CN105334472A - Online remaining life prediction method for mining intrinsic safety power supply - Google Patents
Online remaining life prediction method for mining intrinsic safety power supply Download PDFInfo
- Publication number
- CN105334472A CN105334472A CN201510705414.9A CN201510705414A CN105334472A CN 105334472 A CN105334472 A CN 105334472A CN 201510705414 A CN201510705414 A CN 201510705414A CN 105334472 A CN105334472 A CN 105334472A
- Authority
- CN
- China
- Prior art keywords
- output voltage
- wavelet
- renyi
- elm
- intrinsic safety
- 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
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an online remaining life prediction method for a mining intrinsic safety power supply. The online remaining life prediction method comprises the steps of: calculating output voltage and current signals of a circuit denoised by utilizing an autocorrelation method to obtain an output voltage mean value, an output current mean value, load resistance, an output voltage ripple ratio and Renyi wavelet packet energy entropy; constructing a relation function of the output voltage ripple ratio delta, the load resistance Z<L> and the Renyi wavelet packet energy entropy pr on the basis of an extreme learning machine: delta=f<1>(p<r>, Z<L>); constructing a prediction model of the p<r> based on historical and the current p<r> by utilizing an ELM-PF algorithm, and predicting future p<r>; and solving the output voltage ripple ratio at a qth moment in further by utilizing the load resistance, the predicted p<r>(q) at the qth moment in further and the relation function delta=f<1>(p<r>, Z<L>), and acquiring remaining life of the intrinsic safety power supply by utilizing a specified invalid threshold value of delta. The online remaining life prediction method for the mining intrinsic safety power supply can be applied to the technical field of switching power supply fault diagnosis or fault prediction, and can evaluate the remaining life of the mining intrinsic safety power supply online accurately.
Description
Technical field
The present invention relates to fault prognostics and health management technical field, particularly relate to the online method for predicting residual useful life of a kind of mine intrinsic safety electric source.
Background technology
Intrinsic safety electric source is indispensable part in coal mine intrinsic safety system, and its quality and reliability directly affects the stability of load equipment, reliability and accuracy, is even related to the production safety of whole mine.As the power supply core of electrically energized machine, failure prediction and residual life evaluation are accurately and effectively carried out to mine intrinsic safety electric source, the health status (as fault, health or inferior health etc.) in current or precognition power circuit future can be determined, the maintenance and repair that can be equipment provides decision-making foundation, thus reduction failure risk, for the safety of system, reliability service provide safeguard.
There is strong electromagnetic in mine intrinsic safety electric source working environment, power circuit signal is both relevant with circuit health situation, again by load effect, therefore, comparatively difficult to effective Accurate Prediction of the residual life of mine intrinsic safety electric source.For this reason, the present invention, in conjunction with the performance index requirement of intrinsic safety electric source, proposes online life prediction new method of a kind of mine intrinsic safety electric source life-span.The method is by carrying out auto-correlation denoising to the power circuit signal of on-line monitoring, calculate electric power output voltage Renyi wavelet-packet energy entropy, build intrinsic safety electric source Life Prediction Model, set up output voltage Renyi wavelet-packet energy entropy, loaded impedance with output voltage ripple than relation function, final assessment power supply residual life to be measured.Strong electromagnetic, the fluctuation of load etc. that the present invention can solve mine intrinsic safety electric source affect the problem of mine intrinsic safety electric source predicting residual useful life accuracy, can realize high precision predicting residual useful life and the reliability assessment of mine intrinsic safety electric source.
Summary of the invention
The object of the present invention is to provide the online method for predicting residual useful life of a kind of mine intrinsic safety electric source, for judging the health status of intrinsic safety electric source, for the prediction and health control that realize efficiently and accurately provide safeguard, to take measures in time, ensure the safe and reliable operation of using electricity system.
In order to reach above-mentioned purpose, solution of the present invention is:
The online method for predicting residual useful life of mine intrinsic safety electric source, comprises the following steps (1) ~ (7):
(1) on-line checkingi electric power output voltage signal u
o(t), output current signal i
ot (), utilizes correlation method respectively to surveyed output voltage u
o(t), current signal i
ot () denoising, obtains noiseless output voltage signal u
o *(t) and noiseless output current signal i
o *(t);
(2) u is utilized
o *(t), i
o *t () calculates output voltage average U
o, output current average I
o, loaded impedance Z
l, output voltage ripple than δ, wherein
u
max, u
minbe respectively output voltage maximal value and minimum value;
(3) noiseless output voltage signal u is calculated
o *the Renyi wavelet-packet energy entropy p of (t)
r;
(4) extreme learning machine (ExtremeLearningMachine, ELM) is utilized to build output voltage ripple than δ and loaded impedance Z
l, Renyi wavelet-packet energy entropy p
rrelation function, be δ=f
1(p
r, Z
l);
(5) based on p
rhistory and current data, utilize ELM-MPF (ExtremeLearningMachine-MultiParticleFilter) algorithm to build p
rforecast model, and dope the Renyi wavelet-packet energy entropy p in following q moment
r(q);
(6) loaded impedance Z is utilized
lthe Renyi wavelet-packet energy entropy p in the q moment in future of (q), step (5) prediction gained
r(q), the δ=f set up in step (4)
1(p
r, Z
l), ask for following q moment output voltage ripple than δ (q);
(7) δ failure threshold according to the rules, utilizes the output voltage ripple of gained in step (6) than δ (q), seeks out intrinsic safety electric source residual life.
The online method for predicting residual useful life of mine intrinsic safety electric source of the present invention, utilize in described step (1) correlation method remove output voltage, output current noise concrete steps for (1.1) ~ (1.2):
(1.1) electric power output voltage signal u is asked for respectively
o(t), output current signal i
ot the autocorrelation function of (), is designated as R
u(τ), R
i(τ), that is:
Wherein T is the time interval of intercept signal, and τ is delay parameter;
(1.2) u
o(t), i
ot data that () removes near time delay τ=0 and τ maximal value are noiseless output voltage signal u
o *(t) and noiseless output current signal i
o *(t).
The online method for predicting residual useful life of mine intrinsic safety electric source of the present invention, in described step (3), calculates noiseless output voltage signal u
o *t the calculation procedure of the Renyi wavelet-packet energy entropy of () is (2.1) ~ (2.3):
(2.1) to u
o *t () carries out 4 layers of WAVELET PACKET DECOMPOSITION, wavelet basis function adopts db4 function, and WAVELET PACKET DECOMPOSITION is existing mature technology, specifically repeats no more:
(2.2) each frequency band energy after WAVELET PACKET DECOMPOSITION is calculated, namely
wherein i=1,2 ..., 16,
represent through m coefficient corresponding to WAVELET PACKET DECOMPOSITION posterior nodal point S (j, n), N is signal length;
(2.3) Renyi wavelet-packet energy entropy p is calculated
r, be specially
wherein 0 < α < 1,
preferred α=0.45 herein.
The online method for predicting residual useful life of mine intrinsic safety electric source of the present invention, in described step (4), builds output voltage ripple than δ and Renyi wavelet-packet energy entropy p
r, loaded impedance Z
lthe concrete steps of relation function be (3.1) ~ (3.3):
(3.1) the Renyi wavelet-packet energy entropy p of c group historical juncture is chosen
r, output voltage ripple is than δ and loaded impedance Z
l, be designated as { p
r(k-c+1), p
r(k-c+2) ..., p
r(k) }, δ (k-c+1), δ (k-c+2) ..., δ (k) }, { Z
l(k-c+1), Z
l(k-c+2) ..., Z
l(k) }, wherein 1≤c≤k, k is current time;
(3.2) structure arranging extreme learning machine ELM is input layer, output layer and 1 hidden layer, wherein 2 input quantities, 1 output quantity, and hidden neuron number is 15;
(3.3) with p
r(i), Z
li (), as the input of ELM, the output using δ (i) as extreme learning machine, is total to c group sample training ELM, the ELM after training as output voltage ripple than δ and Renyi wavelet-packet energy entropy p
r, loaded impedance Z
lrelation function δ=f
1(p
r, Z
l), wherein i=k-c+1 ..., k.
The online method for predicting residual useful life of mine intrinsic safety electric source of the present invention, in described step (5), utilizes ELM-MPF algorithm to build p
rforecast model, and dope the Renyi wavelet-packet energy entropy p in following q moment
rq the concrete steps of () are (4.1) ~ (4.3):
(4.1) the output voltage Renyi wavelet-packet energy entropy p of l group historical juncture is chosen
r, be designated as { p
r(k-l+1), p
r(k-l+2) ..., p
r(k) }, wherein 1≤l≤k, k is current time;
(4.2) arranging ELM structure is input layer, output layer and 1 hidden layer, wherein 4 input quantities, and single output quantity, hidden neuron number are 20, with p
r(i-4), p
r(i-3), p
rand p (i-2)
r(i-1) as the input of ELM, with p
ri (), as the output of ELM, ELM, the ELM after training are as Renyi wavelet-packet energy entropy p in training
rstate transition function, i.e. p
r(i)=f
2(p
r(i-1), p
r(i-2), p
r(i-3), p
r(i-4)), wherein i=k-l+1 ..., k;
(4.3) the Renyi wavelet-packet energy entropy p set up in step (4.2) is utilized
rstate transition function p
r(i)=f
2(p
r(i-1), p
r(i-2), p
r(i-3), p
r(i-4) multistage particle filter algorithm (MultiParticleFilter, MPF)), is adopted to predict the Renyi wavelet-packet energy entropy p in q moment
rq (), is implemented as:
(4.3.1) p is used
r(i)=f
2(p
r(i-1), p
r(i-2), p
r(i-3), p
r(i-4)) State Transferring, obtains system state space model such as formula (3),
In formula (3), w
i-1for system noise, v
i-1for observation noise, y (i) is observed reading;
(4.3.2) stochastic generation primary
j=1,2 ..., N
s, N
sfor population, make i=k, k is current time point;
(4.3.3) by step (4.3.2)
substitution formula (3), calculates
y
j(i);
(4.3.4) by known state parameter p
r(i-1), p
r(i-2), p
r(i-3), p
r(i-4) formula (3), is utilized to calculate p
r(i), y (i);
(4.3.5) each particulate errors e is calculated
j=y
j(i)-y (i), j=1,2 ..., N
s;
(4.3.6) by each particulate errors size, each particle weights w is determined
j, that is:
(4.3.7) p is calculated
rthe predicted value of (i)
(4.3.8) particle resampling, particle upgrade, and concrete grammar is:
Number of effective particles is
effective particle threshold
if
then carry out resampling, concrete steps are:
To a jth particle
j=1,2 ..., N
s, calculate weight cumulative sum
n
r=1,2 ..., N
sif,
Then
(4.3.9) make i=i+1, go to step (4.3.3), carry out next step prediction, if carry out forward direction s to walk prediction, then circulation performs s time, obtains predicted value p
r(q)=p
r(k+s)=p
r(i).
The online method for predicting residual useful life of mine intrinsic safety electric source of the present invention, in described step (7), ask for being implemented as of intrinsic safety electric source residual life:
If current time is the k moment, intrinsic safety electric source fault threshold to be measured is set to electric power output voltage ripple and compares δ
0=10%,
(5.1) utilize the ELM-MPF algorithm in the online method for predicting residual useful life step (5) of mine intrinsic safety electric source of the present invention to carry out forward direction s and walk prediction, obtain the output voltage Renyi wavelet-packet energy entropy p in (k+s) moment
r(k+s);
(5.2) utilize output voltage ripple that the online method for predicting residual useful life step (4) of mine intrinsic safety electric source of the present invention is set up than δ and Renyi wavelet-packet energy entropy p
r, loaded impedance Z
lrelation function δ=f
1(p
r, Z
l), the p of step (5.1) gained
r(k+s) output voltage ripple in (k+s) moment, is solved than δ (k+s);
(5.3) the forward direction s calculated in step (5.2) is walked output voltage ripple than predicted value δ (k+s) and δ
0compare, if δ (k+s)>=δ
0, then intrinsic safety electric source residual life to be measured is t
rUL=s Δ h, wherein Δ h is p
rhistory and time interval of current time sequence data.
Accompanying drawing explanation
Fig. 1 is the online method for predicting residual useful life process flow diagram of mine intrinsic safety electric source.
Embodiment
Below in conjunction with accompanying drawing, technical scheme of the present invention is described in detail.
The invention provides the online life-span prediction method of a kind of mine intrinsic safety electric source, its general thought is, first utilize correlation method to circuit output voltage, current signal denoising, utilize the output voltage after denoising and current signal to calculate output voltage average, output current average, loaded impedance, output voltage ripple ratio, Renyi wavelet-packet energy entropy; Then, output voltage ripple is built than δ and loaded impedance Z based on extreme learning machine
l, Renyi wavelet-packet energy entropy p
rrelation function δ=f
1(p
r, Z
l); Based on p
rhistory and current data, utilize ELM-PF (ExtremeLearningMachine-ParticleFilter) algorithm to build p
rforecast model, and dope the Renyi wavelet-packet energy entropy of future time instance; Finally, loaded impedance Z is utilized
l, prediction the Renyi wavelet-packet energy entropy p in q moment in future
r(q), and relation function δ=f
1(p
r, Z
l), ask for following q moment output voltage ripple than δ (q), and δ failure threshold according to the rules, obtain intrinsic safety electric source residual life to be measured.The inventive method is particularly useful for colliery etc. to be existed the intrinsic safety electric source of strong electromagnetic working environment and not by the operating infulence such as circuit input voltage and the fluctuation of load, can realize the Accurate Prediction of the residual life of mine intrinsic safety electric source.
As shown in Figure 1, the online method for predicting residual useful life of mine intrinsic safety electric source of the present invention, concrete enforcement comprises the following steps:
(1) on-line checkingi electric power output voltage signal u
o(t), output current signal i
ot (), utilizes correlation method respectively to surveyed output voltage u
o(t), current signal i
ot () denoising, obtains noiseless output voltage signal u
o *(t) and noiseless output current signal i
o *(t), concrete steps are (1.1) ~ (1.2):
(1.1) electric power output voltage signal u is asked for respectively
o(t), output current signal i
ot the autocorrelation function of (), is designated as R
u(τ), R
i(τ), that is:
Wherein T is the time interval of intercept signal, and τ is delay parameter;
(1.2) u
o(t), i
ot data that () removes near time delay τ=0 and τ maximal value are noiseless output voltage signal u
o *(t) and noiseless output current signal i
o *(t).
(2) u is utilized
o *(t), i
o *t () calculates output voltage average U
o, output current average I
o, loaded impedance Z
l, output voltage ripple than δ, wherein
u
max, u
minbe respectively output voltage maximal value and minimum value;
(3) noiseless output voltage signal u is calculated
o *the Renyi wavelet-packet energy entropy p of (t)
r, specific implementation step is (3.1) ~ (3.3):
(3.1) to u
o *t () carries out 4 layers of WAVELET PACKET DECOMPOSITION, wavelet basis function adopts db4 function, and WAVELET PACKET DECOMPOSITION is existing mature technology, specifically repeats no more;
(3.2) each frequency band energy after WAVELET PACKET DECOMPOSITION is calculated, namely
wherein i=1,2 ..., 16,
represent through m coefficient corresponding to WAVELET PACKET DECOMPOSITION posterior nodal point S (j, n), N is signal length;
(3.3) Renyi wavelet-packet energy entropy p is calculated
r, be specially
wherein 0 < α < 1,
preferred α=0.45 herein.
(4) extreme learning machine (ExtremeLearningMachine, ELM) is utilized to build output voltage ripple than δ and loaded impedance Z
l, Renyi wavelet-packet energy entropy p
rrelation function, be δ=f
1(p
r, Z
l), specific implementation step is (4.1) ~ (4.3):
(4.1) the Renyi wavelet-packet energy entropy p of c group historical juncture is chosen
r, output voltage ripple is than δ and loaded impedance Z
l, be designated as { p
r(k-c+1), p
r(k-c+2) ..., p
r(k) }, δ (k-c+1), δ (k-c+2) ..., δ (k) }, { Z
l(k-c+1), Z
l(k-c+2) ..., Z
l(k) }, wherein 1≤c≤k, k is current time;
(4.2) structure arranging extreme learning machine ELM is input layer, output layer and 1 hidden layer, wherein 2 input quantities, 1 output quantity, and hidden neuron number is 15;
(4.3) with p
r(i), Z
li (), as the input of ELM, the output using δ (i) as extreme learning machine, is total to c group sample training ELM, the ELM after training as output voltage ripple than δ and Renyi wavelet-packet energy entropy p
r, loaded impedance Z
lrelation function δ=f
1(p
r, Z
l), wherein i=k-c+1 ..., k.
(5) based on p
rhistory and current data, utilize ELM-MPF (ExtremeLearningMachine-MultiParticleFilter) algorithm to build p
rforecast model, and dope the Renyi wavelet-packet energy entropy p in following q moment
r(q), specific implementation step is (5.1) ~ (5.3):
(5.1) the output voltage Renyi wavelet-packet energy entropy p of l group historical juncture is chosen
r, be designated as { p
r(k-l+1), p
r(k-l+2) ..., p
r(k) }, wherein 1≤l≤k, k is current time;
(5.2) arranging ELM structure is input layer, output layer and 1 hidden layer, wherein 4 input quantities, and single output quantity, hidden neuron number are 20, with p
r(i-4), p
r(i-3), p
rand p (i-2)
r(i-1) as the input of ELM, with p
ri (), as the output of ELM, ELM, the ELM after training are as Renyi wavelet-packet energy entropy p in training
rstate transition function, i.e. p
r(i)=f
2(p
r(i-1), p
r(i-2), p
r(i-3), p
r(i-4)), wherein i=k-l+1 ..., k;
(5.3) the Renyi wavelet-packet energy entropy p set up in step (5.2) is utilized
rstate transition function p
r(i)=f
2(p
r(i-1), p
r(i-2), p
r(i-3), p
r(i-4) multistage particle filter algorithm (MultiParticleFilter, MPF)), is adopted to predict the Renyi wavelet-packet energy entropy p in q moment
rq (), is implemented as:
(5.3.1) p is used
r(i)=f
2(p
r(i-1), p
r(i-2), p
r(i-3), p
r(i-4)) State Transferring, obtains system state space model such as formula (3),
In formula (3), w
i-1for system noise, v
i-1for observation noise, y (i) is observed reading;
(5.3.2) stochastic generation primary
j=1,2 ..., N
s, N
sfor population, make i=k, k is current time point;
(5.3.3) by step (5.3.2)
substitution formula (3), calculates
y
j(i);
(5.3.4) by known state parameter p
r(i-1), p
r(i-2), p
r(i-3), p
r(i-4) formula (3), is utilized to calculate p
r(i), y (i);
(5.3.5) each particulate errors e is calculated
j=y
j(i)-y (i), j=1,2 ..., N
s;
(5.3.6) by each particulate errors size, each particle weights w is determined
j, that is:
(5.3.7) p is calculated
rthe predicted value of (i)
(5.3.8) particle resampling, particle upgrade, and concrete grammar is:
Number of effective particles is
effective particle threshold
if N
eff< N
th, then carry out resampling, concrete steps are:
To a jth particle
j=1,2 ..., N
s, calculate weight cumulative sum
n
r=1,2 ..., N
sif,
Then
(5.3.9) make i=i+1, go to step (5.3.3), carry out next step prediction, if carry out forward direction s to walk prediction, then circulation performs s time, obtains predicted value p
r(q)=p
r(k+s)=p
r(i).
(6) loaded impedance Z is utilized
lthe Renyi wavelet-packet energy entropy p in the q moment in future of (q), step (5) prediction gained
r(q), to substitute in step (4) set up δ=f based on ELM
1(p
r, Z
l), ask for following q moment output voltage ripple is corresponding ELM output valve than δ (q), δ (q).
(7) δ failure threshold according to the rules, utilize the output voltage ripple of gained in step (6) than δ (q), seek out intrinsic safety electric source residual life, concrete steps are:
If current time is the k moment, intrinsic safety electric source fault threshold to be measured is set to electric power output voltage ripple and compares δ
0=10%,
(7.1) utilize the ELM-MPF algorithm in the online method for predicting residual useful life step (5) of mine intrinsic safety electric source of the present invention to carry out forward direction s and walk prediction, obtain the output voltage Renyi wavelet-packet energy entropy p in (k+s) moment
r(k+s);
(7.2) utilize output voltage ripple that the online method for predicting residual useful life step (4) of mine intrinsic safety electric source of the present invention is set up than δ and Renyi wavelet-packet energy entropy p
r, loaded impedance Z
lrelation function δ=f
1(p
r, Z
l), the p of step (5.1) gained
r(k+s) output voltage ripple in (k+s) moment, is solved than δ (k+s);
(7.3) the forward direction s calculated in step (7.2) is walked output voltage ripple than predicted value δ (k+s) and δ
0compare, if δ (k+s)>=δ
0, then intrinsic safety electric source residual life to be measured is t
rUL=s Δ h, wherein Δ h is p
rhistory and time interval of current time sequence data.
Above embodiment is only and technological thought of the present invention is described, can not limit protection scope of the present invention with this, and every technological thought proposed according to the present invention, any change that technical scheme basis is done, all falls within scope.
Claims (6)
1. the online method for predicting residual useful life of mine intrinsic safety electric source, is characterized in that, comprise the following steps:
(1) on-line checkingi electric power output voltage signal u
o(t), output current signal i
ot (), utilizes correlation method respectively to surveyed output voltage u
o(t), current signal i
ot () denoising, obtains noiseless output voltage signal u
o *(t) and noiseless output current signal i
o *(t);
(2) u is utilized
o *(t), i
o *t () calculates output voltage average U
o, output current average I
o, loaded impedance Z
l, output voltage ripple than δ, wherein
u
max, u
minbe respectively output voltage maximal value and minimum value;
(3) noiseless output voltage signal u is calculated
o *the Renyi wavelet-packet energy entropy p of (t)
r;
(4) extreme learning machine (ExtremeLearningMachine, ELM) is utilized to build output voltage ripple than δ and loaded impedance Z
l, Renyi wavelet-packet energy entropy p
rrelation function, be δ=f
1(p
r, Z
l);
(5) based on p
rhistory and current data, utilize ELM-MPF (ExtremeLearningMachine-MultiParticleFilter) algorithm to build p
rforecast model, and dope the Renyi wavelet-packet energy entropy p in following q moment
r(q);
(6) loaded impedance Z is utilized
lthe Renyi wavelet-packet energy entropy p in the q moment in future of (q), step (5) prediction gained
r(q), the δ=f set up in step (4)
1(p
r, Z
l), ask for following q moment output voltage ripple than δ (q);
(7) δ failure threshold according to the rules, utilizes the output voltage ripple of gained in step (6) than δ (q), seeks out intrinsic safety electric source residual life.
2. the online method for predicting residual useful life of mine intrinsic safety electric source as claimed in claim 1, is characterized in that, utilizes correlation method to remove output voltage, the concrete steps of output current noise are in described step (1):
(2.1) electric power output voltage signal u is asked for respectively
o(t), output current signal i
ot the autocorrelation function of (), is designated as R
u(τ), R
i(τ), that is:
Wherein T is the time interval of intercept signal, and τ is delay parameter;
(2.2) u
o(t), i
ot data that () removes near time delay τ=0 and τ maximal value are noiseless output voltage signal u
o *(t) and noiseless output current signal i
o *(t).
3. the online method for predicting residual useful life of mine intrinsic safety electric source as claimed in claim 1, is characterized in that, in described step (3), calculates noiseless output voltage signal u
o *t the calculation procedure of the Renyi wavelet-packet energy entropy of () is:
(3.1) to u
o *t () carries out 4 layers of WAVELET PACKET DECOMPOSITION, wavelet basis function adopts db4 function, and WAVELET PACKET DECOMPOSITION is existing mature technology, specifically repeats no more;
(3.2) each frequency band energy after WAVELET PACKET DECOMPOSITION is calculated, namely
wherein i=1,2 ..., 16,
represent through m coefficient corresponding to WAVELET PACKET DECOMPOSITION posterior nodal point S (j, n), N is signal length;
(3.3) Renyi wavelet-packet energy entropy p is calculated
r, be specially
wherein 0 < α < 1,
preferred α=0.45 herein.
4. the online method for predicting residual useful life of mine intrinsic safety electric source as claimed in claim 1, is characterized in that, in described step (4), builds output voltage ripple than δ and Renyi wavelet-packet energy entropy p
r, loaded impedance Z
lthe concrete steps of relation function be:
(4.1) the Renyi wavelet-packet energy entropy p of c group historical juncture is chosen
r, output voltage ripple is than δ and loaded impedance Z
l, be designated as { p
r(k-c+1), p
r(k-c+2) ..., p
r(k) }, δ (k-c+1), δ (k-c+2) ..., δ (k) }, { Z
l(k-c+1), Z
l(k-c+2) ..., Z
l(k) }, wherein 1≤c≤k, k is current time;
(4.2) structure arranging extreme learning machine ELM is input layer, output layer and 1 hidden layer, wherein 2 input quantities, 1 output quantity, and hidden neuron number is 15;
(4.3) with p
r(i), Z
li (), as the input of ELM, the output using δ (i) as extreme learning machine, is total to c group sample training ELM, the ELM after training as output voltage ripple than δ and Renyi wavelet-packet energy entropy p
r, loaded impedance Z
lrelation function δ=f
1(p
r, Z
l), wherein i=k-c+1 ..., k.
5. the online method for predicting residual useful life of mine intrinsic safety electric source as claimed in claim 1, is characterized in that, in described step (5), utilizes ELM-MPF algorithm to build p
rforecast model, and dope the Renyi wavelet-packet energy entropy p in following q moment
rq the concrete steps of () are:
(5.1) the output voltage Renyi wavelet-packet energy entropy p of l group historical juncture is chosen
r, be designated as { p
r(k-l+1), p
r(k-l+2) ..., p
r(k) }, wherein 1≤l≤k, k is current time;
(5.2) arranging ELM structure is input layer, output layer and 1 hidden layer, wherein 4 input quantities, and single output quantity, hidden neuron number are 20, with p
r(i-4), p
r(i-3), p
rand p (i-2)
r(i-1) as the input of ELM, with p
ri (), as the output of ELM, ELM, the ELM after training are as Renyi wavelet-packet energy entropy p in training
rstate transition function, i.e. p
r(i)=f
2(p
r(i-1), p
r(i-2), p
r(i-3), p
r(i-4)), wherein i=k-l+1 ..., k;
(5.3) the Renyi wavelet-packet energy entropy p set up in step (5.2) is utilized
rstate transition function p
r(i)=f
2(p
r(i-1), p
r(i-2), p
r(i-3), p
r(i-4) multistage particle filter algorithm (MultiParticleFilter, MPF)), is adopted to predict the Renyi wavelet-packet energy entropy p in q moment
rq (), is implemented as:
(5.3.1) p is used
r(i)=f
2(p
r(i-1), p
r(i-2), p
r(i-3), p
r(i-4)) State Transferring, obtains system state space model such as formula (3),
In formula (3), w
i-1for system noise, v
i-1for observation noise, y (i) is observed reading;
(5.3.2) stochastic generation primary
j=1,2 ..., N
s, N
sfor population, make i=k, k is current time point;
(5.3.3) by step (5.3.2)
substitution formula (3), calculates
y
j(i);
(5.3.4) by known state parameter p
r(i-1), p
r(i-2), p
r(i-3), p
r(i-4) formula (3), is utilized to calculate p
r(i), y (i);
(5.3.5) each particulate errors e is calculated
j=y
j(i)-y (i), j=1,2 ..., N
s;
(5.3.6) by each particulate errors size, each particle weights w is determined
j, that is:
(5.3.7) p is calculated
rthe predicted value of (i)
(5.3.8) particle resampling, particle upgrade, and concrete grammar is:
Number of effective particles is
effective particle threshold
if N
eff< N
th, then carry out resampling, concrete steps are:
To a jth particle
j=1,2 ..., N
s, calculate weight cumulative sum
n
r=1,2 ..., N
sif,
Then
(5.3.9) make i=i+1, go to step (5.3.3), carry out next step prediction, if carry out forward direction s to walk prediction, then circulation performs s time, obtains predicted value p
r(q)=p
r(k+s)=p
r(i).
6. the online method for predicting residual useful life of mine intrinsic safety electric source as claimed in claim 1, is characterized in that, in described step (7), ask for being implemented as of intrinsic safety electric source residual life:
If current time is the k moment, power fail threshold value to be measured is set to output voltage ripple and compares δ
0=10%,
(6.1) utilize the ELM-MPF algorithm in step (5) to carry out forward direction s and walk prediction, obtain the output voltage Renyi wavelet-packet energy entropy p in (k+s) moment
r(k+s);
(6.2) output voltage ripple utilizing step (4) to set up is than δ and Renyi wavelet-packet energy entropy p
r, loaded impedance Z
lrelation function δ=f
1(p
r, Z
l), the p of step (6.1) gained
r(k+s) output voltage ripple in (k+s) moment, is solved than δ (k+s);
(6.3) the forward direction s calculated in step (6.2) is walked output voltage ripple than predicted value δ (k+s) and δ
0compare, if δ (k+s)>=δ
0, then power supply residual life to be measured is t
rUL=s Δ h, wherein Δ h is p
rhistory and time interval of current time sequence data.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510705414.9A CN105334472B (en) | 2015-10-26 | 2015-10-26 | The online method for predicting residual useful life of mine intrinsic safety electric source |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510705414.9A CN105334472B (en) | 2015-10-26 | 2015-10-26 | The online method for predicting residual useful life of mine intrinsic safety electric source |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105334472A true CN105334472A (en) | 2016-02-17 |
CN105334472B CN105334472B (en) | 2018-06-26 |
Family
ID=55285115
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510705414.9A Active CN105334472B (en) | 2015-10-26 | 2015-10-26 | The online method for predicting residual useful life of mine intrinsic safety electric source |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105334472B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107491840A (en) * | 2017-08-21 | 2017-12-19 | 浙江理工大学 | The prediction of flowing wearing character and lifetime estimation method based on ELM neural network models |
CN109615003A (en) * | 2018-12-06 | 2019-04-12 | 哈尔滨工业大学 | A kind of power source failure prediction method based on ELM-CHMM |
CN111308960A (en) * | 2020-01-20 | 2020-06-19 | 广西玉柴机器股份有限公司 | Load monitoring method and system |
CN113689044A (en) * | 2021-08-26 | 2021-11-23 | 北京航空航天大学 | Method and system for predicting residual service life of switching power supply |
CN116882303A (en) * | 2023-09-06 | 2023-10-13 | 深圳市联明电源有限公司 | Laser power supply life prediction method, system and storage medium |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20060250137A1 (en) * | 2003-06-25 | 2006-11-09 | Bernd Frey | Method for predicting the residual service life of an electric energy accumulator |
CN102749572A (en) * | 2012-06-26 | 2012-10-24 | 南京航空航天大学 | IECMAC parameter identification-based power electronic circuit failure predicting method |
CN102867132A (en) * | 2012-10-16 | 2013-01-09 | 南京航空航天大学 | Aviation direct-current converter online fault combined prediction method based on fractional order wavelet transformation |
CN102879728A (en) * | 2012-10-16 | 2013-01-16 | 南京航空航天大学 | Health evaluation index and failure predication method for DC (Direct Current)-DC convertor |
CN102968573A (en) * | 2012-12-14 | 2013-03-13 | 哈尔滨工业大学 | Online lithium ion battery residual life predicting method based on relevance vector regression |
CN103033761A (en) * | 2012-12-17 | 2013-04-10 | 哈尔滨工业大学 | Lithium ion battery residual life forecasting method of dynamic gray related vector machine |
CN103399279A (en) * | 2013-08-01 | 2013-11-20 | 哈尔滨工业大学 | Method for predicting cycle life of fused lithium ion battery based on EKF (Extended Kalman Filter) method and AR (AutoRegressive) model |
CN104537444A (en) * | 2015-01-13 | 2015-04-22 | 安徽理工大学 | Gas outburst predicting method based on EMD and ELM |
-
2015
- 2015-10-26 CN CN201510705414.9A patent/CN105334472B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20060250137A1 (en) * | 2003-06-25 | 2006-11-09 | Bernd Frey | Method for predicting the residual service life of an electric energy accumulator |
CN102749572A (en) * | 2012-06-26 | 2012-10-24 | 南京航空航天大学 | IECMAC parameter identification-based power electronic circuit failure predicting method |
CN102867132A (en) * | 2012-10-16 | 2013-01-09 | 南京航空航天大学 | Aviation direct-current converter online fault combined prediction method based on fractional order wavelet transformation |
CN102879728A (en) * | 2012-10-16 | 2013-01-16 | 南京航空航天大学 | Health evaluation index and failure predication method for DC (Direct Current)-DC convertor |
CN102968573A (en) * | 2012-12-14 | 2013-03-13 | 哈尔滨工业大学 | Online lithium ion battery residual life predicting method based on relevance vector regression |
CN103033761A (en) * | 2012-12-17 | 2013-04-10 | 哈尔滨工业大学 | Lithium ion battery residual life forecasting method of dynamic gray related vector machine |
CN103399279A (en) * | 2013-08-01 | 2013-11-20 | 哈尔滨工业大学 | Method for predicting cycle life of fused lithium ion battery based on EKF (Extended Kalman Filter) method and AR (AutoRegressive) model |
CN104537444A (en) * | 2015-01-13 | 2015-04-22 | 安徽理工大学 | Gas outburst predicting method based on EMD and ELM |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107491840A (en) * | 2017-08-21 | 2017-12-19 | 浙江理工大学 | The prediction of flowing wearing character and lifetime estimation method based on ELM neural network models |
CN107491840B (en) * | 2017-08-21 | 2020-06-05 | 浙江理工大学 | Flow wear characteristic prediction and service life evaluation method based on ELM neural network model |
CN109615003A (en) * | 2018-12-06 | 2019-04-12 | 哈尔滨工业大学 | A kind of power source failure prediction method based on ELM-CHMM |
CN109615003B (en) * | 2018-12-06 | 2022-11-08 | 哈尔滨工业大学 | Power failure prediction method based on ELM-CHMM |
CN111308960A (en) * | 2020-01-20 | 2020-06-19 | 广西玉柴机器股份有限公司 | Load monitoring method and system |
CN111308960B (en) * | 2020-01-20 | 2024-01-12 | 广西玉柴机器股份有限公司 | Load monitoring method and system |
CN113689044A (en) * | 2021-08-26 | 2021-11-23 | 北京航空航天大学 | Method and system for predicting residual service life of switching power supply |
CN116882303A (en) * | 2023-09-06 | 2023-10-13 | 深圳市联明电源有限公司 | Laser power supply life prediction method, system and storage medium |
CN116882303B (en) * | 2023-09-06 | 2023-12-22 | 深圳市联明电源有限公司 | Laser power supply life prediction method, system and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN105334472B (en) | 2018-06-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Park et al. | LSTM-based battery remaining useful life prediction with multi-channel charging profiles | |
CN105334472A (en) | Online remaining life prediction method for mining intrinsic safety power supply | |
CN105277896B (en) | Lithium battery method for predicting residual useful life based on ELM MUKF | |
CN102867132B (en) | The online fault combination forecasting method of aviation DC converter based on fractional wavelet transform | |
CN102663412B (en) | Power equipment current-carrying fault trend prediction method based on least squares support vector machine | |
CN103711523B (en) | Based on the gas density real-time predicting method of local decomposition-Evolutionary Neural Network | |
CN103793854B (en) | The overhead transmission line operation risk informatization evaluation method that Multiple Combination is optimized | |
CN102779230B (en) | State analysis and maintenance decision judging method of power transformer system | |
CN104318347A (en) | Power transmission line icing state assessment method based on information fusion of multiple sensors | |
CN109978229A (en) | The method that the full battery core multi-point temperature of a kind of pair of power battery pack and tie point temperature carry out thermal runaway prediction | |
CN104217104A (en) | Power transformer service life analysis method and system based on risk evaluation | |
Zhang et al. | A Wiener process model with dynamic covariate for degradation modeling and remaining useful life prediction | |
CN103810328A (en) | Transformer maintenance decision method based on hybrid model | |
CN105300692A (en) | Bearing fault diagnosis and prediction method based on extended Kalman filtering algorithm | |
CN110287543B (en) | Method for predicting service life of relay protection device | |
CN103675525A (en) | DC-DC converter health monitoring and fault prediction method based on multiple SVDD models | |
CN103942735A (en) | Method for evaluating relay protection states | |
CN104915552A (en) | Method and device for predicting system faults | |
CN104976139A (en) | Mechanical equipment state diagnosis method based on Gaussian model | |
CN105116344A (en) | Battery open circuit voltage estimation method based on binary coding | |
CN106203693A (en) | A kind of system and method for Power Output for Wind Power Field climbing event prediction | |
CN104615866A (en) | Service life prediction method based on physical statistic model | |
CN104318485A (en) | Power transmission line fault identification method based on nerve network and fuzzy logic | |
CN112883634A (en) | DC measurement system state prediction method and system based on multi-dimensional analysis | |
Wei et al. | Remaining useful life estimation based on gamma process considered with measurement error |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |