A kind of Cnc ReliabilityintelligeNetwork Network model modelling approach and system based on energy consumption characters
Technical field
The present invention relates to a kind of reliability model modeling method and system, more particularly to a kind of number based on energy consumption characters
Control lathe reliability model modeling method and system.
Background technology
When the reliability model of Digit Control Machine Tool is process-cycle and working time and the failure for describing Digit Control Machine Tool
Between probability density function.The reliability model of Digit Control Machine Tool is management, production line Performance Evaluation and the property to Digit Control Machine Tool
The basis that can optimize.In general, before lathe dispatches from the factory, manufacturer can provide the reference parameter of the lathe in production explanation.To making
For user, staff's early stage provides the nominal value of parameter by modes such as records.But it is due to that machine maintenance personnel can be right
Digit Control Machine Tool is adjusted and is negligent of record, and these adjustment are likely to produce influence to parameter.In the course of time, these change day
The product moon is tired to have ultimately resulted in the situation that nominal value is not inconsistent with actual value.Accordingly, it would be desirable to carry out carefully measurement to them so that ensure can
By the precision of property model.The measurement of these current parameters mostly manually regularly counts to complete, this artificial periodic statistical
Method cause data feedback not in time, easily error and the problems such as labor intensive resource.
The content of the invention
, can it is an object of the invention to provide a kind of Digit Control Machine Tool based on energy consumption characters for above-mentioned the deficiencies in the prior art
By property model modelling approach.It is a further object to provide a kind of Cnc ReliabilityintelligeNetwork Network model based on energy consumption characters
Modeling.
The technical scheme is that such:A kind of Cnc ReliabilityintelligeNetwork Network model modeling side based on energy consumption characters
Method, comprises the following steps,
S01, the power data for gathering lathe, calculate power difference sequence;
S02, by power difference sequence segmentation obtain the power fragment based on conditions of machine tool;
S03, the feature database of its power characteristic can be characterized to each one group of power snippet extraction, the feature database is included and machine
The different running status correlations of bed and some features changed with the change of state;
S04, with the corresponding conditions of machine tool transfer path of the procedure of the first string representation one;
S05, all power fragments are selected to power data in proportion carry out status indicator, based on feature database training
CART decision trees, the classification by the decision tree after training to power fragment;
It S06, will be arranged by the sorted power fragments of step S05 according to acquisition time order, and and use the second character string
Represent;
S07, using the character string of Knuth-Morris-Pratt algorithmic match first and the second character string, obtain power data
Power sequence during middle processing, subtracts initial time by the end time of power sequence when processing and obtains process-cycle duration;
S08, one time slip-window of setting, count process-cycle quantity in the time window, and during all process-cycles
Long, the arithmetic average for calculating all process-cycle durations in the time window is used as the process-cycle of work at present day;It is described to slide
Time window is moved rearwards by a working day, obtains next workaday process-cycle;
S09, given threshold, think working condition more than threshold value by lathe operation time power data, lathe are worked
Moment power data thinks malfunction less than threshold value, and stream time and its quantity are counted in the time slip-window
And calculate arithmetic average, statistics fault time and its quantity and calculate arithmetic mean;The time slip-window is moved rearwards by one
On working day, obtain next workaday working time arithmetic mean and fault time arithmetic mean;
S10, the different operating day for obtaining step S08 process-cycle, the work for the different operating day that step S09 is obtained
Time arithmetic mean and fault time arithmetic mean are used as lathe reliability model parameter.
Further, the conditions of machine tool is divided into standby, startup, four states of zero load and cutting.
Further, the calculating power difference sequence is to use P={ Pi, i=1,2 ..., n } and represent power data, n tables
Show the quantity of power collecting point, PiRepresent i-th of power collecting point, power difference sequence D={ Di, i=1,2 ..., n } table
Show, Di=Pi+1-Pi。
Further, when splitting power difference sequence in the step S02, rule of thumb Bayes's threshold method determines lathe
Power data critical point when state is shifted, the cut-point of power data critical point position power difference sequence.
Further, feature described in the step S03 includes maximum, minimum value, mean absolute deviation, geometric average
With three rank autoregressive coefficients.
Further, the maximum value calculation formula is pmax=max (pi, i=1,2 ..., m), minimum value calculation formula
For pmin=min (pi, i=1,2 ..., m), mean absolute deviation calculation formula is pmad=mediani(|pi-medianj(pj)
|), geometric average calculation formula isThree rank autoregressive coefficient calculation formula areM is power data collection number, p in power fragmentiI-th of power collecting point is represented, y is the work(
The geometric average of rate fragment, ω is intercept, βtIt is regression coefficient,It is noise parameter.
A kind of Cnc ReliabilityintelligeNetwork Network model modeling system based on energy consumption characters, including on-site data gathering module:Adopt
Collect machine power data and pack and upload onto the server;Base module, the spy for setting up lathe running power characteristic
Levy the conditions of machine tool transfer path of storehouse and process;Analysis module, carry out power difference sequence segmentation, the classification of power fragment with
And the calculating of lathe reliability model parameter;Setup module, sets time slip-window and threshold value.
Further, the on-site data gathering module includes some intelligent electric meters composition RS-485 networks.
The advantage of technical scheme provided by the present invention is, machine is realized to machine power data acquisition using intelligent electric meter
The automatic calculating of bed reliability model parameter is handled compared to artificial periodic statistical, and accuracy is higher, can avoid human error, number
According to not in time, the problem of taking time and effort.
Brief description of the drawings
Fig. 1 is the Cnc ReliabilityintelligeNetwork Network model modeling system construction drawing based on energy consumption characters.
Cnc ReliabilityintelligeNetwork Network model modelling approach flow charts of the Fig. 2 based on energy consumption characters.
Fig. 3 is that power when Digit Control Machine Tool processes certain process changes over time curve.
Fig. 4 is the state recognition result of embodiment Digit Control Machine Tool turning process.
Embodiment
With reference to embodiment, the invention will be further described, but not as a limitation of the invention.
Fig. 1 is referred to, the Cnc ReliabilityintelligeNetwork Network model modeling system based on energy consumption characters includes:On-site data gathering mould
Block:Every Digit Control Machine Tool configures an intelligent electric meter, and power data is gathered with certain frequency, and power data converges in data acquisition unit
Always, during then packing uploads onto the server, connected between intelligent electric meter and data acquisition unit using RS485 buses;Knowledge base mould
Block, for setting up the feature database of lathe running power characteristic and the conditions of machine tool transfer path of process;Analysis module, enters
The segmentation of row power difference sequence, the classification of power fragment and the calculating of lathe reliability model parameter;Setup module, sets and slides
Dynamic time window and threshold value.Can be networked three kinds of sides between data acquisition unit and server using wired networking, Wireless Networking and GPRS
Formula.Between server and analysis module, base module and setup module by the way of wired connection.
Incorporated by reference to Fig. 2, by taking a numerically controlled lathe as an example, the Cnc ReliabilityintelligeNetwork Network model modelling approach based on energy consumption characters
It is as follows:
S01, the power data of on-site data gathering module collection lathe are uploaded onto the server, and analysis module is carried from server
Power data is taken to calculate power difference sequence:With P={ Pi, i=1,2 ..., n } and power data is represented, n represents power collecting point
Quantity, PiRepresent i-th of power collecting point, power difference sequence D={ Di, i=1,2 ..., n } represent, Di=Pi+1-Pi。
S02, power difference sequence pass through document " Empirical Bayes Selection of as observation
The Empirical Bayes threshold method that Wavelet Thresholds.Annals of Statistics " (2005) are proposed determines lathe
Power data critical point when state is shifted, so as to split the power fragment obtained based on conditions of machine tool.
S03, base module set up lathe running power features storehouse.According to document " Multi-objective
teaching–learning-based optimization algorithm for reducing carbon emissions
And operation time in turning operations.Engineering Optimization " (2015), by number
Control lathe can be divided into four states such as standby, startup, unloaded and cutting.Lathe can be divided into standby, startup, unloaded and cutting
Etc. state.According to the characteristic of lathe running power, running statuses correlations different from lathe are extracted and with the change of state
And the multiple features changed, form power features storehouse.The feature of its power characteristic can be characterized to every piece of one group of power snippet extraction
Storehouse, including maximum (pmax), minimum value (pmin), mean absolute deviation (pmad), geometric average (pgm), three rank autoregressive coefficients
(β1,β2,β3) etc. seven features.Common m collection point in the power fragment, each collection point pi(i=1,2 ..., m), feature
Calculation formula it is as follows:
pmax=max (pi, i=1,2 ..., m)
pmin=min (pi, i=1,2 ..., m)
pmad=mediani(|pi-medianj(pj)|)
M is power data collection number, p in power fragmentiI-th of power collecting point is represented, y is the several of the power fragment
What is average, and ω is intercept, βtIt is regression coefficient,It is noise parameter.
S04, the processing sequence by process, can obtain its state transition path.Fig. 3 gives the machine tooling one
Power during procedure changes over time curve, its state transition path for it is standby->Start->Unloaded->Cutting->It is unloaded.Four
Individual machining state is indicated with a character respectively, as shown in table 1.Signable state transition path then in Fig. 3 is the first character
String R:“ITACA”
Table 1, machining state and character corresponding table
State |
Character |
It is standby |
I |
Start |
T |
It is unloaded |
A |
Cutting |
C |
S05, all power fragments are selected to certain proportion carry out status indicator, CART decision trees are trained, after training
Decision tree be used for power fragment classification.Fig. 4 provides the state recognition result of certain turning process.
S06, by sorted power fragment according to acquisition time order arrange, with the second character string CpSign.
S07, using Knuth-Morris-Pratt algorithmic match character strings R and Cp, CpThe substring of middle matching is to add
The power sequence in man-hour.The end time of power sequence when processing is subtracted into initial time can obtain the process-cycle.
S08 and S09, setup module, set Cnc ReliabilityintelligeNetwork Network Parameters in Mathematical Model time slip-window length and
Threshold value a.Time slip-window is extracted from setup module, up to the termination time since the initial time of time window, if with R
Match somebody with somebody, then counter values add 1, while the duration for recording the substring is a process-cycle.In this way, can be somebody's turn to do
Quantity (uses n the process-cycle in time window1Represent), and all process-cycle duration τi(i=1,2 ..., n1).Calculate the time
The arithmetic average of all process-cycles in window, as the process-cycle of work at present day.
Similarly, time slip-window is moved rearwards by a working day, it is possible to obtain next workaday process-cycle.
Designated firing duration and repair time:Give a threshold value a, when certain normal work moment power data be more than a, then
Think that lathe is in running order;When power data is less than a, then it is assumed that lathe is in malfunction.Give a sliding time
Window, counts stream time t in the time windowup,iAnd quantity nup, also fault time tdown,iAnd quantity ndown.Meter
Calculate the arithmetic average of working time and fault time in the time window.Similarly, time slip-window is moved rearwards by a working day, and
Next workaday t can be obtainedup, tdown。
S10, the reliability model relevant parameter that can obtain by above step the Digit Control Machine Tool are as shown in table 2.From table
It can be seen that, in less time dimension, the process-cycle is held essentially constant, but average operation time and mean down time are then
In the presence of certain fluctuation.As can be seen here, the industrial plant Cnc ReliabilityintelligeNetwork Network model modelling approach that the present invention is described is to mould
The lasting monitoring of the relevant parameter of type and analyze meaningful to industrial plant production line Performance Evaluation.
Table 2, Cnc ReliabilityintelligeNetwork Network Parameters in Mathematical Model
Relevant parameter |
1st day |
2nd day |
3rd day |
4th day |
5th day |
τ(s) |
72 |
73 |
72 |
72 |
72 |
tup(h) |
38.4 |
39.1 |
40.5 |
39.7 |
41.3 |
tdown(h) |
4.8 |
4.3 |
4.9 |
5.2 |
4.8 |