CN113191568A - Meteorological-based urban operation management big data analysis and prediction method and system - Google Patents
Meteorological-based urban operation management big data analysis and prediction method and system Download PDFInfo
- Publication number
- CN113191568A CN113191568A CN202110558687.0A CN202110558687A CN113191568A CN 113191568 A CN113191568 A CN 113191568A CN 202110558687 A CN202110558687 A CN 202110558687A CN 113191568 A CN113191568 A CN 113191568A
- Authority
- CN
- China
- Prior art keywords
- model
- event
- data
- weather
- meteorological
- Prior art date
- Legal status (The legal status 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 status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 78
- 238000007405 data analysis Methods 0.000 title claims abstract description 26
- 230000000694 effects Effects 0.000 claims abstract description 17
- 238000012549 training Methods 0.000 claims description 67
- 238000005065 mining Methods 0.000 claims description 13
- 238000005457 optimization Methods 0.000 claims description 2
- 238000004422 calculation algorithm Methods 0.000 abstract description 16
- 238000001556 precipitation Methods 0.000 abstract description 13
- 238000010801 machine learning Methods 0.000 abstract description 6
- 238000003066 decision tree Methods 0.000 description 23
- 238000007726 management method Methods 0.000 description 21
- 230000008569 process Effects 0.000 description 15
- 238000004458 analytical method Methods 0.000 description 14
- 230000000875 corresponding effect Effects 0.000 description 14
- 238000005516 engineering process Methods 0.000 description 9
- 238000010276 construction Methods 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 238000007781 pre-processing Methods 0.000 description 5
- 238000002790 cross-validation Methods 0.000 description 4
- 230000001186 cumulative effect Effects 0.000 description 4
- 238000009826 distribution Methods 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 4
- 230000002159 abnormal effect Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000018109 developmental process Effects 0.000 description 3
- 238000005096 rolling process Methods 0.000 description 3
- 230000002123 temporal effect Effects 0.000 description 3
- 241001123248 Arma Species 0.000 description 2
- 230000002596 correlated effect Effects 0.000 description 2
- 238000005520 cutting process Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000013180 random effects model Methods 0.000 description 2
- 230000008439 repair process Effects 0.000 description 2
- 238000012731 temporal analysis Methods 0.000 description 2
- 238000000700 time series analysis Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000002354 daily effect Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- 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
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Software Systems (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- Game Theory and Decision Science (AREA)
- General Business, Economics & Management (AREA)
- Operations Research (AREA)
- Development Economics (AREA)
- Quality & Reliability (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Educational Administration (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a weather-based urban operation management big data analysis and prediction method and a system, wherein an event quantity prediction model is established, the model is based on a classic machine learning algorithm gradient lifting model, and a two-step method modeling, a random intercept model and a random effect model are used for improving the gradient lifting model so as to better capture the occurrence rule of an event. In the actual operation of the model, through fusing various data sources (meteorological automatic station data, grid point meteorological element forecast data, grid event data, hot line event data and 110 meteorological disaster data), the forecasting of the occurrence number of the events within 48 hours and every 12 hours and the corresponding risk early warning level are realized. On the basis of the event quantity prediction model, the influence of three meteorological elements, namely wind speed, precipitation and air temperature, on the event quantity is obtained by calculating the contribution value of the meteorological elements to the event quantity prediction value.
Description
Technical Field
The invention relates to the technical field of weather, in particular to a weather-based urban operation management big data analysis and prediction method and system.
Background
In recent years, the development of big data technology is rapid, and the big data technology becomes a hot spot of research and pursuit of all countries in the world. On one hand, the application range of the big data technology is wide, and the big data technology is particularly prominent in the fields of medical treatment, finance, security protection, automobiles and the like. The meteorological application is also an important field of high-performance computing, and the big data technology brings rare opportunities to the development of observation, forecast, service and other businesses and also brings great challenges. Therefore, the characteristics of the development of big data technology also have great influence on the meteorological service. On the other hand, the machine learning, model training processing, computer vision and other important big data technologies have profound influence on weather in different fields and influence the life of audiences to different degrees.
Disclosure of Invention
In order to solve the defects in the prior art, the invention provides a weather-based urban operation management big data analysis and prediction method and system, which can establish a prediction model of the total number of long-time rolling accumulated events and a specific event scene.
The invention provides the following technical scheme:
in a first aspect, a weather-based method for analyzing and predicting big data of city operation management comprises the following steps:
collecting a data source, and generating an event quantity prediction model training set;
based on the training set, adopting a two-step method to model the occurrence probability and the occurrence number of the events respectively;
in the event occurrence probability model, predicting whether each object will have an event within t time by adopting a gradient lifting model, and outputting the probability of whether each object has the event within t time;
in the event occurrence quantity model, predicting the quantity of events occurring in t time once each object has an event by adopting a gradient lifting model;
and multiplying the expected occurrence probability of the event by the expected number of the event once occurring to obtain the number value of the event occurrence, and establishing an event number prediction model.
As a preferred technical solution of the prediction method of the present invention, the data sources include weather automatic station data, grid point weather element forecast data, grid event data, hot line event data, and 110 weather disaster data.
As a preferable technical scheme of the prediction method, the gradient boost model is a LightGBM-based gradient boost model, and a random intercept model and a random effect model are introduced to optimize and upgrade the model on the basis.
As a preferable technical scheme of the prediction method, the method further comprises the steps of carrying out feature mining on time data, historical and live meteorological element data and historical and live city operation data (grid data, hot line data and 110 meteorological disaster data), and adding the model training as the input of the comprehensive prediction model of the event number.
As a preferable technical scheme of the prediction method, in the mining of the historical and live city operation data characteristics, the delay of the occurrence number of the events in time is judged by adopting a partial autocorrelation coefficient.
In a preferred embodiment of the prediction method of the present invention, temporal and historical influences of the meteorological elements on the number of events are considered in mining the data characteristics of the historical and live meteorological elements, and the time lag relationship between the number of events and the different meteorological elements is calculated by using a cross-correlation coefficient on the influence of the historical meteorological elements on the number of events.
As a preferred technical scheme of the prediction method, the method further comprises the step of selecting an event scene closely related to meteorological influence by adopting the two-step modeling to establish a scene model while establishing the event quantity prediction model.
As an optimal technical scheme of the prediction method, the method further comprises the following steps of establishing an event risk early warning model: and formulating an event early warning standard by combining the quantile of the occurrence number of the historical events and the absolute threshold of the occurrence number of the events.
As a preferred technical scheme of the prediction method, the method further comprises the following steps of establishing a meteorological influence index model:
calculating the percentage contribution of each meteorological element characteristic to the event quantity prediction based on the event quantity prediction model;
and taking the percentage contribution value of the meteorological element characteristics in the range of the identified meteorological element as a reference, calculating the absolute deviation of the percentage contribution value of the current meteorological element characteristics and the reference, and taking the absolute deviation as the influence index of the meteorological element characteristics.
In a second aspect, the system is used for executing the above method for analyzing and predicting the weather-based urban operation management big data.
Due to the adoption of the technical scheme, the invention has the following beneficial effects:
the event quantity prediction model is based on a Gradient Boosting model (a classical Machine learning algorithm) and is improved by using a two-step modeling method, a random intercept model and a random effect model, so that the occurrence rule of events can be captured better. In the actual operation of the model, through fusing various data sources (meteorological automatic station data, grid point meteorological element forecast data, grid event data, hot line event data and 110 meteorological disaster data), the forecasting of the occurrence number of the events within 48 hours and every 12 hours and the corresponding risk early warning level are realized. On the basis of the event quantity prediction model, the influence indexes of three meteorological elements, namely wind speed, precipitation and air temperature, on the occurrence of the event quantity are obtained by calculating the contribution values of the meteorological elements to the event quantity prediction value.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a general flowchart of the weather-based big data analysis and prediction method for city operation management according to the present invention.
FIG. 2 is a model technology route diagram of the weather-based urban operation management big data analysis and prediction method of the present invention.
FIG. 3 is a flowchart illustrating an embodiment of a weather-based method for analyzing and predicting big data of city operation management according to the present invention.
FIG. 4 is a schematic diagram of a characteristic engineering process of the weather-based urban operation management big data analysis and prediction method of the present invention.
Detailed Description
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail below.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced in other ways than those specifically described herein, and thus the present invention is not limited to the specific embodiments disclosed below.
Referring to fig. 1-4, the embodiment of the invention provides a city operation management big data analysis and prediction method and system based on weather, a method of integrating multiple statistical machine learning models is used, a prediction model of the total number of events in 12-hour sliding and a specific event scene is established for city street township grids/hot lines (such as Shanghai grids, Pudong grids, Xuhui grids, Pudong hot lines and Xuhui hot lines) respectively, and the prediction time is 48 hours; and (3) a prediction model of the number of 1-hour events of rainstorm and strong wind is established in the case of the 110 weather disasters, and the prediction aging is 48 hours. According to the event quantity prediction model, a meteorological influence index composed of three meteorological element characteristics of wind speed, precipitation and air temperature is designed, and the meteorological influence index is used for quantifying the influence degree of different meteorological elements on the urban operation condition. And finally, according to the historical event data distribution, designing an event early warning standard, and forming a risk early warning system by matching with an event quantity prediction model.
The event quantity prediction model is based on a Gradient Boosting model (a classical Machine learning algorithm) and is improved by using a two-step modeling method, a random intercept model and a random effect model, so that the occurrence rule of events can be captured better. In the actual operation of the model, through fusing various data sources (meteorological automatic station data, grid point meteorological element forecast data, grid event data, hot line event data and 110 meteorological disaster data), the forecasting of the occurrence number of the events within 48 hours and every 12 hours and the corresponding risk early warning level are realized. On the basis of the event quantity prediction model, the influence of three meteorological elements, namely wind speed, precipitation and air temperature, on the event quantity is obtained by calculating the contribution value of the meteorological elements to the event quantity prediction value. A prediction model of the quantity of grid events of city, Pudong and Xuhui, a prediction model of the quantity of Hot line events of Pudong and Xuhui and a prediction model of the quantity of meteorological disasters of the city 110 are deployed according to the data types and sources.
Specifically, the city operation management big data analysis and prediction method based on weather comprises the following steps:
step 1: collecting a data source, and generating an event quantity prediction model training set;
step 2: based on the training set, adopting a two-step method to model the event occurrence probability and the event occurrence quantity respectively;
and step 3: in the event occurrence probability model, predicting whether each object will have an event within t time by adopting a gradient lifting model, and outputting the probability of whether each object has the event within t time;
and 4, step 4: in the event occurrence quantity model, predicting the quantity of events occurring in t time once each object has an event by adopting a gradient lifting model;
and 5: and multiplying the expected occurrence probability of the event by the expected number of the event once occurring to obtain the number value of the event occurrence, and establishing an event number prediction model.
The data sources in step 1 include city operation management big data such as weather automatic station data, grid point weather element forecast data, grid event data, hot line event data, 110 weather disaster data and the like.
The gradient boost model in step 2 to step 4 is a LightGBM-based gradient boost model. The object may be a city, a district, or a town.
During the training and real-time online prediction of the model, feature mining is also carried out, and feature mining is carried out on time data, historical and live meteorological element data and historical and live city operation data (grid data, hot line data and 110 meteorological disaster data), and model training and real-time online prediction are added. The method and features of feature mining will be described in detail below.
In addition, when the event quantity prediction model is established, a two-step method modeling, a random intercept model and a random effect model are adopted to upgrade and optimize the model, select event scenes closely related to meteorological influence and establish a scene model. And an event early warning standard is formulated by combining the quantile of the occurrence number of the historical events and the absolute threshold of the occurrence number of the events, and an event risk early warning model is established. And establishing a meteorological influence index model by adopting the percentage contribution degree of each meteorological element characteristic to the event quantity prediction.
The details of the weather-based big data analysis and prediction model for city operation management according to the present invention are further described below for specific model purposes:
purpose of the model: and establishing a prediction model of the total number of the sliding 12-hour events and a specific event scene for the Shanghai grid, the Pudong grid, the Xuhui grid, the Pudong hot line and the Xuhui hot line, and establishing a prediction model of the specific event scene for 1 hour for the 110-weather disaster.
Firstly, a data source used by a model:
1. weather automatic station live data
(1) Data source structure
One of the main meteorological data sources used by the model in training and real-time online prediction is city-wide meteorological automatic station data. The historical data of each station specifically comprises fields such as automatic station names, automatic station longitude and latitude coordinates, cities and streets to which the automatic stations belong, automatic station element numbers and the like. The meteorological element data monitored by each automatic station in real time comprise fields such as temperature, rainfall, wind direction, wind speed, 2-minute wind direction, two-minute wind speed, maximum wind direction and maximum wind speed. When a specific event prediction model is trained and quantity prediction is carried out online in real time, temperature, rainfall and maximum wind speed are selected as main inputs of the model.
(2) Data preprocessing process
The corresponding live meteorological element data of each street town is generated through the following calculation process:
for the case where there are autonomous stations within a zone/street town:
firstly, carrying out space matching on the automatic meteorological station and a street town;
calculating the average rainfall, the maximum wind speed, the lowest air temperature and the highest air temperature of the automatic station in each region/street town;
thirdly, the rainfall, the wind speed and the highest and lowest temperature obtained in the second step are used as meteorological element characteristics of the current street and town corresponding time.
For the case where there are no autonomous stations within a region/street town:
the street town acquires the automatic station which is closest to the street town and has a corresponding observation value by calculating the minimum straight-line distance between each automatic station and each district/street town;
and (4) calculating the first step and the second step to obtain the meteorological element characteristics of the current time of the current region/street town.
2. Lattice meteorological element forecast data
(1) Data source structure
Another major meteorological element data source used by the model for real-time on-line prediction is the grid-spotted meteorological element prediction data. The main form of data exists as formatted meteorological element forecast grid data, the size of the grid is 51 × 57, and the grid covers the urban and suburban areas of the Shanghai. The data storage is in the form of multi-channel raster data of NetCDF. The updating frequency of the grid meteorological element forecast is twice a day, namely eight morning hours and eight night hours each day. And outputting the grid point meteorological element forecast data hour by hour within 48 hours every forecast. And the grid point meteorological element forecast data comprise cloud layer coverage percentage, a wind field with the height of 10 meters, ground rainfall, relative humidity and air temperature. And in the process of the model running on line actually, a wind field with the height of 10 meters, ground rainfall and air temperature are selected as input characteristics of the model.
(2) Data preprocessing process
When the model is operated online in real time, the lattice meteorological element forecast is preprocessed by using the following steps, and the rainfall, air temperature and maximum wind speed characteristics of all streets and towns under the jurisdiction of the city are generated:
carrying out spatial connection according to geographical grid data in grid point forecast data and boundary data of a street town to form a corresponding table of the grid points and the street town;
reading 3-dimensional (time, longitude and latitude) grid data of ground rainfall, wind field and air temperature elements in a NetCDF format;
thirdly, calculating the maximum wind speed corresponding to the grid points according to the U vector and the V vector in the wind field;
converting the unit of the air temperature element from Kelvin to centigrade, and unifying the unit of the real-time monitoring meteorological element of the meteorological automatic station;
calculating the average rainfall, the maximum wind speed, the lowest air temperature and the highest air temperature of the automatic grid points in each county/street town according to the grid point-street town correspondence table in the step 1;
and sixthly, taking the rainfall, the wind speed and the highest and lowest air temperature obtained in the step 5 as meteorological element characteristics of the current street and town corresponding time.
Through the calculation process, the strong wind, precipitation and temperature information obtained by forecasting the grid point meteorological elements of the whole town can be obtained and used for inputting the subsequent meteorological features of the model.
3. Grid live event data
(1) Data source structure
Shanghai grid data used 2019/01/0100: 19:18 to 2020/02/01,2020/05/01-2020/10/19 data as the training set, which involved 4352245 pieces of data. The Shanghai grid original data has 124 fields, and five fields, namely discovery time, INFOSCNAME, STREETNAME (name of affiliated street) and TASKID (task number), are selected for final analysis according to analysis requirements.
The Pudong grid data used 2020/01/0100: 19:18 through 2020/02/01,2020/03/01-2020/10/19 data as the training set, which relates to 1077230 bars of data. The Pudong grid original data has 124 fields, and according to analysis requirements, six fields of DISCOVERTIME (discovery time), INFOSCNAME (subclass name), ISFAST (whether rapid treatment exists), STREETNAME (name of affiliated street) and TASKID (task number) are selected for final analysis.
The xu-hui grid data used 2018/01/15-2020/10/15 data as training sets, which collectively involved 1,054,935 pieces of data. The original data of the xu hui grid has 17 fields, and according to analysis requirements, six fields of CREATETIME (discovery time), SECOND _ TYPE (subclass name), THIRD _ TYPE (subclass name), STREETNAME (name of street to which the Xuhui grid belongs), TASKID (task number) and CASEATTributATE (event TYPE) are selected for final analysis.
(2) Data preprocessing process
When the grid scene model is trained, event types are classified by using a subclass name field in data, and prediction and training targets in the grid scene model are determined by the method.
And when the corresponding label of the non-rapid handling model is determined, selecting an isfast (whether to rapidly handle the class event) field as a screening condition, and filtering the field into a prediction event object of the non-rapid handling class model. And the non-rapid disposal model total amount and the prediction model label of the scene model are output by combining with the standard reaching rule of the scene model.
4. Hotline live event data
(1) Data source structure
The Pudong hotline data uses the data from 2020/01/0100: 19:18 through 2020/02/01,2020/03/01-2020/10/19 as a training set, referring to a total of 586869 pieces of data. The original data of the Pudong hotline has 136 fields, and six fields of DISCOVERTIME (discovery time), INFOSCNAME (subclass name), INFOZCNAME (subclass name), STREETNAME (street name) and TASKID (task number) are selected for final analysis according to analysis requirements.
The xu Hui Hot line data uses 2018/01/15-2020/10/15 data as a training set, which collectively refers to 144,297 pieces of data. The original data of the xu Hui hot line has 17 fields, and according to analysis requirements, six fields of CREATETIME (discovery time), SECOND _ TYPE (subclass name), THIRD _ TYPE (subclass name), STREETNAME (name of street to which the Xuhui hot line belongs), TASKID (task number) and CASEATTributATE (event TYPE) are selected for final analysis.
(2) Data preprocessing process
When the hotline scene model is trained, event types are classified by using subclass name fields in data, and prediction and training targets in the hotline scene model are determined by the method.
5. 110 event weather disaster data
(1) Data source structure
The 110 weather disaster data uses data from 2020/01/0100: 00:00 to 2020/07/2924: 00:00 as a validation set, and relates to 1,998 pieces of data in total.
110 real-time original data of meteorological DISASTERs have 13 fields, according to analysis requirements, OBJECTD (CASE id), DATETIME _ DISASTER (alarm time), TELEPHONE (alarm person TELEPHONE), LONTITUDE (longitude), LATITUDE (LATITUDE), CASE _ ADDR (alarm CASE address), CASE _ DESC (alarm CASE content), DISTRICT (DISTRICT and county) are selected for final analysis.
(2) Data preprocessing process
When the 110 meteorological disaster scene model is trained, the prefecture and county fields are simply cleaned, and the meteorological disaster content fields are structurally processed. The method is characterized in that disaster fields in data are used for classifying weather disaster types, prediction and training targets in a scene model are determined through the method, and finally, the strong wind and rainstorm events with the largest number of events are selected for prediction modeling.
Second, event number prediction model
1. Principle of model
(1) Decision tree model
The decision tree is an algorithm for solving classification and problems, the decision tree algorithm adopts a tree structure, and final classification is realized by using layer-by-layer reasoning. The decision tree is composed of several elements:
root node: containing the corpus of samples
Internal nodes: corresponding feature attribute testing
Leaf node: representing the result of the decision
At each state node, the decision tree traverses each data latitude (feature) and calculates the maximum information gain obtained in the feature as the growth direction of the decision tree, and calculates which specific node in the direction will bring the maximum information gain when the data is segmented. And constructing the internal node of the next level according to the internal node until the highest growth height of the decision tree of the preset quota is reached or information gain cannot be brought by growth in any direction and data segmentation. The finally formed decision tree is a classification and regression model formed by combining a plurality of binary conditional judgment rules. And different information gain functions also determine the difference in the decision tree growing algorithm. Commonly used algorithms are the ID3/C4.5 algorithm based on information entropy and the CART tree which is calculated based on the Gini coefficient. And in a weather-based urban operation management big data analysis and prediction model, the used decision tree algorithm is a CART tree.
Decision trees have many advantages: the algorithm is easy to intuitively understand the internal structure of the model, and can directly reflect the characteristics of data and intuitively explain the logic of making predictions by the model. Moreover, the decision tree is very robust in data preparation, and can simultaneously process numerical and entry type features. However, in practical applications, the disadvantages of the decision tree are also more prominent. When the data dimension is large and the number of samples is unbalanced, the decision tree model generates an excessively complex but unstable judgment structure without constraint. And when the data has a sample unbalanced item type characteristic, the decision tree model is very easy to grow a deep and unbalanced structure. Therefore, in practical projects, the upgrade algorithm of the decision tree is used, and the gradient upgrade model is used for alleviating the disadvantages of the decision tree.
(2) Gradient lifting model
The Gradient Boosting Machine (Gradient Boosting Machine) is a classic Machine learning model algorithm, and a flexible prediction model with excellent fitting capability is obtained by integrating a plurality of simple decision trees. In the model, the residual error of each decision tree fitting the target is used as the learning target of the next decision tree, and the steps are repeated until the model converges or makes a decision. The specific gradient boosting model implementation used in the event prediction model is based on LightGBM. This is a gradient lifting model framework from Microsoft (Microsoft) open sources. The method has the advantages of high training speed, high memory use efficiency, high prediction precision, support for various prediction scenes (support for classification, regression and sequencing) and the like, and is widely applied in the industry.
The LightGBM has two special advantages compared to the general gradient boosting algorithm implementation. First, the LightGBM uses a growth logic generated by leaf nodes, as opposed to a logic generated by a conventional decision tree layer. At each secondary length of the decision tree, one leaf node with the highest splitting gain (generally, the largest data amount) is found from all the leaf nodes at present, and then the leaf nodes are split, and the process is repeated. This growing strategy will bring more information gain improvement while keeping the same number of leaf nodes. And in the aspect of processing the class characteristics, the LightGBM can find out the optimal cutting of the class characteristics, namely, the cutting mode of many-vs-many. This also solves the phenomenon that tree models are prone to grow imbalances when the number of classes in the entry-type feature is large.
Compared with other decision tree algorithms which need to traverse data and sort the data, the LightGBM adopts the histogram algorithm to preprocess the data, so that the memory use efficiency of the model during operation is improved, and the discretization operation of the histogram algorithm on the data is used, so that the sensitivity of the tree model to abnormal points during growth is caused, and the generalization effect of the model is further improved.
The gradient lifting model introduced above is modified, a two-step modeling, a random intercept model and a random effect model are used for resisting the sparsity problem existing in the samples and the unbalanced condition of each street and town sample, and finally a grid event number prediction model is established, and the grid event number prediction model is shown in fig. 2 and fig. 3.
(3) Two step method modeling
Two-step modeling is a modeling method used to account for the impact of data sparsity (the presence of a large number of zero values in the data) on the prediction model. When the grid hot line data is distributed to towns/12 hours, the number of the sample events in the data is zero, namely no event occurs, and more than 70 percent of the sample events in the data are zero. Fitting the model using this data would make the prediction of the model less significant. In order to reduce the influence of the problem on the model, the model is divided into an event occurrence probability model and an event occurrence number model. In the event occurrence probability model, a gradient lifting model is used for predicting whether a corresponding grid hot line event occurs in 12 hours in each street town, and the probability of whether the corresponding event occurs in 12 hours in each street town is taken as main output. In the event occurrence number model, the gradient lifting model is used for predicting the number of events occurring in 12 hours once the events occur in each street town. And multiplying the expected occurrence probability of the event by the expected number of the event once occurring to finally obtain the value of the number of the grid hotline occurrences.
Two-step modeling is widely applied in economics, sociology, and medicine. In economics, two-step modeling is also commonly referred to as a fence model. In economics, he is used to set some hard settings (i.e., fences) for the supply chain, which is common in segment pricing practices. Customers who meet this criteria are discounted prices, called the Price Discrimination fence Model (Hurdle Model of Price Discrimination) by economists. Due to the existence of the barrier, the purchase data of the user usually presents a sparse phenomenon, namely, the real consumption population often only occupies a small part of the used population. This also coincides with temporal sparsity encountered in the mesh hotline scene model. The use of two-step modeling herein mitigates the phenomenon of data sparseness.
(4) Random intercept model
When the event occurrence number Model is trained, in order to solve the problem that the Model is easy to be over-fitted due to the fact that street and town sample imbalance problems (event samples are concentrated on a small number of street and towns) appear in characteristics, a Random-Intercept Model (Random-Intercept Model) is used for estimating a reference level of each street and town corresponding to the occurrence of an event, and the estimated reference level is used as the pre-input of an event number prediction Model in two-step modeling.
(5) Model of random effects
Random effects models (random effects models) are a generalization of the classical linear model, which considers fixed regression coefficients as random variables, generally assumed to be from a normal distribution. If some of the coefficients in the model are random and others are fixed, the model is generally called a mixed model (mixed models). The random effect is introduced to make the individual observation have a certain correlation, so the random effect is a proper choice for fitting the data of the non-independent observation. The grid hotline quantity data is also a non-independent observation data as described above. When the mixed/random effect model is used, due to a compression (shrinkage) phenomenon during random effect fitting, estimated values of individuals with fewer samples during model fitting are close to a middle value of a population, and the phenomenon also limits that part of street towns with fewer samples but more abnormal street towns are not easily influenced by abnormal values which are sporadically appeared during estimation.
2. Feature mining
In conjunction with fig. 4, three features of the same type, time, historical live data, and meteorological elements, are used as model inputs in mining of model inputs.
In the temporal characteristics, the hours of day, whether it belongs to a workday, the months of day, and whether it is on a holiday are used to capture the grid number of events occurring and the relationship between the number of events occurring and the time of day.
In the mining of the historical live data characteristics, the delay of the occurrence number of events in time is judged by using a statistical index of Partial Auto-Correlation (Partial Auto-Correlation). By this index, it can be found that the number of events occurring 12 hours ago, 24 hours ago, 36 hours ago, and 1 week ago are statistically significantly correlated with the number of events occurring at the present time. These features are then also used as inputs to the model.
In the mining of the characteristics of meteorological elements, the instantaneous and historical influences of three meteorological conditions, namely precipitation, air temperature and wind speed, on the occurrence number of events are mainly considered. On the instantaneous meteorological elements, the current hour cumulative rainfall, the current hour maximum wind speed, the current hour maximum temperature, and the current hour minimum temperature are used as inputs to the model. In addition to the statistics (e.g., 12-hour and 24-hour cumulative rainfall for determining whether the storm standard is present) commonly used in the gas phase standard, the influence of the historical meteorological elements on the occurrence number of events is also calculated by using Cross-Correlation coefficients (Cross-Correlation) to calculate the time lag relationship between the occurrence number of events and different meteorological elements, and finally adding statistical indexes (e.g., maximum, minimum, average, cumulative value) of the strong wind, precipitation and atmospheric temperature elements in 6 hours, 36 hours, and 48 hours as the input of the model.
The partial autocorrelation coefficients are common statistical indicators in time series analysis and modeling. It is commonly used to find out whether a sample in a sequence is correlated with a sample that is a period of time ago. And in a specific mathematical definition he measures XtAnd Xt-kThe correlation coefficient after removing the middle k-1 interference term variables is eliminated. Typically, in the classical model autocorrelation-averaging (ARMA) model of time series. The partial autocorrelation coefficients are used to determine the number of autoregressive terms in the ARMA model. In the grid hot line event number prediction model, the time delay post term of the grid hot line event number is taken as a characteristic and put into the gradient lifting model. The cross-correlation coefficient is also a statistical index which is commonly used in time series analysis and modeling. Unlike the partial autocorrelation coefficients, it is often used to mine the correlation between sequences of different equal length. If the sequence A and the sequence B have more significant correlation coefficient on the lag term k, thenThe time lag between sequence a and sequence B can also be said to be K time units. Because the hourly grid hotline event occurrence number is equal to the time sequence of the strong wind, the rainfall and the air temperature, the historical meteorological element data most relevant to the grid hotline event number can be mined and put into the input of the grid hotline event gradient lifting model as the characteristics.
3. Grid quantity prediction model construction and scene model
The gradient lifting model is changed in the model construction process, a two-step method modeling, a random intercept model and a random effect model are used for resisting the sparsity problem existing in the samples and the unbalanced condition of each street and town sample, and finally a grid event number prediction model is established.
Besides predicting the total quantity of grid events, 6 grid event subclass scenes which are closely related to meteorological influences are selected, and a grid scene model is established. The selected subclass of scenes is as follows: street trees, public greenbelts, vehicle moving for help, overhead line falling, greenbelt guardrails and community greening. For grid scenes (such as cell greening, street trees and the like) with small number and unbalanced street-town samples, the phenomenon of over-fitting of the grid scene model can be remarkably reduced by using two-step modeling and a random intercept model.
Unlike the gross model of the grid event quantity prediction model (i.e. the quantity value of event occurrences), the phenomenon of sparse tag data in the grid scene model is particularly obvious. And the sparseness of events of different scene types is also not nearly the same. Different model parameters are also used for fitting and training the model for different scene models. And also to some extent on the training strategy of the model. In the gross model, the two-step modeling is centered on the event occurrence number model, i.e., predicting what the number of events will occur once an event occurs. Due to the sparse data in the scene model, the number of the scene model is often only the same in the event occurrence situation. Therefore, in the scene model, the training center will shift to the direction of whether the event occurs or not (i.e. the center is placed on the event occurrence probability model), i.e. the occurrence probability of the events of the subclass is changed under the current meteorological conditions, time and town. The training parameters of the scene model are also modified according to such model training strategy changes. Through a mode of combining cross validation and manual work, the training parameters of each scene model are determined, and the training parameters of each scene model are as follows:
and (3) street tree: the number of training rounds of the event occurrence probability model is 600 rounds, and the number of training rounds of the event occurrence number model is 300 rounds;
public greenbelt: the number of training rounds of the event occurrence probability model is 500 rounds, and the number of training rounds of the event occurrence number model is 500 rounds;
seeking help when moving the vehicle, wherein the number of training rounds of the event occurrence probability model is 400 rounds, and the number of training rounds of the event occurrence number model is 600 rounds;
the overhead line falls: the number of training rounds of the event occurrence probability model is 700 rounds, and the number of training rounds of the event occurrence number model is 300 rounds;
green land guardrail: the number of training rounds of the event occurrence probability model is 500 rounds, and the number of training rounds of the event occurrence number model is 500 rounds;
greening the residential area: the number of training rounds of the event occurrence probability model is 400 rounds, and the number of training rounds of the event occurrence number model is 300 rounds.
4. Hot line quantity prediction model construction and scene model
The gradient lifting model is changed in the model construction process, a two-step method modeling, a random intercept model and a random effect model are used for resisting the sparsity problem existing in the samples and the unbalanced condition of each street and town sample, and finally a hot line event number prediction model is established.
Besides the prediction of the total quantity of hot-line events, the prediction modeling is also carried out on 5 hot-line sub-class scenes of taxies, road maintenance, fault reporting and repairing, green space greening and drainage and pollution discharge management, and a scene model is established. For grid scenarios (such as drainage management, road maintenance, etc.) where a small number of grid scenarios occur and the samples between streets and towns are unbalanced, the phenomenon of model overfitting can be significantly reduced by using two-step modeling and a random intercept model.
Unlike the total amount model, the phenomenon that the label data is rare in the hotline scene model is particularly obvious. And the sparseness of events of different scene types is also not nearly the same. So we also use different model parameters for fitting and training the model for different scene models. And also to some extent on the training strategy of the model. In the hot-line total model, the two-step modeling is centered on the quantity prediction model, i.e., the quantity of events occurring once an event occurs is predicted. Due to the sparse data in the scene model, the number of the scene model is often only the same in the event occurrence situation. Therefore, in the scene model, the training center will shift to the direction of whether the event occurs, i.e. the current meteorological condition, time, and street town are in, the occurrence probability of the events of small class is changed. The training parameters of the scene model are also modified according to such model training strategy changes. Through a mode of combining cross validation and manual work, the training parameters of each scene model are determined, and the training parameters of each scene model are as follows:
a taxi: the number of training rounds of the event occurrence probability model is 600 rounds, and the number of training rounds of the event occurrence number model is 300 rounds;
road maintenance: the number of training rounds of the event occurrence probability model is 600 rounds, and the number of training rounds of the event occurrence number model is 400 rounds;
and (3) fault repair: the number of training rounds of the event occurrence probability model is 500 rounds, and the number of training rounds of the event occurrence number model is 500 rounds;
green land greening: the number of training rounds of the event occurrence probability model is 600 rounds, and the number of training rounds of the event occurrence number model is 500 rounds;
and (3) drainage and pollution discharge management: the number of training rounds of the event occurrence probability model is 500 rounds, and the number of training rounds of the event occurrence number model is 600 rounds.
5. 110 quantity prediction model construction and scene model
The gradient lifting model is changed in the model construction process, a two-step method modeling, a random intercept model and a random effect model are used for resisting the sparsity problem existing in the samples and the condition that each street and town sample is unbalanced, and finally a 110 meteorological disaster quantity prediction model is established.
Different from a hotline grid event quantity prediction model, the quantity prediction modeling is only carried out on two kinds of disaster scenes, namely strong wind and storm events. In the area range, the street town prediction is different from the grid hot line street town prediction, and the problem of data sparsity of the 110 weather disaster amount is more serious, so that the street town prediction is only carried out on the street town prediction. Over a time range, an hourly forecast is selected.
For scenes with a small number of samples and unbalanced street-town samples, the phenomenon of model overfitting can be remarkably reduced by using a two-step modeling method, a random intercept model and a random effect model. Aiming at the difference of the distribution rule of the 110 meteorological disasters and the distribution rule of grid hot-line events, different model parameters are used for fitting and training the model. The training parameters of the scene model are also modified according to such model training strategy changes. Through a mode of combining cross validation and manual work, the training parameters of each scene model are determined, and the training parameters of each scene model are as follows:
rainstorm scenes: the number of training rounds of the event occurrence probability model is 500 rounds, and the number of training rounds of the event occurrence number model is 500 rounds;
in a strong wind scene: the number of training rounds of the event occurrence probability model is 600 rounds, and the number of training rounds of the event occurrence number model is 500 rounds.
Third, the early warning model of the incident risk
According to statistical analysis on the occurrence number of the historical events and consideration on an actual use scene, an event early warning standard is formulated by combining the quantile of the occurrence number of the historical events and the absolute threshold of the occurrence number of the events. Now, the early warning and the like are in 5 grades, and the grading rule is as follows:
firstly, no early warning is given: less than 80% quantile or less than 9;
secondly, blue early warning: greater than or equal to 80% quantile and greater than or equal to 9;
③ yellow early warning: greater than or equal to 90% quantile and greater than or equal to 15;
orange early warning: greater than or equal to 95% quantile and greater than or equal to 25;
red early warning: 99% or more quantile and 40 or more.
For the 110 meteorological disaster, because the number of events is sparsely distributed and sparse, the early warning standard is modified as follows:
firstly, no early warning is given: less than 80% quantile or less than 9;
secondly, blue early warning: greater than or equal to 80% quantile and greater than or equal to 5;
③ yellow early warning: greater than or equal to 90% quantile and greater than or equal to 10;
orange early warning: greater than or equal to 95% quantile and greater than or equal to 15;
red early warning: 99% or more quantile and 25 or more.
According to this method, the number of predicted events may be increased or decreased from that in the ordinary state on weekdays by using the median of the historical 12-hour cumulative number of events as a comparison criterion for the events.
Fourth, weather influence index model
1. Grid event weather impact index
Because the LightGBM is a gradient lifting model based on a tree model, the information gain of each feature in different value range on the whole model can be obtained in the model training process. By performing aggregate summation calculation on information gains brought by different features and normalizing the contribution degrees of all the features, the percentage contribution degrees brought by different features can be calculated. From a model explanatory point of view, the percentage contribution quantifies the decision logic in the LightGBM model, i.e. which features will give the largest gain to the mode band under the current input conditions, and quantifies the importance of the features from the side.
Based on the grid event prediction model, the percentage contribution degree of the event to the event quantity prediction through calculation of each feature is calculated. Because the input of the model already comprises three meteorological characteristics of strong wind, precipitation and temperature, the percentage contribution values of the strong wind, precipitation and temperature to the event are obtained; and calculating the absolute deviation of the percentage contribution value of the current meteorological feature and the wind and daycare benchmark by taking the percentage contribution value of the three atmospheric meteorological features under the wind and daycare conditions (no rainfall, the wind speed is 2 levels, and the air temperature is 25 ℃), wherein the deviation is the influence index of the three types of meteorological features.
Through the meteorological influence index analysis to the grid scene model, public greenbelt and vehicle moving help seeking can be found to be influenced by precipitation more, overhead lines fall, greenbelt guardrails are influenced by strong wind more, and residential greening is influenced by temperature more. The street trees are closely related to strong wind, precipitation and air temperature.
2. Hotline event weather impact index
Similar to the calculation method of the grid model meteorological influence index, the meteorological influence index of the hotline model is based on the hotline event prediction model. And based on a deviation method, the influence of three weather features of strong wind, precipitation and air temperature on the total hot line quantity and the sub-events is calculated.
Through the meteorological influence index analysis of the hotline scene model, the drainage and pollution discharge management and road maintenance are influenced by the precipitation, the fault repair and green land greening are influenced by strong wind, and the taxi is influenced by the temperature.
Fifth, model verification mechanism
In order to simulate the scene of the model in real online operation, the prediction capability of the model is verified in a rolling type cross verification mode. The following is a rolling cross validation process:
now, let t be the time. The data of (t-1, t-2, …, t-12) is used as a training set and the event prediction model M is trainedt；
② use the model MtMaking a prediction result on the data set with the time t and recording the prediction result;
updating time t, namely t is t + 1;
and fourthly, repeating the step one until t is the latest time point, and counting the error rates of all the prediction results.
Sixthly, model updating optimization
The model adopts an automatic updating training strategy, the updating frequency is a daily level, namely the model is retrained by using latest updating data every day. If adjustments to the update frequency of the model are needed, modifications can be made to the cron syntax in this line of commands in the file model _ crontab.
If the model needs to be updated and replaced, the updated model with the same name can be directly placed into the corresponding folder. If the model needs to be trained manually in the deployed server, the following commands can be executed to manually update the model.
It is worth mentioning that the invention also provides a city operation management big data analysis and prediction system based on weather, so as to support the realization of the city operation management big data analysis and prediction method based on weather. The system can be stored in a computer, and when the computer runs the system, the steps of the weather-based urban operation management big data analysis and prediction method are executed.
Similarly, it should be noted that in the preceding description of embodiments of the application, various features are sometimes grouped together in a single embodiment, figure, or description thereof for the purpose of streamlining the disclosure aiding in the understanding of one or more of the embodiments. This method of disclosure, however, is not intended to require more features than are expressly recited in the claims. Indeed, the embodiments may be characterized as having less than all of the features of a single embodiment disclosed above.
Although the present application has been described with reference to the present specific embodiments, it will be recognized by those skilled in the art that the foregoing embodiments are merely illustrative of the present application and that various changes and substitutions of equivalents may be made without departing from the spirit of the application, and therefore, it is intended that all changes and modifications to the above-described embodiments that come within the spirit of the application fall within the scope of the claims of the application.
The techniques, shapes, and configurations not described in detail in the present invention are all known techniques.
Claims (10)
1. A weather-based urban operation management big data analysis and prediction method is characterized by comprising the following steps:
collecting a data source, and generating an event quantity prediction model training set;
based on the training set, adopting a two-step method to model the occurrence probability and the occurrence number of the events respectively;
in the event occurrence probability model, predicting whether each object will have an event within t time by adopting a gradient lifting model, and outputting the probability of whether each object has the event within t time;
in the event occurrence quantity model, predicting the quantity of events occurring in t time once each object has an event by adopting a gradient lifting model;
and multiplying the expected occurrence probability of the event by the expected number of the event once occurring to obtain the number value of the event occurrence, and establishing an event number prediction model.
2. The weather-based big data analysis and prediction method for city operation management as claimed in claim 1, wherein the data sources include weather automatic station data, grid point weather element forecast data, grid event data, hotline event data and 110 event weather disaster data.
3. The weather-based urban operation management big data analysis and prediction method according to claim 1, wherein the gradient boost model is a LightGBM-based gradient boost model, and a random intercept model and a random effect model are introduced on the basis of the gradient boost model to perform optimization and upgrade on the model.
4. The weather-based big data analysis and prediction method for urban operation management as claimed in claim 1, further comprising feature mining for time data, historical and live meteorological element data and historical and live urban operation data, and adding model training as input to the event quantity prediction model.
5. The weather-based big data analysis and prediction method for urban operation management as claimed in claim 4, wherein in the mining of the historical and live urban operation data characteristics, a partial autocorrelation coefficient is used to determine the time delay of the occurrence number of events.
6. A weather-based urban operation management big data analysis and prediction method as claimed in claim 4, wherein the mining of the data characteristics of historical and live meteorological elements takes into account the instantaneous and historical influence of each meteorological element on the number of events, and the influence of the historical meteorological elements on the number of events is used to calculate the time lag relationship between the number of events and different meteorological elements by using cross-correlation coefficients.
7. The weather-based big data analysis and prediction method for urban operation management, according to claim 1, wherein the method further comprises selecting an event scene closely related to weather influence by using the two-step modeling to establish a scene model while establishing the event quantity prediction model.
8. The weather-based urban operation management big data analysis and prediction method according to claim 1, further comprising establishing an event risk early warning model: and formulating an event early warning standard by combining the quantile of the occurrence number of the historical events and the absolute threshold of the occurrence number of the events.
9. The weather-based big data analysis and prediction method for urban operation management as claimed in claim 1, further comprising establishing a weather influence index model:
calculating the percentage contribution of each meteorological element characteristic to the event quantity prediction based on the event quantity prediction model;
and taking the percentage contribution value of the meteorological element characteristics under the condition of determining the meteorological element range as a reference, calculating the absolute deviation of the percentage contribution value of the current meteorological element characteristics and the reference, and taking the absolute deviation as the influence index of the meteorological element characteristics.
10. A city operation management big data analysis prediction system based on weather is characterized in that: the system is used for executing the weather-based urban operation management big data analysis and prediction method according to any claim 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110558687.0A CN113191568B (en) | 2021-05-21 | 2021-05-21 | Meteorological-based urban operation management big data analysis and prediction method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110558687.0A CN113191568B (en) | 2021-05-21 | 2021-05-21 | Meteorological-based urban operation management big data analysis and prediction method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113191568A true CN113191568A (en) | 2021-07-30 |
| CN113191568B CN113191568B (en) | 2024-02-02 |
Family
ID=76984763
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110558687.0A Active CN113191568B (en) | 2021-05-21 | 2021-05-21 | Meteorological-based urban operation management big data analysis and prediction method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113191568B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116862071A (en) * | 2023-07-18 | 2023-10-10 | 杭州博晟科技有限公司 | Urban operation index data prediction method based on mixing model |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105260603A (en) * | 2015-10-14 | 2016-01-20 | 成都信息工程大学 | Climatic event risk evaluation method and system |
| CN109359770A (en) * | 2018-10-11 | 2019-02-19 | 中国疾病预防控制中心环境与健康相关产品安全所 | A kind of model and method based on machine learning prediction heatstroke generation |
| CN109472283A (en) * | 2018-09-13 | 2019-03-15 | 中国科学院计算机网络信息中心 | A kind of hazardous weather event prediction method and apparatus based on Multiple Incremental regression tree model |
| CN109978222A (en) * | 2019-01-04 | 2019-07-05 | 国网山东省电力公司电力科学研究院 | A kind of wind-powered electricity generation climbing probability of happening prediction technique and system based on Bayesian network |
| CN110290120A (en) * | 2019-06-12 | 2019-09-27 | 西安邮电大学 | A kind of timing evolved network safe early warning method of cloud platform |
| CN110706823A (en) * | 2019-11-15 | 2020-01-17 | 广州地理研究所 | Method for predicting respiratory system disease morbidity based on lag analysis and LSTM |
| CN111612899A (en) * | 2019-02-25 | 2020-09-01 | 中国石油化工股份有限公司 | Geological modeling method for fractured-vuggy carbonate reservoir |
| CN111612233A (en) * | 2020-05-13 | 2020-09-01 | 云南电网有限责任公司电力科学研究院 | Method and device for obtaining importance scores of power distribution network line re-jump influence factors |
| CN111929748A (en) * | 2020-09-17 | 2020-11-13 | 南京浦蓝大气环境研究院有限公司 | Meteorological element forecasting method and system |
| CN112285807A (en) * | 2019-11-12 | 2021-01-29 | 北京嘉韵楷达气象科技有限公司 | Meteorological information prediction method and device |
| WO2021077729A1 (en) * | 2019-10-23 | 2021-04-29 | 国网电力科学研究院武汉南瑞有限责任公司 | Lightning prediction method |
-
2021
- 2021-05-21 CN CN202110558687.0A patent/CN113191568B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105260603A (en) * | 2015-10-14 | 2016-01-20 | 成都信息工程大学 | Climatic event risk evaluation method and system |
| CN109472283A (en) * | 2018-09-13 | 2019-03-15 | 中国科学院计算机网络信息中心 | A kind of hazardous weather event prediction method and apparatus based on Multiple Incremental regression tree model |
| CN109359770A (en) * | 2018-10-11 | 2019-02-19 | 中国疾病预防控制中心环境与健康相关产品安全所 | A kind of model and method based on machine learning prediction heatstroke generation |
| CN109978222A (en) * | 2019-01-04 | 2019-07-05 | 国网山东省电力公司电力科学研究院 | A kind of wind-powered electricity generation climbing probability of happening prediction technique and system based on Bayesian network |
| CN111612899A (en) * | 2019-02-25 | 2020-09-01 | 中国石油化工股份有限公司 | Geological modeling method for fractured-vuggy carbonate reservoir |
| CN110290120A (en) * | 2019-06-12 | 2019-09-27 | 西安邮电大学 | A kind of timing evolved network safe early warning method of cloud platform |
| WO2021077729A1 (en) * | 2019-10-23 | 2021-04-29 | 国网电力科学研究院武汉南瑞有限责任公司 | Lightning prediction method |
| CN112285807A (en) * | 2019-11-12 | 2021-01-29 | 北京嘉韵楷达气象科技有限公司 | Meteorological information prediction method and device |
| CN110706823A (en) * | 2019-11-15 | 2020-01-17 | 广州地理研究所 | Method for predicting respiratory system disease morbidity based on lag analysis and LSTM |
| CN111612233A (en) * | 2020-05-13 | 2020-09-01 | 云南电网有限责任公司电力科学研究院 | Method and device for obtaining importance scores of power distribution network line re-jump influence factors |
| CN111929748A (en) * | 2020-09-17 | 2020-11-13 | 南京浦蓝大气环境研究院有限公司 | Meteorological element forecasting method and system |
Non-Patent Citations (4)
| Title |
|---|
| LIU, TIANZE: "Short-Term Wind Speed and Wind Power Prediction Based on Meteorological Model Modification", 《9TH IEEE DATA DRIVEN CONTROL AND LEARNING SYSTEMS CONFERENCE (DDCLS)》 * |
| 付超: "基于沉积模拟的虚拟井建立与深水峡谷相控建模策略——以琼东南盆地中央峡谷陵水段为例", 《海相油气地质》, no. 2 * |
| 曾钦文;曾思亮;王辉;刘东瑶;: "龙川县森林火险等级预报方法的建立及应用", 广东气象, no. 04 * |
| 李根银: "风电场功率预测的研究进展及发展趋势", 《排灌机械工程学报》, no. 3 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116862071A (en) * | 2023-07-18 | 2023-10-10 | 杭州博晟科技有限公司 | Urban operation index data prediction method based on mixing model |
| CN116862071B (en) * | 2023-07-18 | 2024-02-13 | 杭州博晟科技有限公司 | Urban operation index data prediction method based on mixing model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113191568B (en) | 2024-02-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Dong et al. | Hourly energy consumption prediction of an office building based on ensemble learning and energy consumption pattern classification | |
| Luu et al. | Analyzing Vietnam's national disaster loss database for flood risk assessment using multiple linear regression-TOPSIS | |
| CN112070286B (en) | Precipitation forecast and early warning system for complex terrain river basin | |
| Tsakiris et al. | A system-based paradigm of drought analysis for operational management | |
| CN107045673B (en) | Public bicycle flow variation prediction method based on stack model fusion | |
| Lu et al. | Vulnerability analysis of transportation network under scenarios of sea level rise | |
| Serrano-Guerrero et al. | A new interval prediction methodology for short-term electric load forecasting based on pattern recognition | |
| CN106507315A (en) | A kind of urban traffic accident Forecasting Methodology and system based on network social intercourse media data | |
| CN111861027A (en) | Urban traffic flow prediction method based on deep learning fusion model | |
| CN107180270A (en) | Passenger flow forecasting and system | |
| CN113496314B (en) | Method for predicting road traffic flow by neural network model | |
| CN115983522B (en) | Rural habitat quality assessment and prediction method | |
| CN111815098A (en) | Traffic information processing method and device based on extreme weather, storage medium and electronic equipment | |
| Taylor et al. | Dynamic modeling of the effects of vegetation management on weather-related power outages | |
| CN117494034A (en) | Air quality prediction method based on traffic congestion index and multi-source data fusion | |
| Hou et al. | Spatial distribution assessment of power outage under typhoon disasters | |
| CN114414491B (en) | Grass ecology dynamic monitoring and analysis system | |
| CN113191568B (en) | Meteorological-based urban operation management big data analysis and prediction method and system | |
| Soldatenko et al. | Managing climate risks associated with socio-economic development of the Russian Arctic | |
| CN115829150A (en) | Accumulated water prediction system | |
| CN112578479B (en) | Urban inland inundation warning device and method based on rainfall forecast | |
| KR102308748B1 (en) | Heavy Rain Disaster Vulnerability Analysis Advanced System | |
| Rahman et al. | Understanding network wide hurricane evacuation traffic pattern from large-scale traffic detector data | |
| Chao | Machine learning-based intelligent weather modification forecast in smart city potential area | |
| CN113112794A (en) | Traffic accident occurrence rate prediction method based on space-time meteorological grid |
Legal Events
| 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |