the final influencing factors.
本发明提出的基于深度学习的多特征充电选址影响因素分析方法，包括以下步骤：The method for analyzing the influencing factors of multi-feature charging location selection based on deep learning proposed by the present invention includes the following steps:
步骤A：获取公开的站点信息特征数据及站点附近POI的特征数据，形成初步的特征数据集；Step A: Obtain the public site information feature data and feature data of POIs near the site to form a preliminary feature data set;
步骤B：筛选特征并对特征数据进行清理及标准化；Step B: Screening features and cleaning and standardizing feature data;
步骤C：持续收集所述站点信息，将站点使用情况转换为利用率数据；Step C: continuously collect the site information, and convert the site usage into utilization data;
步骤D：采用MLP构建分类模型，将选址问题转化为机器学习多分类预测问题；Step D: Using MLP to construct a classification model, transforming the location selection problem into a machine learning multi-classification prediction problem;
步骤E：完成利用率分类之后，对特征的重要性进行排序，探索选址的影响因素。Step E: After completing the utilization classification, sort the importance of the features and explore the influencing factors of site selection.
本发明步骤A中，获取网络上公开的站点信息特征数据，包括站点的位置信息，站点内充电桩的数量、类型、充电费用以及充电桩的使用情况；从地图API获取站点附近POI的特征数据，形成初步的特征数据集。In step A of the present invention, obtain the characteristic data of site information disclosed on the network, including the location information of the site, the number, type, charging fee and usage of charging piles in the site; obtain the characteristic data of POIs near the site from the map API , forming a preliminary feature data set.
本发明步骤B中，通过筛选得到五大类型的特征，包括：In the step B of the present invention, five types of features are obtained by screening, including:
站点利用率；site utilization;
站点附近重要的POI；Important POIs near the site;
充电桩的类型，分为直流桩和交流桩；The types of charging piles are divided into DC piles and AC piles;
充电费用，包括直流充电费用和交流充电费用；Charging costs, including DC charging costs and AC charging costs;
站点的使用类型，分为公共和专用。The usage type of the site, divided into public and private.
本发明步骤B中，对特征数据进行清洗包括：删除其中离群值的噪声数据，采用平均值填充法补全空值数据。In step B of the present invention, cleaning the feature data includes: deleting outlier noise data among them, and using an average value filling method to fill in null data.
本发明步骤B中，数据标准化为将特征值标准化为0至1范围内的值。In step B of the present invention, the data is normalized to normalize the feature value to a value ranging from 0 to 1.
本发明步骤C中，将站点数据集划分为郊区和城区两部分，以分析在不同地区充电站点的利用率情况。In step C of the present invention, the station data set is divided into two parts: suburban area and urban area, so as to analyze the utilization rate of charging stations in different regions.
本发明步骤C中，将数据集划分为白天，夜晚，工作日和周末四部分，以分析充电站点在不同时间段的利用率情况。In step C of the present invention, the data set is divided into four parts: daytime, nighttime, weekdays and weekends, so as to analyze the utilization rate of charging stations in different time periods.
本发明步骤D中，所述利用率低，中，高三档分别对应利用率的取值为0-20％，20％-50％以及50％-100％。In step D of the present invention, the low, medium and high utilization ratios respectively correspond to the utilization ratios of 0-20%, 20%-50% and 50%-100%.
本发明步骤E中，采用梯度提升决策树算法，计算出每个特征对结果的贡献度得分，得到特征与利用率的相关性；根据所述相关性，对特征按所得的贡献度分数进行排序，分数高者为对利用率影响较大的特征，从而得出影响站点利用率的重要因素。In step E of the present invention, the gradient lifting decision tree algorithm is used to calculate the contribution score of each feature to the result, and the correlation between the feature and the utilization rate is obtained; according to the correlation, the features are sorted according to the obtained contribution score , the one with the higher score is the feature that has a greater impact on the utilization rate, and thus the important factors that affect the site utilization rate can be obtained.
基于以上方法，本发明还提出了一种基于深度学习的多特征充电选址影响因素分析系统，所述系统包括：Based on the above method, the present invention also proposes a multi-feature charging location influencing factor analysis system based on deep learning, the system includes:
特征数据获取模块，用于获取公开的站点信息特征数据及站点附近POI的特征数据，形成初步的特征数据集；The feature data acquisition module is used to acquire the feature data of public site information and the feature data of POIs near the site to form a preliminary feature data set;
数据清洗及标准化模块，用于筛选特征并对特征数据进行清理及标准化；Data cleaning and standardization module, used to filter features and clean and standardize feature data;
划分数据集模块，用于持续收集所述站点信息，将站点使用情况转换为利用率数据；A data set module is used to continuously collect the site information and convert site usage into utilization data;
利用率分类模块，用于采用MLP构建分类模型，将选址问题转化为机器学习多分类预测问题；The utilization rate classification module is used to construct a classification model using MLP, and convert the location selection problem into a machine learning multi-classification prediction problem;
特征重要性排序模块，用于完成利用率分类之后，对特征的重要性进行排序。The feature importance sorting module is used to sort the importance of features after the utilization rate classification is completed.
本发明的有益效果在于：通过特征筛选，数据清洗，数据集划分，模型选择以及实现利用率分类和特征重要性排序两项任务，利用了数据挖掘与深度学习的思想和方法，最终能够得到经过筛选的特征哪些是决定站点利用率较高的关键因素，从而得出运营商在进行站点选址时所考虑的重要因素。相较于现有研究，本发明更加关注站点本身的特征，充分利用了站点的位置信息、充电类型、充电费用以及周围环境等与站点利用率非常相关的特征，使用了深度学习网络对不同地区和不同时间段的站点数据分别进行利用率分类任务，同时，将特征重要性排序加入模型结构中，能够有效得出各个特征影响站点利用率的量化表示，从而分析得出布设站点时需要考虑的重要因素。The beneficial effects of the present invention are: through feature screening, data cleaning, data set division, model selection, and realization of two tasks of utilization rate classification and feature importance ranking, the thought and method of data mining and deep learning can be used to finally obtain the Which of the screened features is the key factor to determine the high utilization rate of the site, so as to obtain the important factors that the operator considers when selecting the site location. Compared with the existing research, the present invention pays more attention to the characteristics of the site itself, makes full use of the site's location information, charging type, charging cost and surrounding environment, etc. The utilization rate classification task is carried out separately with the site data of different time periods. At the same time, the feature importance ranking is added to the model structure, which can effectively obtain the quantitative representation of each feature's influence on the site utilization rate, and thus analyze the factors that need to be considered when laying out the site. Key factor.
附图说明Description of drawings
图1为数据处理和系统运行的模型图。Figure 1 is a model diagram of data processing and system operation.
图2为站点及附近POI的示意图。Fig. 2 is a schematic diagram of a station and nearby POIs.
图3为分类和特征排序模型的架构图。Figure 3 is an architecture diagram of the classification and feature ranking model.
图4为城区站点利用率分类特征排序结果。Figure 4 shows the ranking results of urban site utilization classification features.
具体实施方式Detailed ways
结合以下具体实施例和附图，对发明作进一步的详细说明。实施本发明的过程、条件、实验方法等，除以下专门提及的内容之外，均为本领域的普遍知识和公知常识，本发明没有特别限制内容。In conjunction with the following specific embodiments and accompanying drawings, the invention will be further described in detail. The process, conditions, experimental methods, etc. for implementing the present invention, except for the content specifically mentioned below, are common knowledge and common knowledge in this field, and the present invention has no special limitation content.
实施例1Example 1
参考图1，说明了整个系统的流程。Referring to Figure 1, the flow of the entire system is illustrated.
本实施例所述基于深度学习的多特征充电选址因素分析方法，包括以下步骤：The multi-feature charging site selection factor analysis method based on deep learning described in this embodiment includes the following steps:
(1)特征数据的获取(1) Acquisition of characteristic data
针对不同类型的特征，形成恰当的表示形式，并进行特征融合等操作。For different types of features, form appropriate representations, and perform operations such as feature fusion.
首先，获取运营商在网络上公开的站点信息特征数据，包括站点的位置信息，站点内充电桩的数量、类型、充电费用以及充电桩的使用情况等。此外，还从百度地图API获取站点附近POI的特征数据，形成初步的特征数据集。First, obtain the characteristic data of the site information disclosed by the operator on the network, including the location information of the site, the number and type of charging piles in the station, charging fees, and the usage of charging piles. In addition, feature data of POIs near the site are obtained from Baidu Map API to form a preliminary feature data set.
(2)数据处理(2) Data processing
(2.1)特征筛选及分类(2.1) Feature screening and classification
在诸多种类的特征中，通过筛选得到五大类型的特征，包括：Among the many types of features, five types of features are obtained through screening, including:
站点利用率；site utilization;
站点附近重要的POI(即兴趣点，可表示一栋房子，一个商铺，一家商场等)，如公司、房地产、医院、地铁站、购物中心、大学等；Important POIs near the site (that is, points of interest, which can represent a house, a shop, a shopping mall, etc.), such as companies, real estate, hospitals, subway stations, shopping centers, universities, etc.;
充电桩的类型，分为直流桩和交流桩；The types of charging piles are divided into DC piles and AC piles;
充电费用，包括直流充电费用和交流充电费用；Charging costs, including DC charging costs and AC charging costs;
站点的使用类型，分为公共和专用。The usage type of the site, divided into public and private.
(2.2)数据清洗及标准化(2.2) Data cleaning and standardization
对特征数据进行清洗，包括：删除其中离群值的噪声数据，采用平均值填充法补全空值数据；Clean the characteristic data, including: delete the noise data of outliers, and use the average value filling method to fill in the null data;
数据标准化：将特征值标准化为0至1范围内的值，以便模型得到更加优秀的效果。Data standardization: standardize the feature values to values in the range of 0 to 1, so that the model can get better results.
(3)划分数据集(3) Divide the data set
将站点数据集划分为郊区和城区两部分，以分析在不同地区充电站点的利用率情况。The station dataset is divided into suburban and urban areas to analyze the utilization of charging stations in different regions.
将数据集划分为白天，夜晚，工作日和周末四部分，以分析充电站点在不同时间段的利用率情况。Divide the data set into four parts: daytime, nighttime, weekdays and weekends to analyze the utilization rate of charging stations in different time periods.
(4)利用率分类(4) Utilization rate classification
采用MLP构建分类模型，将步骤(3)获得的利用率分为低，中，高三档，分别对应利用率的取值为0-20％，20％-50％以及50％-100％。以将选址问题转化为机器学习多分类预测问题。Using MLP to build a classification model, the utilization rate obtained in step (3) is divided into low, medium and high levels, corresponding to the utilization rate values of 0-20%, 20%-50% and 50%-100%. In order to transform the location selection problem into a machine learning multi-class prediction problem.
(5)特征重要性排序(5) Ranking of feature importance
在完成利用率分类之后，为了得到各个特征对实验结果的影响，对特征的重要性进行排序，探索选址的影响因素。After completing the utilization classification, in order to obtain the impact of each feature on the experimental results, the importance of the features is sorted to explore the influencing factors of site selection.
实施例2Example 2
参考图2，是某个站点及其一定范围内部分POI数量的示意图。Referring to FIG. 2 , it is a schematic diagram of a site and the number of POIs within a certain range.
在每个站点周围设置一个半径，然后在半径范围内收集POI。在实验的基础上，选择半径300米为宜。收集工作总共得到80种不同类型的POI。其中一些POI彼此非常相似，因此，进一步地对80个POI进行分组，最终得到10组POI分类。Set a radius around each station, then collect POIs within the radius. On the basis of experiments, it is advisable to choose a radius of 300 meters. The collection effort resulted in a total of 80 different types of POIs. Some of these POIs are very similar to each other, therefore, the 80 POIs are further grouped, and finally 10 groups of POI classifications are obtained.
在利用率预测中，需要考虑距离。如果计划去的目的地较远，人们就不会选择停车充电。在本系统中，考虑距离地铁站、金融中心和主要功能建筑较近的充电站会更频繁地被使用。因此，选择以下距离最近的POI数量作为我们的特征：公司、房地产、医院、地铁站、购物中心和大学。In utilization forecasting, distance needs to be considered. If the planned destination is far away, people will not choose to stop and charge. In this system, it is considered that charging stations closer to subway stations, financial centers and main functional buildings will be used more frequently. Therefore, the following number of POIs with the closest distance are selected as our features: company, real estate, hospital, subway station, shopping mall, and university.
通过进一步的数据挖掘，发现收费价格与使用率之间存在0.3的相关性。由于充电桩有两种类型：直流和交流。因此，把两种类型充电桩的数量和价格也选取为特征之一。Through further data mining, it is found that there is a correlation of 0.3 between charging price and usage rate. There are two types of charging piles: DC and AC. Therefore, the quantity and price of the two types of charging piles are also selected as one of the features.
通过观察收集到的数据，可以得到大多数充电站都是私用充电站，这意味着它们通常被电力公交车和出租车使用，或者被特定公司的员工使用。此外，由于私用站点被更多的规律用户使用，因此其使用率比公共站点高。基于以上区别，将充电站点为私用或公用也作为本实施例所考虑的特征之一。By observing the collected data, it can be concluded that most of the charging stations are private charging stations, which means that they are usually used by electric buses and taxis, or by employees of a specific company. Also, since private sites are used by more regular users, their usage rates are higher than public sites. Based on the above differences, whether the charging station is private or public is also considered as one of the features considered in this embodiment.
通过一个月内持续收集运营商公布的站点信息，将充电桩的使用情况转换为实验所需的利用率数据。以一日内站点利用率值的获取为例。将一日内某个充电桩被使用的时长与24小时取比值，可得到一日内该充电桩的利用率值。对于一个站点，对该站点内充电桩的利用率值取平均，便可得到该站点一日内的利用率值。By continuously collecting the site information announced by the operator within a month, the usage of charging piles is converted into the utilization data required for the experiment. Take the acquisition of site utilization value within a day as an example. The utilization rate value of the charging pile in a day can be obtained by taking the ratio of the time that a charging pile is used in a day to 24 hours. For a site, the utilization rate value of the site within a day can be obtained by averaging the utilization rate values of the charging piles in the site.
实施例3Example 3
基于实验需求，要根据地区和时间段对数据集进行划分。在地区划分中，将数据集分为城区和郊区两部分，其中城区包含七个地区；郊区包含九个地区。在时间段划分中，设置了早晨、晚上、工作日、周末四个时间段，每个时间段的数据集包含在该时段所有充电站的特征信息，区别仅在于每个时段站点利用率会有所不同。Based on the experimental needs, the data set should be divided according to the region and time period. In the regional division, the data set is divided into two parts: urban area and suburban area. The urban area contains seven areas; the suburban area contains nine areas. In the division of time periods, four time periods of morning, evening, weekdays, and weekends are set. The data set of each time period contains the characteristic information of all charging stations in this period. different.
实施例4Example 4
原始数据集存在一些异常值和缺失值，因此需要一个数据清理过程来处理这些无效的数据。如果检测到离群值，就将它们从数据集中删除；对于缺失值，使用中值填充方法用“插补器”函数计算的中值填充它。The original dataset has some outliers and missing values, so a data cleaning process is needed to deal with these invalid data. If outliers are detected, they are removed from the dataset; for missing values, they are filled with the median value calculated by the "Imputator" function using the median filling method.
在“距离最近的重要POI”中，每个特征的值都超过100，而其他特征(如“交流桩充电费用”)中，平均值仅为每小时0.85。为了使用机器学习算法获得更好的性能，我们采用特征缩放来规范化特征值，使其范围在[0,1]之内。In "Nearest important POI", the value of each feature exceeds 100, while in other features (such as "AC pile charging cost"), the average value is only 0.85 per hour. In order to achieve better performance with machine learning algorithms, we employ feature scaling to normalize the feature values so that they are in the range [0,1].
实施例5Example 5
参考图3，是进行利用率分类和特征重要性排序所使用的模型架构。Referring to Figure 3, it is the model architecture used for utilization classification and feature importance ranking.
本实施例中，定义站点充电桩的利用率是决定站点选址好坏的关键因素，因此对站点选址因素的分析转变为对站点利用率的影响因素(特征)的分析。In this embodiment, it is defined that the utilization rate of charging piles at a site is a key factor that determines whether a site is located well, so the analysis of site site selection factors is transformed into an analysis of factors (features) that affect site utilization.
使用数据集划分工作中得到的地区和时间段数据集进行利用率分类和特征重要性排序工作。在地区分类预测中，对于城区和郊区的数据集，分别随机选择80％的数据作为训练集，剩下20％作为测试集。采用多标签分类任务中常用的平均精度(MAP)作为评价指标。分类预测完成之后，进行特征重要性排序，特征所得分数越高，重要性越高。Use the regional and time period data sets obtained in the data set division work to carry out utilization classification and feature importance ranking. In the regional classification prediction, for the urban and suburban data sets, 80% of the data are randomly selected as the training set, and the remaining 20% are used as the test set. The average precision (MAP), which is commonly used in multi-label classification tasks, is used as the evaluation index. After the classification prediction is completed, the feature importance is sorted. The higher the score of the feature, the higher the importance.
具体地，采用梯度提升决策树算法，计算出每个特征对结果的贡献度得分，便可得到特征与利用率的相关性，并根据相关性，对特征按所得的贡献度分数进行排序，分数高者即为对利用率影响较大的特征，从而得出哪些是影响站点利用率的重要因素。Specifically, the gradient boosting decision tree algorithm is used to calculate the contribution score of each feature to the result, and the correlation between the feature and the utilization rate can be obtained, and according to the correlation, the features are sorted according to the obtained contribution score, and the score The higher one is the feature that has a greater impact on the utilization rate, so it can be concluded which are the important factors that affect the site utilization rate.
参考图4，为对城区站点进行利用率分类所得的特征排序结果。Referring to Fig. 4, it is the feature sorting result obtained by classifying the utilization rate of urban sites.
对于这两个预测任务，可以看到，POI“地铁站”在站点利用率中都起着最重要的作用，在城市地区，使用率受“医院”的影响大于“购物中心”，而在郊区，这两个特征处于相反的位置。在时间段分类预测中，对于四个时间段的数据集，同样分别随机选择80％的数据作为训练集，剩下20％作为测试集。模型在早晨、晚上、工作日、周末数据集上的分类任务都达到了较高的准确率。在特征重要性排序上，对四个时间段站点利用率影响最大的特征分别为“购物中心”、“医院”、“大学”、“地铁站”，从而可进一步得出站点附近重要的POI数量很大程度上影响着站点的利用率情况。For both prediction tasks, it can be seen that POI "subway station" plays the most important role in site utilization, and in urban areas, utilization is more affected by "hospital" than "shopping mall", while in suburban , these two features are in opposite positions. In the time period classification prediction, for the data sets of the four time periods, 80% of the data are also randomly selected as the training set, and the remaining 20% are used as the test set. The classification tasks of the model on the morning, evening, weekday, and weekend datasets have all achieved high accuracy. In terms of feature importance ranking, the features that have the greatest impact on site utilization in the four time periods are "shopping mall", "hospital", "university", and "subway station", so that the number of important POIs near the site can be further obtained It greatly affects the utilization rate of the site.
最终实验结果表明，不论是在不同地区分区的数据集还是不同时间段的数据集上，充电站点周围重要的POI数量都是决定站点利用率高低的关键因素，也说明运营商在进行充电站选址时主要考虑的因素是目标地点附近的商圈、居民区、学校、医院等的数量。The final experimental results show that the number of important POIs around the charging station is a key factor in determining the utilization rate of the station, whether it is in the data sets of different regions or different time periods. The main factor to consider when choosing a location is the number of business districts, residential areas, schools, hospitals, etc. near the target location.
本发明的保护内容不局限于以上实施例。在不背离发明构思的精神和范围下，本领域技术人员能够想到的变化和优点都被包括在本发明中，并且以所附的权利要求书为保护范围。The protection content of the present invention is not limited to the above embodiments. Without departing from the spirit and scope of the inventive concept, changes and advantages conceivable by those skilled in the art are all included in the present invention, and the appended claims are the protection scope.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910444304.XACN110263783A (en) | 2019-05-27 | 2019-05-27 | Multiple features charging addressing analysis of Influential Factors method and system based on deep learning |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910444304.XACN110263783A (en) | 2019-05-27 | 2019-05-27 | Multiple features charging addressing analysis of Influential Factors method and system based on deep learning |
| Publication Number | Publication Date |
|---|---|
| CN110263783Atrue CN110263783A (en) | 2019-09-20 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910444304.XAPendingCN110263783A (en) | 2019-05-27 | 2019-05-27 | Multiple features charging addressing analysis of Influential Factors method and system based on deep learning |
| Country | Link |
|---|---|
| CN (1) | CN110263783A (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111967773A (en)* | 2020-08-18 | 2020-11-20 | 湖南汽车工程职业学院 | Charging facility site selection system of charging automobile based on big data |
| CN112561322A (en)* | 2020-12-14 | 2021-03-26 | 广东电网有限责任公司珠海供电局 | Charging pile site selection method based on machine learning |
| CN112785049A (en)* | 2021-01-12 | 2021-05-11 | 北京首汽智行科技有限公司 | Shared automobile city opening site selection method and system based on POI characteristics |
| CN114693042A (en)* | 2020-12-31 | 2022-07-01 | 奥动新能源汽车科技有限公司 | Site selection method and device for power conversion station, electronic equipment and storage medium |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105023196A (en)* | 2015-07-02 | 2015-11-04 | 国家电网公司 | Analysis method and device for charging transaction data of charging stations |
| CN107870810A (en)* | 2017-10-31 | 2018-04-03 | 广东欧珀移动通信有限公司 | Application cleaning method, device, storage medium and electronic device |
| CN109003128A (en)* | 2018-07-07 | 2018-12-14 | 太原理工大学 | Based on improved random forest public bicycles website Demand Forecast method |
| CN109636031A (en)* | 2018-12-10 | 2019-04-16 | 国家电网有限公司 | A kind of city charging station integrated planning method |
| CN109740879A (en)* | 2018-12-20 | 2019-05-10 | 国网北京市电力公司 | Method and device for processing charging pile data |
| CN109754168A (en)* | 2018-12-20 | 2019-05-14 | 国网北京市电力公司 | Method and device for site selection of charging station |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105023196A (en)* | 2015-07-02 | 2015-11-04 | 国家电网公司 | Analysis method and device for charging transaction data of charging stations |
| CN107870810A (en)* | 2017-10-31 | 2018-04-03 | 广东欧珀移动通信有限公司 | Application cleaning method, device, storage medium and electronic device |
| CN109003128A (en)* | 2018-07-07 | 2018-12-14 | 太原理工大学 | Based on improved random forest public bicycles website Demand Forecast method |
| CN109636031A (en)* | 2018-12-10 | 2019-04-16 | 国家电网有限公司 | A kind of city charging station integrated planning method |
| CN109740879A (en)* | 2018-12-20 | 2019-05-10 | 国网北京市电力公司 | Method and device for processing charging pile data |
| CN109754168A (en)* | 2018-12-20 | 2019-05-14 | 国网北京市电力公司 | Method and device for site selection of charging station |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111967773A (en)* | 2020-08-18 | 2020-11-20 | 湖南汽车工程职业学院 | Charging facility site selection system of charging automobile based on big data |
| CN112561322A (en)* | 2020-12-14 | 2021-03-26 | 广东电网有限责任公司珠海供电局 | Charging pile site selection method based on machine learning |
| CN114693042A (en)* | 2020-12-31 | 2022-07-01 | 奥动新能源汽车科技有限公司 | Site selection method and device for power conversion station, electronic equipment and storage medium |
| CN112785049A (en)* | 2021-01-12 | 2021-05-11 | 北京首汽智行科技有限公司 | Shared automobile city opening site selection method and system based on POI characteristics |
| Publication | Publication Date | Title |
|---|---|---|
| Salon | Neighborhoods, cars, and commuting in New York City: A discrete choice approach | |
| Zhao et al. | Identification of land-use characteristics using bicycle sharing data: A deep learning approach | |
| Zhang et al. | Urban spatial structure and travel patterns: Analysis of workday and holiday travel using inhomogeneous Poisson point process models | |
| CN110263783A (en) | Multiple features charging addressing analysis of Influential Factors method and system based on deep learning | |
| Yang et al. | How to improve urban transportation planning in big data era? A practice in the study of traffic analysis zone delineation | |
| CN105355049A (en) | Highway running state evaluation method based on macroscopic fundamental diagram | |
| CN114048920A (en) | Site selection layout method, device, equipment and storage medium for charging facility construction | |
| CN110836675A (en) | Decision tree-based automatic driving search decision method | |
| CN118072508B (en) | Urban traffic intelligent planning management system based on data analysis | |
| Xu et al. | Cross-region traffic prediction for china on openstreetmap | |
| Tu et al. | Research on intelligent calculation method of intelligent traffic flow index based on big data mining | |
| CN107590766A (en) | A kind of method of discrimination of the land used combination form related to road traffic accident risk | |
| CN108122131A (en) | Travel pattern and the recognition methods of duty residence based on public bicycles brushing card data | |
| Li et al. | Spatial accessibility to shopping malls in Nanjing, China: Comparative analysis with multiple transportation modes | |
| Li et al. | Effects of the built environment on travel distance in bus-oriented, medium-sized cities in China | |
| CN114186146B (en) | Network taxi service network site selection method and system with capacity limitation | |
| Kurte et al. | Regional-scale spatio-temporal analysis of impacts of weather on traffic speed in Chicago using probe data | |
| Talaat et al. | Mobile monitoring and spatial prediction of black carbon in Cairo, Egypt | |
| Sun et al. | An AHP-Fuzzy comprehensive evaluation model for parking lots in urban CBD area | |
| CN110807078A (en) | Citizen track big data service platform serving for civil planning | |
| Kong et al. | Charging pile siting recommendations via the fusion of points of interest and vehicle trajectories | |
| Crawford et al. | Analysing spatial intrapersonal variability of road users using point-to-point sensor data | |
| Gao et al. | iTA: inferring traffic accident hotspots with vehicle trajectories and road environment data | |
| Chen et al. | An overall analysis method of urban road parking lots based on data mining | |
| Borowska-Stefańska et al. | Patterns of temporal and spatial variability of parking in a large City in the context of road network configuration–The case of Łódź, Poland |
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | Application publication date:20190920 | |
| WD01 | Invention patent application deemed withdrawn after publication |