CN109636055A - A kind of non-coal mine Safety Risk in Production prediction and warning platform - Google Patents
A kind of non-coal mine Safety Risk in Production prediction and warning platform Download PDFInfo
- Publication number
- CN109636055A CN109636055A CN201811574195.5A CN201811574195A CN109636055A CN 109636055 A CN109636055 A CN 109636055A CN 201811574195 A CN201811574195 A CN 201811574195A CN 109636055 A CN109636055 A CN 109636055A
- Authority
- CN
- China
- Prior art keywords
- data
- prediction
- module
- hidden danger
- model
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Mining
-
- 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
- G06Q50/265—Personal security, identity or safety
Abstract
The present invention relates to a kind of non-coal mine Safety Risk in Production prediction and warning platforms comprising data acquisition module, ETL process processing module sum number are it is predicted that module and warning module;After the collected non-coal mine production target data analyzed are transmitted to the ETL process processing module processing by the data acquisition module, treated, and achievement data is transmitted to the data prediction module, Safety Risk in Production prediction is carried out according to the data received by the data prediction module, and prediction result is transmitted in the warning module, realize the risk profile keep the safety in production to non-coal mine and early warning.The present invention can keep the safety in production to non-coal mine and Risk Monitoring data are collected arrangement and prediction and warning is analyzed, and provide macro-level policy-making for government regulator and support, provide Risk-warning service for enterprise.
Description
Technical field
The present invention relates to a kind of risk profile early warning platforms, predict especially with regard to a kind of non-coal mine Safety Risk in Production
Early warning platform.
Background technique
In recent years, the situation of production of China's non-coal mine increasingly becomes government and public's focus of attention.With
The reinforcement of safety law enforcement and the standardization of security production supervision and management, the number of casualties be although declined slightly, but total safety is raw
It is still severe to produce situation.The mine disaster to rise one after another, the employment injuries accident to emerge one after another, it appears that at China's rapid economic development
Accompaniment.Carry out the safeguard work of non-coal mine safety in production, it is ensured that the safety in production of non-coal mine has become major enterprise
The priority research areas of industry.
Currently, the safety in production in order to ensure non-coal mine, many enterprises have been equipped with the safety in production of non-coal mine
Monitoring system is sent to administrative staff, to the pipe of safety in production by acquiring the operation data of non-coal mine in the form of statements
Reason has certain booster action.But current Safety Production Monitoring System have the disadvantage that (1) these monitoring systems it is big
It is mounted in locally and just for the application in single mine more, is unfavorable for safety supervision mechanisms access at different levels, and the data collected
It is unfavorable for summarizing for use in big data excavation and forecast analysis.(2) monitoring system relies primarily on the dynamic of sensor collection at present
State data are simultaneously confined to data statistic analysis, the basic functions such as report is checked, lack the related function such as data mining and prediction and warning
Energy.
For big data in the application of non-coal mine risk profile warning aspect, most basic function is exactly to supervise from the risk of magnanimity
Rule, the prediction future developing trend that non-coal mine operation is found in measured data, to targetedly be safeguarded, be improved, in advance
The generation of anti-non-coal mine accident.This project is acquired using multi-source data and processing, big data prediction and warning, visual analyzing etc.
Key technology establishes non-coal mine Safety Risk in Production prediction and warning platform.Platform using different zones non-coal mine operation and
Monitoring data carry out big data mining analysis and obtain disaster pests occurrence rule, and are used for that production safety management is instructed to work, and change
Become the mode that traditional safety in production " subsequent management " is " preventing in advance ", innovation security control method for monitoring and means.Therefore,
It is necessary to establish non-coal mine Safety Risk in Production prediction and warning platform.
Summary of the invention
In view of the above-mentioned problems, the object of the present invention is to provide a kind of non-coal mine Safety Risk in Production prediction and warning platform,
It can keep the safety in production to non-coal mine and Risk Monitoring data are collected arrangement and prediction and warning is analyzed, and is government regulator
It provides macro-level policy-making to support, provides Risk-warning service for enterprise.
To achieve the above object, the present invention takes following technical scheme: a kind of non-coal mine Safety Risk in Production prediction is pre-
Alert platform comprising data acquisition module, ETL process processing module sum number are it is predicted that module and warning module;The data are adopted
The collected non-coal mine production target data analyzed are transmitted at the ETL process processing module by collection module
After reason, treated, and achievement data is transmitted to the data prediction module, by the data prediction module according to the number received
According to progress Safety Risk in Production prediction, and prediction result is transmitted in the warning module, is realized raw safely to non-coal mine
The risk profile of production and early warning.
Further, the ETL process processing module is for lacking the achievement data that data acquisition module is transmitted to
Value processing, outlier processing and data normalization pretreatment;And the pretreated data of selected section carry out ETL processing, obtain
Training sample set;Meanwhile pretreated data are extracted in real time as sample data to be identified.
Further, the missing values processing is to specify its corresponding missing values processing method for each specific targets;
Missing values processing method includes that missing values are filled with default value filling, missing values with average value, missing values are filled with median, are lacked
Mistake value is by mode is filled, the data record comprising missing values is filled in a manner of customized and ignored to missing values.
Further, the outlier processing is to specify its corresponding outlier processing method for each specific targets;
Outlier processing method includes that exceptional value is replaced with to default value and ignores the data record comprising exceptional value.
Further, the method that the data prediction module carries out Safety Risk in Production prediction to the data received are as follows: right
After monitoring index data to be analyzed carry out data prediction, using training data, pass through Linear Regression Forecasting Model, BP nerve net
Network prediction model and regression tree prediction model establish prediction model;Then it is commented using training pattern of the test data to generation
Estimate, the smallest model of Select Error carries out prediction and warning to target as optimum prediction model, using optimum prediction model.
Further, the warning module is used to carry out prediction and warning point to non-coal mine safety in production and Risk Monitoring data
Analysis, including index trend analysis, clustering, Time-Series analysis, analysis of Influential Factors, correlation analysis and hidden danger classification.
Further, the Time Series Analysis Method are as follows: be analysed to after monitoring index data carry out data prediction, using more
Kind Time Series Analysis Method carries out Time-Series analysis, selects optimal Time Series Analysis Method and parameter setting, obtains optimum timing analysis
Prediction result;Wherein, optimal Time Series Analysis Method uses exponential smoothing.
Further, the index trend analysis are as follows: after being analysed to monitoring index data progress data prediction, benefit
With the data generated after pretreatment, determine each factor to the significance level of key index by Information Gain Method;Wherein, information
Gain method is to find out an attribute R, and the information gain of this attribute R division front and back is bigger than other attributes.
Further, the correlation analysis are as follows: after being analysed to monitoring index data progress data prediction, calculate
Related coefficient between each index;By the value range of following related coefficient absolute value come the correlation intensity between judgment variable:
The extremely strong correlation of 0.8-1.0;0.6-0.8 strong correlation;The moderate correlation of 0.4-0.6;The weak correlation of 0.2-0.4;The extremely weak phase of 0.0-0.2
It closes or without correlation.
Further, the hidden danger classification method are as follows: acquisition arranges incipient fault data and carries out data prediction to it: including class
Not Xuan Ze, data balancing;Then it carries out word segmentation processing and obtains term vector model using deep learning method;Utilize term vector mould
Type calculates the hidden danger vector of every incipient fault data, and then obtains hidden danger disaggregated model using the training of hidden danger vector, is classified by hidden danger
Model carries out hidden danger classification;Term vector model is a kind of efficient algorithm model that word is characterized as to real number value vector, use
Term vector generation method is Word2vec;The calculating of hidden danger vector: describing term vector to all hidden danger and be overlapped, to obtain
Hidden danger describes corresponding hidden danger vector;Hidden danger disaggregated model is classified to incipient fault data: using in decision Tree algorithms
C4.5 algorithm is classified.
The invention adopts the above technical scheme, which has the following advantages: the present invention is towards non-coal mine operational process
In risk adopted around the operating parameter and specific monitoring data of non-coal mine using comprehensive monitoring model, multi-source data
The key technologies such as collection and processing scheme, big data prediction and warning method, visual analyzing, establish non-coal mine risk profile early warning
Platform promotes the risk perceptions ability in non-coal mine production safety management.
Specific embodiment
The present invention is described in detail below with reference to embodiment.
The present invention provides a kind of non-coal mine Safety Risk in Production prediction and warning platform comprising data acquisition module, number
According to ETL processing module sum number it is predicted that module and warning module.Data acquisition module by it is collected analyze it is non-
After Coal Mines production target data are transmitted to the processing of ETL process processing module, treated, and achievement data is transmitted to data prediction
Module carries out Safety Risk in Production prediction according to the data that receive by data prediction module, and prediction result is transmitted to pre-
In alert module, the risk profile keep the safety in production to non-coal mine and early warning are realized.
In above-described embodiment, it is preferred that ETL process processing module is used for the achievement data being transmitted to data acquisition module
Carry out the pretreatment such as missing values processing, outlier processing and data normalization.And the pretreated data of selected section carry out ETL
Processing, obtains training sample set;Meanwhile pretreated data are extracted in real time as sample data to be identified.Wherein:
Missing values processing is to specify its corresponding missing values processing method for each specific targets.Missing values processing
Method includes:
1) missing values are filled with default value;
2) missing values are filled with average value;
3) missing values are filled with median;
4) missing values are filled with mode;
5) missing values are filled in a manner of customized;
6) ignore the data record comprising missing values.
Outlier processing is to specify its corresponding outlier processing method for each specific targets.Outlier processing
Method includes: that exceptional value 1) is replaced with default value;2) ignore the data record comprising exceptional value.
Data normalization is to be allowed to fall into a small preset section by data bi-directional scaling.Data mark
Standardization can have dimension expression formula to become dimensionless expression formula with the convergence rate of lift scheme, the precision of lift scheme and handle
Deng.
In the present embodiment, data normalization method is standardized using 0-1.0-1 standardization is also referred to as deviation standardization, it is
Linear transformation is carried out to initial data, result is made to fall on [0,1] section.
In above-described embodiment, it is preferred that data prediction module carries out Safety Risk in Production prediction to the data received
Method are as follows:
After treating research and application achievement data progress data prediction, using training data, built by a variety of prediction techniques
Vertical prediction model.Then it is assessed using training pattern of the test data to generation, the smallest model of Select Error is as most
Excellent prediction model carries out prediction and warning to target using optimum prediction model.
In a preferred embodiment, a variety of prediction techniques include Linear Regression Forecasting Model, BP neural network prediction
Model and regression tree prediction model.Wherein:
(1) linear regression model (LRM) refers to that the relationship between dependent variable and independent variable is linear.It adopts in the present embodiment
Multivariate regression models are as follows:
yi=β0+β1x1+β2x2+…+βpxp
Multivariate regression models is a combinatorial problem, it is known that how some data seek the unknown parameter of the inside, provide one
Optimal solution.And multivariate regression models is solved using least square method or gradient descent method.
In the present embodiment, selection participates in the index of goaf Pillar Stability forecast analysis, and determines that prediction target is
Pillar Stability.
(2) BP neural network prediction model is a multilayer feedforward neural networks, learning process by signal forward-propagating and mistake
Two process compositions of backpropagation of difference.When forward-propagating, input sample is incoming from input layer, after hidden layer is successively handled,
It is transmitted to output layer.If the reality output of output layer is not inconsistent with desired output, the back-propagation phase of turning error.Error it is anti-
It is output error to be passed through to hidden layer with some form to the layer-by-layer anti-pass of input layer, and give error distribution to all of each layer to propagation
Unit, to obtain the error signal of each layer unit, this error signal is the foundation as amendment each unit weight.Weight is continuous
Adjustment process, that is, the learning training process of network.The error that this process is performed until network output is reduced to and can connect
The degree received, or until proceeding to preset study number.
BP neural network model topology structure includes input layer (input), hidden layer (hide layer) and output layer
(output layer).Its workflow is yes: input learning sample, and stimulation is passed to hidden layer by input layer, and hidden layer is logical
It crosses the intensity (weight) contacted between neuron and delivery rules (activation primitive) and output layer is passed into stimulation, output layer arranges hidden
Stimulation after hiding layer processing generates final result.If have correctly as a result, so by correct result and the result of generation into
Row compares, and obtains error, then backstepping carries out feedback modifiers to the link weight in nerve net, thus to complete the process of study.
And the study mechanism fed back backward is used, the weight in Lai Xiuzheng nerve net is finally reached the purpose of output correct result.
In the present embodiment, selection participates in the index of forecast analysis, and determines prediction target.
(3) regression tree prediction model uses CART algorithm, and CART is one kind of decision tree, and is very important decision
Tree.
When creating regression tree, observed value value be it is continuous, without tag along sort, the value that only data obtain according to the observation
To create the rule of a prediction.In this case, with regard to helpless, CART is then used most the optimal dividing rule of classification tree
Small residual variance (Squared Residuals Minimization) determines the optimal dividing of Regression Tree, should
Criteria for classifying is that the subtree error variance after expectation divides is minimum.When creating model tree, each leaf node is then a machine
Device learning model, such as linear regression model (LRM).
CART algorithm includes two points, single argument divides and Pruning strategy three parts:
1) two points (Binary Split): being all that two points are carried out to observation variable in each deterministic process.
CART algorithm uses a kind of technology of two points of recursive subdivisions, and current sample set is always divided into two increments by algorithm
This collection, so that only there are two branches for the non-leaf node of each of decision tree generated.Therefore the decision tree of CART algorithm generation is
Binary tree simple for structure.Therefore it is to be or non-scene that CART algorithm, which is suitable for the value of sample characteristics, for continuous feature
Processing it is then similar to C4.5 algorithm.
2) single argument segmentation (Split Based on One Variable): each optimal dividing becomes both for single
Amount.
3) Pruning strategy: the key point of CART algorithm, and the committed step of entire Tree-Based algorithm.
Beta pruning process is especially important, so occupying an important position in optimum decision tree generating process.Some researches show that cut
The importance of branch process is more even more important than tree generating process, for the maximal tree that the different criteria for classifying generates, beta pruning it
After can retain most important Attribute transposition, difference is little.It is that pruning method is more crucial for the generation of optimal tree instead.
In the present embodiment, selection participates in the index of forecast analysis, and determines prediction target.
In above-described embodiment, in the present embodiment, the smallest regression tree model of Select Error is as optimum prediction model.
In above-described embodiment, it is preferred that warning module is used to carry out non-coal mine safety in production and Risk Monitoring data
Prediction and warning analysis, including index trend analysis, clustering, Time-Series analysis, analysis of Influential Factors, correlation analysis and hidden danger
Classification.Wherein:
(1) Time-Series analysis refers to one group of data being sequentially arranged, and is formed using known time series data
Different mathematics, and be used to the method predicted and be referred to as time series analysis abbreviation Time-Series analysis.
Time Series Analysis Method are as follows: after being analysed to monitoring index data progress data prediction, utilize a variety of Time-Series analyses
Method carries out Time-Series analysis, selects optimal Time Series Analysis Method and parameter setting, obtains optimum timing analysis prediction result.
Optimal Time Series Analysis Method uses exponential smoothing, and exponential smoothing rule has been compatible with that full period is average and rolling average
The chief does not give up past data, but is given only the influence degree gradually weakened, i.e., separate with data, assigns gradually
Converge to zero flexible strategy.Exponential smoothing is a kind of Time Series Analysis Forecasting to grow up on the basis of the method for moving average
Method, it is to cooperate regular hour sequential forecasting models to predict the future of phenomenon by gauge index smooth value.
Exponential smoothing formula are as follows:
Ft+1=axt+(1-a)Ft
Wherein, a is known as smoothing factor, value 0 < a < 1.In general, a value chooses whether proper, directly influences prediction
Result and precision.
By exponential smoothing formula qualitative analysis:
As a=1, then there is Ft+1=xt, indicate that the predicted value of next phase is equal to the actual value of current period;
As a=0, then there is Ft+1=Ft, indicate that the predicted value of next phase is equal to the predicted value of current period.
It can be seen that a value is bigger, the influence and effect of more paying attention to Recent data are indicated.A value is in principle according to the wave of sequence
The choosing of dynamic and trend.If sequence variation is gentle or irregular fluctuation, a value should be selected smaller to eliminate irregular variation influence.Such as
Sequence variation, which has, significantly rises or falls trend, and a value should be selected greatly, so that there are Recent data larger flexible strategy to reflect to prediction
As a result in.
(2) index trend analysis carries out comprehensive assessment to several correlative factors for influencing non-coal mine key index, and measures
Change each factor to the influence power size of key index, provides macro-level policy-making for supervision department and support.
Index trend analysis are as follows: after being analysed to monitoring index data progress data prediction, after pretreatment
The data of generation determine each factor to the significance level of key index by Information Gain Method.
Information Gain Method is to find out an attribute R, and the information gain of this attribute R division front and back is than other attributes
Greatly.Information before division calculates as follows:
In formula, m represents classification number, piThis kind of other probabilities of occurrence are represented, info (D) indicates the information of data set D
Amount.
Assuming that selecting attribute R as Split Attribute, in data set D, R has k different value { V1,V2,…,Vk, then
D can be divided into k group { D according to the value of R1,D2,…,Dk, after being divided by R, the information content summation of data set D difference group are as follows:
And information gain is to divide front and back, the difference of two information content:
Gain (R)=info (D)-infoR(D)
Information gain Gain (R) indicates that attribute R gives classification bring information content.If Gain (R) is bigger, show that R can make
Classify pure as far as possible, most probable that different classes is separated, i.e., R is more important for objective attribute target attribute.Due to all attributes
Info (D) is the same, and institute can be converted into the hope of maximum Gain (R) seeks the smallest infoR(D)。
In the present embodiment, several index significance levels analysis for carrying out influencing Pillar Stability determines that analysis target is
Pillar Stability simultaneously carries out discretization to target.Then importance analysis is carried out to each index, there are the times to ore pillar for dead zone
Stability influence is maximum, and the significance level of other factors successively reduces.
(3) correlation analysis is quantified between the degree of correlation non-coal mine monitoring index and the situation that influences each other
Analysis, some potential rules between allowing user to better understand non-coal mine monitoring index provide macroscopic view for supervision department
Decision support.
Correlation analysis are as follows: after being analysed to monitoring index data progress data prediction, calculate between each index
Related coefficient.
Related coefficient is the index of degree of correlation between variable.The value of related coefficient is typically in the range of between -1~1, related
The the absolute value of coefficient the big, and it is higher to represent its degree of correlation.It is called positive correlation when related coefficient is greater than zero, conversely, then claiming it
For negative correlation.The present invention uses Pearson correlation coefficient, the expression formula of related coefficient are as follows:
By the value range of following related coefficient absolute value come the correlation intensity between judgment variable:
The extremely strong correlation of 0.8-1.0;0.6-0.8 strong correlation;The moderate correlation of 0.4-0.6;The weak correlation of 0.2-0.4;0.0-
0.2 is extremely weak related or without correlation.
Correlation analysis is carried out to Tailings Dam index of correlation monitoring data, correlation matrix can be obtained, by related coefficient
Matrix, what can be will be apparent that sees the degree of correlation between each index.
(4) hidden danger classification is the analysis to incipient fault data in production environment, and training obtains hidden danger disaggregated model, realizes hidden danger
The automatic classification of description, significantly reduces manual analysis cost.
Hidden danger classification method are as follows: acquisition arrange incipient fault data and it is carried out data prediction (including classification selection, data
It is balanced), it then carries out word segmentation processing and obtains term vector model using deep learning method;Every is calculated using term vector model
The hidden danger vector of incipient fault data, and then hidden danger disaggregated model is obtained using the training of hidden danger vector, it is carried out by hidden danger disaggregated model hidden
Suffer from classification.
In data prediction, the selection of hidden danger classification: the selection biggish several hidden danger classifications of data volume describe data.For number
According to small hidden danger classification is measured, default is given up.
Data balancing: by the lesser hidden danger classification of data volume, the more hidden danger numbers of the category are obtained by way of duplication
According to, and the incipient fault data quantity of each classification is substantially the same.
Term vector model is a kind of efficient algorithm model that word is characterized as to real number value vector, can by training,
Vector operation in K dimensional vector space is reduced to the processing of content of text, and the similarity in vector space can be used to table
Show similar on text semantic.The term vector generation method that the present invention uses is Word2vec: being mapped each word by training
At K tie up real vector (K is generally the hyper parameter in model), by the distance between word (such as cosine similarity, Euclidean away from
From etc.) judge the semantic similarity between them.It uses one three layers of neural network, input layer-hidden layer-output layer.
It is encoded according to word frequency with Huffman, so that the content of the similar word hidden layer activation of all word frequency is almost the same, the frequency of occurrences is got over
High word, the hiding number of layers that they activate is fewer, effectively reduces the complexity of calculating in this way.
The calculating of hidden danger vector: describing term vector to all hidden danger and be overlapped, thus obtain hidden danger describe it is corresponding hidden
Suffer from vector.
Hidden danger disaggregated model is classified to incipient fault data, and decision Tree algorithms, including decision tree, Bayes can be used
With neural network etc..In the present embodiment, using the C4.5 algorithm in decision Tree algorithms.
In conclusion the present invention is towards the risk in non-coal mine operational process, to protect people's life's property to pacify
Complete and periphery ecological environment is core objective, comprehensive around the operating parameter of non-coal mine and specific monitoring data, utilization
Monitoring model, multi-source data acquire and the key technologies such as processing scheme, big data prediction and warning method, visual analyzing, establishes
Non-coal mine risk profile early warning platform, improves the risk perceptions ability in non-coal mine production safety management, is government
Supervision department provides macro-level policy-making and supports, provides Risk-warning service for enterprise.
The various embodiments described above are merely to illustrate the present invention, and the structure and setting steps of each component are all that can be varied
, based on the technical solution of the present invention, the improvement and equivalents that all principles according to the present invention carry out separate step,
It should not exclude except protection scope of the present invention.
Claims (10)
1. a kind of non-coal mine Safety Risk in Production prediction and warning platform, it is characterised in that: including data acquisition module, data
ETL processing module sum number is it is predicted that module and warning module;The data acquisition module is analyzed collected
After non-coal mine production target data are transmitted to the ETL process processing module processing, treated, and achievement data is transmitted to institute
Data prediction module is stated, carries out Safety Risk in Production prediction according to the data received by the data prediction module, and will be pre-
It surveys result to be transmitted in the warning module, realizes the risk profile keep the safety in production to non-coal mine and early warning.
2. early warning platform as described in claim 1, it is characterised in that: the ETL process processing module is used for data acquisition module
The achievement data that block is transmitted to carries out missing values processing, outlier processing and data normalization pretreatment;And selected section is located in advance
Data after reason carry out ETL processing, obtain training sample set;Meanwhile pretreated data are extracted in real time as sample to be identified
Notebook data.
3. early warning platform as claimed in claim 2, it is characterised in that: the missing values processing is to be directed to each specific targets,
Specify its corresponding missing values processing method;Missing values processing method includes that missing values are filled with default value, missing values are with average
Value filling, missing values are filled by median, missing values are by mode is filled, missing values are filled in a manner of customized and is ignored comprising lacking
The data record of mistake value.
4. early warning platform as claimed in claim 2, it is characterised in that: the outlier processing is to be directed to each specific targets,
Specify its corresponding outlier processing method;Outlier processing method includes that exceptional value is replaced with to default value and is ignored comprising different
The data record of constant value.
5. early warning platform as described in claim 1, it is characterised in that: the data prediction module pacifies the data received
The method of full production risk prediction are as follows: after treating research and application achievement data progress data prediction, using training data, pass through
Linear Regression Forecasting Model, BP neural network prediction model and regression tree prediction model establish prediction model;Then test is utilized
Data assess the training pattern of generation, and the smallest model of Select Error utilizes optimum prediction as optimum prediction model
Model carries out prediction and warning to target.
6. early warning platform as described in claim 1, it is characterised in that: the warning module be used for non-coal mine safety in production and
Risk Monitoring data carry out prediction and warning analysis, including index trend analysis, clustering, Time-Series analysis, analysis of Influential Factors,
Correlation analysis and hidden danger classification.
7. early warning platform as claimed in claim 6, it is characterised in that: the Time Series Analysis Method are as follows: be analysed to monitoring index
After data carry out data prediction, Time-Series analysis is carried out using a variety of Time Series Analysis Methods, selects optimal Time Series Analysis Method
And parameter setting, obtain optimum timing analysis prediction result;Wherein, optimal Time Series Analysis Method uses exponential smoothing.
8. early warning platform as claimed in claim 6, it is characterised in that: the index trend analysis are as follows: be analysed to monitor
After achievement data carries out data prediction, using the data generated after pretreatment, each factor pair is determined by Information Gain Method
The significance level of key index;Wherein, Information Gain Method is to find out an attribute R, the information of this attribute R division front and back
Gain is bigger than other attributes.
9. early warning platform as claimed in claim 6, it is characterised in that: the correlation analysis are as follows: be analysed to monitoring and refer to
After marking data progress data prediction, related coefficient between each index is calculated;
By the value range of following related coefficient absolute value come the correlation intensity between judgment variable:
The extremely strong correlation of 0.8-1.0;0.6-0.8 strong correlation;The moderate correlation of 0.4-0.6;The weak correlation of 0.2-0.4;The pole 0.0-0.2
It is weak related or without correlation.
10. early warning platform as claimed in claim 6, it is characterised in that: the hidden danger classification method are as follows: acquisition arranges incipient fault data
And data prediction is carried out to it: including classification selection, data balancing;Then it carries out word segmentation processing and utilizes deep learning method
Obtain term vector model;The hidden danger vector of every incipient fault data is calculated using term vector model, and then utilizes the training of hidden danger vector
Hidden danger disaggregated model is obtained, hidden danger classification is carried out by hidden danger disaggregated model;
Term vector model is a kind of efficient algorithm model that word is characterized as to real number value vector, the term vector generation method of use
For Word2vec;
The calculating of hidden danger vector: describing term vector to all hidden danger and be overlapped, thus obtain hidden danger describe corresponding hidden danger to
Amount;
Hidden danger disaggregated model is classified to incipient fault data: being classified using the C4.5 algorithm in decision Tree algorithms.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811574195.5A CN109636055A (en) | 2018-12-21 | 2018-12-21 | A kind of non-coal mine Safety Risk in Production prediction and warning platform |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811574195.5A CN109636055A (en) | 2018-12-21 | 2018-12-21 | A kind of non-coal mine Safety Risk in Production prediction and warning platform |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109636055A true CN109636055A (en) | 2019-04-16 |
Family
ID=66076466
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811574195.5A Pending CN109636055A (en) | 2018-12-21 | 2018-12-21 | A kind of non-coal mine Safety Risk in Production prediction and warning platform |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109636055A (en) |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110065866A (en) * | 2019-04-25 | 2019-07-30 | 永大电梯设备(中国)有限公司 | A kind of integrated forecasting system of the exception monitoring based on SMT |
CN110148290A (en) * | 2019-05-24 | 2019-08-20 | 烟台市牟金矿业有限公司 | Information-based big data system is supervised in the early warning of Intellisense Mine Safety in Production and prevention and control |
CN110751807A (en) * | 2019-10-23 | 2020-02-04 | 智洋创新科技股份有限公司 | Method for determining visual smoke foreign matter continuous alarm of power transmission line channel |
CN113095544A (en) * | 2021-03-09 | 2021-07-09 | 中国气象局公共气象服务中心(国家预警信息发布中心) | Marine information early warning method and device and electronic equipment |
CN113202560A (en) * | 2021-06-01 | 2021-08-03 | 桂林慧谷人工智能产业技术研究院 | Coal mine risk prevention and rescue decision-making system based on data mining |
CN113688169A (en) * | 2021-08-11 | 2021-11-23 | 北京科技大学 | Mine potential safety hazard identification and early warning system based on big data analysis |
CN115375137A (en) * | 2022-08-22 | 2022-11-22 | 中国安全生产科学研究院 | Safety risk early warning prediction system of non-coal mine mountain tailing mine base |
CN116363600A (en) * | 2023-06-01 | 2023-06-30 | 深圳恒邦新创科技有限公司 | Method and system for predicting maintenance operation risk of motor train unit |
CN116579601A (en) * | 2023-03-27 | 2023-08-11 | 中国安全生产科学研究院 | Mine safety production risk monitoring and early warning system and method |
CN116862703A (en) * | 2023-07-18 | 2023-10-10 | 中国安全生产科学研究院 | Multi-port non-coal mine safety monitoring information control system and method |
CN116911618A (en) * | 2023-09-07 | 2023-10-20 | 北京网藤科技有限公司 | Artificial intelligence decision-making system and method for safety production risk |
CN117391458A (en) * | 2023-12-08 | 2024-01-12 | 四川省寰宇众恒科技有限公司 | Safety production risk detection and early warning method and system based on data analysis |
CN117635219A (en) * | 2024-01-26 | 2024-03-01 | 长春黄金设计院有限公司 | Intelligent analysis system and method for big data of metal mine production |
CN117635219B (en) * | 2024-01-26 | 2024-04-26 | 长春黄金设计院有限公司 | Intelligent analysis system and method for big data of metal mine production |
Citations (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004033673A (en) * | 2002-06-21 | 2004-02-05 | Trustees Of The Univ Of Pennsylvania | Unified probability framework for predicting and detecting intracerebral stroke manifestation and multiple therapy device |
KR101210729B1 (en) * | 2012-03-06 | 2012-12-11 | 한국가스안전공사 | Predicting method and system of real-time risk sign for energy plants using intelligent risk sign pattern model |
CN103559551A (en) * | 2013-09-23 | 2014-02-05 | 北京中安健科安全技术咨询有限公司 | Production-enterprise-oriented potential safety hazard quantitative assessment and early warning system |
CN103647276A (en) * | 2013-12-10 | 2014-03-19 | 国家电网公司 | Electric energy quality early warning system and method thereof |
US20160019554A1 (en) * | 2014-07-15 | 2016-01-21 | Emily M. MacDonald-Korth | Standard System and Method for Assigning Ratings to Art Materials and Works of Art Based on the Projected Stability of the Constituents |
CN105678446A (en) * | 2015-12-31 | 2016-06-15 | 浙江图讯科技股份有限公司 | Method used for enterprise safety production risk early warning |
CN106194263A (en) * | 2016-08-29 | 2016-12-07 | 中煤科工集团重庆研究院有限公司 | Coal mine gas disaster monitoring early-warning system and method for early warning |
CN106682394A (en) * | 2016-11-30 | 2017-05-17 | 北京拓明科技有限公司 | Big data analyzing method and system of survival risk |
US20170337478A1 (en) * | 2016-05-22 | 2017-11-23 | Microsoft Technology Licensing, Llc | Self-Learning Technique for Training a PDA Component and a Simulated User Component |
CN107503797A (en) * | 2017-08-25 | 2017-12-22 | 合肥明信软件技术有限公司 | Mine Methane tendency early warning system based on 3D emulation platforms |
CN107728234A (en) * | 2017-09-17 | 2018-02-23 | 北京工业大学 | A kind of intensity of lightning value Forecasting Methodology based on atmospheric electric field data |
CN108257673A (en) * | 2018-01-12 | 2018-07-06 | 南通大学 | Risk value Forecasting Methodology and electronic equipment |
CN108447565A (en) * | 2018-03-23 | 2018-08-24 | 北京工业大学 | A kind of small for gestational age infant disease forecasting method based on improvement noise reduction autocoder |
CN108510146A (en) * | 2017-12-28 | 2018-09-07 | 国家安全生产监督管理总局通信信息中心 | Safety of Coal Mine Production method for prewarning risk and system |
AU2018101514A4 (en) * | 2018-10-11 | 2018-11-15 | Chi, Henan Mr | An automatic text-generating program for Chinese Hip-hop lyrics |
CN108959329A (en) * | 2017-05-27 | 2018-12-07 | 腾讯科技(北京)有限公司 | A kind of file classification method, device, medium and equipment |
CN108981785A (en) * | 2018-06-19 | 2018-12-11 | 江苏高远智能科技有限公司 | A kind of intelligent Detection of coal breaker equipment safety |
CN109034612A (en) * | 2018-07-24 | 2018-12-18 | 山西精英科技股份有限公司 | A kind of security risk diagnostic method based on coal mine early warning analysis Yu prevention and control system |
-
2018
- 2018-12-21 CN CN201811574195.5A patent/CN109636055A/en active Pending
Patent Citations (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004033673A (en) * | 2002-06-21 | 2004-02-05 | Trustees Of The Univ Of Pennsylvania | Unified probability framework for predicting and detecting intracerebral stroke manifestation and multiple therapy device |
KR101210729B1 (en) * | 2012-03-06 | 2012-12-11 | 한국가스안전공사 | Predicting method and system of real-time risk sign for energy plants using intelligent risk sign pattern model |
CN103559551A (en) * | 2013-09-23 | 2014-02-05 | 北京中安健科安全技术咨询有限公司 | Production-enterprise-oriented potential safety hazard quantitative assessment and early warning system |
CN103647276A (en) * | 2013-12-10 | 2014-03-19 | 国家电网公司 | Electric energy quality early warning system and method thereof |
US20160019554A1 (en) * | 2014-07-15 | 2016-01-21 | Emily M. MacDonald-Korth | Standard System and Method for Assigning Ratings to Art Materials and Works of Art Based on the Projected Stability of the Constituents |
CN105678446A (en) * | 2015-12-31 | 2016-06-15 | 浙江图讯科技股份有限公司 | Method used for enterprise safety production risk early warning |
US20170337478A1 (en) * | 2016-05-22 | 2017-11-23 | Microsoft Technology Licensing, Llc | Self-Learning Technique for Training a PDA Component and a Simulated User Component |
CN106194263A (en) * | 2016-08-29 | 2016-12-07 | 中煤科工集团重庆研究院有限公司 | Coal mine gas disaster monitoring early-warning system and method for early warning |
CN106682394A (en) * | 2016-11-30 | 2017-05-17 | 北京拓明科技有限公司 | Big data analyzing method and system of survival risk |
CN108959329A (en) * | 2017-05-27 | 2018-12-07 | 腾讯科技(北京)有限公司 | A kind of file classification method, device, medium and equipment |
CN107503797A (en) * | 2017-08-25 | 2017-12-22 | 合肥明信软件技术有限公司 | Mine Methane tendency early warning system based on 3D emulation platforms |
CN107728234A (en) * | 2017-09-17 | 2018-02-23 | 北京工业大学 | A kind of intensity of lightning value Forecasting Methodology based on atmospheric electric field data |
CN108510146A (en) * | 2017-12-28 | 2018-09-07 | 国家安全生产监督管理总局通信信息中心 | Safety of Coal Mine Production method for prewarning risk and system |
CN108257673A (en) * | 2018-01-12 | 2018-07-06 | 南通大学 | Risk value Forecasting Methodology and electronic equipment |
CN108447565A (en) * | 2018-03-23 | 2018-08-24 | 北京工业大学 | A kind of small for gestational age infant disease forecasting method based on improvement noise reduction autocoder |
CN108981785A (en) * | 2018-06-19 | 2018-12-11 | 江苏高远智能科技有限公司 | A kind of intelligent Detection of coal breaker equipment safety |
CN109034612A (en) * | 2018-07-24 | 2018-12-18 | 山西精英科技股份有限公司 | A kind of security risk diagnostic method based on coal mine early warning analysis Yu prevention and control system |
AU2018101514A4 (en) * | 2018-10-11 | 2018-11-15 | Chi, Henan Mr | An automatic text-generating program for Chinese Hip-hop lyrics |
Non-Patent Citations (1)
Title |
---|
杨晔: "基于行为的恶意代码检测方法研究", no. 03, pages 138 - 213 * |
Cited By (22)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110065866A (en) * | 2019-04-25 | 2019-07-30 | 永大电梯设备(中国)有限公司 | A kind of integrated forecasting system of the exception monitoring based on SMT |
CN110148290A (en) * | 2019-05-24 | 2019-08-20 | 烟台市牟金矿业有限公司 | Information-based big data system is supervised in the early warning of Intellisense Mine Safety in Production and prevention and control |
CN110751807A (en) * | 2019-10-23 | 2020-02-04 | 智洋创新科技股份有限公司 | Method for determining visual smoke foreign matter continuous alarm of power transmission line channel |
CN110751807B (en) * | 2019-10-23 | 2021-09-07 | 智洋创新科技股份有限公司 | Method for determining visual smoke foreign matter continuous alarm of power transmission line channel |
CN113095544A (en) * | 2021-03-09 | 2021-07-09 | 中国气象局公共气象服务中心(国家预警信息发布中心) | Marine information early warning method and device and electronic equipment |
CN113095544B (en) * | 2021-03-09 | 2024-04-16 | 中国气象局公共气象服务中心(国家预警信息发布中心) | Marine information early warning method and device and electronic equipment |
CN113202560A (en) * | 2021-06-01 | 2021-08-03 | 桂林慧谷人工智能产业技术研究院 | Coal mine risk prevention and rescue decision-making system based on data mining |
CN113688169B (en) * | 2021-08-11 | 2023-08-08 | 北京科技大学 | Mine potential safety hazard identification and early warning system based on big data analysis |
CN113688169A (en) * | 2021-08-11 | 2021-11-23 | 北京科技大学 | Mine potential safety hazard identification and early warning system based on big data analysis |
CN115375137A (en) * | 2022-08-22 | 2022-11-22 | 中国安全生产科学研究院 | Safety risk early warning prediction system of non-coal mine mountain tailing mine base |
CN116579601A (en) * | 2023-03-27 | 2023-08-11 | 中国安全生产科学研究院 | Mine safety production risk monitoring and early warning system and method |
CN116579601B (en) * | 2023-03-27 | 2024-03-05 | 中国安全生产科学研究院 | Mine safety production risk monitoring and early warning system and method |
CN116363600B (en) * | 2023-06-01 | 2023-08-01 | 深圳恒邦新创科技有限公司 | Method and system for predicting maintenance operation risk of motor train unit |
CN116363600A (en) * | 2023-06-01 | 2023-06-30 | 深圳恒邦新创科技有限公司 | Method and system for predicting maintenance operation risk of motor train unit |
CN116862703B (en) * | 2023-07-18 | 2024-01-23 | 中国安全生产科学研究院 | Multi-port non-coal mine safety monitoring information control system and method |
CN116862703A (en) * | 2023-07-18 | 2023-10-10 | 中国安全生产科学研究院 | Multi-port non-coal mine safety monitoring information control system and method |
CN116911618B (en) * | 2023-09-07 | 2023-12-05 | 北京网藤科技有限公司 | Artificial intelligence decision-making system and method for safety production risk |
CN116911618A (en) * | 2023-09-07 | 2023-10-20 | 北京网藤科技有限公司 | Artificial intelligence decision-making system and method for safety production risk |
CN117391458B (en) * | 2023-12-08 | 2024-02-27 | 四川省寰宇众恒科技有限公司 | Safety production risk detection and early warning method and system based on data analysis |
CN117391458A (en) * | 2023-12-08 | 2024-01-12 | 四川省寰宇众恒科技有限公司 | Safety production risk detection and early warning method and system based on data analysis |
CN117635219A (en) * | 2024-01-26 | 2024-03-01 | 长春黄金设计院有限公司 | Intelligent analysis system and method for big data of metal mine production |
CN117635219B (en) * | 2024-01-26 | 2024-04-26 | 长春黄金设计院有限公司 | Intelligent analysis system and method for big data of metal mine production |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109636055A (en) | A kind of non-coal mine Safety Risk in Production prediction and warning platform | |
CN105589806B (en) | A kind of software defect tendency Forecasting Methodology based on SMOTE+Boosting algorithms | |
CN108447534A (en) | A kind of electronic health record data quality management method based on NLP | |
Zhang et al. | A novel case adaptation method based on an improved integrated genetic algorithm for power grid wind disaster emergencies | |
CN108536123A (en) | The method for diagnosing faults of the train control on board equipment of the long neural network of memory network combination in short-term | |
CN109800875A (en) | Chemical industry fault detection method based on particle group optimizing and noise reduction sparse coding machine | |
CN109657947A (en) | A kind of method for detecting abnormality towards enterprises ' industry classification | |
Kirange et al. | Sentiment Analysis of news headlines for stock price prediction | |
CN113222140B (en) | C4.5 algorithm and BP neuron-based power distribution network fault auxiliary decision-making method | |
CN109559045A (en) | A kind of method and system of personnel's intelligence control | |
Dimitriou et al. | Fuzzy modeling of freeway accident duration with rainfall and traffic flow interactions | |
CN106127242A (en) | Year of based on integrated study Extreme Precipitation prognoses system and Forecasting Methodology thereof | |
Ayon | Neural network based software defect prediction using genetic algorithm and particle swarm optimization | |
Deo et al. | A white-box SVM framework and its swarm-based optimization for supervision of toothed milling cutter through characterization of spindle vibrations | |
CN109101584A (en) | A kind of sentence classification improved method combining deep learning with mathematical analysis | |
Li et al. | Risk analysis of dam based on artificial bee colony algorithm with fuzzy c-means clustering | |
Hatim et al. | Addressing challenges and demands of intelligent seasonal rainfall forecasting using artificial intelligence approach | |
CN104732067A (en) | Industrial process modeling forecasting method oriented at flow object | |
Wang et al. | A new approach of obtaining reservoir operation rules: Artificial immune recognition system | |
CN104966106A (en) | Biological age step-by-step predication method based on support vector machine | |
CN117349782B (en) | Intelligent data early warning decision tree analysis method and system | |
CN107066468A (en) | A kind of case search method based on genetic algorithm and nearest neighbor algorithm | |
Sharma et al. | Classification algorithms on a large continuous random dataset using rapid miner tool | |
CN110335160A (en) | A kind of medical treatment migratory behaviour prediction technique and system for improving Bi-GRU based on grouping and attention | |
CN113743461B (en) | Unmanned aerial vehicle cluster health degree assessment method and device |
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 |