CN104751242A

CN104751242A - Method and device for predicting air quality index

Info

Publication number: CN104751242A
Application number: CN201510142623.7A
Authority: CN
Inventors: 张震
Original assignee: Beijing Qihoo Technology Co Ltd; Qizhi Software Beijing Co Ltd
Current assignee: Beijing Qihoo Technology Co Ltd; Qizhi Software Beijing Co Ltd
Priority date: 2015-03-27
Filing date: 2015-03-27
Publication date: 2015-07-01
Also published as: WO2016155372A1

Abstract

The invention provides a method and a device for predicting air quality index. The method includes the following steps: acquiring a corresponding prediction model according to a to-be-predicted geographical position and current time; acquiring meteorological data of the to-be-predicted geographical position; according to the meteorological data and on the basis of the prediction model, performing operation processing, and determining the air quality index of the to-be-predicted geographical position within to-be-predicted time. By the method and the device, prediction of future air quality index can be completed only through meteorological information which is little and easy to acquire, so that realization difficulty and access threshold of AQI prediction are lowered. In addition, corresponding prediction models are built aiming at each geographical position in different prediction time intervals, so that weather characteristics of local areas can be reflected better, and a powerful guarantee is provided for determining AQI high in accuracy.

Description

The method and apparatus of prediction air quality index

Technical field

The present invention relates to technical field of industrial automatic control, specifically, the present invention relates to a kind of method and apparatus predicting air quality index.

Background technology

AQI (Air Quality Index, air quality index) is the zero dimension index of quantitative description Air Quality.Air quality subindex is also specify for individual event pollutant.The major pollutants participating in Air Quality Evaluation are fine particle, pellet, sulphuric dioxide, nitrogen dioxide, ozone, carbon monoxide etc. six.

In prior art, be the AQI prediction mode based on SPRINTARS Atmospheric General Circulation Model (AGCM) to the major prognostic mode of AQI.SPRINTARS (Spectral Radiation-Transport Model forAerosol Species) is with global scale simulated atmosphere suspended particulate substance, i.e. atmospheric aerosol, the impact that weather system is caused and State of Air pollution and the numerical model developed.Based on the coupled air-sea model MIROC that SPRINTARS develops by Atmosphere and Ocean research institute of Tokyo University (weather system research system), Japanese National Institute for Environmental Studies and Japanese ocean research and development organization (earth environment variation field), to the self-assembling formation be present in troposphere and the artificial main atmospheric aerosol formed, comprise black carbon, organism, sulfate, soil particle and sea salt particle, study, these particles are also classified as PM10 and PM2.5.SPRINTARS can calculate the moving process of atmospheric aerosol, comprises the process of generation, convection current, diffusion, wet deposition, dry deposition, gravity settling; Also can calculate the direct effect that atmospheric aerosol produces, if atmospheric aerosol is to sunlight and ultrared scattering and absorption, with indirectly-acting, if atmospheric aerosol is as the function of cloud condensation nucl and ice-nucleus.

SPRINTARS constructs a comparatively perfect, complicated Atmospheric General Circulation Model (AGCM), calculates the diffusion tendency of global pollution composition granule with this.Which has the plains region that season is clearly demarcated and well predicts the outcome, but for areas such as basin with a varied topography, hills, plateau, mountain regions, may be complicated due to weather, steam is prosperous, thus causes being difficult to provide predicting accurately.Such as, SPRINTARS cannot distinguish haze and the water smoke of Southwest China well, Chongqing in China area as foggy for few wind often provides the prediction of lasting severe contamination, but this urban air-quality actual is in excellent condition always, the AQI forecast based on SPRINTARS is made not have reference significance in Southwest China.

Therefore, in prior art, there is following problem in the AQI prediction mode based on SPRINTARS Atmospheric General Circulation Model (AGCM):

1) SPRINTARS Atmospheric General Circulation Model (AGCM) mainly considers the overall factor of general circulation, analyzes the diffusion mode of pollutant, and be difficult to detailed differentiation for the concrete climatic condition in certain concrete city from the dimension of general circulation.Due to the concrete climatic condition in same city, can because of Various Seasonal, different time sections, even human factor and change to some extent, such as, before and after somewhere is newly-built chemical plant, the discharge of pollutant and accumulation are certain to change to some extent, therefore, the AQI prediction mode based on SPRINTARS Atmospheric General Circulation Model (AGCM) is lower for the concrete forecasting accuracy in concrete city.

2) the data acquisition amount of SPRINTARS Atmospheric General Circulation Model (AGCM) is very huge, at least need to collect a large amount of pollution source specifying informations and satellite meteorolo-gy information, simultaneously, the calculated amount of SPRINTARS Atmospheric General Circulation Model (AGCM) is also very huge, needing high performance hardware device to support, there is very high technology entry threshold according to collection capacity and huge calculated amount for common entity and individual in googol.

Summary of the invention

For overcoming above-mentioned technical matters or solving the problems of the technologies described above at least in part, the following technical scheme of special proposition:

Embodiments of the invention propose a kind of method predicting air quality index, comprising:

According to geographic position to be predicted and current time, obtain corresponding forecast model;

Obtain the weather data in described geographic position to be predicted;

According to described weather data information, carry out calculation process based on described forecast model, determine the air quality index of geographic position to be predicted in the time to be predicted.

Embodiments of the invention also proposed a kind of device predicting air quality index, comprising:

Model acquisition module, for according to geographic position to be predicted and current time, obtains corresponding forecast model;

Weather data acquisition module, for obtaining the weather data in described geographic position to be predicted;

Index determination module, for according to described weather data information, carries out calculation process based on described forecast model, determines the air quality index of geographic position to be predicted in the time to be predicted.

In embodiments of the invention, can all set up corresponding forecast model for each geographic position under current time; When performing prediction, first, obtaining the weather data in geographic position to be predicted, obtaining the public informations such as the weather forecast that this weather data can be issued from meteorological department; Subsequently, according to weather data, carry out calculation process based on corresponding forecast model, determine the air quality index of geographic position to be predicted in the time to be predicted.The present invention is relative to the advantage of prior art: first, the present invention only needs weather information that is less and that be easy to obtain can complete the prediction of following air quality index, because the solution of the present invention does not need to obtain the complicated general circulation parameter in prior art needed for Atmospheric General Circulation Model (AGCM), and only need the basic weather data being easy to acquisition, degree/day as every in history, humidity, air pressure, wind-force, wind direction, AQI numerical value etc., prediction can be completed, therefore the Public meteorology information that any entity or individual can utilize meteorological department to issue carries out AQI prediction based on technical solutions according to the invention, simultaneously, based on forecast model, calculated amount as the forecast model based on artificial neural network is less, without the need to the high cost investment of high-performance hardware equipment, what reduce prediction realizes difficulty and entry threshold.In addition, under different predicted time intervals, corresponding forecast model is all set up owing to the present invention is directed to each geographic position, thus the climatic characteristic of this area can be reflected better, better Regional suitability is had, for determining that the AQI that degree of accuracy is higher provides strong guarantee relative to the Atmospheric General Circulation Model (AGCM) of prior art.

The aspect that the present invention adds and advantage will part provide in the following description, and these will become obvious from the following description, or be recognized by practice of the present invention.

Accompanying drawing explanation

The present invention above-mentioned and/or additional aspect and advantage will become obvious and easy understand from the following description of the accompanying drawings of embodiments, wherein:

Fig. 1 is the schematic flow sheet of an embodiment of the method for prediction air quality index in the present invention;

Fig. 2 is the schematic flow sheet of the method preferred embodiment predicting air quality index in the present invention;

Fig. 3 is the schematic flow sheet of the prediction air quality index specific embodiment in the present invention;

Fig. 4 is the structural representation of the device embodiment predicting air quality index in the present invention;

Fig. 5 is the structural representation of the device preferred embodiment predicting air quality index in the present invention.

Embodiment

Be described below in detail embodiments of the invention, the example of described embodiment is shown in the drawings, and wherein same or similar label represents same or similar element or has element that is identical or similar functions from start to finish.Being exemplary below by the embodiment be described with reference to the drawings, only for explaining the present invention, and can not limitation of the present invention being interpreted as.

Those skilled in the art of the present technique are appreciated that unless expressly stated, and singulative used herein " ", " one ", " described " and " being somebody's turn to do " also can comprise plural form.Should be further understood that, the wording used in instructions of the present invention " comprises " and refers to there is described feature, integer, step, operation, element and/or assembly, but does not get rid of and exist or add other features one or more, integer, step, operation, element, assembly and/or their group.Should be appreciated that, when we claim element to be " connected " or " coupling " to another element time, it can be directly connected or coupled to other elements, or also can there is intermediary element.In addition, " connection " used herein or " coupling " can comprise wireless connections or wirelessly to couple.Wording "and/or" used herein comprises one or more whole or arbitrary unit listing item be associated and all combinations.

Those skilled in the art of the present technique are appreciated that unless otherwise defined, and all terms used herein (comprising technical term and scientific terminology), have the meaning identical with the general understanding of the those of ordinary skill in field belonging to the present invention.It should also be understood that, those terms defined in such as general dictionary, should be understood to that there is the meaning consistent with the meaning in the context of prior art, unless and by specific definitions as here, otherwise can not explain by idealized or too formal implication.

Within a context alleged " computer equipment ", also referred to as " computer ", refer to the intelligent electronic device that can be performed the predetermined process such as numerical evaluation and/or logical calculated process by operation preset program or instruction, it can comprise processor and storer, the survival instruction that prestores in memory is performed to perform predetermined process process by processor, or perform predetermined process process by the hardware such as ASIC, FPGA, DSP, or combined by said two devices and realize.Computer equipment includes but not limited to server, PC, notebook computer, panel computer, smart mobile phone etc.

Described computer equipment comprises subscriber equipment and the network equipment.Wherein, described subscriber equipment includes but not limited to computer, smart mobile phone, PDA etc.; The described network equipment includes but not limited to the server group that single network server, multiple webserver form or the cloud be made up of a large amount of computing machine or the webserver based on cloud computing (Cloud Computing), wherein, cloud computing is the one of Distributed Calculation, the super virtual machine be made up of a group loosely-coupled computing machine collection.Wherein, described computer equipment isolated operation can realize the present invention, also accessible network by realizing the present invention with the interactive operation of other computer equipments in network.Wherein, the network residing for described computer equipment includes but not limited to internet, wide area network, Metropolitan Area Network (MAN), LAN (Local Area Network), VPN etc.

Described artificial neural network is that a kind of application class is similar to the structure that cerebral nerve cynapse connects and carries out the mathematical model of information processing, engineering and academia also often direct referred to as neural network or neural network.Artificial neural network is a kind of operational model, by a large amount of nodes, or claim neuron, between be interconnected to constitute.A kind of specific output function of each node on behalf, is called excitation function (activation function).Every two internodal connections all represent one and are referred to as weight for the weighted value by this connection signal, are equivalent to the memory of artificial neural network.The output of artificial neural network is then different from the difference of weight and excitation function between the connected mode of artificial neural network, node.And artificial neural network self is all approaching certain algorithm of nature or function usually, also may be the expression to a kind of logic strategy.

Fig. 1 is the schematic flow sheet of an embodiment of the method for prediction air quality index in the present invention.

This method is performed by computer equipment; In step s 110, computer equipment, according to geographic position to be predicted and current time, obtains corresponding forecast model; In the step s 120, the weather data in geographic position to be predicted is obtained; In step s 130, which, according to weather data, and carry out calculation process based on forecast model, determine the air quality index of geographic position to be predicted in the time to be predicted.

The forecast model that in this programme, forecast model can be " based on artificial neural network ", carries out writing of specific embodiment by using the forecast model based on artificial neural network as an implementation below.

In step s 110, computer equipment, according to geographic position to be predicted and current time, obtains corresponding forecast model.

Particularly, the process setting up forecast model can be to be set up in real time and sets up in advance.

When setting up the process based on the forecast model of artificial neural network for setting up in real time, step S110 (with reference to Fig. 1) comprises step S111 (not shown), step S112 (not shown), step S113 (not shown) and step S114 (not shown); In step S111, determine that the predicted time at current time place is interval; Determine in step S112 and the meteorological index that geographic position to be predicted and described predicted time interval match; In step S113, according to geographic position to be predicted and described predicted time interval, obtain the meteorological sample data of history under meteorological index; In step S114, carry out machine learning according to the meteorological sample data of history, determine the forecast model corresponding with geographic position to be predicted and current time; Wherein, step S111, step S112, step S113 are similar to the mode setting up forecast model in advance to the mode setting up forecast model in real time in step S114, in this reference with the embodiment of following embodiment.

Preferably (with reference to Fig. 1), the method also comprises: judge whether geographic position to be predicted exists sample filtering rule in predicted time interval; If exist, in step S113, according to geographic position to be predicted and predicted time interval, based on sample filtering rule, obtain the meteorological sample data of history under meteorological index.

Such as, the policies such as large area shut-down and vehicle restricted driving are implemented in early November, 2014 in Beijing area, cause the discharge capacity of pollutant significantly to decline, be then preset at when choosing the meteorological sample data of history during Beijing area November, the meteorological sample data in-10 days on the 1st November need be filtered.

When setting up the process based on the forecast model of artificial neural network for setting up in advance, step S110 (with reference to Fig. 1) comprises step S115 (not shown) and step S116 (not shown); In step sl 15, determine that the predicted time at current time place is interval; In step S116, according to geographic position to be predicted and predicted time interval, in forecasting model database, carry out matching inquiry, obtain and the geographic position to be predicted and interval corresponding forecast model of predicted time.

Preferably, set up each geographic position corresponding forecast model in different predicted time interval in advance, forecast model is stored in forecasting model database, and the corresponding relation preserved between geographic position, predicted time interval and forecast model, as the corresponding relation between three is stored in model corresponding lists, for inquiry.

As shown in Figure 2, in a preferred embodiment, according to geographic position to be predicted and predicted time interval, matching inquiry is carried out in forecasting model database, before obtaining the forecast model corresponding with geographic position to be predicted and current time, also comprise step S240, step S250, step S260 and step S270; In step S240, determine and the meteorological index that geographic position to be predicted and predicted time interval match; In step s 250, according to geographic position to be predicted and predicted time interval, obtain the meteorological sample data of history under meteorological index; In step S260, carry out machine learning according to the meteorological sample data of history, determine and geographic position to be predicted and the interval corresponding forecast model of predicted time; In step S270, be saved to forecasting model database by with geographic position to be predicted and the interval corresponding forecast model of predicted time.

Preferably, when forecasting model database has existed corresponding forecast model interval with geographic position to be predicted and predicted time, to replace the corresponding forecast model in previously already present and geographic position to be predicted and predicted time interval with geographic position to be predicted and predicted time interval corresponding forecast model with up-to-date.

In step S240, determine and the meteorological index that geographic position to be predicted and predicted time interval match;

Wherein, meteorological index includes but not limited to:

Temperature index, can comprise max. daily temperature, Daily minimum temperature;

Humidity index, can comprise per day humidity;

Wind-force index, can comprise a day maximum wind power, day cardinal wind maximum wind power, day cardinal wind average wind;

Wind direction index, can comprise day maximum wind direction;

Air pressure index, can comprise per day air pressure;

Rainfall amount index, can comprise per day rainfall amount, day maximum rainfall;

Dew point index, can comprise per day dew point;

Air quality index index, i.e. AQI.

Such as, geographic position to be predicted is Beijing area, predicted time interval is " the 1-3 month ", carry out match query in table 1 below, determine the meteorological index matched with " Beijing area " and predicted time interval " the 1-3 month " comprise max. daily temperature, Daily minimum temperature, per day humidity, day cardinal wind maximum wind power and AQI.

Table 1:

In step s 250, according to geographic position to be predicted and predicted time interval, obtain the meteorological sample data of history under meteorological index.

Such as, connect example, geographic position to be predicted is Beijing area, predicted time interval is " the 1-3 month ", and the meteorological index matched comprise max. daily temperature, Daily minimum temperature, per day humidity, day cardinal wind maximum wind power and AQI, therefore obtain every max. daily temperature of the history of the Beijing area 1-3 month in 2015 and the 2012-2014 1-3 month, every Daily minimum temperature, every day medial humidity, every day cardinal wind maximum wind power and every day AQI as the meteorological sample data of history; Wherein, the factors such as season need be considered when obtaining history meteorology sample data, the standard obtaining the meteorological sample data of history can comprise: recent meteorological sample data, as the meteorological sample data of nearly 3 months, and the meteorological sample data of history same period, as the meteorological sample data of history same period of nearly 3 years.

Preferably (with reference to Fig. 2), the method also comprises step S280 (not shown); In step S280, judge whether geographic position to be predicted exists sample filtering rule in predicted time interval; If exist, in step s 250, according to geographic position to be predicted and predicted time interval, based on sample filtering rule, obtain the meteorological sample data of history under meteorological index.

In step S260, carry out machine learning according to the meteorological sample data of history, determine and geographic position to be predicted and the interval corresponding forecast model of predicted time.

Particularly, step S260 (with reference to Fig. 2) comprises step S261 (not shown) and step S262 (not shown); In step S261, according to the meteorological sample data of history, carry out machine learning based on artificial neural network, determine weight between the node with geographic position to be predicted and the interval corresponding artificial neural network of predicted time; In step S262, according to weight between the node of artificial neural network, set up corresponding forecast model.

Particularly, using meteorological for history sample data as the input data of artificial neural network, carry out machine learning based on artificial neural network, determine weight between the node that machine learning terminates rear artificial neural network; Subsequently, according to weight between the node of artificial neural network, corresponding forecast model is set up.

Wherein, step S261 (not shown) comprises step S2611 (not shown), step S2612 (not shown) and step S2613 (not shown): in step S2611, according to the meteorological sample data of history, carry out machine learning based on artificial neural network, determine the air quality index learning outcome of artificial neural network; In step S2612, calculate the error amount of the history air quality index in air quality index learning outcome and the meteorological sample data of history; In step S2613, when error amount is less than predictive error threshold value, weight between the node of extraction artificial neural network.

Particularly, using meteorological for history sample data as the input data of artificial neural network, carry out machine learning based on artificial neural network, obtain the output data of artificial neural network, i.e. air quality index learning outcome; Then, calculate and export data and the error amount of air quality index in input data, be i.e. the error amount of air quality index learning outcome and the air quality index in history meteorology sample data; When error amount is less than predictive error threshold value, determine that machine learning terminates, weight between the node of extraction artificial neural network.

In step S270, be saved to forecasting model database by with geographic position to be predicted and the interval corresponding forecast model of predicted time.

Produce along with constantly there being new meteorological sample data, every the scheduled update time interval, an execution capable of circulation machine learning, weight between the node extracting new artificial neural network for each geographic position, and be stored in renewal forecasting model database, specifically can be extract the artificial neural network after machine learning again for each geographic position node between weight, specifically can be the data record deleting weight between original node in forecasting model database, write up-to-date data.

In the process of concrete machine learning, according to the actual conditions of diverse geographic location in different predicted time interval, choose different meteorological index to select the meteorological sample data of history for machine learning.Such as: mountain area local wind is to indefinite, and wind-force is very little, index of just can keeping watch is removed, to prevent from causing interference to machine learning; And by wind impact in plains region is very large, wind index can be used to choose sample data.Therefore, in preferred embodiment, artificial neural network carries out machine learning according to the meteorological sample data of the history of diverse geographic location in different predicted time interval and obtains, namely each geographic position have a set of unique people's neural network node between weight.There is a set of artificial neural network being suitable for this area in each geographic position, can reflect the climatic characteristic of this area better, for determining that the AQI that degree of accuracy is higher provides strong guarantee.

With reference to Fig. 1, in the step s 120, obtain the weather data in geographic position to be predicted.

Particularly, step S120 comprises step S121 (not shown), step S122 (not shown) and step S123 (not shown); In step S121, determine that the predicted time at current time place is interval; In step S122, determine and the meteorological index that geographic position to be predicted and described predicted time interval match; In step S123, according to geographic position to be predicted, determine the weather data under meteorological index.

Particularly, first, according to upper table 1, determine that predicted time corresponding to current time is interval, then, determine and meteorological index that geographic position to be predicted and predicted time interval match; Subsequently, according to geographic position to be predicted, determine the weather data under meteorological index.

When the time to be predicted is tomorrow, weather data comprises:

The weather data of today and the prediction weather data of tomorrow; Or

The true weather data of front predetermined number of days, the weather data of today and the prediction weather data of tomorrow;

Wherein, the weather data of today comprises:

The true weather data of today;

If the true weather data of today is imperfect, then can comprise the true weather data of today and the prediction weather data of today.

Wherein, predict that weather data includes but not limited to:

Temperature information; Humidity information; Wind direction information; Wind-force information; Pressure information; Rainfall amount information; Dew point information.

Such as, the time to be predicted is tomorrow; Then weather data can comprise real air quality index today, temperature information, humidity information, wind direction information, wind-force information, pressure information, rainfall amount information, dew point information etc., and the temperature information of tomorrow prediction, humidity information, wind direction information, wind-force information, pressure information, rainfall amount information, dew point information etc.

In one example, geographic position to be predicted is " Beijing area ", today is on January 14th, 2015, time to be predicted is January 15 2015 tomorrow, the weather forecast of then issuing by weather bureau obtains the true weather data such as real maximum temperature, minimum temperature, relative humidity, maximum wind velocity, maximum wind velocity wind direction, AQI on the same day on January 14th, 2015, and can obtain the prediction such as maximum temperature, minimum temperature, relative humidity, maximum wind velocity, the maximum wind velocity wind direction weather data of prediction on January 15th, 2015.

In step s 130, which, according to weather data, carry out calculation process based on forecast model, determine the air quality index of geographic position to be predicted in the time to be predicted.

Such as, as shown in table 2, on January 14th, 2015 is the same day true weather data of Beijing area from actual observation, and on January 15th, 2015 and later date are the prediction weather datas from weather forecast.

Table 2

Time	AQI	Maximum temperature	Minimum temperature	Relative humidity	Maximum wind velocity	Maximum wind velocity wind direction
							2015/1/14	267	3	-6	69	8	340
2015/1/15		6	-5	79	16	155
							2015/1/16		4	-7	41	40	329
2015/1/17		6	-4	33	16	223
							2015/1/18		9	-5	36	24	283
2015/1/19		8	-6	43	16	113
							2015/1/20		9	-4	51	16	330
2015/1/21		7	-6	37	16	258
							2015/1/22		8	-4	39	16	268
2015/1/23		7	-5	31	16	290

Geographic position to be predicted is " Beijing area ", today is on January 14th, 2015, weather data is as above shown in table 2, according to true weather data and the prediction weather data of the tomorrow on January 15th, 2015 of today on January 14th, 2015, carry out calculation process based on the forecast model matched with this geographic position and current time, determine the air quality index of Beijing area in tomorrow on the 15th January in 2015.

As shown in Figure 3, in an embody rule scene, build the stage in advance at forecast model, geographic position to be predicted is city A, performs the forecast model based on artificial neural network that machine learning obtains city A in advance; First inquiry obtains the historical climate sample data of city A, judge whether there is sample filtering rule for this city A, if no, then obtain according to concrete predicted time interval the historical climate sample data that city A gives tacit consent to, and as the input data of artificial neural network; If have, then and sample filtering rule interval according to concrete predicted time, obtains the historical climate sample data after the filtration of city A, and as the input data of artificial neural network; Subsequently, this historical climate sample data carries out machine learning training at artificial neural network, obtains the output data of artificial neural network, and exporting data is air quality index learning outcome; Then, compare the history air quality index in the input data of artificial neural network and the error amount of the air quality index learning outcome of output data, if when error amount is greater than predetermined threshold, continue to perform machine learning; If when error amount is less than predetermined threshold, machine learning is complete, exports and stores city A weight between the node in this predicted time interval, thus setting up the forecast model of city A in this predicted time interval.At forecast period, first the weather data of city A is read, the forecast model of the city A matched is loaded according to current time, and using the input data of the weather data of city A as artificial neural network, carry out calculation process by artificial neural network, obtain the output data of artificial neural network, be prediction AQI value, utilize the AQI of up-to-date acquisition subsequently, circulation performs this calculation step.Such as, as shown in table 2, first round computation process, according to today January 14 in 2015 real maximum temperature, minimum temperature, relative humidity, maximum wind velocity, maximum wind velocity wind direction, the prediction of AQI and 2,015 15, tomorrow of on January maximum temperature, minimum temperature, relative humidity, maximum wind velocity, maximum wind velocity wind direction, by carrying out the prediction AQI that day tomorrow January 15 in 2015 is determined in computing based on the forecast model of artificial neural network; Second takes turns computation process, utilize maximum temperature, minimum temperature, relative humidity, maximum wind velocity, the maximum wind velocity wind direction of the prediction in January 15 in 2015 and on January 16th, 2015, and predict the AQI in the 15 days January in 2015 obtained, the prediction AQI on January 16th, 2015 is determined by computing; Carry out computing by that analogy, can determine that the AQI of 15 ~ 22 days predicts the outcome.

Fig. 4 is the structural representation of the device embodiment predicting air quality index in the present invention.

Described device can be in computer equipment, also can be the functional module in computer equipment; This device comprises model acquisition module 410, weather data acquisition module 420 and index determination module 430.

First, model acquisition module 410, according to geographic position to be predicted and current time, obtains corresponding forecast model; Subsequently, weather data acquisition module 420 obtains the weather data in geographic position to be predicted; Then, index determination module 430 according to weather data, and carries out calculation process based on forecast model, determines the air quality index of geographic position to be predicted in the time to be predicted.

First, model acquisition module 410, according to geographic position to be predicted and current time, obtains corresponding forecast model.

When setting up the process based on the forecast model of artificial neural network for setting up in real time, model acquisition module 410 (with reference to Fig. 4) comprises predicted time determining unit (not shown), the first index determining unit (not shown), the first sample acquisition unit (not shown) and the first model determining unit (not shown); First, the predicted time at predicted time determining unit determination current time place is interval; First index determining unit is determined and the meteorological index that geographic position to be predicted and described predicted time interval match; Subsequently, the first sample acquisition unit, according to geographic position to be predicted and described predicted time interval, obtains the meteorological sample data of history under meteorological index; Then, the first model determining unit carries out machine learning according to the meteorological sample data of history, determines the forecast model corresponding with geographic position to be predicted and current time; Wherein, predicted time determining unit, the first index determining unit, the first sample acquisition unit and the mode setting up forecast model in real time performed by the first model determining unit are similar to the mode setting up forecast model in advance, in this reference with the embodiment of following embodiment.

Preferably (with reference to Fig. 4), this device also comprises the first judge module (not shown), and the first judge module judges whether geographic position to be predicted exists sample filtering rule in predicted time interval; If exist, the first sample acquisition unit, according to geographic position to be predicted and predicted time interval, based on sample filtering rule, obtains the meteorological sample data of history under meteorological index.

When setting up the process based on the forecast model of artificial neural network for setting up in advance, model acquisition module 410 (with reference to Fig. 4) comprises predicted time determining unit (not shown) and the second model acquiring unit (not shown); The predicted time at predicted time determining unit determination current time place is interval; Subsequently, the second model acquiring unit, according to geographic position to be predicted and predicted time interval, carries out matching inquiry in forecasting model database, obtains and the geographic position to be predicted and interval corresponding forecast model of predicted time.

Preferably, set up the forecast model of each geographic position accordingly based on artificial neural network in different predicted time interval in advance, forecast model is stored in forecasting model database, and the corresponding relation preserved between geographic position, predicted time interval and forecast model, as the corresponding relation between three is stored in model corresponding lists, for inquiry.

As shown in Figure 5, in a preferred embodiment, predict that the device of air quality index also comprises the second index determination module 540, second sample acquisition module 550, the pre-modeling block 560 of model and memory module 570; Second index determination module 540 is determined and the meteorological index that geographic position to be predicted and predicted time interval match; Second sample acquisition module 550, according to geographic position to be predicted and predicted time interval, obtains the meteorological sample data of history under meteorological index; Subsequently, the pre-modeling block 560 of model carries out machine learning according to the meteorological sample data of history, determines and geographic position to be predicted and the interval corresponding forecast model of predicted time; Memory module 570 is saved to forecasting model database by with geographic position to be predicted and the interval corresponding forecast model of predicted time.

Preferably, this device also comprises update module, when having there is the forecast model corresponding with geographic position to be predicted and predicted time interval in forecasting model database, in update module, to replace the corresponding forecast model in previously already present and geographic position to be predicted and predicted time interval with geographic position to be predicted and predicted time interval corresponding forecast model with up-to-date.

Second index determination module 540 is determined and the meteorological index that geographic position to be predicted and predicted time interval match;

Wherein, meteorological index includes but not limited to:

Humidity index, can comprise per day humidity;

Wind direction index, can comprise day maximum wind direction;

Air pressure index, can comprise per day air pressure;

Dew point index, can comprise per day dew point;

Air quality index index, i.e. AQI.

Such as, geographic position to be predicted is Beijing area, predicted time interval is " the 1-3 month ", in upper table 1, carry out match query, determine the meteorological index matched with " Beijing area " and predicted time interval " the 1-3 month " comprise max. daily temperature, Daily minimum temperature, per day humidity, day cardinal wind maximum wind power and AQI.

Subsequently, the second sample acquisition module 550, according to geographic position to be predicted and predicted time interval, obtains the meteorological sample data of history under meteorological index.

Preferably (with reference to Fig. 5), this device also comprises the second judge module (not shown); Judge module judges whether geographic position to be predicted exists sample filtering rule in predicted time interval; If exist, then the second sample acquisition module 550 is according to geographic position to be predicted and predicted time interval, based on sample filtering rule, obtains the meteorological sample data of history under meteorological index.

Then, the pre-modeling block 560 of model carries out machine learning according to the meteorological sample data of history, determines and geographic position to be predicted and the interval corresponding forecast model of predicted time.

Particularly, the pre-modeling block of model 560 (with reference to Fig. 5) comprises weight determining unit (not shown) and unit (not shown) set up by model; First, weight determining unit, according to the meteorological sample data of history, carries out machine learning based on artificial neural network, determines weight between the node with geographic position to be predicted and the interval corresponding artificial neural network of predicted time; Subsequently, model sets up unit according to weight between the node of artificial neural network, sets up corresponding forecast model.

Wherein, weight determining unit, according to the meteorological sample data of history, carries out machine learning based on artificial neural network, determines the air quality index learning outcome of artificial neural network; Subsequently, the error amount of the air quality index in air quality index learning outcome and the meteorological sample data of history is calculated; Then, when error amount is less than predictive error threshold value, weight between the node of extraction artificial neural network.

Memory module 570 is saved to forecasting model database by with geographic position to be predicted and the interval corresponding forecast model of predicted time.

Preferably, when forecasting model database has existed corresponding forecast model interval with geographic position to be predicted and predicted time, update module to replace the corresponding forecast model in previously already present and geographic position to be predicted and predicted time interval with geographic position to be predicted and predicted time interval corresponding forecast model with up-to-date.。

With reference to Fig. 4, weather data acquisition module 420 obtains the weather data in geographic position to be predicted.

Particularly, weather data acquisition module 420 comprises the 3rd index determining unit (not shown) and data determination unit (not shown); 3rd index determining unit is determined and the meteorological index that geographic position to be predicted and current time match; Data determination unit, according to geographic position to be predicted, determines the weather data under meteorological index.

When the time to be predicted is tomorrow, weather data comprises:

The weather data of today and the prediction weather data of tomorrow; Or

Wherein, the weather data of today comprises:

The true weather data of today;

Wherein, predict that weather data includes but not limited to:

Index determination module 430, according to weather data, carries out calculation process based on forecast model, determines the air quality index of geographic position to be predicted in the time to be predicted.

Such as, as above shown in table 2, on January 14th, 2015 be today Beijing area from the true weather data of actual observation, on January 15th, 2015 and later date are the prediction weather datas from weather forecast.

Those skilled in the art of the present technique are appreciated that the one or more equipment that the present invention includes and relate to for performing in operation described in the application.These equipment for required object and specialized designs and manufacture, or also can comprise the known device in multi-purpose computer.These equipment have storage computer program within it, and these computer programs optionally activate or reconstruct.Such computer program can be stored in equipment (such as, computing machine) in computer-readable recording medium or be stored in and be suitable for store electrons instruction and be coupled in the medium of any type of bus respectively, described computer-readable medium includes but not limited to that the dish of any type (comprises floppy disk, hard disk, CD, CD-ROM, and magneto-optic disk), ROM (Read-Only Memory, ROM (read-only memory)), RAM (Random Access Memory, storer immediately), EPROM (Erasable Programmable Read-Only Memory, Erarable Programmable Read only Memory), EEPROM (Electrically Erasable ProgrammableRead-Only Memory, EEPROM (Electrically Erasable Programmable Read Only Memo)), flash memory, magnetic card or light card.Namely, computer-readable recording medium comprises and being stored or any medium of transmission information with the form that can read by equipment (such as, computing machine).

Those skilled in the art of the present technique are appreciated that the combination that can realize the frame in each frame in these structural drawing and/or block diagram and/or flow graph and these structural drawing and/or block diagram and/or flow graph with computer program instructions.Those skilled in the art of the present technique are appreciated that, the processor that these computer program instructions can be supplied to multi-purpose computer, special purpose computer or other programmable data disposal routes realizes, thus is performed the scheme of specifying in the frame of structural drawing disclosed by the invention and/or block diagram and/or flow graph or multiple frame by the processor of computing machine or other programmable data disposal routes.

Those skilled in the art of the present technique are appreciated that various operations, method, the step in flow process, measure, the scheme discussed in the present invention can be replaced, changes, combines or delete.Further, there is various operations, method, other steps in flow process, measure, the scheme discussed in the present invention also can be replaced, change, reset, decompose, combine or delete.Further, of the prior art have also can be replaced with the step in operation various disclosed in the present invention, method, flow process, measure, scheme, changed, reset, decomposed, combined or deleted.

The above is only some embodiments of the present invention; it should be pointed out that for those skilled in the art, under the premise without departing from the principles of the invention; can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.

The invention discloses A1, a kind of method predicting air quality index, comprising:

Obtain the weather data in described geographic position to be predicted; And

According to described weather data, and carry out calculation process based on described forecast model, determine the air quality index of geographic position to be predicted in the time to be predicted.

The method of A2, prediction air quality index according to claim 1, is characterized in that, described according to geographic position to be predicted and current time, the step obtaining corresponding forecast model comprises further:

Determine that the predicted time at current time place is interval;

Determine and the meteorological index that geographic position to be predicted and predicted time interval match;

According to described geographic position to be predicted and described predicted time interval, obtain the meteorological sample data of history under described meteorological index; And

Carry out machine learning according to the meteorological sample data of described history, determine and described geographic position to be predicted and the interval corresponding forecast model of described predicted time.

The method of A3, prediction air quality index according to claim A2, the method also comprises:

Judge whether described geographic position to be predicted exists sample filtering rule in described predicted time interval;

If exist, then described according to described geographic position to be predicted and described predicted time interval, the step obtaining the meteorological sample data of history under described meteorological index comprises further:

According to described geographic position to be predicted and described predicted time interval, based on described sample filtering rule, obtain the meteorological sample data of history under described meteorological index.

The method of A4, prediction air quality index according to claim A1, wherein, described according to geographic position to be predicted and current time, the step obtaining corresponding forecast model comprises further:

Determine that the predicted time at current time place is interval; And

According to described geographic position to be predicted and described predicted time interval, in forecasting model database, carry out matching inquiry, obtain the forecast model corresponding with described geographic position to be predicted and described current time.

The method of A5, prediction air quality index according to claim A4, wherein, described according to geographic position to be predicted and predicted time interval, matching inquiry is carried out in forecasting model database forecasting model database, before obtaining the step of the forecast model corresponding with geographic position to be predicted and current time, the method also comprises:

According to described geographic position to be predicted and described predicted time interval, obtain the meteorological sample data of history under described meteorological index;

Machine learning is carried out, to determine and described geographic position to be predicted and the interval corresponding forecast model of described predicted time according to the meteorological sample data of described history; And

Described forecasting model database is saved to by with described geographic position to be predicted and the interval corresponding forecast model of described predicted time.

The method of A6, prediction air quality index according to claim A5, wherein, the method also comprises:

When described forecasting model database has existed the forecast model corresponding with described geographic position to be predicted and described predicted time interval, previously already present and described geographic position to be predicted and the interval corresponding forecast model of described predicted time are replaced with up-to-date with described geographic position to be predicted and the interval corresponding forecast model of described predicted time.

The method of A7, prediction air quality index according to claim A5 or A6, the method also comprises:

The method of A8, prediction air quality index according to any one of claim A1-A7, it is characterized in that, described forecast model is the forecast model of " based on artificial neural network ".

The method of A9, prediction air quality index according to any one of claim A2-A8, wherein, describedly carry out machine learning according to the meteorological sample data of described history, determine to comprise further with the step of described geographic position to be predicted and the interval corresponding forecast model of described predicted time:

According to the meteorological sample data of described history, carry out machine learning based on artificial neural network, determine weight between the node with described geographic position to be predicted and the interval corresponding artificial neural network of described predicted time;

According to weight between the node of described artificial neural network, set up corresponding forecast model.

The method of A10, prediction air quality index according to claim A9, wherein, described according to the meteorological sample data of described history, carry out machine learning based on artificial neural network, between the node determining corresponding artificial neural network interval with described geographic position to be predicted and described predicted time, the step of weight comprises further:

According to the meteorological sample data of described history, carry out machine learning based on artificial neural network, determine the air quality index learning outcome of described artificial neural network;

Calculate the error amount of the history air quality index in described air quality index learning outcome and the meteorological sample data of described history;

When described error amount is less than predictive error threshold value, weight between the node extracting described artificial neural network.

The method of A11, prediction air quality index according to any one of claim A1-A10, wherein, the step of the weather data in described acquisition geographic position to be predicted comprises further:

Determine and the meteorological index that geographic position to be predicted and current time match;

According to described geographic position to be predicted, determine the weather data under described meteorological index.

The method of A12, prediction air quality index according to any one of claim A1-A11, wherein, described meteorological index comprises following at least any one:

Temperature index; Humidity index; Wind direction index; Wind-force index; Air pressure index; Rainfall amount index; Dew point index; Air quality index index.

The method of A13, prediction air quality index according to any one of claim A1-A12, wherein, the described time to be predicted is tomorrow, and described weather data comprises following at least any one:

The weather data of today and the prediction weather data of tomorrow;

Wherein, the weather data of today comprises following at least any one:

The true weather data of today; The prediction weather data of today;

Wherein, predict that weather data comprises following at least any one:

Present invention also offers A14, a kind of device predicting air quality index, comprising:

Weather data acquisition module, for obtaining the weather data in described geographic position to be predicted; And

Index determination module, for according to described weather data, and carries out calculation process based on described forecast model, determines the air quality index of geographic position to be predicted in the time to be predicted.

The device of A15, prediction air quality index according to claim A14, wherein, described model acquisition module comprises further:

Predicted time determining unit, interval for determining the predicted time at current time place;

First index determining unit, for determining and the meteorological index that geographic position to be predicted and predicted time interval match;

First sample acquisition unit, for according to described geographic position to be predicted and described predicted time interval, obtains the meteorological sample data of history under described meteorological index; And

First model determining unit, for carrying out machine learning according to the meteorological sample data of described history, determines and described geographic position to be predicted and the interval corresponding forecast model of described predicted time.

The device of A16, prediction air quality index according to claim A15, this device also comprises:

First judge module, for judging whether described geographic position to be predicted exists sample filtering rule in described predicted time interval;

If exist, described first sample acquisition module is used for according to described geographic position to be predicted and described predicted time interval, based on described sample filtering rule, obtains the meteorological sample data of history under described meteorological index.

The device of A17, prediction air quality index according to claim A14, wherein, described model acquisition module comprises further:

Predicted time determining unit, interval for determining the predicted time at current time place; And

Second model acquiring unit, for according to described geographic position to be predicted and described predicted time interval, carries out matching inquiry, obtains the forecast model corresponding with described geographic position to be predicted and described current time in forecasting model database.

The device of A18, prediction air quality index according to claim A17, wherein, this device also comprises:

Second index determination module, for determining and the meteorological index that described geographic position to be predicted and described predicted time interval match;

Second sample acquisition module, for according to described geographic position to be predicted and described predicted time interval, obtains the meteorological sample data of history under described meteorological index;

The pre-modeling block of model, for carrying out machine learning according to the meteorological sample data of described history, determines and described geographic position to be predicted and the interval corresponding forecast model of described predicted time; And

Memory module, for being saved to described forecasting model database by with described geographic position to be predicted and the interval corresponding forecast model of described predicted time.

The device of A19, prediction air quality index according to claim A18, wherein, this device also comprises:

Update module, during for there is the forecast model corresponding with described geographic position to be predicted and described predicted time interval when described forecasting model database, replace previously already present and described geographic position to be predicted and the interval corresponding forecast model of described predicted time with up-to-date with described geographic position to be predicted and the interval corresponding forecast model of described predicted time.

The device of A20, prediction air quality index according to claim A18 or A19, this device also comprises:

Second judge module, for judging whether described geographic position to be predicted exists sample filtering rule in described predicted time interval;

If exist, described second sample acquisition module is used for according to described geographic position to be predicted and described predicted time interval, based on described sample filtering rule, obtains the meteorological sample data of history under described meteorological index.

The device of A21, prediction air quality index according to any one of claim A14-A20, it is characterized in that, described forecast model is the forecast model of " based on artificial neural network ".

The device of A22, prediction air quality index according to any one of claim A15-A20, wherein, the pre-modeling block of described model comprises:

Weight determining unit, for according to the meteorological sample data of described history, carries out machine learning based on artificial neural network, determines weight between the node with described geographic position to be predicted and the interval corresponding artificial neural network of described predicted time; And

Unit set up by model, for according to weight between the node of described artificial neural network, sets up corresponding forecast model.

The device of A23, prediction air quality index according to claim A22, wherein, described weight determining unit is used for, according to the meteorological sample data of described history, carrying out machine learning, determine the learning outcome data of described artificial neural network based on artificial neural network; Calculate the error amount of the history air quality index in described air quality index learning outcome and the meteorological sample data of described history; And when described error amount is less than predictive error threshold value, weight between the node extracting described artificial neural network.

The device of A24, prediction air quality index according to any one of claim A14-A23, wherein, described weather data acquisition module comprises:

3rd index determining unit, for determining and the meteorological index that geographic position to be predicted and current time match;

Data determination unit, for according to described geographic position to be predicted, determines the weather data under described meteorological index.

The method of A25, prediction air quality index according to any one of claim A14-A24, wherein, described meteorological index comprises following at least any one:

The method of A26, prediction air quality index according to any one of claim A14-A25, wherein, the described time to be predicted is tomorrow, and described weather data comprises following at least any one:

The weather data of today and the prediction weather data of tomorrow;

Wherein, the weather data of today comprises following at least any one:

The true weather data of today; The prediction weather data of today;

Wherein, predict that weather data comprises following at least any one:

Claims

1. predict a method for air quality index, comprising:

Obtain the weather data in described geographic position to be predicted; And

2. the method for prediction air quality index according to claim 1, is characterized in that, described according to geographic position to be predicted and current time, the step obtaining corresponding forecast model comprises further:

Determine that the predicted time at current time place is interval;

3. the method for prediction air quality index according to claim 1, wherein, described according to geographic position to be predicted and current time, the step obtaining corresponding forecast model comprises further:

Determine that the predicted time at current time place is interval; And

4. the method for the prediction air quality index according to any one of claim 1-3, is characterized in that, described forecast model is the forecast model of " based on artificial neural network ".

5. the method for the prediction air quality index according to any one of claim 1-4, wherein, the step of the weather data in described acquisition geographic position to be predicted comprises further:

6. predict a device for air quality index, comprising:

7. the device of prediction air quality index according to claim 6, wherein, described model acquisition module comprises further:

8. the device of prediction air quality index according to claim 6, wherein, described model acquisition module comprises further:

9. the device of the prediction air quality index according to any one of claim 6-8, is characterized in that, described forecast model is the forecast model of " based on artificial neural network ".

10. the device of the prediction air quality index according to any one of claim 7-9, wherein, described weather data acquisition module comprises: