CN108399470B - Indoor PM2.5 prediction method based on multi-example genetic neural network - Google Patents
Indoor PM2.5 prediction method based on multi-example genetic neural network Download PDFInfo
- Publication number
- CN108399470B CN108399470B CN201810134218.4A CN201810134218A CN108399470B CN 108399470 B CN108399470 B CN 108399470B CN 201810134218 A CN201810134218 A CN 201810134218A CN 108399470 B CN108399470 B CN 108399470B
- Authority
- CN
- China
- Prior art keywords
- indoor
- neural network
- current
- data
- genetic neural
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Operations Research (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Air Conditioning Control Device (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an indoor PM2.5 prediction method based on a multi-example genetic neural network, which comprises the following steps: (1) data acquisition and feature selection; (2) constructing a multi-example genetic neural network prediction model; (3) and predicting the indoor PM2.5 concentration m according to a multi-example genetic neural network prediction model. Through the mode, the indoor PM2.5 prediction method based on the multi-example genetic neural network carries out model training by selecting 7 characteristics of ventilation rate, air temperature, relative humidity and the like closely related to the indoor PM2.5, applies the multi-example neural network and the genetic algorithm to the prediction of the indoor air quality, obtains better prediction precision, greatly shortens the modeling time compared with an empirical model, and has the advantages of high reliability, high precision, high efficiency, strong practicability and the like.
Description
Technical Field
The invention relates to the technical field of particulate matter concentration detection, in particular to an indoor PM2.5 prediction method based on a multi-example genetic neural network.
Background
The empirical model and the statistical model are the main methods for predicting the indoor air quality in the early years, and later, with the development of atmospheric physics and chemical mechanism research, the mechanism model gradually replaces the previous prediction method. In the research of the air quality prediction method based on the mechanism model, the main idea is to implement abstract simulation on the processes of the propagation, diffusion, chemical reaction and the like of pollutants in the air and predict the future air quality condition by researching the physical and chemical characteristics of the atmospheric pollutants and the conversion rule of the atmospheric pollutants under specific conditions.
Most of air quality prediction models commonly adopted in the world are mechanism models, but for PM2.5 which is one of important pollutants in indoor air in China, the sources are diverse, the forming mechanism is complex, and the difficulty in researching the diffusion and complex transformation mechanism of the PM2.5 in the indoor air and effectively modeling the PM2.5 is high, so that the accurate prediction of the PM2.5 concentration by using the traditional mechanism model is difficult, the data sources needed by the mechanism model are various, the operation process is time-consuming, machine learning is considered as a brand-new learning approach by numerous experts, the problems can be effectively solved, and along with the development of machine learning, new machine learning theories and methods are continuously born, many researchers begin to try to use the machine learning method to research and explore the air quality prediction, however, the current research is mostly focused on the prediction of outdoor atmospheric pollutants, the prediction of indoor air pollutants by using a machine learning method is only rarely mentioned.
Disclosure of Invention
The invention mainly solves the technical problem of providing an indoor PM2.5 prediction method based on a multi-example genetic neural network.
In order to solve the technical problems, the invention adopts a technical scheme that:
provided is an indoor PM2.5 prediction method based on a multi-example genetic neural network, comprising the following steps:
(1) data acquisition and feature selection: setting sampling time and sampling time interval, and acquiring current outdoor temperature w corresponding to current time toutCurrent indoor temperature winCurrent outdoor humidity soutCurrent indoor humidity sinCurrent outdoor PM2.5 concentration poutData, current outdoor temperature woutCurrent indoor temperature winCurrent outdoor humidity soutCurrent indoor humidity sinCurrent outdoor PM2.5 concentration poutTaking input parameter data collected in each hour as a sample for inputting parameters, and deleting noise data caused by instability during initial detection of the sensor;
(2) constructing a multi-example genetic neural network prediction model;
(3) predicting indoor PM2.5 concentration m according to a multi-example genetic neural network prediction model;
the specific steps for constructing the multi-example genetic neural network prediction model are as follows:
step 1: defining a predictive model
m=F(t,pout,win,wout,sin,sout,v),
Wherein m represents the PM2.5 concentration in the current room, v represents the ventilation rate in the current room,
v=S×|wout-win|,
wherein S is the area of the opened window;
step 2: normalizing each input parameter data in the sample to be distributed in the interval of [ -1,1],
Wherein A represents all data under the same input parameter in the sample, and a represents single data under the input parameter of A;
and step 3: setting example BijThreshold e1 and threshold e 2: n samples are taken as N data packets, each data packet has M examples, each example is a 7-dimensional feature vector, and the ith data packet BiThe j' th example in (1) is
[Bij1,Bij2,...,Bij7]T;
And 4, step 4: initializing a network structure, then initializing a population, and performing regression training on a data set by using individuals in the population to obtain a predicted value YijCalculating the fitness function f as 1/SSE,
wherein L isiThe average value of the example actual values in the ith data packet is obtained;
and 5: performing evolution operation to generate a next generation population, when the minimum value of the fitness function f is smaller than a threshold value e1 or the evolution algebra reaches 100 generations, selecting a network weight corresponding to the minimum value of the fitness function f as an optimal weight, and otherwise, returning to the step 4 to perform the next round of optimization;
step 6: calculate the ith packet BiError E ofi,
And 7: a global error function E is calculated which is,
and when the E is smaller than the threshold E2, ending the cycle, otherwise, correcting the optimal individual network weight according to the E, and returning to the step 6.
In a preferred embodiment of the invention, the sampling time is 7 consecutive days in summer, autumn and winter respectively, and the sampling interval is 24 hours every other 1 minute all day.
In a preferred embodiment of the present invention, the step of initializing the network structure comprises setting the number of input layer and output layer neurons.
In a preferred embodiment of the invention, the chamber is a breathing zone area with a room height of 1.5 meters.
The invention has the beneficial effects that: the indoor PM2.5 prediction method based on the multi-example genetic neural network is provided, model training is carried out by selecting 7 characteristics such as ventilation rate, air temperature and relative humidity closely related to the indoor PM2.5, the multi-example neural network and the genetic algorithm are applied to prediction of indoor air quality, good prediction accuracy is obtained, meanwhile, compared with an empirical model, modeling time is greatly shortened, and the method has the advantages of being high in reliability, accuracy, efficiency, practicability and the like.
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 are briefly introduced below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without inventive efforts, wherein:
FIG. 1 is a block diagram of a multi-example genetic neural network model of the present invention;
FIG. 2 is a flow chart of the genetic neural network training algorithm of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1 to 2, an embodiment of the present invention includes:
an indoor PM2.5 prediction method based on a multi-example genetic neural network comprises the following steps:
(1) data acquisition and feature selection:
setting sampling time and sampling time interval, and acquiring current outdoor temperature w corresponding to current time toutCurrent indoor temperature winCurrent outdoor humidity soutCurrent indoor humidity sinCurrent outdoor PM2.5 concentration poutData, current outdoor temperature woutCurrent indoor temperature winCurrent outdoor humidity soutCurrent indoor humidity sinCurrent outdoor PM2.5 concentration poutIs an input parameter;
respectively setting monitoring points indoors and outdoors in a test room, arranging the outdoor monitoring points at 3 layers of a hospital building, arranging the indoor monitoring points at a breathing zone area with the height of 1.5 meters, continuously acquiring 7-day data in 3 seasons of 2017, 7 months, 10 months and 2018, namely 3 seasons of summer, autumn and winter, forming 3 data sets, sampling every 24 hours every 1 minute at the sampling interval, wherein the sampling interval is 24 hours all day, the acquired data respectively comprise indoor and outdoor air temperature, indoor and outdoor relative humidity and indoor and outdoor PM2.5 concentration, and totally 7, 24 and 60 totally comprise 10080 data, but noise data is generated due to instability when a sensor is just initially detected, the original data is partially rejected, finally, each sample set comprises 9600 data, and an acquisition time point t is also taken as one of model characteristics, so that each data is (t, p) out,win,wout,sin,sout,v)T;
(2) Constructing a multi-example genetic neural network prediction model;
(3) and predicting the indoor PM2.5 concentration m according to a multi-example genetic neural network prediction model.
The specific steps for constructing the multi-example genetic neural network prediction model are as follows:
step 1: defining a predictive model
m=F(t,pout,win,wout,sin,sout,v),
Current indoor PM2.5 concentration and current outdoor temperature woutCurrent indoor temperature winCurrent outdoor humidity soutCurrent indoor humidity sinCurrent outdoor PM2.5 concentration poutThe current time point t, the current indoor ventilation rate v are related to 7 parameters, corresponding to 7 inputs of the example subnet, and the current indoor PM2.5 concentration corresponds to the output of each data packet, the characteristic inputs of the example subnet are as follows:
table 1 input profile data table for exemplary subnets
Wherein m represents the PM2.5 concentration in the current room, v represents the ventilation rate in the current room,
v=S×|wout-win|,
wherein S is the area of the opened window;
step 2: normalizing each input parameter in the sample to be distributed in the interval of [ -1,1]
Wherein A represents all data under the same input parameter in the sample, and a represents single data under the input parameter of A;
and step 3: setting example BijThreshold e1 and threshold e 2: n samples are taken as N data packets, each data packet has M examples, each example is a 7-dimensional feature vector, and the ith data packet B iThe j' th example in (1) is
[Bij1,Bij2,...,Bij7]T;
And 4, step 4: initializing a network structure, then initializing a population, and performing regression training on a data set by using individuals in the population to obtain a predicted value YijCalculating the fitness function f as 1/SSE,
wherein L isiThe average value of the example actual values in the ith data packet is obtained;
and 5: performing evolution operation to generate a next generation population, when the minimum value of the fitness function f is smaller than a threshold value e1 or the evolution algebra reaches 100 generations, selecting a network weight corresponding to the minimum value of the fitness function f as an optimal weight, and otherwise, returning to the step 4 to perform the next round of optimization;
step 6: calculate the ith packet BiError E ofi,
And 7: real value L of data packet in regression learning problem of multi-example neural networkiIs known, and therefore, utilizes the actual output of the training packet, based on the data packet, according to the error function EiA global error function E is calculated which is,
and when the E is smaller than the threshold E2, ending the cycle, otherwise, correcting the optimal individual network weight according to the E, and returning to the step 6.
The indoor PM2.5 prediction method based on the multi-example genetic neural network has the beneficial effects that:
(1) the multi-example neural network and the genetic algorithm are creatively combined and applied to the prediction of the indoor air quality, so that better prediction precision is obtained, and meanwhile, compared with an empirical model, the modeling time can be greatly shortened;
(2) Data are collected for 1 time every minute, and 60 pieces of data in each hour are combined into one packet as 1 sample, so that possible hidden relations of continuous moments in one hour are considered, and relatively more meaningful hour-level prediction is achieved.
The above description is only an embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by the present specification, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.
Claims (4)
1. An indoor PM2.5 prediction method based on a multi-example genetic neural network is characterized by comprising the following steps:
(1) data acquisition and feature selection: setting sampling time and sampling time interval, and acquiring current outdoor temperature w corresponding to current time toutCurrent indoor temperature winCurrent outdoor humidity soutCurrent indoor humidity sinCurrent outdoor PM2.5 concentration poutData, current outdoor temperature woutCurrent indoor temperature winCurrent outdoor humidity soutCurrent indoor humidity sinCurrent outdoor PM2.5 concentration poutTaking input parameter data collected in each hour as a sample for inputting parameters, and deleting noise data caused by instability during initial detection of the sensor;
(2) Constructing a multi-example genetic neural network prediction model;
(3) predicting indoor PM2.5 concentration m according to a multi-example genetic neural network prediction model;
the specific steps for constructing the multi-example genetic neural network prediction model are as follows:
step 1: defining a predictive model
m=F(t,pout,win,wout,sin,sout,v),
Wherein m represents the PM2.5 concentration in the current room, v represents the ventilation rate in the current room,
v=S×|wout-win|,
wherein S is the area of the opened window;
step 2: normalizing each input parameter data in the sample to be distributed in the interval of [ -1,1],
wherein A represents all data under the same input parameter in the sample, and a represents single data under the input parameter of A;
and step 3: setting example BijThreshold e1 and threshold e 2: n samples are taken as N data packets, each data packet has M examples, each example is a 7-dimensional feature vector, and the ith data packet BiThe j' th example in (1) is
[Bij1,Bij2,...,Bij7]T;
And 4, step 4: initializing a network structure, then initializing a population, and performing regression training on a data set by using individuals in the population to obtain a predicted value YijCalculating the fitness function f as 1/SSE,
wherein L isiThe average value of the example actual values in the ith data packet is obtained;
and 5: performing evolution operation to generate a next generation population, when the minimum value of the fitness function f is smaller than a threshold value e1 or the evolution algebra reaches 100 generations, selecting a network weight corresponding to the minimum value of the fitness function f as an optimal weight, and otherwise, returning to the step 4 to perform the next round of optimization;
Step 6: calculate the ith packet BiError E ofi,
And 7: a global error function E is calculated which is,
2. The method for predicting indoor PM2.5 based on multi-example genetic neural network as claimed in claim 1, wherein the sampling time is 7 consecutive days in summer, autumn and winter, and the sampling time interval is 24 hours every 1 minute all day.
3. The method according to claim 1, wherein the step of initializing the network structure comprises setting the number of input layer and output layer neurons.
4. The method for predicting indoor PM2.5 based on the multi-example genetic neural network is characterized in that the indoor is a breathing zone area with the indoor height of 1.5 meters.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810134218.4A CN108399470B (en) | 2018-02-09 | 2018-02-09 | Indoor PM2.5 prediction method based on multi-example genetic neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810134218.4A CN108399470B (en) | 2018-02-09 | 2018-02-09 | Indoor PM2.5 prediction method based on multi-example genetic neural network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108399470A CN108399470A (en) | 2018-08-14 |
CN108399470B true CN108399470B (en) | 2021-10-08 |
Family
ID=63096576
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810134218.4A Active CN108399470B (en) | 2018-02-09 | 2018-02-09 | Indoor PM2.5 prediction method based on multi-example genetic neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108399470B (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110895390A (en) * | 2018-09-12 | 2020-03-20 | 珠海格力电器股份有限公司 | Method and device for determining placement position of household appliance |
CN109784553B (en) * | 2018-12-29 | 2022-12-02 | 沈阳建筑大学 | Indoor PM2.5 concentration estimation method |
CN109951413B (en) * | 2019-03-19 | 2021-11-23 | 浙江工业大学 | PM2.5 pollution detection method based on multi-antenna WLAN |
CN110009245A (en) * | 2019-04-12 | 2019-07-12 | 阳江职业技术学院 | Indoor air quality prediction technique, device and electronic equipment neural network based |
CN112304831B (en) * | 2020-10-08 | 2021-09-24 | 大连理工大学 | Mass conservation-based student behavior vs. PM in classroom2.5Concentration calculating method |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103955742A (en) * | 2014-04-28 | 2014-07-30 | 淮阴工学院 | PM2.5 forecasting method based on ensemble learning |
CN106056210A (en) * | 2016-06-07 | 2016-10-26 | 浙江工业大学 | PM 2.5 concentration value prediction method based on hybrid neural network |
CN106991491A (en) * | 2017-02-21 | 2017-07-28 | 南京邮电大学 | A kind of environment pollution monitoring system of the RBF neural based on genetic algorithm optimization |
-
2018
- 2018-02-09 CN CN201810134218.4A patent/CN108399470B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103955742A (en) * | 2014-04-28 | 2014-07-30 | 淮阴工学院 | PM2.5 forecasting method based on ensemble learning |
CN106056210A (en) * | 2016-06-07 | 2016-10-26 | 浙江工业大学 | PM 2.5 concentration value prediction method based on hybrid neural network |
CN106991491A (en) * | 2017-02-21 | 2017-07-28 | 南京邮电大学 | A kind of environment pollution monitoring system of the RBF neural based on genetic algorithm optimization |
Non-Patent Citations (1)
Title |
---|
基于改进遗传算法的多示例神经网络优化;蔡自兴 等;《计算机应用》;20051031;第25卷(第10期);第2387-2412页 * |
Also Published As
Publication number | Publication date |
---|---|
CN108399470A (en) | 2018-08-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108399470B (en) | Indoor PM2.5 prediction method based on multi-example genetic neural network | |
CN110084367B (en) | Soil moisture content prediction method based on LSTM deep learning model | |
CN109214592B (en) | Multi-model-fused deep learning air quality prediction method | |
CN111210082B (en) | Precipitation prediction method based on optimized BP neural network algorithm | |
CN108900346B (en) | Wireless network flow prediction method based on LSTM network | |
CN108009674A (en) | Air PM2.5 concentration prediction methods based on CNN and LSTM fused neural networks | |
CN110807554B (en) | Generation method and system based on wind power/photovoltaic classical scene set | |
CN108320016A (en) | A kind of building energy consumption short term prediction method | |
CN112434787B (en) | Terminal space energy consumption prediction method, medium and equipment based on total energy consumption of building | |
CN107346459B (en) | Multi-mode pollutant integrated forecasting method based on genetic algorithm improvement | |
CN113837499A (en) | Ultra-short-term wind power prediction method and system | |
CN113361761A (en) | Short-term wind power integration prediction method and system based on error correction | |
CN116167531A (en) | Photovoltaic power generation prediction method based on digital twin | |
CN114694767B (en) | PM2.5 concentration prediction method based on space-time diagram ordinary differential equation network | |
CN111594996A (en) | Method for predicting air supply quantity of variable air volume air conditioner based on deep belief neural network | |
CN112418495A (en) | Building energy consumption prediction method based on longicorn stigma optimization algorithm and neural network | |
CN112183877A (en) | Photovoltaic power station fault intelligent diagnosis method based on transfer learning | |
CN116525135B (en) | Method for predicting epidemic situation development situation by space-time model based on meteorological factors | |
CN113642255A (en) | Photovoltaic power generation power prediction method based on multi-scale convolution cyclic neural network | |
CN112365082A (en) | Public energy consumption prediction method based on machine learning | |
KR102067110B1 (en) | Energy data processing apparatus and method with high accuracy for constructed building | |
CN110909492A (en) | Sewage treatment process soft measurement method based on extreme gradient lifting algorithm | |
CN115796327A (en) | Wind power interval prediction method based on VMD (vertical vector decomposition) and IWOA-F-GRU (empirical mode decomposition) -based models | |
CN114065646B (en) | Energy consumption prediction method based on hybrid optimization algorithm, cloud computing platform and system | |
CN114881338A (en) | Power distribution network line loss prediction method based on maximum mutual information coefficient and deep learning |
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 |