CN103473439A - Early failure prediction methodorientedtoComplex electrochemical equipment low signal-noise ratio information - Google Patents
Early failure prediction methodorientedtoComplex electrochemical equipment low signal-noise ratio information Download PDFInfo
- Publication number
- CN103473439A CN103473439A CN2013103597503A CN201310359750A CN103473439A CN 103473439 A CN103473439 A CN 103473439A CN 2013103597503 A CN2013103597503 A CN 2013103597503A CN 201310359750 A CN201310359750 A CN 201310359750A CN 103473439 A CN103473439 A CN 103473439A
- Authority
- CN
- China
- Prior art keywords
- data
- rough
- noise ratio
- low signal
- ratio information
- 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.)
- Granted
Links
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to an early failure prediction methodorientedto complex electrochemical equipment low signal-noise ratio information. The method comprises the following steps: (1) acquiring vibration data, acoustic emission data, cutting force data, noise data and temperature data; (2) respectively performing discretization on continuous related data by using a hierarchical clustering algorithm to obtain N groups of discrete data; (3) forming a decision table DT from each group of discrete data after discretization as an input layer Xi(t) of a rough function adaptive method; (4) performing prediction analysis on the decision table DT by using the rough function adaptive method to obtain an optimal target prediction model; (5) taking the obtained optimal target prediction model as a precise prediction model of the moment and displaying a prediction value of a next moment through monitoring equipment so as to realize early failure prediction of the complex electrochemical equipment low signal-noise ratio information. According to the method, the early failure prediction of the low signal-noise ratio information can be realized; the method can be widely applied to complex electrochemical equipment.
Description
Technical field
The present invention relates to a kind of failure prediction method of complicated electromechanical equipment, particularly about a kind of fault forecast method towards complicated electromechanical equipment low signal-to-noise ratio information.
Background technology
High-grade turning center is as typical complicated electromechanical equipment, become one of manufacturing main production equipments of modernization, can realize the disposable clamping of processing work, and outfit automatic tool changer, reduce positioning error and lay time, made the machining precision of workpiece and working (machining) efficiency greatly improve.Along with coming into operation of high-grade turning center, the research of machine failure forecasting techniques starts to come into one's own.At present, the failure prediction technology is mainly used in the aspects such as power equipment, large rotating machinery, for numerical control device, relates to less.There are maximization, integrated, precise treatment and the characteristics such as intelligent at aspects such as mechanism, functions due to high-grade turning center; make and usually can run into full accuracy index deficiency in process; precision can not be guaranteed for a long time; the problems such as failure rate height, seriously restricted the numerically-controlled machine effect of bringing into normal play.Carrying out the research of diagnosing faults of numerical control machine forecasting techniques is one of guarantee numerically-controlled machine reliability service, the state-of-the-art technology that improves lathe military service performance and core technology, is also the focal issue of studying both at home and abroad.The extensive application of high-grade, digitally controlled machine tools and the deficiency of diagnosis and maintenance technology have caused huge lathe early warning and diagnostic requirements at present, become one of bottleneck of current machine tool technology development.
Mainly contain linear prediction method, non-linear prediction method and neural network prediction in traditional failure prediction method.Be mainly by adopting correct signal processing method to extract different fault signatures, then these features being carried out to state recognition, this is the key element that improves fault pre-alarming and performance evaluation accuracy.For example numerically-controlled machine adds and carries out slice the dint to monitor man-hour, and utilizes off-line data to be trained the BP neural network, makes the tool wear prediction accuracy reach more than 97%.Such Forecasting Methodology has obtained satisfied effect to a certain extent.Carry out monitoring, diagnosing but utilize this method to carry out the electromechanical equipment that high-grade turning center etc. has a highly flexible mechanism, tend to run into the situation that quantity of information is few, noise is large in the collection signal process, had a strong impact on the precision of Fault diagnosis and forecast.Simultaneously, because the signal gathered often has the characteristics such as intermittence, ambiguity, time variation, cause the uncertain factor of failure prediction to increase, and it is very large that processing operating mode variation makes in process random factors affect meeting, causes the accuracy of failure prediction and the confidence level of conclusion all to descend to some extent.
Summary of the invention
For the problems referred to above, the purpose of this invention is to provide a kind of fault forecast method towards complicated electromechanical equipment low signal-to-noise ratio information, the method can realize the fault forecast to low signal-to-noise ratio information.
For achieving the above object, the present invention takes following technical scheme: a kind of fault forecast method towards complicated electromechanical equipment low signal-to-noise ratio information, and it comprises the following steps: continuous and vibration data equipment operation condition, acoustic emission data, cutting force data, noise data and the temperature data that 1) by the on-line monitoring center, obtain a series of long courses that can represent equipment operation condition; 2) utilize hierarchical clustering algorithm that the whole continuous datas that obtain in step 1) are carried out respectively to the discretize processing, obtain N group discrete data; Wherein, N is natural number; 3) every group of discrete data after discretize formed to decision table DT, and the input layer X using decision table DT as the rough function adaptive approach
i(t), i=1,2,3 ... n; Wherein n means number of probes; 4) utilize the rough function adaptive approach to carry out forecast analysis to decision table DT, obtain characterizing the complicated electromechanical equipment running status optimal objective forecast model of development trend in the future; 5) the accurately predicting model as this moment according to the optimal objective forecast model obtained, and show next predicted value constantly by monitoring equipment, realize the fault forecast to complicated electromechanical equipment low signal-to-noise ratio information.
In described step 4), the rough function adaptive approach step of utilization when told decision table DT is carried out to forecast analysis is as follows: (1) objective definition rough function F (M) is:
F(M)＝f(m,w,θ)＝R[f(m|θ)×f(ω|θ)]，
Wherein, M is the optimal objective forecast model; M, for prediction rough function model, is the list entries after discretize; R means the coarse relation of f (m) * f (ω), in difference, connects under weights θ and calculates corresponding rough function value; (2) the input sample is: X
i(t)=(X
1(t), X
2(t) ..., X
n(t)), X
i(t) be every group of data after discretize, and according to the upper and lower approximate definition of rough set to X
i(t) carry out upper and lower approximate treatment, obtain approximate data
with lower approximate data
t is the sampling time; (3) take off approximate data
calculate rough function, and suppose two tolerance
with
and real function f:R
t→ R
m, R wherein
t=[0t,,
real function f is f for the rough function of tolerance d and e
*(n)=e
*(f (x)),
(4) any given connection weights θ
i(t), its scope is 0<θ
i(t)<1, by with rough function f
*(n) multiply each other and obtain output layer
j=0,1,2 ..., n; (5) revise and connect weights θ
i(t) make prediction accuracy w reach more than 85%,
=0,1,2 ..., n; Wherein, prediction accuracy is
Y={X
i(t) | i=1,2 ..., n} is the list entries after discretize;
for predicted data; η ∈ (0,1) is learning rate; (6) return to step (2), repeat, utilize the input sample data constantly to regulate and connect weights θ
i(t) size, make prediction accuracy w reach predefined standard, until connect weights θ
i(t) to all sample standard deviations stablize constant till, finally obtain target rough function F (M)=f (m, w, θ), and then obtain optimal objective forecast model M.
The present invention is owing to taking above technical scheme, it has the following advantages: 1, the present invention, owing to adopting a kind of rough function adaptive forecasting method that is suitable for analyzing Low SNR signal, has avoided the generation of unsteady phenomenas such as lacking because quantity of information appears in process status, noise is large.2, the present invention is owing to by regulating rough function, connecting weights θ
i(t) can adaptation equipment the variation of processing operating mode, reduced some uncertain factors impacts of failure prediction, improved the precision of failure prediction and the confidence level of conclusion.3, the failure prediction method based on rough function provided by the invention, solved because the signal gathered has ambiguity and the time variation characteristics cause implicit data to be difficult to characterize the problem of failure symptom information.The present invention can extensively apply in complicated electromechanical equipment.
The accompanying drawing explanation
Fig. 1 is overall flow schematic diagram of the present invention.
Embodiment
Below in conjunction with drawings and Examples, the present invention is described in detail.
As shown in Figure 1, the present invention is based on rough function towards the fault forecast method of complicated electromechanical equipment low signal-to-noise ratio information and carries out failure prediction, and it comprises the following steps:
1) obtain continuous and equipment operation condition related data, such as vibration data, acoustic emission data, cutting force data, noise data and temperature data etc. of a series of long courses of the equipment operation condition that can represent axis system, tooling system, kinematic train, feed system and electrical system by existing on-line monitoring center;
2) utilize hierarchical clustering algorithm that the whole serial correlation data that obtain in step 1) are carried out respectively to the discretize processing, obtain N group discrete data; Wherein, N is natural number;
3) every group of discrete data after discretize formed to decision table DT, and the input layer X using decision table DT as the rough function adaptive approach
i(t), i=1,2,3 ... n; Wherein n means number of probes;
4) utilize the rough function adaptive approach to carry out forecast analysis to decision table DT, obtain characterizing complicated electromechanical equipment (for example high-grade turning center) the running status optimal objective forecast model of development trend in the future;
The rough function adaptive approach step of utilization while wherein, decision table DT being carried out to forecast analysis is as follows:
(1) objective definition rough function F (M) is: F (M)=f (m, w, θ)=R[f (m| θ) * f (ω | θ)], wherein, M is the optimal objective forecast model; M, for prediction rough function model, is the list entries after discretize; R means the coarse relation of f (m) * f (ω), in difference, connects under weights θ and calculates corresponding rough function value.
(2) the input sample is: X
i(t)=(X
1(t), X
2(t) ..., X
n(t)), X
i(t) be every group of data after discretize, and according to the upper and lower approximate definition of rough set to X
i(t) carry out upper and lower approximate treatment, obtain approximate data
with lower approximate data
t is the sampling time;
(3) take off approximate data
calculate rough function, and suppose two tolerance
with
and real function f:R
t→ R
m, R wherein
t=[0, t],
real function f is f for the rough function of tolerance d and e
*(n)=e
*(f (x)),
(4) any given connection weights θ
i(t), its scope is 0<θ
i(t)<1.By with rough function f
*(n) multiply each other and can obtain output layer
j=0,1,2 ..., n;
(5) revise and connect weights θ
i(t) make prediction accuracy w reach more than 85%,
i=0,1,2,…,n。
Wherein, prediction accuracy
y={X
i(t) | i=1,2 ..., n} is the list entries after discretize;
for predicted data; η ∈ (0,1) is learning rate, also referred to as Learning Step, for controlling erection rate, if η too conference affect θ
i(t) stablize, the too little meeting of η makes θ
i(t) speed of convergence is too slow.
(6) return to step (2), repeat, utilize the input sample data constantly to regulate and connect weights θ
i(t) size, make prediction accuracy w reach predefined standard, until connect weights θ
i(t) to all sample standard deviations stablize constant till, prediction accuracy w meets in the specified value scope, finally obtains target rough function F (M) to be: F (M)=f (m, w, θ), and then obtain optimal objective forecast model M.Wherein, M is the optimal objective forecast model; M is prediction rough function model.
5) the accurately predicting model as this moment according to the optimal objective forecast model M obtained, and show next predicted value constantly by existing monitoring equipment, realize the fault forecast to complicated electromechanical equipment low signal-to-noise ratio information.
The various embodiments described above are only for illustrating the present invention; the connection of each parts and structure all can change to some extent; on the basis of technical solution of the present invention; all improvement and equivalents of connection and the structure of indivedual parts being carried out according to the principle of the invention, all should not get rid of outside protection scope of the present invention.
Claims (2)
1. the fault forecast method towards complicated electromechanical equipment low signal-to-noise ratio information, it comprises the following steps:
1) obtain continuous and vibration data equipment operation condition, acoustic emission data, cutting force data, noise data and the temperature data of a series of long courses that can represent equipment operation condition by the on-line monitoring center;
2) utilize hierarchical clustering algorithm that the whole continuous datas that obtain in step 1) are carried out respectively to the discretize processing, obtain N group discrete data; Wherein, N is natural number;
3) every group of discrete data after discretize formed to decision table DT, and the input layer X using decision table DT as the rough function adaptive approach
i(t), i=1,2,3 ... n; Wherein n means number of probes;
4) utilize the rough function adaptive approach to carry out forecast analysis to decision table DT, obtain characterizing the complicated electromechanical equipment running status optimal objective forecast model of development trend in the future;
5) the accurately predicting model as this moment according to the optimal objective forecast model obtained, and show next predicted value constantly by monitoring equipment, realize the fault forecast to complicated electromechanical equipment low signal-to-noise ratio information.
2. a kind of fault forecast method towards complicated electromechanical equipment low signal-to-noise ratio information as claimed in claim 1, it is characterized in that: in described step 4), the rough function adaptive approach step of utilization when told decision table DT is carried out to forecast analysis is as follows:
(1) objective definition rough function F (M) is: F (M)=f (m, w, θ)=R[f (m| θ) * f (ω | θ)], wherein, M is the optimal objective forecast model; M, for prediction rough function model, is the list entries after discretize; R means the coarse relation of f (m) * f (ω), in difference, connects under weights θ and calculates corresponding rough function value;
(2) the input sample is: X
i(t)=(X
1(t), X
2(t) ..., X
n(t)), X
i(t) be every group of data after discretize, and according to the upper and lower approximate definition of rough set to X
i(t) carry out upper and lower approximate treatment, obtain approximate data
with lower approximate data
t is the sampling time;
(3) take off approximate data
calculate rough function, and suppose two tolerance
with
and real function f:R
t→ R
m, R wherein
t=[0, t],
real function f is f for the rough function of tolerance d and e
*(n)=e
*(f (x)),
(4) any given connection weights θ
i(t), its scope is 0<θ
i(t)<1, by with rough function f
*(n) multiply each other and obtain output layer
j=0,1,2 ..., n;
(5) revise and connect weights θ
i(t) make prediction accuracy w reach more than 85%,
i=0,1,2,…,n；
Wherein, prediction accuracy
y={X
i(t) | i=1,2 ..., n} is the list entries after discretize;
for predicted data; η ∈ (0,1) is learning rate;
(6) return to step (2), repeat, utilize the input sample data constantly to regulate and connect weights θ
i(t) size, make prediction accuracy w reach predefined standard, until connect weights θ
i(t) to all sample standard deviations stablize constant till, finally obtain target rough function F (M)=f (m, w, θ), and then obtain optimal objective forecast model M.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310359750.3A CN103473439B (en) | 2013-08-17 | 2013-08-17 | A kind of fault forecast method towards complicated electromechanical equipment low signal-to-noise ratio information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310359750.3A CN103473439B (en) | 2013-08-17 | 2013-08-17 | A kind of fault forecast method towards complicated electromechanical equipment low signal-to-noise ratio information |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103473439A true CN103473439A (en) | 2013-12-25 |
| CN103473439B CN103473439B (en) | 2016-04-06 |
Family
ID=49798286
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310359750.3A Active CN103473439B (en) | 2013-08-17 | 2013-08-17 | A kind of fault forecast method towards complicated electromechanical equipment low signal-to-noise ratio information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103473439B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117168865A (en) * | 2023-11-03 | 2023-12-05 | 南通宝佳精密机械有限公司 | Protection system for electromechanical equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101770219A (en) * | 2010-01-29 | 2010-07-07 | 北京信息科技大学 | Knowledge acquisition method of fault diagnosis knowledge base of turn-milling combined machine tool |
| CN102736562A (en) * | 2012-07-10 | 2012-10-17 | 北京信息科技大学 | Knowledge base construction method oriented to fault diagnosis and fault prediction of numerical control machine tool |
-
2013
- 2013-08-17 CN CN201310359750.3A patent/CN103473439B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101770219A (en) * | 2010-01-29 | 2010-07-07 | 北京信息科技大学 | Knowledge acquisition method of fault diagnosis knowledge base of turn-milling combined machine tool |
| CN102736562A (en) * | 2012-07-10 | 2012-10-17 | 北京信息科技大学 | Knowledge base construction method oriented to fault diagnosis and fault prediction of numerical control machine tool |
Non-Patent Citations (4)
| Title |
|---|
| ANANTHANARAYANA V S，ET AL.: "Tree structure for efficient data mining using rough sets", 《PATTERN RECOGNITION LETTERS》 * |
| 于达仁，等.: "融合粗糙集和模糊聚类的连续数据知识发现", 《中国电机工程学报》 * |
| 姚鑫骅，等.: "基于粗糙集理论的数控机床智能故障诊断研究", 《浙江大学学报（工学版）》 * |
| 牛东晓，等.: "基于粗糙集和决策树的自适应神经网络短期负荷预测方法", 《电力自动化设备》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117168865A (en) * | 2023-11-03 | 2023-12-05 | 南通宝佳精密机械有限公司 | Protection system for electromechanical equipment |
| CN117168865B (en) * | 2023-11-03 | 2024-04-12 | 南通宝佳精密机械有限公司 | Electromechanical device protection system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103473439B (en) | 2016-04-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109822399B (en) | Numerical control machine tool cutter wear state prediction method based on parallel deep neural network | |
| CN104390657B (en) | A kind of Generator Unit Operating Parameters measurement sensor fault diagnosis method and system | |
| CN103064340B (en) | Failure prediction method facing to numerically-controlled machine tool | |
| CN108375476B (en) | Hydroelectric generating set health assessment method | |
| CN104808585B (en) | A kind of quick inspection method of lathe health status | |
| CN101520651B (en) | Analysis method for reliability of numerical control equipment based on hidden Markov chain | |
| CN107016235B (en) | Equipment running state health degree evaluation method based on multi-feature adaptive fusion | |
| CN108255649B (en) | Diagnosis strategy design method based on modeling simulation cooperative analysis | |
| CN105718876B (en) | A kind of appraisal procedure of ball-screw health status | |
| CN102520697B (en) | Onsite information preprocessing method of remote cooperative diagnosis | |
| CN106662072A (en) | Wind turbine condition monitoring method and system | |
| CN102819239A (en) | Intelligent fault diagnosis method of numerical control machine tool | |
| CN104808587A (en) | Utilization statistical approach based on operation states of machining equipment | |
| CN107065834B (en) | The method for diagnosing faults of concentrator in hydrometallurgy process | |
| CN111812427A (en) | Health state assessment method and system for power electronic device | |
| CN102848266B (en) | Machine tool spindle accuracy prediction method | |
| CN103034170A (en) | Numerical control machine tool machining performance prediction method based on intervals | |
| CN108875281A (en) | A kind of hybrid intelligent health status diagnostic method and device based on deep learning | |
| US20220179394A1 (en) | Machine tool management method, machine tool management system and medium | |
| CN105243393A (en) | Characteristic-based fault forecasting method for complex electromechanical system | |
| CN111860839A (en) | Shore bridge fault monitoring method based on multi-signal fusion and Adam optimization algorithm | |
| CN108196525A (en) | The operational safety risk dynamic analysing method of Train Running Control System for High Speed | |
| CN114559298B (en) | Cutter wear monitoring method based on physical information fusion | |
| CN106567964B (en) | A kind of flow control valve monitoring method based on DCS data | |
| CN109492286B (en) | Numerical integration early termination method based on disturbed trajectory dynamic characteristics |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |