CN114418179A - Construction raise dust monitoring and predicting method, device and system - Google Patents
Construction raise dust monitoring and predicting method, device and system Download PDFInfo
- Publication number
- CN114418179A CN114418179A CN202111530458.4A CN202111530458A CN114418179A CN 114418179 A CN114418179 A CN 114418179A CN 202111530458 A CN202111530458 A CN 202111530458A CN 114418179 A CN114418179 A CN 114418179A
- Authority
- CN
- China
- Prior art keywords
- model
- construction
- training
- term memory
- bidirectional
- 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.)
- Pending
Links
- 238000010276 construction Methods 0.000 title claims abstract description 74
- 239000000428 dust Substances 0.000 title claims abstract description 59
- 238000012544 monitoring process Methods 0.000 title claims abstract description 51
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000012549 training Methods 0.000 claims abstract description 55
- 230000001537 neural effect Effects 0.000 claims abstract description 38
- 238000003066 decision tree Methods 0.000 claims abstract description 33
- 238000012360 testing method Methods 0.000 claims abstract description 26
- 230000002457 bidirectional effect Effects 0.000 claims abstract description 7
- 238000012545 processing Methods 0.000 claims abstract description 4
- 230000000737 periodic effect Effects 0.000 claims description 3
- 125000004122 cyclic group Chemical group 0.000 claims 3
- 238000010410 dusting Methods 0.000 claims 1
- 239000003344 environmental pollutant Substances 0.000 description 10
- 230000006870 function Effects 0.000 description 10
- 231100000719 pollutant Toxicity 0.000 description 10
- 238000004364 calculation method Methods 0.000 description 7
- 235000011034 Rubus glaucus Nutrition 0.000 description 6
- 235000009122 Rubus idaeus Nutrition 0.000 description 6
- 240000007651 Rubus glaucus Species 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000004891 communication Methods 0.000 description 5
- 238000005286 illumination Methods 0.000 description 5
- 230000002159 abnormal effect Effects 0.000 description 3
- 230000004913 activation Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000007405 data analysis Methods 0.000 description 3
- 239000002689 soil Substances 0.000 description 3
- 238000012800 visualization Methods 0.000 description 3
- 238000004140 cleaning Methods 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 244000235659 Rubus idaeus Species 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000003749 cleanliness Effects 0.000 description 1
- 239000000356 contaminant Substances 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000013079 data visualisation Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000003628 erosive effect Effects 0.000 description 1
- 230000002401 inhibitory effect Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 238000005070 sampling 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
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N15/00—Investigating characteristics of particles; Investigating permeability, pore-volume or surface-area of porous materials
- G01N15/06—Investigating concentration of particle suspensions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/08—Construction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
- G08B21/12—Alarms for ensuring the safety of persons responsive to undesired emission of substances, e.g. pollution alarms
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Computational Linguistics (AREA)
- Economics (AREA)
- Artificial Intelligence (AREA)
- Tourism & Hospitality (AREA)
- Biomedical Technology (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Primary Health Care (AREA)
- Chemical & Material Sciences (AREA)
- Development Economics (AREA)
- Analytical Chemistry (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Pathology (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Immunology (AREA)
- Biochemistry (AREA)
- Environmental & Geological Engineering (AREA)
- Toxicology (AREA)
- Emergency Management (AREA)
- Educational Administration (AREA)
Abstract
The invention discloses a construction raise dust monitoring and predicting method, a device and a system, belonging to the technical field of construction, wherein the method comprises the steps of obtaining construction site data; constructing a construction monitoring and predicting model, wherein the construction monitoring and predicting model comprises a bidirectional-circulation long-short term memory neural model and a gradient lifting decision tree model; training the bidirectional-circulation long-short term memory neural model and the gradient lifting decision tree model to obtain a test network; inputting a test sample into the output value of the long-short term memory neural model of the bidirectional cycle after training and the output value of the gradient lifting decision tree model for weighting processing to obtain prediction data; the device comprises an acquisition module, a construction module, a training module and a variable weight combination module; the prediction method can combine the characteristics of the LSTM network and the LightGBM model, not only can consider the correlation among time sequence data, but also can mine effective information of discontinuous characteristics.
Description
Technical Field
The invention belongs to the technical field of construction, and particularly relates to a construction raise dust monitoring and predicting method, device and system.
Background
At present, products related to construction raise dust monitoring systems are more and more, the precision of monitoring equipment is improved, the types of monitoring meteorological information are more and more, the diversity and the portability of the functions of the equipment are better and better, the functions of the monitoring systems are more and more abundant, so that the hardware has been developed sufficiently, but the purpose of raising dust monitoring is to accurately, efficiently and stably obtain the ground meteorological information of areas near monitoring points, more importantly, abnormal values are monitored, corresponding measures need to be taken to recover to normal values after the abnormal values are detected, the pollution of the raise dust to the atmospheric environment is reduced, the existing raise dust monitoring systems start to take measures to inhibit the raise dust and reduce the pollution after the abnormal values are monitored, but the process has obvious hysteresis, measures for inhibiting the generation of the pollutants can only be taken after the pollution occurs, and the generation of the pollutants can not be completely stopped, a certain amount of contaminants may also be generated into the atmospheric system.
Disclosure of Invention
In order to solve the defects, the invention provides a construction raise dust monitoring and predicting method, a device and a system.
In order to solve the technical problems, the technical scheme adopted by the invention is as follows:
a construction raise dust monitoring and predicting method comprises the following steps:
acquiring construction site data, and dividing the construction site data into training samples and testing samples;
constructing a construction monitoring and predicting model, wherein the construction monitoring and predicting model comprises a bidirectional-circulation long-short term memory neural model and a gradient lifting decision tree model;
training the bidirectional-circulation long-short term memory neural model and the gradient lifting decision tree model to obtain a test network;
and inputting the test sample into the output value of the long-short term memory neural model of the bidirectional cycle after training and the output value of the gradient lifting decision tree model for weighting processing to obtain PM2.5 and PM10 prediction data.
The method is further improved in that: the training and test sample data includes PM2.5, PM10, temperature, humidity, air pressure, light, and noise.
The method is further improved in that: the bidirectional-cycle long-short term memory neural model comprises an input layer, a hidden layer and an output layer.
The method is further improved in that: the gradient boosting decision tree model parameters comprise the maximum depth of the decision tree, the number of leaf nodes, the learning rate, the maximum characteristic value of a tool box, the minimum number of data on one leaf and the randomly selected characteristic proportion in iteration.
The method is further improved in that: the training bidirectional-cycle long-short term memory neural model comprises:
decomposing PM2.5 and PM10 numerical time sequences in a training sample into a plurality of inherent mode components with frequencies from high to low respectively by adopting a self-adaptive noise complete set empirical mode, and removing data noise with higher frequency to obtain reconstructed PM2.5 and PM10 numerical time sequences of low frequency components;
extracting a plurality of maximum correlation variables from the training sample by using the Pearson correlation coefficient as characteristic variables;
training a bidirectional-cycle long-short-term memory neural model by using the plurality of correlation degree maximum variables as characteristic variables, and the PM2.5 and PM10 numerical time sequences of the low-frequency components;
a test network is obtained.
The method is further improved in that: the training of the bidirectional-cycle long-short term memory neural model further comprises:
and setting the training parameters of the bidirectional-cycle long-short-term memory neural model, wherein the parameters comprise the number of training samples, batch size, learning rate and periodic period.
A construction raise dust monitoring and predicting device, comprising:
the acquisition module is used for acquiring construction site data and dividing the construction site data into training samples and testing samples;
the construction module is used for constructing a construction monitoring and predicting model, wherein the construction monitoring and predicting model comprises a bidirectional-circulation long-short term memory neural model and a gradient lifting decision tree model;
the training module is used for training the bidirectional-circulation long-short term memory neural model and the gradient lifting decision tree model to obtain a test network;
and the variable weight combination module is used for weighting the output value of the trained bidirectional-circulation long-short term memory neural model and the output value of the gradient lifting decision tree model by adopting a test sample to obtain prediction data.
The utility model provides a construction raise dust monitoring and prediction system which characterized in that, includes construction raise dust monitoring and prediction device.
Due to the adoption of the technical scheme, the invention has the technical progress that:
the intelligent construction raise dust monitoring system is used for monitoring intelligent construction raise dust, not only can monitor meteorological information of a construction site in real time, but also can take dust reduction measures in advance, and effectively prevents atmospheric pollutants from being generated in the construction site. In order to solve the problem of dust fall measure hysteresis, the change trend of atmospheric pollutants on a construction site needs to be predicted, the invention predicts the change condition of the dust pollutants through an LSTM-LightGBM variable weight combined model, and the prediction method can be combined with the characteristics of an LSTM network and a LightGBM model, not only can consider the correlation among time sequence data, but also can mine effective information of discontinuous characteristics. Compared with a single model, the combined model can effectively reduce the risk of extreme prediction errors. Higher prediction accuracy is also achieved compared to other combination models. After prediction is finished, whether the atmospheric pollutant concentration exceeds an alarm threshold value or not can be judged according to a prediction result, an alarm can be given after the atmospheric pollutant concentration exceeds the alarm threshold value, a display screen is used for warning construction units and urban residents, dust fall measures are taken before the atmospheric pollutant concentration reaches the threshold value, the atmospheric pollutant concentration is informed to a construction site in advance, field personnel are instructed to conduct regulation and control, dust fall equipment such as a fog gun machine can be automatically controlled to operate, and pollution generation is controlled from the root.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a structural frame diagram of the present invention;
Detailed Description
Herein "LSTM" is a bidirectional-cyclic long-short term memory neural model;
"LightGBM" herein is a gradient boosting decision tree model;
"CEEMDAN" herein is an adaptive noise complete set empirical mode;
the invention is described in further detail below with reference to the accompanying drawings:
FIG. 1 is a flow chart of the method of the present invention, which includes the following:
a construction raise dust monitoring and predicting method comprises the following steps:
acquiring construction site data, and dividing the construction site data into training samples and testing samples;
constructing a construction monitoring and predicting model, wherein the construction monitoring and predicting model comprises a bidirectional-circulation long-short term memory neural model and a gradient lifting decision tree model;
training the bidirectional-circulation long-short term memory neural model and the gradient lifting decision tree model to obtain a test network;
and inputting the test sample into the output values of the trained bidirectional-circulation long-short term memory neural model and the gradient lifting decision tree model for weighting processing to obtain prediction data.
Specifically, the change condition of dust pollutants is predicted by constructing a dust monitoring and predicting model, the predicting method can be combined with the characteristics of a long and short memory neural model network and a gradient lifting decision tree model of bidirectional circulation, not only can the correlation among time sequence data be considered, but also effective information of discontinuous characteristics can be mined.
Further, training samples and test sample data include PM2.5, PM10, temperature, humidity, barometric pressure, light, and noise.
Further, the bidirectional-cycle long-short term memory neural model comprises an input layer, a hidden layer and an output layer; in this embodiment, the method for constructing the bidirectional-circulation long-short term memory neural model is as follows:
1. establishing a bidirectional circulation LSTM neural network model with the number of stacked LSTM layers (num _ layers) being 2;
2. the first layer is the input layer, the input _ size: the characteristic dimension is the number of the extracted characteristic variables and the number of PM values, and the length of the input time series (time _ step): the time series length is the amount of data collected by the sensor for one hour (hour), in the method, one piece of data is updated every 5 minutes, so the input time series length is set to 12;
3. the second layer and the third layer are hidden layers (hidden _ size), the characteristic dimension is set to be 14, and bidirectional loop calculation is adopted;
4. the fourth layer outputs the layer (output _ size), and this model is used to predict values for PM2.5 and PM10, so the feature dimension is 2.
Further, the gradient boosting decision tree model parameters comprise the maximum depth of the decision tree, the number of leaf nodes, the learning rate, the maximum characteristic value of a tool box, the minimum number of data on one leaf and the randomly selected characteristic proportion in iteration; in this embodiment, the method for constructing the gradient lifting decision tree model is as follows:
constructing a LightGBM model and defining parameters: the maximum depth (max _ depth) of the decision tree in the decision random forest is set to 4; the number of leaf nodes (num _ leaves) is set to 16; learning rate (learning _ rate) is 0.1 as a default value; the maximum eigenvalue number for the toolbox (max _ bin) is set to 255; the minimum number of data on one leaf (min _ data _ in _ leaf) is set to 20; the proportion of randomly selected features (feature _ fraction) in each iteration is set to 0.7.
Further, training a bidirectional-circulation long-short term memory neural model and a gradient lifting decision tree model to obtain a test network, wherein the training process is as follows:
decomposing PM2.5 and PM10 numerical time sequences in a training sample into a plurality of inherent mode components with frequencies from high to low respectively by adopting a self-adaptive noise complete set empirical mode, and removing data noise with higher frequency to obtain reconstructed PM2.5 and PM10 numerical time sequences of low frequency components;
a plurality of maximum variables of the correlation degree are extracted from a training sample by adopting a Pearson correlation coefficient as characteristic variables, and the Pearson correlation coefficient formula is as follows:
wherein x and y represent two attribute objects, the temperature, the humidity, the illumination, the noise and the air pressure are taken into x, the PM2.5 and the PM10 are taken into y, N represents the total attribute quantity, 7 is taken out, calculation is respectively carried out, and 5 variables with the maximum correlation degree are selected as characteristic variables.
Training a bidirectional-circulation long-short-term memory neural model and a gradient lifting decision tree model by using the plurality of maximum correlation degree variables as characteristic variables and the PM2.5 and PM10 numerical time sequences of the low-frequency components, wherein 5 maximum correlation degree variables can be selected as the characteristic variables;
the bidirectional cycle model training method comprises forward calculation and backward calculation, wherein the forward calculation uses an activation function, and the backward calculation uses BPTT gradient calculation;
using the ELU function as an activation function for hidden layer neurons, the activation function formula is:
setting training parameters of a bidirectional-circulation long-short term memory neural model, wherein the number of training samples is the data volume (yearn) collected by the sensor in one year, and the training samples are set to be 96768; the batch size (batch _ size) is set to the amount of data collected by the sensor per hour (hour), set to 12; the learning rate (learning _ rate) is set to 1 e-6; the styling period (epochs) is set to 100; parameters include training sample number, batch size, learning rate, and periodic period.
When training the LSTM neural network model, data is sent into the model for training with the batch size of 12, an Adam algorithm is selected as an optimizer, and training parameters m are adjusted0Initialized to 0, beta1Default is 0.9, beta2Default to 0.999, epsilon is set to 1e-8, and the square root of the gradient variance is avoided to become 0. And (4) calculating the updated step length, and performing self-adaptive adjustment from two angles of the mean gradient value and the square gradient value to obtain a better training result.
Setting LightGBM model parameters and training: loading a data set collected by a sensor in one year into a model, wherein a training set comprises 96768 samples, the number of boosting iterations (num _ boost _ round) is set to 20, the sample sampling rate (bagging _ fraction) is set to 0.8, training is carried out, and the construction dust monitoring and predicting model is obtained after the training is finished.
The device based on the construction raise dust monitoring and predicting method comprises the following steps:
the acquisition module is used for acquiring construction site data and dividing the construction site data into training samples and testing samples;
the construction module is used for constructing a construction monitoring and predicting model, wherein the construction monitoring and predicting model comprises a bidirectional-circulation long-short term memory neural model and a gradient lifting decision tree model;
the training module is used for training the bidirectional-circulation long-short term memory neural model and the gradient lifting decision tree model to obtain a test network;
and the variable weight combination module is used for weighting the output value of the trained bidirectional-circulation long-short term memory neural model and the output value of the gradient lifting decision tree model by adopting a test sample to obtain prediction data.
System based on above-mentioned construction raise dust monitoring and prediction device includes:
the construction raise dust monitoring and predicting device is used for acquiring construction site data from the cloud server to obtain PM2.5 and PM10 prediction data;
the cloud server is used for receiving the data transmitted by the controller and transmitting the data to the construction raise dust monitoring and predicting device;
the controller comprises a wireless communication module and is used for receiving the data of the acquisition module and forwarding the data to the cloud server;
the acquisition module is used for acquiring construction site data and transmitting the data to the controller;
and the display module is used for displaying the data of the construction raise dust monitoring and predicting device.
The system aims to collect, store, clean, analyze and visualize construction raise dust monitoring data in the environment and carry out online prediction and control on the environment.
The system mainly comprises data acquisition equipment, a storage server, an analysis center and a visualization, prediction and control system. The implementation mainly comprises a data acquisition integration module, a raspberry group, a 5G communication module, a database server and a visualization and prediction control platform.
The data acquisition module adopts industrial data acquisition integrated equipment and can acquire multi-dimensional data such as PM2.5, PM10, temperature, humidity, air pressure, illumination, noise, wind direction, wind speed, pictures, videos, audio and the like related to the generation of construction raise dust;
the storage server is a local storage and cloud storage dual-storage server, the local server is based on a raspberry card type microcomputer, the cloud adopts an Ariiyun database server, and the dual-storage simultaneously stores the acquired multidimensional mass data;
the analysis center is based on a raspberry pi card type microcomputer, the data is cleaned and analyzed, and the data is embedded into an internal algorithm of the raspberry pi through a data mining technology to predict the data so as to achieve a data prediction effect;
decision-making system includes visual, control module, and is visual to contain webpage display and cell-phone APP, can look over data anytime and anywhere, and control divide into automatic control and artificial control, and automatic control can automatic coordinated control environment dust fall equipment promptly on the basis of prediction, and artificial control is promptly through display terminal operation control environment dust fall equipment promptly.
The construction site dust emission belongs to an unorganized emission source, and dust sources comprise stockpile dust, transportation dust, mechanical dust, road wind erosion dust and the like, so that the affected area is large, the factors influencing the dust are more, and the treatment, prevention and control difficulty is large. Therefore, the data acquisition module needs to acquire multiple factors, including the controllable camera and the meteorological multi-element louver containing monitoring air pressure, temperature and humidity, noise, illumination and PM value.
The acquired data presents the characteristics of multiple dimensions and large data volume, and the data are respectively processed to improve the accuracy of a subsequent prediction model, wherein the adjustable camera is used for photographing and acquiring the site environment, personnel information and vehicle information, and the site environment mainly comprises a construction temporary road, a non-paved road, a built roadbed, a stockpile storage place and the surrounding environment of the road; the personnel information mainly comprises wearing information of the operating personnel; the vehicle information mainly includes information such as a license plate, a model of the vehicle, a type of the vehicle, a traveling speed, a degree of cleanliness of the vehicle, and the like.
The multi-element meteorological shelter uses an RS-BYH-M meteorological multi-element shelter, adopts a standard MODBUS-RTU communication protocol, outputs RS485 signals, and monitors and collects the air pressure, the temperature, the humidity, the illumination intensity, the noise, the air speed, the PM2.5 and the PM10 of a construction area.
The storage server is a local storage and cloud storage dual-storage server, the local server is based on a raspberry card type microcomputer, the cloud adopts an Ariiyun database server, and the dual-storage simultaneously stores the acquired multidimensional mass data.
The data analysis uses Raspberry Pi4B as an analysis processor, and has a television output interface with video analog signals, an HDMI high-definition video output interface and an Ethernet interface, and the data analysis has a storage function.
The data analysis of the site environment is to identify images by using a deep learning algorithm, and identify construction roads, non-paved roads, built roadbed, stockpile storage, vehicle information and other related environments of a work place, which is important for preventing and controlling dust emission, and whether the stockpile is subjected to green net protection or not is judged by identifying the site information, so that dust emission of the stockpile caused by wind power can be effectively reduced; identifying whether the worker wears the working clothes and the safety helmet or not, and finishing the comparison of the worker information in the background database to ensure the life safety of the worker; the information such as vehicle license plate, vehicle model and the like is mastered by recognizing the vehicle information, the information such as vehicle length, vehicle weight and the like is judged, the vehicle running speed is calculated by using an algorithm, and the following dust raising and rising amount, concentration and trend can be analyzed by combining the soil information returned by the sensor.
Under the condition that a plurality of independent variables influence the dust concentration, the dust concentration is predicted by adopting an LSTM-LightGBM variable weight combined model based on CEEMDAN, and for different construction procedures of the whole road section, interpolation operation is carried out by adopting an interpolation method on the basis of data of each detection device to calculate the dust distribution of the whole road section.
The decision-making system makes a decision on the construction environment on the basis of the analysis system, and comprises visualization and control functions. The data visualization is divided into three display modes of site display, a PC end and a mobile phone APP, a display screen is adopted on site to display basic information such as mileage, time, PM values, noise, temperature and humidity, the PC end and the mobile phone APP display the same information, and the display mainly aims at decision making and field situation understanding of background personnel, including basic information of prediction analysis results, personnel, environment and vehicles.
The control function is divided into automatic control and artificial control, the automatic control can automatically control the environmental dust-settling equipment in a linkage way on the basis of prediction, and the artificial control is display terminalAnd the terminal operates and controls the environment dust settling equipment, and carries a voice communication system to realize real-time command and control of field personnel. When a decision maker makes a decision, the decision maker can also indicate field personnel through a display screen, so that the working efficiency is improved, measures for preventing and controlling the dust are based on monitoring data, the dust is managed in a grading way and divided into four stages I, II, III and IV from low to high, and the AQI of the stage I is 0-50ug/m3The quality of II-grade air is 51-100ug/m3The AQI of grade III is 101-150ug/m3, the AQI of grade IV is 151-200ug/m3Different decisions are made according to different levels. The system contrasts and analyzes factors influencing dust emission, and a communication system is started to instruct field personnel to cover green cloths under the condition that the green cloths are not covered on the piled materials; directing a sprinkler to perform sprinkling operation on an operation area under the conditions of strong illumination intensity and low soil moisture content; the vehicle cleaning device commands the fog gun machine to operate under the condition that vehicle transportation and wind power greatly cause large flying dust, and commands the vehicle to decelerate at a high vehicle speed by cleaning the vehicle carrying more soil. And the voice reminding device is started when the operator does not wear the voice reminding device into the operation area according to the regulations. And a command decision part system self-learns the problem solution by using a self-learning algorithm and records the problem solution in a problem solution association table, so that an automatic control function is better provided, and three-party interconnection of the monitoring equipment, the dust settling equipment and the terminal equipment is realized.
The above-mentioned embodiments are merely illustrative of the preferred embodiments of the present invention, and do not limit the scope of the present invention, and various modifications and improvements of the technical solution of the present invention by those skilled in the art should fall within the protection scope defined by the claims of the present invention without departing from the spirit of the present invention.
Claims (8)
1. A construction raise dust monitoring and predicting method is characterized by comprising the following steps:
acquiring construction site data, and dividing the construction site data into training samples and testing samples;
constructing a construction monitoring and predicting model, wherein the construction monitoring and predicting model comprises a bidirectional-circulation long-short term memory neural model and a gradient lifting decision tree model;
training the bidirectional-circulation long-short term memory neural model and the gradient lifting decision tree model to obtain a test network;
and inputting the test sample into the output value of the long-short term memory neural model of the bidirectional cycle after training and the output value of the gradient lifting decision tree model for weighting processing to obtain PM2.5 and PM10 prediction data.
2. A construction fugitive dust monitoring and prediction method according to claim 1, wherein the training sample and test sample data include PM2.5, PM10, temperature, humidity, air pressure, light and noise.
3. The construction dust monitoring and predicting method according to claim 1, wherein the bi-directional cyclic long-short term memory neural model comprises an input layer, a hidden layer and an output layer.
4. The construction dust monitoring and predicting method according to claim 1, wherein the gradient boosting decision tree model parameters comprise maximum depth of decision tree, number of leaf nodes, learning rate, maximum feature value of tool box, minimum number of data on one leaf and randomly selected feature ratio in iteration.
5. A construction dust monitoring and prediction method according to claim 1 or 2, characterized in that the training of the bi-directional cyclic long-short term memory neural model comprises:
decomposing PM2.5 and PM10 numerical time sequences in a training sample into a plurality of inherent mode components with frequencies from high to low respectively by adopting a self-adaptive noise complete set empirical mode, and removing data noise with higher frequency to obtain reconstructed PM2.5 and PM10 numerical time sequences of low frequency components;
extracting a plurality of maximum correlation variables from the training sample by using the Pearson correlation coefficient as characteristic variables;
training a bidirectional-cycle long-short-term memory neural model by using the plurality of correlation degree maximum variables as characteristic variables, and the PM2.5 and PM10 numerical time sequences of the low-frequency components;
a test network is obtained.
6. The method of claim 2, wherein the training of the bi-directional cyclic long-short term memory neural model further comprises:
and setting the training parameters of the bidirectional-cycle long-short-term memory neural model, wherein the parameters comprise the number of training samples, batch size, learning rate and periodic period.
7. The utility model provides a construction raise dust monitoring and prediction device which characterized in that includes:
the acquisition module is used for acquiring construction site data and dividing the construction site data into training samples and testing samples;
the construction module is used for constructing a construction monitoring and predicting model, wherein the construction monitoring and predicting model comprises a bidirectional-circulation long-short term memory neural model and a gradient lifting decision tree model;
the training module is used for training the bidirectional-circulation long-short term memory neural model and the gradient lifting decision tree model to obtain a test network;
and the variable weight combination module is used for weighting the output value of the trained bidirectional-circulation long-short term memory neural model and the output value of the gradient lifting decision tree model by adopting a test sample to obtain prediction data.
8. A construction dusting monitoring and prediction system comprising the apparatus of claim 7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111530458.4A CN114418179A (en) | 2021-12-14 | 2021-12-14 | Construction raise dust monitoring and predicting method, device and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111530458.4A CN114418179A (en) | 2021-12-14 | 2021-12-14 | Construction raise dust monitoring and predicting method, device and system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114418179A true CN114418179A (en) | 2022-04-29 |
Family
ID=81267969
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111530458.4A Pending CN114418179A (en) | 2021-12-14 | 2021-12-14 | Construction raise dust monitoring and predicting method, device and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114418179A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116147712A (en) * | 2023-04-18 | 2023-05-23 | 石家庄铁道大学 | Space-time restriction-free three-dimensional construction environment monitoring device and prediction method |
CN116295604A (en) * | 2023-01-04 | 2023-06-23 | 中铁十一局集团有限公司 | Intelligent dust real-time monitoring and control system |
CN116340768A (en) * | 2023-02-28 | 2023-06-27 | 江苏省环境工程技术有限公司 | Intelligent road dust accumulation load monitoring method and monitoring device |
-
2021
- 2021-12-14 CN CN202111530458.4A patent/CN114418179A/en active Pending
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116295604A (en) * | 2023-01-04 | 2023-06-23 | 中铁十一局集团有限公司 | Intelligent dust real-time monitoring and control system |
CN116295604B (en) * | 2023-01-04 | 2024-02-06 | 中铁十一局集团有限公司 | Intelligent dust real-time monitoring and control system |
CN116340768A (en) * | 2023-02-28 | 2023-06-27 | 江苏省环境工程技术有限公司 | Intelligent road dust accumulation load monitoring method and monitoring device |
CN116340768B (en) * | 2023-02-28 | 2023-10-20 | 江苏省环境工程技术有限公司 | Intelligent road dust accumulation load monitoring method and monitoring device |
CN116147712A (en) * | 2023-04-18 | 2023-05-23 | 石家庄铁道大学 | Space-time restriction-free three-dimensional construction environment monitoring device and prediction method |
CN116147712B (en) * | 2023-04-18 | 2023-07-07 | 石家庄铁道大学 | Space-time restriction-free three-dimensional construction environment monitoring device and prediction method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN114418179A (en) | Construction raise dust monitoring and predicting method, device and system | |
CN104103005B (en) | The source tracing method of sudden water environment event pollution sources under a kind of limited conditions | |
CN110502991A (en) | Internal combustion engine health monitor method and system based on random convolutional neural networks structure | |
CN112541028B (en) | Water environment big data monitoring system and method | |
CN109631998A (en) | For environment artificial intelligence monitors system and environment artificial intelligence monitors method | |
CN107784395A (en) | A kind of electric power line ice-covering thickness Forecasting Methodology based on LSTM artificial neural networks | |
CN114518143A (en) | Intelligent environment sensing system | |
US20230290122A1 (en) | Unmanned aerial vehicle based system to track solar panel system construction and commissioning | |
CN106292282A (en) | Reading intelligent agriculture environmental monitoring systems based on big data | |
CN115876655A (en) | Whole process monitoring system and method for explosion-related dust in limited space | |
CN116341776A (en) | Forest fire prevention intelligent detection system and method based on big data | |
CN106989778A (en) | A kind of Tailings Dam on-line monitoring system | |
CN112576312A (en) | Data collection and processing method for electric-hydraulic control support of intelligent fully-mechanized coal mining face | |
CN116826958A (en) | Intelligent safety inspection method for digital transmission channel | |
CN106227127A (en) | Generating equipment intelligent monitoring and controlling device and monitoring method | |
CN113349188B (en) | Lawn and forage precise weeding method based on cloud weeding spectrum | |
CN116168019B (en) | Power grid fault detection method and system based on machine vision technology | |
CN117094532A (en) | Orchard intelligent monitoring system | |
CN106404055A (en) | Power transmission network tower monitoring system | |
CN110211336A (en) | The method of sensor-based landslide data intelligence processing | |
CN111597464B (en) | Data display optimization method and device based on Internet of things | |
CN114359797A (en) | Construction site night abnormity real-time detection method based on GAN network | |
CN113064940A (en) | Highway intelligence real-time charging analytic system based on big data | |
CN117284817B (en) | Dust removing method and system for belt conveyor | |
CN117456363A (en) | Orchard monitoring system based on digital twinning |
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 |