The content of the invention
The present invention is in order to solve problems of the prior art, there is provided a kind of industry control signal detection system.
In order to realize above-mentioned purpose, the technical scheme is that:A kind of industry control signal detection system, including:
Data acquisition module, the real-time status data of collecting device electrical characteristic;
And it is connected to store the history for the real-time status data that data collecting module collected arrives with data acquisition module
Process database;
Also include be connected with historical process data storehouse and set up electric spy by the data in historical process data storehouse
Property in signal input and output corresponding relation signal process forecast model;The signal process forecast model is according to data acquisition
The real-time input signal collected in module, obtain predicting output signal, and the prediction output signal is sent to
Traffic filter, the traffic filter receive prediction output signal and number from signal process forecast model
The real time output that is collected according to acquisition module is simultaneously compared, if corresponding real time output and prediction output signal feature
It is consistent, then regards as effectively can perform signal;It is on the contrary then regard as it is invalid refusal perform signal.
Preferably, the data acquisition module includes:
Digital data acquisition submodule, gather the digital information of industry control signal;
Analog acquisition submodule, gather the analog quantity information of industry control signal;
Logical signal gathers submodule, gathers the data message in industry control signal logic bus.
Preferably, in addition to traffic filter the accident analysis judgment models being connected.
Preferably, the signal process forecast model is established according to neutral net, SVMs or PCA methods.
Present invention also offers a kind of detection method of industry control signal, comprise the following steps:
Step S1:The real-time status data of data collecting module collected equipment electrical characteristic;And it is stored in historical process data
Storehouse;
Step S2:Using the data in historical process data storehouse as foundation, it is established that signal process forecast model;
Step S3:Input signal to be checked in real time is sent to signal process forecast model, and exports prediction output signal;
Step S4:Traffic filter will predict that output signal feature carries out uniformity contrast with real time output feature;
If corresponding real time output is consistent with prediction output signal feature, regard as effectively can perform signal;It is if corresponding defeated in real time
Go out signal and prediction output signal feature is inconsistent, then regard as invalid refusal and perform signal.
Preferably, also include after step s4:
Step S5:Invalid refusal is performed into signal and is uploaded to accident analysis judgment models, accident analysis judges diagnostic module
Accident analysis occurs and carried out for failure judgement.
Preferably, in the step S1, digital information, simulation in data collecting module collected equipment electrical characteristic
Data message in amount information, logic bus.
The detecting system and detection method of the present invention, the final output work(of subsystems from each industrial control system
It can detect and set out, by off-line training, believable one is carried out to the real time output and prediction output signal for being detected examining system
The contrast of cause property, system invasion and failure as caused by information security are found in time.Fundamentally information security is prevented to cause work
Industry control system safety, change the industrial control information traditional method puted prevention first safely, it is anti-with the result of function accuracy
To the method for backstepping, fundamentally solves the information security issue of industrial control system.
Embodiment
In order that present invention solves the technical problem that, use technical scheme, obtain technique effect it can be readily appreciated that below
With reference to specific accompanying drawing, the embodiment of the present invention is described further.
With reference to figure 1, the invention provides a kind of industry control signal detection system, including:Data acquisition module 1, for gathering
The real-time status data of equipment electrical characteristic;The real-time shape of the electrical characteristic of hardware device in the data acquisition module acquisition system
State, the main voltage including equipment, electric current, frequency etc., these values can be acquired by PLC or DCS system, it is only necessary to
Train control system provides acquisition interface can.Currently preferred to be, the data acquisition module 1 includes digital data acquisition submodule
Block 10, analog acquisition submodule 11 and logical signal collection submodule 12.Wherein, digital data acquisition submodule 10 gathers number
Word amount information, such as switch closed condition;Analog acquisition submodule 11 gathers analog quantity information, such as voltage, electric current;Logic
Data message on the data acquisition bus of signal acquisition submodule 12 etc..
The industry control signal detection system of the present invention also includes being connected to store data acquisition module with data acquisition module 1
The historical process data storehouse 2 for the real-time status data that collection 1 is arrived;Data basis is provided for the foundation of following model.Also include with
It is that historical process data storehouse 2 connects and by the data in historical process data storehouse 2 set up in electrical characteristic signal input and
Export the signal process forecast model 3 of corresponding relation.
These data and output collected have corresponding relation, in this case, as long as providing substantial amounts of history number
According to signal process prediction module carries out off-line training modeling to it, can find out the relation between wherein input and output, use y=f
(x) represent, wherein y represents output quantity;X represents input variable.Y may have multiple output quantity y1, y2...ym;X may also have
Multiple input quantity x1, x2...xn.
By different types of historical data source, continuous, discrete, batch production process can be analyzed with rapid build model.
Data preparation, visual check data can be carried out, rule model can also be established.Show that the influence between parameters is closed
It is to predict the value of inconvenient measurement parameter, the offline or online optimal control of implementation, the knowledge excavated from these models,
Help is assessed the improved income of process, and whole process is also very simple and convenient, workload is seldom.Use this model system
System, the reason for production process fluctuates can be found out, and be adjusted, reached stabilized product quality, improve the target of yield.
The signal process forecast model 3 that input and output correspond to relation can be built by existing technology.Such as basis
Neutral net, SVMs or PCA methods.
Artificial neural network (Artificial Neural Networks, be abbreviated as ANNs) is also referred to as neutral net
(NNs) or link model (Connection Model) is made, it is a kind of imitation animal nerve network behavior feature, is divided
The algorithm mathematics model of cloth parallel information processing.This network relies on the complexity of system, by adjusting internal a large amount of sections
The relation being connected with each other between point, so as to reach the purpose of processing information.
The algorithm that neutral net is used is exactly vector multiplication, and widely used sign function and its various is approached.Parallel,
It is fault-tolerant, can be realized with hardware and self-teaching characteristic, be several principal advantages of neutral net, and neural computing side
Where the difference of method and conventional method.
Support vector machine method is built upon in the VC dimensions theory and Structural risk minization basis of Statistical Learning Theory
, according to limited sample information is in the complexity (the study precision i.e. to specific training sample) of model and learning ability is (i.e.
Identify the ability of arbitrary sample without error) between seek optimal compromise, in the hope of obtaining best Generalization Ability.
SVMs will be in DUAL PROBLEMS OF VECTOR MAPPING to the space of a more higher-dimension, and being established in this space has between a maximum
Every hyperplane.Two hyperplane parallel to each other are had on the both sides of the hyperplane of separated data.Direction is established suitably to separate
Hyperplane maximizes the distance between two hyperplane parallel with it.It is assumed to, distance or gap between parallel hyperplane
Bigger, the overall error of grader is smaller.
Principal component analysis (Principal Component Analysis, PCA) or pivot analysis.It is a kind of grasp thing
The statistical analysis technique of thing principal contradiction, it can parse major influence factors from polynary things, disclose the essence of things,
Simplify the problem of complicated.The purpose for calculating principal component is to project to high dimensional data compared with lower dimensional space.Give m of n variable
Observed value, forms n ' m data matrix, and n is generally bigger.If the main aspect of things is just embodied in several main
On variable, we only need to come out these separating variables, carry out labor.But in general, can not be straight
Connect and find out such key variables.At this moment we can represent the main aspect of things, PCA with the linear combination of original variable
It is exactly such a analysis method.
The industry control signal detection system of the present invention, the signal process forecast model 3 gather according in data acquisition module 1
The real-time input signal arrived, obtain predicting output signal, and the prediction output signal is sent to traffic filter 4, the letter
Number wave filter 4 receives prediction output signal from signal process forecast model 3 and data acquisition module 1 collect it is real-time
Output signal is simultaneously compared, if corresponding real time output is consistent with prediction output signal feature, regards as effectively holding
Row signal;It is on the contrary then regard as it is invalid refusal perform signal.
Accident analysis judgment models 5 that are currently preferred to be, in addition to being connected with traffic filter 4, when signal filters
When device 4 judges that real time output is not consistent with prediction output signal feature, the signal is transferred to accident analysis judgment models 5
In, accident analysis occurs and carried out for the failure judgement of accident analysis judgment models 5.Accident analysis judgment models 5 are carried out to data
Analysis, establishes operational model using neural Meta algorithm, finds out the various factors related to failure.Failure can be reported after occurring
Alert, user defines the rank of alarm according to the threshold value of data.Alert levels are divided into critical, advanced, intermediate, rudimentary, normal
State and unknown accident.
Present invention also offers a kind of detection method of industry control signal, with reference to figure 2, comprise the following steps:
Step S1:The real-time status data of data collecting module collected equipment electrical characteristic;And it is stored in historical process data
Storehouse;
Step S2:Using the data in historical process data storehouse as foundation, it is established that signal process forecast model;
Step S3:Input signal to be checked in real time is sent to signal process forecast model, and exports prediction output signal;
Step S4:Traffic filter will predict that output signal feature carries out uniformity contrast with real time output feature;
If corresponding real time output is consistent with prediction output signal feature, regard as effectively can perform signal;It is if corresponding defeated in real time
Go out signal and prediction output signal feature is inconsistent, then regard as invalid refusal and perform signal.
Preferably, also include after step s4:
Step S5:Invalid refusal is performed into signal and is uploaded to accident analysis judgment models, accident analysis judges diagnostic module
Accident analysis occurs and carried out for failure judgement.
Preferably, in the step S1, digital information, simulation in data collecting module collected equipment electrical characteristic
Data message in amount information, logic bus.
The detecting system and detection method of the present invention, the final output work(of subsystems from each industrial control system
It can detect and set out, by off-line training, believable one is carried out to the real time output and prediction output signal for being detected examining system
The contrast of cause property, system invasion and failure as caused by information security are found in time.Fundamentally information security is prevented to cause work
Industry control system safety, change the industrial control information traditional method puted prevention first safely, it is anti-with the result of function accuracy
To the method for backstepping, fundamentally solves the information security issue of industrial control system.
The present invention has carried out detailed explanation by preferred embodiment.However, by studying carefully above, to each
The change and increase of embodiment are obvious for those of ordinary skill in the art.It is intended that institute
These changes and increase having all fall in the scope that the claims in the present invention are protected.