CN110135564A - Coal mine gas sensing data method for detecting abnormality based on edge calculations - Google Patents
Coal mine gas sensing data method for detecting abnormality based on edge calculations Download PDFInfo
- Publication number
- CN110135564A CN110135564A CN201910402329.3A CN201910402329A CN110135564A CN 110135564 A CN110135564 A CN 110135564A CN 201910402329 A CN201910402329 A CN 201910402329A CN 110135564 A CN110135564 A CN 110135564A
- Authority
- CN
- China
- Prior art keywords
- data
- layer
- edge calculations
- neuron
- sample
- 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
- 238000004364 calculation method Methods 0.000 title claims abstract description 62
- 239000003245 coal Substances 0.000 title claims abstract description 61
- 230000005856 abnormality Effects 0.000 title claims abstract description 38
- 238000000034 method Methods 0.000 title claims abstract description 36
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 claims abstract description 75
- 238000001514 detection method Methods 0.000 claims abstract description 40
- 238000012549 training Methods 0.000 claims abstract description 17
- 238000013528 artificial neural network Methods 0.000 claims abstract description 14
- 239000010410 layer Substances 0.000 claims description 161
- 210000002569 neuron Anatomy 0.000 claims description 50
- 238000012544 monitoring process Methods 0.000 claims description 26
- 230000007613 environmental effect Effects 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 11
- 238000005070 sampling Methods 0.000 claims description 10
- 230000008569 process Effects 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 6
- 239000011229 interlayer Substances 0.000 claims description 5
- 238000002372 labelling Methods 0.000 claims description 5
- 239000011159 matrix material Substances 0.000 claims description 5
- 238000005259 measurement Methods 0.000 claims description 5
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 238000013500 data storage Methods 0.000 claims description 3
- 238000004891 communication Methods 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000013135 deep learning Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 230000002265 prevention Effects 0.000 description 3
- 230000009467 reduction Effects 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 210000005036 nerve Anatomy 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000000739 chaotic effect Effects 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012880 independent component analysis Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000009423 ventilation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/22—Fuels; Explosives
- G01N33/225—Gaseous fuels, e.g. natural gas
-
- 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
- G06N3/045—Combinations of 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
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Food Science & Technology (AREA)
- Pathology (AREA)
- Immunology (AREA)
- Biochemistry (AREA)
- Analytical Chemistry (AREA)
- Medicinal Chemistry (AREA)
- Oil, Petroleum & Natural Gas (AREA)
- General Chemical & Material Sciences (AREA)
- Chemical Kinetics & Catalysis (AREA)
- Testing Or Calibration Of Command Recording Devices (AREA)
Abstract
The invention discloses a kind of coal mine gas sensing data method for detecting abnormality based on edge calculations obtains firedamp sensor data, is labeled to data；Multilayer RBM network structure is established, firedamp sensor data are reconstructed by RBM network, obtains a coding of data, the multilayer that firedamp sensor data are obtained by way of successively training indicates；Increase the final layer for representing desired output variable in multilayer RBM network structure, final layer is exported into the input as BP neural network, using having the BP neural network of supervision for error back propagation, it is top-down to be finely adjusted, depth confidence network (DBN) model of firedamp sensor data exception detection is obtained, coal mine gas sensing data is carried out abnormality detection by obtained DBN model.Network bandwidth load is reduced, the real-time of coal mine gas data exception detection is improved.
Description
Technical field
The present invention relates to the intelligent information processing technology field of edge calculations, more particularly to a kind of based on edge calculations
Coal mine gas sensing data method for detecting abnormality.
Background technique
Coal mine gas disaster is one of the main reason for causing casualties, using firedamp sensor to underground coal mine emphasis
Region carries out gas data monitoring, and is detected extremely using monitoring data to gas, is the weight of pre- preventing coal mine gas disaster
Want means.Currently, carrying out coal mine gas sensing data abnormality detection using the method for machine learning obtains good effect, needle
The prediction and coal and gas prominent risk to gas emission are mostly concentrated on to the research in terms of coal mine gas hazard prediction
Evaluation and prediction etc..Gao Li with gas time series be specific research object, propose based on small echo radial basis function when
Between sequence prediction method.Cheng Jian et al. analyzes the protrusion mechanism and every influence factor of coal and gas, establishes Fei Sheer
(Fisher) discriminant analysis model, it was confirmed that the chaotic characteristic of gas density signal, and built on the incident space of gas density
Least square method supporting vector machine prediction model has been found, has been realized in coal mine gas concentration, long-term forecast.Yin Hongsheng is based on
The Multivariate Time Series dimensionality reduction of KPCA/KICA and feature extraction are theoretical, have carried out dimensionality reduction to multidimensional time-series and feature mentions
It takes, using LS-SVM algorithm, gas time series is classified.Peng Hong is using the method for independent component analysis to gas calamity
Evil information has carried out signature analysis, and is based respectively on maximum entropy and support vector machines establishes Gas Disaster Feature Selection Model.
Kozielski et al. proposes a kind of prediction of rule-based homing method progress gas density.Although these methods all take
Certain effect was obtained, however, Gas Disaster monitoring data influence factor is various due to the complexity in coal mine roadway space, number
It is big according to amount, there is nonlinearity, it is difficult to effectively be used.When carrying out sample characteristics design by artificial experience, it is desirable to
Researcher has very deep understanding to Gas Disaster genesis mechanism.Therefore, engineer's gas data characteristics, there are extendible
Property difference limitation.Deep learning is the learning method that machine learning field is emerging in recent years, 2006, what Hinton et al. was proposed
Unsupervised learning algorithm for deep trust network (Deep Belief Network, DBN) solves deep learning model optimization
Difficult problem.After there is initiative document to deliver, a large amount of scholars conduct extensive research deep learning to mention
It is high and apply depth learning technology.In terms of gas data processing, Krzysztof et al. uses the depth based on recurrent neural network
It spends learning model and carries out gas data classification, good effect is achieved on about coal mine data mining.
But with the development of mine Internet of Things, the sensing data of magnanimity is linked into coal mine and calculates cloud platform, causes network
Communications burden and platform calculate pressure, are unable to satisfy requirement of the industry spot to real-time, are mainly manifested in:
(1) data of mine Internet of Things access include video monitoring, personnel positioning, device status monitoring, environmental parameter prison
Control etc. brings difficulty with connective stability to the communication network bandwidth load above and below coal mine.
(2) large-scale data processing analysis and storage give the mine Internet of things system based on cloud platform to bring problem, difficult
To meet requirement of the industry spot to real-time.
Traditional coal mine gas sensing data abnormality detection based on mine cloud platform is unable to satisfy due to above
The requirement of real-time of industry spot.The present invention is therefore.
Summary of the invention
In order to solve the above-mentioned technical problem, the invention proposes a kind of coal mine gas sensing data based on edge calculations
A kind of method for detecting abnormality, solution of new prevention Gas Disaster, can find coal mine gas data exception in time, pass through
Coal mine gas sensing data is carried out in edge calculations equipment and carries out preliminary treatment, shunts the calculating task of mine cloud platform
Amount reduces the data volume for uploading to cloud platform, reduces network bandwidth load, improves the real-time of coal mine gas data exception detection
Property.
The technical scheme adopted by the invention is that:
A kind of coal mine gas sensing data method for detecting abnormality based on edge calculations, comprising the following steps:
S01: obtaining firedamp sensor data, be labeled to data, and the label of mark includes " warning " and " normal "；
S02: establishing multilayer RBM network structure, is reconstructed by RBM network to firedamp sensor data, obtains data
One coding, successively train by way of obtain firedamp sensor data multilayer indicate；
S03: increase the final layer for representing desired output variable in multilayer RBM network structure, final layer is exported
It is top-down to be finely adjusted using having the BP neural network of supervision for error back propagation as the input of BP neural network, it obtains
Depth confidence network (DBN) model detected to firedamp sensor data exception, passes through obtained firedamp sensor data exception
Depth confidence network (DBN) model of detection carries out abnormality detection coal mine gas sensing data.
In preferred technical solution, the step S01 includes after judging data sample time section to data mask method
Whether the firedamp sensor data in certain time are more than alarm threshold, if being more than, the classification for marking the sample is " warning ",
Otherwise it is labeled as " normal ".
In preferred technical solution, in the step S01, obtained by the coal-mine gas monitoring system comprising based on edge calculations
Firedamp sensor data, the coal-mine gas monitoring system comprising based on edge calculations include believing for obtaining environment in coal mine roadway
The sensor node of breath, the sensor node are communicated with edge calculations node, and the edge calculations node is set as relaying
The standby mine environmental monitoring data by sensor node acquisition are uploaded to mine cloud platform, and the mine cloud platform is at data
Reason and data storage；The model training process of depth confidence network (DBN) model of the firedamp sensor data exception detection
It is deployed in the progress of mine cloud platform, the model parameter of generation is back to edge calculations node, examines extremely for real-time perfoming
It surveys.
In preferred technical solution, when detecting abnormal conditions, exception reporting is sent to by edge calculations node to be located at
The data of mine cloud platform and control centre, while the alarm for being located at bottom is driven, carry out methane prediction.
In preferred technical solution, in the step S02, the RBM network includes visible layer and hidden layer, and layer is interior without even
It connects, interlayer connects entirely, and the vector of input is each sample vector, i.e. vi=xi, firedamp sensor data sample xi=(x1,
x2,...,xm) as input, wherein nv=m, label yi=1 is expressed as " alerting ", yi=0 is expressed as " normal ", output vector
hiIndicate sample xiOne recompile；Visible layer unit of the firedamp sensor data measurement as RBM network, hidden layer
The property detector of firedamp sensor data after the corresponding coding of unit；Between visible layer unit and hiding layer unit, exist
Energy function:
nv,nh, respectively indicate the number in visible layer and hidden layer comprising neuron；
The state vector of visible layer, viIndicate the state of i-th of neuron in visible layer；
The state vector of hidden layer, hjIndicate the state of j-th of neuron in hidden layer；
The bias vector of visible layer, aiIndicate the inclined of i-th neuron in visible layer
It sets；
The bias vector of hidden layer, bjIndicate j-th neuron in visible layer
Biasing；
Weight matrix between hidden layer and visible layer, wi,jIt indicates in hidden layer i-th
Connection weight in neuron and visible layer between j-th of neuron；
The firedamp sensor data sample given for one, binary condition hjThe probability for being set as 1 is σ (aj+∑iviwij), wherein σ (x) is logistic function 1/ (1+exp (- x)), after one group of binary condition of hidden layer Unit selection,
Each viThe probability for being set as 1 is σ (bi+∑jhjwij), hide the update of layer unit again according to reconstructed error progress.
In preferred technical solution, sample is trained using contrast divergence algorithm, comprising the following steps:
S21: sample is assigned to visible layer v1, calculate the probability P (h that each neuron is activated in hidden layer1|v1)；
S22: Gibbs sampling is taken to extract a sample h from the probability distribution being calculated1~P (h1|v1)；
S23: h is used1Visible layer is reconstructed, visible layer is pushed away by the way that hidden layer is counter, calculates each neuron in visible layer and be activated
Probability p (v2|h1)；
S24: Gibbs sampling is taken to extract a sample v from the probability distribution being calculated2~P (v2|h1)；
S25: pass through v2Calculate the probability P (h that each neuron is activated in hidden layer2|v2)；
S26: weight is updated:
W←W+λ(P(h1|v1)v1-P(h2|v2)v2)；
a←a+λ(v1-v2)
b←b+λ(h1-h2)
λ is learning rate, the amplitude that every subparameter updates.
The invention also discloses a kind of coal mine gas sensing data abnormality detection system based on edge calculations, comprising:
Data acquisition and labeling module obtain firedamp sensor data, are labeled to data, the label of mark includes
" warning " and " normal "；
RBM network reconfiguration module establishes multilayer RBM network structure, carries out weight to firedamp sensor data by RBM network
Structure, obtains a coding of data, and the multilayer that firedamp sensor data are obtained by way of successively training indicates；
Model foundation and detection module increase by one in multilayer RBM network structure and represent the last of desired output variable
Final layer is exported the input as BP neural network by layer, using having the BP neural network of supervision for error back propagation, is pushed up certainly
It is finely adjusted downwards, obtains depth confidence network (DBN) model of firedamp sensor data exception detection, pass through obtained gas
Depth confidence network (DBN) model of sensing data abnormality detection carries out abnormality detection coal mine gas sensing data.
In preferred technical solution, supervised in the data acquisition and labeling module by the coal mine gas based on edge calculations
Examining system obtains firedamp sensor data, and the coal-mine gas monitoring system comprising based on edge calculations includes for obtaining coal mine lane
The sensor node of environmental information in road, the sensor node are communicated with edge calculations node, the edge calculations node
The mine environmental monitoring data that sensor node acquires are uploaded to mine cloud platform, the mine cloud platform as trunking
It is stored for data processing and data；The mould of depth confidence network (DBN) model of the firedamp sensor data exception detection
Type training process is deployed in the progress of mine cloud platform, and the model parameter of generation is back to edge calculations node, for real-time
It carries out abnormality detection.
In preferred technical solution, in the RBM network reconfiguration module, the RBM network includes visible layer and hidden layer,
Connectionless in layer, interlayer connects entirely, and the vector of input is each sample vector, i.e. vi=xi, firedamp sensor data sample xi
=(x1,x2,...,xm) as input, wherein nv=m, label yi=1 is expressed as " alerting ", yi=0 is expressed as " normal ", defeated
Outgoing vector hiIndicate sample xiOne recompile；Visible layer unit of the firedamp sensor data measurement as RBM network,
The property detector of firedamp sensor data after hiding the corresponding coding of layer unit；Visible layer unit and hiding layer unit it
Between, there are energy functions:
nv,nh, respectively indicate the number in visible layer and hidden layer comprising neuron；
The state vector of visible layer, viIndicate the state of i-th of neuron in visible layer；
The state vector of hidden layer, hjIndicate the state of j-th of neuron in hidden layer；
The bias vector of visible layer, aiIndicate the inclined of i-th neuron in visible layer
It sets；
The bias vector of hidden layer, bjIndicate j-th neuron in visible layer
Biasing；
Weight matrix between hidden layer and visible layer, wi,jIt indicates in hidden layer i-th
Connection weight in neuron and visible layer between j-th of neuron；
The firedamp sensor data sample given for one, binary condition hjThe probability for being set as 1 is σ (aj+∑iviwij), wherein σ (x) is logistic function 1/ (1+exp (- x)), after one group of binary condition of hidden layer Unit selection,
Each viThe probability for being set as 1 is σ (bi+∑jhjwij), hide the update of layer unit again according to reconstructed error progress.
In preferred technical solution, sample is trained using contrast divergence algorithm, comprising the following steps:
S21: sample is assigned to visible layer v1, calculate the probability P (h that each neuron is activated in hidden layer1|v1)；
S22: Gibbs sampling is taken to extract a sample h from the probability distribution being calculated1~P (h1|v1)；
S23: h is used1Visible layer is reconstructed, visible layer is pushed away by the way that hidden layer is counter, calculates each neuron in visible layer and be activated
Probability p (v2|h1)；
S24: Gibbs sampling is taken to extract a sample v from the probability distribution being calculated2~P (v2|h1)；
S25: pass through v2Calculate the probability P (h that each neuron is activated in hidden layer2|v2)；
S26: weight is updated:
W←W+λ(P(h1|v1)v1-P(h2|v2)v2)；
a←a+λ(v1-v2)
b←b+λ(h1-h2)
λ is learning rate, the amplitude that every subparameter updates.
Compared with prior art, the beneficial effects of the present invention are:
The invention proposes a kind of solutions of new prevention Gas Disaster, can find that coal mine gas data are different in time
Often.Preliminary treatment is carried out by carrying out coal mine gas sensing data in edge calculations equipment, shunts the meter of mine cloud platform
Task amount is calculated, the data volume for uploading to cloud platform is reduced, reduces network bandwidth load, improves the detection of coal mine gas data exception
Real-time, also alleviate the pressure of mine network communication.
Detailed description of the invention
The invention will be further described with reference to the accompanying drawings and embodiments:
Fig. 1 is that the present invention is based on the coal-mine gas monitoring system comprising block diagrams of edge calculations；
Fig. 2 is face gas sensor arrangement figure of the present invention；
Fig. 3 is that the present invention is based on the flow charts of the coal mine gas sensing data method for detecting abnormality of edge calculations；
Fig. 4 is RBM schematic network structure of the present invention；
Fig. 5 is the firedamp sensor data reconstruction schematic diagram based on RBM；
Fig. 6 is that the firedamp sensor data exception based on DBN detects network model schematic diagram.
Specific embodiment
In order to make the objectives, technical solutions and advantages of the present invention clearer, With reference to embodiment and join
According to attached drawing, the present invention is described in more detail.It should be understood that these descriptions are merely illustrative, and it is not intended to limit this hair
Bright range.In addition, in the following description, descriptions of well-known structures and technologies are omitted, to avoid this is unnecessarily obscured
The concept of invention.
Embodiment
A kind of coal mine gas sensing data method for detecting abnormality based on edge calculations, by edge calculations equipment
It carries out coal mine gas sensing data and carries out preliminary treatment, shunt the calculating task amount of mine cloud platform, it is flat that reduction uploads to cloud
The data volume of platform reduces network bandwidth load, improves the real-time of coal mine gas data exception detection, to timely discovery coal mine
Gas data exception, prevention Gas Disaster propose new solution.
1, the coal-mine gas monitoring system comprising frame based on edge calculations
Coal-mine gas monitoring system comprising based on edge calculations 3 parts as shown in Figure 1, be specifically made of, including sensor section
Point, edge calculations node and mine cloud platform.
(1) sensor node
In coal-mine gas monitoring system comprising, sensor node is set for obtaining the basis of environmental information in coal mine roadway
It is standby.The sensor that the present invention uses includes firedamp sensor, air velocity transducer, temperature sensor, humidity sensor and air pressure transmission
Sensor.Main method is to be detected by the data of other sensors come the firedamp sensor data to upper corner position.
The arrangement of firedamp sensor is as shown in Figure 2.Firedamp sensor T is arranged in upper corner in U-shaped draft type0Or portable gas explosion detection
Firedamp sensor T is arranged in alarm, working face1, tailentry setting firedamp sensor T2；If coal and gas outburst mine
Firedamp sensor T1It not can control whole non-intrinsically safe type electrical equipments in coal working face air inlet, then set in air inlet
Set firedamp sensor T3；When present low gas and highly gassy mine coal working face use series ventilation, by the air inlet of string working face
Firedamp sensor T is set4.In the coal-mine gas monitoring system comprising frame based on edge calculations, sensor node is mainly and edge
Calculate node is communicated, and the communication with cloud platform is also carried out by edge calculations node as relaying.On the one hand, it can reduce pair
The requirement of sensor node communication capacity；On the other hand, the data-handling capacity of edge calculations node can be made full use of.
(2) edge calculations node
In coal-mine gas monitoring system comprising, the intelligent gateway disposed in coal mine roadway can be used as the section of edge calculations
Point.The intelligent mine lamp that miner carries can be used as mobile edge calculations node and provide corresponding edge calculations service.From system tray
From the point of view of structure angle, edge calculations node exists as middle layer, provides local service to sensor node to improve service quality
And the real-time of information feedback, and then reduce time delay brought by the remote interaction with mine cloud platform.At the same time, Neng Gouti
For the preliminary treatment of data, the calculating task of mine cloud platform is shunted, the data volume for being uploaded to mine cloud platform is reduced, to subtract
The bandwidth load of light backbone.
In method for detecting abnormality proposed by the present invention, edge calculations node is of crucial importance.It is initial in system
The mine environmental monitoring data that sensor node acquires are uploaded to mine cloud platform as trunking by the change stage.In system
The conventional operation stage carries out abnormality detection the data of acquisition.Once detect abnormal conditions, edge calculations node can immediately by
Exception reporting gives data and control centre positioned at mine cloud platform, while driving the alarm for being located at bottom, and it is pre- to carry out gas
It is alert.
(3) mine cloud platform
Mine cloud platform plays teledata and control with the memory space of its superpower computing capability and super large in systems
The effect at center processed is mainly used for the data processing and data storage of extensive and high complexity.In the abnormality detection side of proposition
In method, captured using RBM model and using environmental monitoring data itself spatial coherence.RBM is used for feature as a kind of
The artificial nerve network model of extraction, the training of model parameter complexity with higher.Sensor node and edge calculations section
Point is difficult to provide resource consumption brought by corresponding computing capability and complicated calculations, and therefore, model training process is disposed
It is carried out in mine cloud platform.The model parameter of generation is back to edge calculations node, is used for real-time abnormality detection.
2, coal mine gas data exception detection method, as shown in Figure 3, comprising:
(1) firedamp sensor data prediction
Coal mine gas hazard prediction is usually using the historical data that coal mine gas sensor monitors come to progress in future
Short-term or long-term prediction.Firedamp sensor monitoring data are typical time series datas, when using classification learning method, are learned
The sample of habit and corresponding labeling requirement are manually divided and are marked.
The present invention carries out the detection of gas data exception using the method based on deep learning, from coal-mine gas monitoring system comprising
Take the firedamp sensor time series of a period of time as data set.Time series is divided into data sample by sliding window
This, is divided into training set and test set.All samples are labeled, label is divided into two kinds, is " warning " and " normal " respectively.Into
The foundation of row classification mark is T0, T1And T2The specific value of these three firedamp sensor data.The standard of division refers to trained sample
Behind this period in 3 to 6 minutes whether be more than alarm threshold, be more than then mark the sample classification be " warning ", otherwise
Just it is labeled as " normal ".
(2) the firedamp sensor data reconstruction based on RBM
The present invention is reconstructed firedamp sensor data by the network of RBM, obtains a coding of data, establishes
Multilayer RBM network structure, the multilayer that gas data are obtained by way of successively training indicate.Assuming that a firedamp sensor number
It is X according to samplei=(x1,x2,...,xm), label yi=1 is expressed as " alerting ", yi=0 is expressed as " normal ".One RBM packet
Containing two layers: visible layer and hidden layer.Connection between neuron has a characteristic that connectionless in layer, and interlayer connects entirely.Cause
This, the corresponding figure of RBM is a bipartite graph.The schematic network structure of RBM is as shown in Figure 4.xi=(x1,x2,...,xm) conduct
It inputs, wherein nv=m, the vector of input are each sample vector, i.e. vi=xi, output vector hiIndicate sample xiA weight
It is newly encoded.Visual element of the firedamp sensor data measurement as RBM, Hidden unit correspond to the firedamp sensor number after coding
According to property detector.Between visual element and hiding layer unit, there are energy functions:
nv,nh: respectively indicate the number in visible layer and hidden layer comprising neuron.
The state vector of visible layer, viIndicate the state of i-th of neuron in visible layer.
The state vector of hidden layer, hjIndicate the state of j-th of neuron in hidden layer.
The bias vector of visible layer, aiIndicate the inclined of i-th neuron in visible layer
It sets；
The bias vector of hidden layer, bjIndicate j-th neuron in visible layer
Biasing；
Weight matrix between hidden layer and visible layer, wi,jIndicate i-th of mind in hidden layer
Through the connection weight between j-th of neuron in member and visible layer.
The firedamp sensor data sample given for one, binary condition hjAccording to probability σ (aj+∑iviwij), if
It is 1, wherein σ (x) is logistic function 1/ (1+exp (- x)), bjIt is the biasing of j, viIt is i-th of firedamp sensor numerical value
State, wijIt is the weight between i and j.Once one group of binary condition of hidden layer Unit selection, will according to probability σ (bi+
∑jhjwij) to each viIt is set as 1, aiIt is the biasing of i, hiding layer unit can carry out update again according to reconstructed error, such as
Shown in Fig. 5.
For a sample data x=(x1,x2,…,xm), it is trained using contrast divergence algorithm:
X is assigned to aobvious layer v by 11, calculate the probability P (h that each neuron in hidden layer is activated1|v1)；
2 take Gibbs sampling to extract a sample from the probability distribution of calculating:
h1~P (h1|v1)
3 use h1Aobvious layer is reconstructed, and pushes away aobvious layer by the way that hidden layer is counter, calculates the Probability p (v that each neuron is activated in aobvious layer2|
h1)；
4 similarly, and Gibbs sampling is taken to extract a sample from the probability distribution being calculated
v2~P (v2|h1)
5 pass through v2The probability that each neuron is activated in hidden layer is calculated again, obtains probability distribution P (h2|v2)；
6 update weight
W←W+λ(P(h1|v1)v1-P(h2|v2)v2)；
a←a+λ(v1-v2)
b←b+λ(h1-h2)
λ is learning rate, the amplitude that every subparameter updates.
(3) gas time series abnormality detection
The effective training process of RBM makes it suitable for the composition module as DBN.The implicit unit of each layer of RBM passes through
It practises, expresses the feature of original input data higher degree relationship.DBN may be considered with the nerve for having trained initial weight
Network.When not using class label and consequent propagation in DBN structure, DBN can be used to carry out dimension about to subtract.When by classification
When label and feature vector connect, DBN can be used to classify.
As shown in fig. 6, increasing the final layer for representing desired output variable on the basis of multilayer RBM.By last
Input of the output of layer RBM network as BP neural network.Using having the BP neural network of supervision for error back propagation, push up certainly
Entire model is finely tuned downwards.Input of the information obtained by the multilayer RBM network optimization as BP neural network solves BP mind
The problems such as being easily trapped into local minimum and slow convergence rate due to random starting values through network.
The available input data feature of feature extraction algorithm of the previous section based on multilayer RBM, and DBN has preferably
Classifying quality.Therefore, the present invention is based on DBN algorithms carries out abnormality detection coal mine gas sensing data, and entire algorithm is main
Process description is as follows:
The first step selects suitable training sample set and test set.
Second step carries out data prediction.
Third step, individually each layer of RBM of unsupervised training.
Training set is inputted DBN, obtains feature vector, then verified, adjust the parameter of DBN, Zhi Daoda by the 4th step
To satisfied feature vector.
5th step, by training set, feature set be passed to DBN, classify, inspection-classification as a result, amendment DBN parameters,
Ultimate analysis classification results.
It should be understood that above-mentioned specific embodiment of the invention is used only for exemplary illustration or explains of the invention
Principle, but not to limit the present invention.Therefore, that is done without departing from the spirit and scope of the present invention is any
Modification, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.In addition, appended claims purport of the present invention
Covering the whole variations fallen into attached claim scope and boundary or this range and the equivalent form on boundary and is repairing
Change example.
Claims (10)
1. a kind of coal mine gas sensing data method for detecting abnormality based on edge calculations, which is characterized in that including following step
It is rapid:
S01: obtaining firedamp sensor data, be labeled to data, and the label of mark includes " warning " and " normal "；
S02: establishing multilayer RBM network structure, is reconstructed by RBM network to firedamp sensor data, obtains the one of data
A coding, the multilayer that firedamp sensor data are obtained by way of successively training indicate；
S03: increase the final layer for representing desired output variable in multilayer RBM network structure, final layer is exported into conduct
The input of BP neural network, it is top-down to be finely adjusted using having the BP neural network of supervision for error back propagation, it obtains watt
Depth confidence network (DBN) model of this sensing data abnormality detection is detected by obtained firedamp sensor data exception
Depth confidence network (DBN) model coal mine gas sensing data is carried out abnormality detection.
2. the coal mine gas sensing data method for detecting abnormality according to claim 1 based on edge calculations, feature
It is, the step S01 includes the methane sensing in the certain time after judging data sample time section to data mask method
Whether device data are more than alarm threshold, if being more than, the classification for marking the sample is " warning ", are otherwise labeled as " normal ".
3. the coal mine gas sensing data method for detecting abnormality according to claim 1 based on edge calculations, feature
It is, in the step S01, firedamp sensor data is obtained by the coal-mine gas monitoring system comprising based on edge calculations, it is described
Coal-mine gas monitoring system comprising based on edge calculations includes the sensor node for obtaining environmental information in coal mine roadway, the biography
Sensor node is communicated with edge calculations node, what the edge calculations node acquired sensor node as trunking
Mine environmental monitoring data are uploaded to mine cloud platform, and the mine cloud platform is stored for data processing and data；Described watt
The model training process of depth confidence network (DBN) model of this sensing data abnormality detection be deployed in mine cloud platform into
Row, the model parameter of generation are back to edge calculations node, are used for real-time perfoming abnormality detection.
4. the coal mine gas sensing data method for detecting abnormality according to claim 3 based on edge calculations, feature
It is, when detecting abnormal conditions, exception reporting is sent to data and control positioned at mine cloud platform by edge calculations node
Center processed, while the alarm for being located at bottom is driven, carry out methane prediction.
5. the coal mine gas sensing data method for detecting abnormality according to claim 1 based on edge calculations, feature
It is, in the step S02, the RBM network includes visible layer and hidden layer, and connectionless in layer, interlayer connects entirely, input
Vector is each sample vector, i.e. vi=xi, firedamp sensor data sample xi=(x1,x2,...,xm) as input, wherein
nv=m, label yi=1 is expressed as " alerting ", yi=0 is expressed as " normal ", output vector hiIndicate sample xiOne again
Coding；Visible layer unit of the firedamp sensor data measurement as RBM network, the gas after hiding the corresponding coding of layer unit pass
The property detector of sensor data；Between visible layer unit and hiding layer unit, there are energy functions:
nv,nh, respectively indicate the number in visible layer and hidden layer comprising neuron；
The state vector of visible layer, viIndicate the state of i-th of neuron in visible layer；
The state vector of hidden layer, hjIndicate the state of j-th of neuron in hidden layer；
The bias vector of visible layer, aiIndicate the biasing of i-th of neuron in visible layer；
The bias vector of hidden layer, bjIndicate the biasing of j-th of neuron in visible layer；
Weight matrix between hidden layer and visible layer, wi,jIndicate i-th of neuron in hidden layer
With the connection weight in visible layer between j-th of neuron；
The firedamp sensor data sample given for one, binary condition hjThe probability for being set as 1 is σ (aj+∑iviwij),
Middle σ (x) is logistic function 1/ (1+exp (- x)), after one group of binary condition of hidden layer Unit selection, each viIt is set as
1 probability is σ (bi+∑jhjwij), hide the update of layer unit again according to reconstructed error progress.
6. the coal mine gas sensing data method for detecting abnormality according to claim 5 based on edge calculations, feature
It is, sample is trained using contrast divergence algorithm, comprising the following steps:
S21: sample is assigned to visible layer v1, calculate the probability P (h that each neuron is activated in hidden layer1|v1)；
S22: Gibbs sampling is taken to extract a sample h from the probability distribution being calculated1~P (h1|v1)；
S23: h is used1Visible layer is reconstructed, the anti-probability for pushing away visible layer, calculating that each neuron is activated in visible layer of hidden layer is passed through
p(v2|h1)；
S24: Gibbs sampling is taken to extract a sample v from the probability distribution being calculated2~P (v2|h1)；
S25: pass through v2Calculate the probability P (h that each neuron is activated in hidden layer2|v2)；
S26: weight is updated:
W←W+λ(P(h1|v1)v1-P(h2|v2)v2)；
a←a+λ(v1-v2)
b←b+λ(h1-h2)
λ is learning rate, the amplitude that every subparameter updates.
7. a kind of coal mine gas sensing data abnormality detection system based on edge calculations characterized by comprising
Data acquisition and labeling module obtain firedamp sensor data, are labeled to data, and the label of mark includes " warning "
" normal "；
RBM network reconfiguration module establishes multilayer RBM network structure, and firedamp sensor data are reconstructed by RBM network,
A coding of data is obtained, the multilayer that firedamp sensor data are obtained by way of successively training indicates；
Model foundation and detection module increase the final layer for representing desired output variable in multilayer RBM network structure, will
Final layer exports input as BP neural network, top-down using having the BP neural network of supervision for error back propagation
It is finely adjusted, obtains depth confidence network (DBN) model of firedamp sensor data exception detection, pass through obtained methane sensing
Depth confidence network (DBN) model of device data exception detection carries out abnormality detection coal mine gas sensing data.
8. the coal mine gas sensing data abnormality detection system according to claim 7 based on edge calculations, feature
It is, firedamp sensor is obtained by the coal-mine gas monitoring system comprising based on edge calculations in the data acquisition and labeling module
Data, the coal-mine gas monitoring system comprising based on edge calculations include the sensor section for obtaining environmental information in coal mine roadway
Point, the sensor node are communicated with edge calculations node, and the edge calculations node is as trunking by sensor
The mine environmental monitoring data of node acquisition are uploaded to mine cloud platform, and the mine cloud platform is deposited for data processing and data
Storage；The model training process of depth confidence network (DBN) model of the firedamp sensor data exception detection is deployed in mine
Mountain cloud platform carries out, and the model parameter of generation is back to edge calculations node, is used for real-time perfoming abnormality detection.
9. the coal mine gas sensing data abnormality detection system according to claim 7 based on edge calculations, feature
It is, in the RBM network reconfiguration module, the RBM network includes visible layer and hidden layer, and connectionless in layer, interlayer connects entirely
It connects, the vector of input is each sample vector, i.e. vi=xi, firedamp sensor data sample xi=(x1,x2,...,xm) conduct
It inputs, wherein nv=m, label yi=1 is expressed as " alerting ", yi=0 is expressed as " normal ", output vector hiIndicate sample xi's
One recompiles；Visible layer unit of the firedamp sensor data measurement as RBM network, after hiding the corresponding coding of layer unit
Firedamp sensor data property detector；Between visible layer unit and hiding layer unit, there are energy functions:
nv,nh, respectively indicate the number in visible layer and hidden layer comprising neuron；
The state vector of visible layer, viIndicate the state of i-th of neuron in visible layer；
The state vector of hidden layer, hjIndicate the state of j-th of neuron in hidden layer；
The bias vector of visible layer, aiIndicate the biasing of i-th of neuron in visible layer；
The bias vector of hidden layer, bjIndicate the biasing of j-th of neuron in visible layer；
Weight matrix between hidden layer and visible layer, wi,jIndicate i-th of neuron in hidden layer
With the connection weight in visible layer between j-th of neuron；
The firedamp sensor data sample given for one, binary condition hjThe probability for being set as 1 is σ (aj+∑iviwij),
Middle σ (x) is logistic function 1/ (1+exp (- x)), after one group of binary condition of hidden layer Unit selection, each viIt is set as
1 probability is σ (bi+∑jhjwij), hide the update of layer unit again according to reconstructed error progress.
10. the coal mine gas sensing data abnormality detection system according to claim 9 based on edge calculations, feature
It is, sample is trained using contrast divergence algorithm, comprising the following steps:
S21: sample is assigned to visible layer v1, calculate the probability P (h that each neuron is activated in hidden layer1|v1)；
S22: Gibbs sampling is taken to extract a sample h from the probability distribution being calculated1~P (h1|v1)；
S23: h is used1Visible layer is reconstructed, the anti-probability for pushing away visible layer, calculating that each neuron is activated in visible layer of hidden layer is passed through
p(v2|h1)；
S24: Gibbs sampling is taken to extract a sample v from the probability distribution being calculated2~P (v2|h1)；
S25: pass through v2Calculate the probability P (h that each neuron is activated in hidden layer2|v2)；
S26: weight is updated:
W←W+λ(P(h1|v1)v1-P(h2|v2)v2)；
a←a+λ(v1-v2)
b←b+λ(h1-h2)
λ is learning rate, the amplitude that every subparameter updates.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910402329.3A CN110135564A (en) | 2019-05-15 | 2019-05-15 | Coal mine gas sensing data method for detecting abnormality based on edge calculations |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910402329.3A CN110135564A (en) | 2019-05-15 | 2019-05-15 | Coal mine gas sensing data method for detecting abnormality based on edge calculations |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110135564A true CN110135564A (en) | 2019-08-16 |
Family
ID=67574151
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910402329.3A Pending CN110135564A (en) | 2019-05-15 | 2019-05-15 | Coal mine gas sensing data method for detecting abnormality based on edge calculations |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110135564A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111031107A (en) * | 2019-11-29 | 2020-04-17 | 武汉智菱物联科技有限公司 | Geological disaster monitoring system and method based on low-power-consumption communication network |
| CN111126658A (en) * | 2019-11-15 | 2020-05-08 | 太原理工大学 | Coal mine gas prediction method based on deep learning |
| CN111582016A (en) * | 2020-03-18 | 2020-08-25 | 宁波送变电建设有限公司永耀科技分公司 | Intelligent maintenance-free power grid monitoring method and system based on cloud edge collaborative deep learning |
| CN112769750A (en) * | 2020-12-11 | 2021-05-07 | 广东电力通信科技有限公司 | Protocol stack sending method suitable for intelligent gateway data management |
| CN112966714A (en) * | 2021-02-02 | 2021-06-15 | 湖南大学 | Edge time sequence data anomaly detection and network programmable control method |
| CN114046179A (en) * | 2021-09-15 | 2022-02-15 | 山东省计算中心（国家超级计算济南中心） | Method for intelligently identifying and predicting underground safety accident based on CO monitoring data |
| CN114320469A (en) * | 2021-12-27 | 2022-04-12 | 中国矿业大学 | Cloud-edge intelligence-based underground hazard source detection method |
| WO2023075690A3 (en) * | 2021-10-27 | 2023-06-15 | Envision Digital International Pte. Ltd. | Method and apparatus for transmitting information, and system |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101446205A (en) * | 2008-12-25 | 2009-06-03 | 山东大学 | Coal-mine gas monitoring system comprising wireless sensor network |
| CN106600041A (en) * | 2016-12-02 | 2017-04-26 | 天津大学 | Restricted Boltzmann machine-based photovoltaic power generation short-term power probability prediction method |
| CN108449399A (en) * | 2018-03-09 | 2018-08-24 | 武汉邮电科学研究院有限公司 | It is suitable for the edge calculations node of intelligent production |
| CN108958217A (en) * | 2018-06-20 | 2018-12-07 | 长春工业大学 | A kind of CAN bus message method for detecting abnormality based on deep learning |
| CN109141884A (en) * | 2018-08-17 | 2019-01-04 | 江南大学 | Method for Bearing Fault Diagnosis based on EEMD-AR model and DBN |
| CN109147817A (en) * | 2018-08-29 | 2019-01-04 | 昆明理工大学 | A kind of denoising audio feature extraction algorithm based on the limited Boltzmann machine that makes a variation |
| CN109612513A (en) * | 2018-12-17 | 2019-04-12 | 安徽农业大学 | A kind of online method for detecting abnormality towards extensive higher-dimension sensing data |
| CN109635958A (en) * | 2018-12-12 | 2019-04-16 | 成都航天科工大数据研究院有限公司 | A kind of predictive industrial equipment maintaining method and maintenance system based on edge calculations |
-
2019
- 2019-05-15 CN CN201910402329.3A patent/CN110135564A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101446205A (en) * | 2008-12-25 | 2009-06-03 | 山东大学 | Coal-mine gas monitoring system comprising wireless sensor network |
| CN106600041A (en) * | 2016-12-02 | 2017-04-26 | 天津大学 | Restricted Boltzmann machine-based photovoltaic power generation short-term power probability prediction method |
| CN108449399A (en) * | 2018-03-09 | 2018-08-24 | 武汉邮电科学研究院有限公司 | It is suitable for the edge calculations node of intelligent production |
| CN108958217A (en) * | 2018-06-20 | 2018-12-07 | 长春工业大学 | A kind of CAN bus message method for detecting abnormality based on deep learning |
| CN109141884A (en) * | 2018-08-17 | 2019-01-04 | 江南大学 | Method for Bearing Fault Diagnosis based on EEMD-AR model and DBN |
| CN109147817A (en) * | 2018-08-29 | 2019-01-04 | 昆明理工大学 | A kind of denoising audio feature extraction algorithm based on the limited Boltzmann machine that makes a variation |
| CN109635958A (en) * | 2018-12-12 | 2019-04-16 | 成都航天科工大数据研究院有限公司 | A kind of predictive industrial equipment maintaining method and maintenance system based on edge calculations |
| CN109612513A (en) * | 2018-12-17 | 2019-04-12 | 安徽农业大学 | A kind of online method for detecting abnormality towards extensive higher-dimension sensing data |
Non-Patent Citations (2)
| Title |
|---|
| 王书芹: "基于深度学习的瓦斯时间序列预测与异常检测", 《中国博士学位论文全文数据库 信息科技辑》 * |
| 胡明春 等: "《雷达目标识别原理与实验技术 第1版》", 31 December 2017 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111126658A (en) * | 2019-11-15 | 2020-05-08 | 太原理工大学 | Coal mine gas prediction method based on deep learning |
| CN111031107A (en) * | 2019-11-29 | 2020-04-17 | 武汉智菱物联科技有限公司 | Geological disaster monitoring system and method based on low-power-consumption communication network |
| CN111031107B (en) * | 2019-11-29 | 2022-08-05 | 武汉智菱物联科技有限公司 | Geological disaster monitoring system and method based on low-power-consumption communication network |
| CN111582016A (en) * | 2020-03-18 | 2020-08-25 | 宁波送变电建设有限公司永耀科技分公司 | Intelligent maintenance-free power grid monitoring method and system based on cloud edge collaborative deep learning |
| CN112769750A (en) * | 2020-12-11 | 2021-05-07 | 广东电力通信科技有限公司 | Protocol stack sending method suitable for intelligent gateway data management |
| CN112966714A (en) * | 2021-02-02 | 2021-06-15 | 湖南大学 | Edge time sequence data anomaly detection and network programmable control method |
| CN112966714B (en) * | 2021-02-02 | 2022-05-31 | 湖南大学 | Edge time sequence data anomaly detection and network programmable control method |
| CN114046179A (en) * | 2021-09-15 | 2022-02-15 | 山东省计算中心（国家超级计算济南中心） | Method for intelligently identifying and predicting underground safety accident based on CO monitoring data |
| CN114046179B (en) * | 2021-09-15 | 2023-09-22 | 山东省计算中心（国家超级计算济南中心） | Method for intelligently identifying and predicting underground safety accidents based on CO monitoring data |
| WO2023075690A3 (en) * | 2021-10-27 | 2023-06-15 | Envision Digital International Pte. Ltd. | Method and apparatus for transmitting information, and system |
| CN114320469A (en) * | 2021-12-27 | 2022-04-12 | 中国矿业大学 | Cloud-edge intelligence-based underground hazard source detection method |
| CN114320469B (en) * | 2021-12-27 | 2023-04-07 | 中国矿业大学 | Cloud-edge intelligence-based underground hazard source detection method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110135564A (en) | Coal mine gas sensing data method for detecting abnormality based on edge calculations | |
| CN111212038B (en) | Open data API gateway system based on big data artificial intelligence | |
| CN109698836B (en) | Wireless local area network intrusion detection method and system based on deep learning | |
| US9852342B2 (en) | Surveillance system | |
| CN106888205B (en) | Non-invasive PLC anomaly detection method based on power consumption analysis | |
| US20180278894A1 (en) | Surveillance system | |
| CN104933841B (en) | A kind of fire prediction method based on self organizing neural network | |
| CN111928888A (en) | Intelligent monitoring and analyzing method and system for water pollution | |
| CN107331132B (en) | A kind of method and system of Urban Fires hidden danger dynamic prediction monitoring | |
| CN210691345U (en) | Cloud intelligence safety inspection auxiliary system based on degree of depth learning | |
| KR102091076B1 (en) | Intelligent security control system and method using mixed map alert analysis and non-supervised learning based abnormal behavior detection method | |
| CN117095506B (en) | Fire safety monitoring system and method based on alarm area model | |
| CN111223263A (en) | Full-automatic comprehensive fire early warning response system | |
| KR102096466B1 (en) | Device and method for remote control and alarm using real time database | |
| CN108664875A (en) | Underground belt-conveying monitoring method based on image recognition | |
| CN113642403B (en) | Crowd abnormal intelligent safety detection system based on edge calculation | |
| CN115640915A (en) | Intelligent gas pipe network compressor safety management method and Internet of things system | |
| CN112367303B (en) | Distributed self-learning abnormal flow collaborative detection method and system | |
| Zhang et al. | Data-driven AI emergency planning in process industry | |
| CN114118202A (en) | Early warning method for abnormal events of urban underground comprehensive pipe gallery | |
| CN108898083A (en) | A kind of construction is long-range to be monitored and management system and management method | |
| CN109871711B (en) | Ocean big data sharing and distributing risk control model and method | |
| CN110689140A (en) | Method for intelligently managing rail transit alarm data through big data | |
| CN117833464A (en) | Online operation state safety monitoring method for electricity consumption information acquisition terminal | |
| CN109614932B (en) | Environment identification method based on electroencephalogram field change mechanism, mining helmet and cloud platform |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190816 |
|
| RJ01 | Rejection of invention patent application after publication |