CN103336200A - Device and method for predicting power distribution cabinet electric health index - Google Patents

Abstract

The invention provides a device and a method for predicting a power distribution cabinet electric health index, and belongs to the technical field of power distribution cabinets. The device comprises a current sensor, a voltage sensor, a frequency sensor, an insulation test instrument, a data acquisition chip, a central processing unit, an industrial personal computer and a wireless communication module. According to the invention, the health status of a power distribution cabinet can be known, so that untimely or unnecessary black out test and overhaul are reduced, repair as required is realized, and reliability and economical efficiency of an electric power system are remarkably improved; prediction analysis is performed on the electric health index of a novel indoor power distribution cabinet which is manufactured and researched and developed independently by manufacturing enterprises, so that maintenance personnel can overhaul conveniently.

Description

The electric health index prediction unit of a kind of power distribution cabinet and method

Technical field

Belong to the power distribution cabinet technical field, particularly the electric health index prediction unit of a kind of power distribution cabinet and method.

Background technology

Relevant statistics shows, power distribution network maintenance cost over half be flower on power distribution cabinet, and 30% be light maintenance and regular maintenance for power distribution cabinet wherein; In addition according to statistics, 15% power distribution cabinet fault is because due to the incorrect maintenance, the overhaul of power distribution cabinet is disintegrated fully, both time-consuming, expense is very high again, can reach 1/3-1/2 of whole power distribution cabinet, and disintegrate and ressemble and can cause a lot of defectives, consequent accident example is too numerous to enumerate especially, for which parts (or critical elements) of power distribution cabinet, how long operation needs to change, be still the problem of a dispute, in fact in relatively more conservative at present scheduled overhaul, it is still functional when the back was upgraded in a lot of years that many parts operations take place often, and owing to find in time that not a certain parts defective occurs and cause the situation of power grid accident also to happen occasionally.So the prediction of the electric health index of power distribution cabinet plays significant role to power grid security.

Summary of the invention

At the deficiencies in the prior art, the present invention proposes the electric health index prediction unit of a kind of power distribution cabinet and method, reduces too early or unnecessary power failure test and maintenance to reach, and improves the purpose of the convenience of Power System Reliability, economy and maintenance.

The electric health index prediction unit of a kind of power distribution cabinet, comprise current sensor, voltage sensor, frequency sensor, insulation tester, data acquisition chip, central processing unit, industrial computer and wireless communication module, wherein, the output terminal of current sensor, the output terminal of voltage sensor, the output terminal of frequency sensor and the output terminal of insulation tester are connected four road input ends of data acquisition chip respectively, the output terminal of data acquisition chip connects the input end of central processing unit, and the two-way output terminal of central processing unit connects the input end of input end of industrial control machine and wireless communication module respectively.

Adopt the electric health index prediction unit of power distribution cabinet to carry out forecast method, may further comprise the steps:

Step 1, five parameters of insulating coefficient, operating voltage, dropout current, frequency, open and close times of gathering power distribution cabinet, and above-mentioned five parameters are sent in the data acquisition chip;

Step 2, data acquisition chip carry out analog to digital conversion to five parameters of gathering, and five parameters after the analog to digital conversion are sent in the data processor;

Step 3, data processor are predicted the electric health index of power distribution cabinet according to five parameters of gathering;

Step 3-1, employing empirical mode decomposition method are carried out resolution process to the time series that five parameters of insulating coefficient, operating voltage, dropout current, frequency, open and close times by power distribution cabinet constitute;

Step 3-2, employing recurrent neural network method are predicted decomposing the back time series, and based on the phase space reconfiguration theory, time series after decomposing is embedded in the phase space, employing mutual information method is determined the time delay in the phase space, adopt embedding dimension algorithm to determine the correlation dimension of phase space, and then obtain a plurality of different anticipation components;

Step 3-3, the Dynamical Recurrent Neural Networks structure is set, the a plurality of different anticipation component input Dynamical Recurrent Neural Networks that step 3-2 obtains is trained, and adopt the method for linear combination that all the different components after training are carried out linear combination, namely obtain the electric health index anticipation function of power distribution cabinet;

Step 4, data processor are sent to the electric health index of predicting of power distribution cabinet in the industrial computer and store, and are sent to the remote dispatching terminal by wireless communication module, so that the maintenance personal in time overhauls.

The described employing empirical mode decomposition of step 3-1 method is carried out resolution process to the time series that five parameters of insulating coefficient, operating voltage, dropout current, frequency, open and close times by power distribution cabinet constitute, and concrete steps are as follows:

Step 3-1-1, determine the Local Extremum that time series signal is all, and adopt the cubic spline interpolation curve that all local maximum points are coupled together to form the coenvelope line;

Step 3-1-2, adopt the cubic spline interpolation curve that all local minizing points are coupled together to form the lower envelope line, determine all data points of envelope envelope according to envelope up and down;

Step 3-1-3, the mean value of envelope is designated as m up and down ₁, obtain difference h ₁:

h ₁=z(t）-m ₁ (1)

Wherein, z (t)=[z ₁(t), z ₂(t) ..., z ₅(t)]; z ₁(t) data of the 1st input quantity in expression t five parameters of gathering constantly; z ₂(t) data of the 2nd input quantity in expression t five parameters of gathering constantly; z ₅(t) data of the 5th input quantity in expression t five parameters of gathering constantly;

If h ₁Maximal value h _MaxWith minimum value h _MinDifference less than a predefined value ε, ε≤0.01

|h _max-h _min|≤ε (2)

H then ₁A mode component for z (t); Otherwise, execution in step 3-1-4;

Step 3-1-4, with h ₁As another primary data, repeating step 3-1-1 recirculates k time, until h to step 3-1-3 _k=h _K-1-m _kMiddle h _kSatisfy the condition of formula (2), note c ₁=h _k, c then ₁First mode component for signal z (t);

Step 3-1-5, with c ₁From z (t), separate, obtain z (t) remainder r ₁:

r ₁=z(t)-c ₁ (3)

With r ₁To step 3-1-4, n recirculates as another primary data repeating step 3-1-1 _tInferior, obtain the n of signal z (t) _tIndividual mode component;

When the remainder of z (t) was a monotonic quantity, circulation finished, and signal z (t) is decomposed into n _tIndividual mode component With a residual amount

Sum, that is:

$z\left(t\right)=\underset{{\mathrm{<figure-callout\; id="1"\; label="i"\; filenames="HDA00003390198800032.png,HDA00003390198800042.png"\; state="\{\{state\}\}">i</figure-callout>}}_1=1}{\overset{n_t}{\mathrm{\&Sigma;}}}{c}_{i_1}+r_{n_t}$ (4)

Wherein, component

Represent the characteristic signal of the different time that comprises among the original signal z (t), established x (t), and order

X (t) has represented the central tendency of signal z (t).

Step 3-2 is described based on the phase space reconfiguration theory, time series after decomposing is embedded in the phase space, time series x (t) after being about to decompose is embedded into phase space X (t)={ x (t), x (t-τ), ..., x (t-(m-1) τ) } in, line number is the number of parameters of gathering in the described phase space, columns is correlation dimension;

Wherein, the dynamics state of etching system during X (t) expression t, τ is time delay, τ=1,2 ..., n ₁, n ₁Be natural number; M is the dimension of embedded space;

Described employing recurrent neural network method predicts decomposing the back time series, determines to certainly exist function F () in the phase space and make time series state after the time-delay And current state X(t) satisfy between:

$\hat{x}(t+\mathrm{\&tau;})=F\left(X\left(t\right)\right)$ (14)

F () is the electric health index anticipation function of power distribution cabinet to be sought;

And adopt interpolation method that sequence is expanded, obtain a plurality of different anticipation components:

$\hat{x}(t^{\&prime;}+(\mathrm{\&lambda;}+1)p)=F(x^{\&prime;}(t^{\&prime;}),x^{\&prime;}(t^{\&prime;}-\mathrm{\&tau;}),...,x^{\&prime;}(t^{\&prime;}-(m-1)\mathrm{\&tau;}))$ (16)

Wherein, λ is the interpolation multiple, and t' is the moment after the interpolation, p is the prediction step number, x'(t') be time series after the interpolation, x ' (t '-τ) be the time series that postpones the τ time after the interpolation, x ' (t'-(m-1) τ) be the time series of delay (m-1) τ time after the interpolation.

Advantage of the present invention:

The electric health index prediction unit of a kind of power distribution cabinet of the present invention and method, can understand the health status of power distribution cabinet, reduce too early or unnecessary power failure test and maintenance, accomplish to answer Xiu Zexiu, Power System Reliability and economy have been significantly improved, the electric health index of novel indoor power distribution cabinet of manufacturing enterprise being researched and developed and makes production voluntarily carries out forecast analysis, can the convenient for maintaining personnel overhaul.

Description of drawings

The electric health index prediction unit of the power distribution cabinet of Fig. 1 an embodiment of the present invention structured flowchart;

The electric health index prediction unit of the power distribution cabinet of Fig. 2 an embodiment of the present invention work synoptic diagram;

The A/D converter of the electric health index prediction unit of the power distribution cabinet of Fig. 3 an embodiment of the present invention and processor circuit schematic diagram;

The level-conversion circuit schematic diagram of Fig. 4 an embodiment of the present invention;

The electric health index Forecasting Methodology of the power distribution cabinet of Fig. 5 an embodiment of the present invention general flow chart;

The Dynamical Recurrent Neural Networks structural representation that adopts in Fig. 6 an embodiment of the present invention;

The employing EMD DRNN mixture model structure of Fig. 7 an embodiment of the present invention is carried out the process flow diagram of the electric health index prediction of power distribution cabinet;

The prediction health index curve of Fig. 8 an embodiment of the present invention and actual health index curve map.

Embodiment

Below in conjunction with accompanying drawing an embodiment of the present invention is described further.

As shown in Figure 1, the electric health index prediction unit of a kind of power distribution cabinet, comprise current sensor, voltage sensor, frequency sensor, insulation tester, data acquisition chip, central processing unit, industrial computer and wireless communication module, wherein, the output terminal of current sensor, the output terminal of voltage sensor, the output terminal of frequency sensor, the output terminal of insulation tester connects four road input ends of data acquisition chip respectively, the output terminal of data acquisition chip connects the input end of central processing unit, and the two-way output terminal of central processing unit connects the input end of input end of industrial control machine and wireless communication module respectively.

In the embodiment of the invention, power distribution cabinet cabinet model is ZGSM, and this power distribution cabinet used 10 years; As shown in Figure 2, by to the prediction of all-closed gas insulating power distribution cabinet electrical endurance, prediction result is sent by wireless module, be sent to the remote dispatching terminal by 3G network.

Signal acquisition module in the embodiment of the invention comprises voltage transformer (VT), current transformer, frequency sensor, insulation tester, wherein, voltage sensor adopts JDG4-0.5 1000/100 model, current sensor adopts LZJC-10Q1000/5 model current transformer, frequency sensor is selected the JLF21 model for use, be installed to the power distribution cabinet end of incoming cables, insulation tester is selected DL09-SDM50 for use, is positioned on the insulation shell of power distribution cabinet and gathers insulating coefficient.

In the embodiment of the invention, as shown in Figure 3, described data acquisition chip, i.e. A/D converter is selected the 12 bits serial A/D converter of TLC2044 model for use, and wherein, AIN0-AIN10 is analog input end; CS is sheet choosing end; DIN is the serial data input end; DOUT is the ternary serial output terminal of A/D transformation result; EOC is the EOC end; CLK is the I/O clock; REF+ is positive reference voltage terminal; REF-is negative reference voltage terminal; VCC is power supply; GND is ground.

In the embodiment of the invention, as shown in Figure 3, it is the single-chip microcomputer of CS8051 that central processing unit is selected model for use, and the serial port that uses single-chip microcomputer to carry can be realized the serial communication with industrial computer; The COM1 that present PC provides, COM2 adopts the RS-232 interface standard, and RS-232 comes the presentation logic state with generating positive and negative voltage, come the regulation of presentation logic state different with TTL with high-low level, therefore, in order to be connected with computer interface or with the TTL device (as single-chip microcomputer) of terminal, must between RS-232 and TTL circuit, carry out the conversion of level and logical relation, as shown in Figure 4, level-conversion circuit in the present embodiment is selected MAX232 for use, this device comprises 2 drivers, 2 receivers and a voltage generator circuit, this voltage generator circuit provides TIA/EIA-232-F level, this device meets the TIA/EIA-232-F standard, each receiver becomes 5V TTL/CMOS level with the TIA/EIA-232-F level conversion, each generator becomes the TIA/EIA-232-F level with the TTL/CMOS level conversion, single-chip microcomputer is the core of whole device, serial a/d converter TLC2044 gathers the simulating signal of input, sampling resolution, ALT-CH alternate channel and output polarity are selected by software, owing to be the serial input structure, can save 51 series monolithic I/O resources, the data that single-chip microcomputer is gathered are by serial ports RXD, TXD pin (10,11 pins) convert realization transmission between RS232 level and industrial computer to through MAX232.

In the embodiment of the invention, industrial computer is selected for use and is adopted UNO-3072 Series P entium M/Celeron M built-in industrial control machine.

In the embodiment of the invention, wireless communication module adopts H7000 series wireless communication system.

As shown in Figure 3, the output terminal of voltage transformer (VT), current transformer, frequency sensor, insulation tester is connected respectively to the input end AIN0-AIN3 of A/D converter TLC2044, output terminal EOC, the CLK of A/D converter TLC2044, DIN, DOUT are connected respectively to P10, P11, P12, the P13 of single-chip microcomputer, 10 pins (RXD) of single-chip microcomputer CS8051,11 pins (TXD) are connected with the T2in pin with the R2out pin of translation circuit MAX232, and the input end of industrial computer input end and wireless communication module is connected with single-chip microcomputer output terminal P01, P00 successively; The electric information of power distribution cabinet carries out synchronized sampling, maintenance, A/D conversion via corresponding device by sampling A, become digital signal, sending into single-chip microcomputer calculates with data and handles, link to each other with industrial computer and data are delivered to transport module by communication interface, for ready with the remote dispatching communication.

Adopt the electric health index prediction unit of described power distribution cabinet to carry out forecast method, process flow diagram may further comprise the steps as shown in Figure 5:

Step 1, five parameters of insulating coefficient, operating voltage, dropout current, frequency, open and close times of gathering power distribution cabinet, and above-mentioned five parameters are sent in the data acquisition chip;

Step 2, data acquisition chip carry out analog to digital conversion to five parameters of gathering, and five parameters after the analog to digital conversion are sent in the data processor;

Step 3, data processor are predicted the health index (residual life) of power distribution cabinet according to five parameters of gathering;

Step 3-1, adopt the empirical mode decomposition method (empirical mode decomposition, EMD) time series that five parameters of insulating coefficient, operating voltage, dropout current, frequency, open and close times by power distribution cabinet are constituted is carried out resolution process;

Gather closing time, operating voltage, dropout current, the frequency of power distribution cabinet, open and close times is as input quantity z (t)=[z ₁(t), z ₂(t) ..., z ₅(t)]; Wherein, z ₁(t) data of expression t the 1st input quantity of gathering constantly; z ₂(t) data of expression t the 2nd input quantity of gathering constantly; z ₅(t) data of expression t the 5th input quantity of gathering constantly.

In the embodiment of the invention, gather sample value and see Table 1;

Table 1

Gather sample	The collection value
		Insulating coefficient	1.02
Operating voltage	38kv
		Dropout current	35kA
Frequency	49.8hz
		Open and close times	2000 times

In the embodiment of the invention, adopt empirical mode decomposition method EMD method with time series z (t)=[z ₁(t), z ₂(t) ..., z ₅(t)] normalize between [0,1], and then decompose this sequence, become limited basic model component

N wherein _tBe natural number; Before to each basic model component prediction, still need each component is carried out normalized, to improve precision of prediction.

Adopt the empirical mode decomposition method to carry out resolution process, specific as follows:

Step 3-1-1, determine the Local Extremum that time series signal is all, and adopt the cubic spline interpolation curve that all local maximum points are coupled together to form the coenvelope line;

Step 3-1-2, adopt the cubic spline interpolation curve that all local minizing points are coupled together to form the lower envelope line, determine all data points of envelope envelope according to envelope up and down;

Step 3-1-3, the mean value of envelope is designated as m up and down ₁, obtain difference h ₁:

h ₁=z(t)-m ₁ （1）

If h ₁Maximal value h _MaxWith minimum value h _MinDifference less than a predefined value ε, ε≤0.01

|h _max-h _min|≤ε （2）

H then ₁A mode component for z (t); Otherwise execution in step 3-1-4;

Step 3-1-4, if h ₁Do not satisfy the condition of formula (2), then with h ₁As another primary data, repeating step 3-1-1 obtains the mean value m of envelope up and down to step 3-1-3 ₂, judge h again ₂=h ₁-m ₂Whether satisfy the condition of (2) formula, if do not satisfy, then recirculate k time, until h _k=h _K-1-m _kMake h _k(t) satisfy the condition of formula (2), note c ₁=h _k, c then ₁For first of signal z (t) satisfies the mode component of formula (2) condition;

Step 3-1-5, with c ₁After from z (t), separating, obtain z (t) remainder r ₁:

r ₁=z(t)-c ₁ （3）

With r ₁As the process of raw data repeating step 3-1-1 to step 3-1-4, obtain the 2nd the component c that satisfies formula (2) condition of z (t) ₂, n recirculates _tInferior, obtain the n of signal z (t) _tThe individual component that satisfies formula (2) condition, when the remainder of z (t) became a monotonic quantity and can not therefrom extract the component that satisfies formula (2) condition again, circulation finished, and any one signal z (t) is decomposed into n _tIndividual mode component With a residual amount

Sum, that is:

$z\left(t\right)=\underset{i_1=1}{\overset{n_{\mathrm{<figure-callout\; id="1"\; label="t\; c\; i"\; filenames="HDA00003390198800032.png,HDA00003390198800042.png"\; state="\{\{state\}\}">t</figure-callout>}}}{\mathrm{\&Sigma;}}}{c}_{i_{}}{{}_1}_{}+r_{n_t}$ （4）

Wherein, component Represent the characteristic signal of the different time that comprises in the original signal, established x (t), and order

X (t) has represented the central tendency of signal z (t).

Step 3-2, employing recurrent neural network method are predicted decomposing the back time series, and based on the phase space reconfiguration theory, time series after decomposing is embedded in the phase space, employing mutual information method is determined the time delay in the phase space, adopt embedding dimension algorithm to determine the correlation dimension of phase space, and then obtain a plurality of different anticipation components;

In the embodiment of the invention, carry out the Dynamical Recurrent Neural Networks chaos analysis, through Chaotic Time Series Analysis, the evolution of arbitrary component is determined by interactional other component with it in the system, implying the full detail of system in the evolutionary process of each component, unique observable be single argument one dimension time series x (t), D is the dimension of attractor, and the one dimension time series is embedded in the m-dimensional space:

X(t)={x(t),x(t-τ),...,x(t-(m-1)τ)} （5）

Wherein, the dynamics state of etching system during X (t) expression t, τ is time delay, τ=1,2 ..., n ₁, n ₁Be natural number; M is the dimension of embedded space, thereby can set up phase space R ^mTo embedded space R ^mMapping, m should satisfy during phase space reconstruction under having the situation of noise: m 〉=2D+1;

Step 3-2-1, employing mutual information method are determined delay time T;

The mutual information method be the time-delay that reaches for the first time hour with mutual information as the time delay of phase space reconfiguration, represent the uncertain degree of discrete random variable x with the closely related H of information (x),

$H\left(x\right)=-\underset{i}{\mathrm{\&Sigma;}}P\left(x_i\left(t\right)\right)\log P\left(x_i\left(t\right)\right)$ （6）

Wherein, x _i(t) i data volume in the expression x (t), i=1,2 ... 5; P (x _i(t)) be generation event x _i(t) probability is for two different stochastic variable X and Y, wherein

X={x ₁(t),x ₂(t),...,x ₅(t)}，Y={y ₁(t),y ₂(t),...,y ₅(t)}，y _j(t)=x(t+jτ)，j=1,2,...5；

The closely related H of conditional information (X|Y) of the Y of definable X is:

$H\left(X\right|Y)=-\underset{i,j}{\mathrm{\&Sigma;}}P\left(y_j\left(t\right)\right)P\left(x_i\left(t\right)\right|y_j\left(t\right))\log P\left(x_i\left(t\right)\right|y_j\left(t\right))$ (7)

Wherein, P (y _j(t)) be the probability of independent generation event, P (x _i(t) | y _j(t)) for being y in the generation event _j(t) event x under the condition _i(t) conditional probability of Fa Shenging, for whole variable X because the generation of variable Y and the correlativity of the two, make its uncertain moisture in the soil value that reduces be called mutual information moisture in the soil I (X, Y):

$I(X,Y)=\underset{x_i\left(t\right),y_i\left(t\right)}{\mathrm{\&Sigma;}}P(x_i\left(t\right),y_j\left(t\right))\ln \frac{P(x_i\left(t\right),y_j\left(t\right))}{P\left(x_i\left(t\right)\right)P\left(y_j\left(t\right)\right)}$ (8)

Wherein, P (x _i(t), y _j(t)) expression x _i(t) event and y _j(t) the simultaneous probability of event;

Mutual information is expressed as the function I (τ) of Yan Xi time τ, can gets:

$I\left(\mathrm{\&tau;}\right)=\underset{i=1\&RightArrow;5}{\mathrm{\&Sigma;}}P(x_i\left(t\right),x_{i+\mathrm{\&tau;}}\left(t\right))\ln \frac{P(x_i\left(t\right),x_{i+\mathrm{\&tau;}}\left(t\right))}{P\left(x_i\left(t\right)\right)P\left(x_{i+\mathrm{\&tau;}}\left(t\right)\right)}$ (9)

Wherein, x _{I+ τ}(t) expression x _i(t) postpone the value of τ after the time;

The τ of I (τ) when minimal value occurring first is time delay:

$\mathrm{\&tau;}=\min \left(\underset{x_i,y_i}{\mathrm{\&Sigma;}}P(x_i\left(t\right),y_j\left(t\right))\ln \frac{P(x_i\left(t\right),y_j\left(t\right))}{P\left(x_i\left(t\right)\right)P\left(x_j\left(t\right)\right)}\right)$ (10)

Step 3-2-2, employing embed the dimension algorithm and determine correlation dimension m;

According to the algorithm that embeds dimension, optional fixed reference point X from phase space M point _{I '}, calculate all the other M-1 points to X _{I '}Distance:

$v_{i^{\&prime;}{j}^{\&prime;}}={\left[\mathrm{\&Sigma;}{(X_{i^{\&prime;}}-X_{j^{\&prime;}})}^2\right]}^{\frac{1}{2}},i^{\&prime;}=\mathrm{1,2},...,M;j^{\&prime;}=\mathrm{1,2},...,M$ (11)

Wherein, M is natural number; Repeat this process, obtain correlation integral, any given distance value v, v=0.3 in the embodiment of the invention, checking has what to point (X _{I '}, X _{J '}) between distance less than v, distance less than v to point all to point in shared proportion be that to represent to have what state points be to be mutually related to correlation function, it is the dense degree of state point in the phase space, thereby also reflected correlation degree and the characteristics of motion degree of system motion, the relation between D and the v satisfies following formula:

$\underset{m\&RightArrow;\&infin;}{\lim }\frac{1}{m}\underset{i^{\&prime;},j^{\&prime;}=1}{\overset{m}{\mathrm{\&Sigma;}}}{H}_{\mathrm{\&theta;}}(v-v_{i^{\&prime;}{j}^{\&prime;}})={3v}^D$ (12)

Wherein, H _θ() is the Heaviside function; Then can be got by formula (11):

$D={\log }_v\frac{\underset{m\&RightArrow;\&infin;}{\lim }\frac{1}{m}\underset{i^{\&prime;}{,j}^{\&prime;}=1}{\overset{m}{\mathrm{\&Sigma;}}}{H}_{\mathrm{\&theta;}}(v-v_{i^{\&prime;}{j}^{\&prime;}})}{3}$ (13)

Then can obtain m=2D+1.

Certainly exist function F () in the phase space and make state after the time-delay

And satisfy between the current state X (t):

$\hat{x}(t+\mathrm{\&tau;})=F\left(X\left(t\right)\right)$ (14)

F () namely is anticipation function to be sought.

Time series X (t) is a limited discrete series, carries out the chaos time sequence multi-step prediction, uses interpolation method that sequence is expanded, and then carries out the analysis of chaos parameter, establishes η=1,2 ..., n ₂, n ₂Be natural number; At R ^mThere is function in the space, satisfies:

T'=t/ η wherein, η τ=p is the prediction step number;

In the embodiment of the invention, use the interpolation sequence spreading, obtain sequence:

X'(t')={x'(t')|t'=1,2,...,n'} (15)

It is carried out the sequence interpolation, and λ is the interpolation multiple, has inserted the value of one times of quantity in λ=1 expression sequence, and original time series X (t) becomes X'(t'), obtain a plurality of different anticipation components:

$\hat{x}(t^{\&prime;}+(\mathrm{\&lambda;}+1)p)=F(x^{\&prime;}(t^{\&prime;}),x^{\&prime;}(t^{\&prime;}-\mathrm{\&tau;}),...,x^{\&prime;}(t^{\&prime;}-(m-1)\mathrm{\&tau;}))$ (16)

Step 3-3, Dynamical Recurrent Neural Networks structure (Dynamical Recurrent Neural Networks is set, DRNN), the a plurality of different anticipation component input Dynamical Recurrent Neural Networks that step 3-2 obtains is trained, and adopt the method for linear combination that all the different components after training are carried out linear combination, namely obtain the electric health index anticipation function of power distribution cabinet

Neural network is simplified to the Dynamical Recurrent Neural Networks that three-layer network is used for direct multi-step prediction as shown in Figure 6, wherein 1 expression recurrence connection, 2 is the feedforward connection, wherein contain an input layer as cushion, a non-linear hidden layer and a linear output layer;

If being input as of network: X ' (t ')=x (t '), x (t '-τ) ..., x (t '-(m-1) τ) }, be output as G (t+p) namely In Dynamical Recurrent Neural Networks, the i of l layer the general matching formula of neuronic input and output is:

$\left\{\begin{array}{c}{s}_{i^{\&prime;\&prime;}}^{\left[l\right]}(t+p)=\underset{i^{\&prime;\&prime;},j^{\&prime;\&prime;}=1}{\overset{N^{\left[l\right]}}{\mathrm{\&Sigma;}}}{w}_{j^{\&prime;\&prime;}}^{\left[l\right]}{w}_{i^{\&prime;\&prime;}}^{\left[l\right]}{o}_{i^{\&prime;\&prime;}}^{\left[l\right]}\left(t\right)+\underset{i^{\&prime;\&prime;},j^{\&prime;\&prime;}=1}{\overset{N^{[l-1]}}{\mathrm{\&Sigma;}}}{w}_{[j^{\&prime;\&prime;},i^{\&prime;\&prime;}]}^{[l-1,l]}{o}_{j^{\&prime;\&prime;}}^{[l-1]}\left(t\right)b_{i^{\&prime;\&prime;}}^{\left[l\right]}\\ o_{i^{\&prime;\&prime;}}^{\left[l\right]}(t+p)={\mathrm{\&theta;}}_{i^{\&prime;\&prime;}}^{\left[l\right]}\left(s_{i^{\&prime;\&prime;}}^{\left[l\right]}(t+p)\right)\end{array}\right.$ (17)

Wherein,

With

Represent i " individual neuronic state and the output of l layer respectively;

Represent neuronic deviation;

Expression is linked at the individual neuron of j ' ' of l layer and the individual interneuronal weight of i ' ' of l-1 layer; , N ^[l]It is the neuronic number of l layer;

Be activation function,

The individual neuronic weight of j ' ' of representing the l layer;

The individual neuronic weight of i ' ' of representing the l layer; Represent the individual neuronic output of j ' ' of l-1 layer; I ' '=1...N ^[l]; J ' '=1...N ^[l];

In the embodiment of the invention, the input layer number of network is m, the last of hidden layer determined and can selected different neuron numbers, relatively determine on the basis of its performance index size, output layer node number is the output variable number of system to be identified, and the recurrent neural network structural model that is used for direct multi-step prediction is:

$\left\{\begin{array}{c}S(t^{\&prime;}+(\mathrm{\&lambda;}+1)p)=\mathrm{\&Theta;}(W^{[1\&RightArrow;2]}{s}_1^{\left[1\right]}\left(t^{\&prime;}\right)+W^{[1\&RightArrow;2]}{x}^{\&prime;}\left(t^{\&prime;}\right)+b^{\left[2\right]})\\ \hat{x}(t^{\&prime;}+(\mathrm{\&lambda;}+1)p)=(W^{[2\&RightArrow;3]}S(t^{\&prime;}+(\mathrm{\&lambda;}+1)p)+b^{\left[3\right]})\end{array}\right.$ (18)

Wherein, N of Θ () ^[2]The vector set of dimension, S (t'+ (λ+1) p) is the state set of hidden neuron;

$b^{\left[2\right]}=[b_1^{\left[2\right]},b_2^{\left[2\right]},...,b_N^{\left[2\right]}];$

$\left\{\begin{array}{c}{W}^{[1\&RightArrow;2]}{=\left[W_{\left[i^{\&prime;\&prime;,}{j}^{\&prime;\&prime;}\right]}^{\left[\mathrm{1,2}\right]}\right|i^{\&prime;\&prime;}=\mathrm{1,2},...,N^{\left[1\right]};j^{\&prime;\&prime;=\mathrm{1,2},...,N^{\left[2\right]}}]}^T\\ W^{[2\&RightArrow;2]}={\left[W_{\left[i^{\&prime;\&prime;,}{j}^{\&prime;}\right]}^{\left[\mathrm{2,2}\right]}\right|i^{\&prime;\&prime;}=\mathrm{1,2},...,N^{\left[2\right]};j^{\&prime;\&prime;}=\mathrm{1,2},...,N^{\left[2\right]}]}^T\\ W^{[2\&RightArrow;3]}={\left[W_{[i,j]}^{\left[\mathrm{2,3}\right]}\right|i=\mathrm{1,2},...,N^{\left[2\right]};j=\mathrm{1,2},...,N^{\left[3\right]}]}^T\end{array}\right.;$

The state set of hidden neuron is expressed as:

$S(t^{\&prime;}+(\mathrm{\&lambda;}+1)p)={[s_1^{\left[2\right]}\left(t^{\&prime;}(\mathrm{\&lambda;}+1)p\right),s_2^{\left[2\right]}\left(t^{\&prime;}(\mathrm{\&lambda;}+1)p\right),...,s_{N^{\left[l\right]}}^{\left[2\right]}\left(t^{\&prime;}(\mathrm{\&lambda;}+1)p\right)]}^T$ (19)

With predicting the outcome of all different components

Be grouped together to predict the next sample point of original time series linearly

$\hat{x}(t^{\&prime;}+p)=\underset{k^{\&prime;}=1}{\overset{n_t+1}{\mathrm{\&Sigma;}}}{\mathrm{<figure-callout\; id="1"\; label="w\; j"\; filenames="HDA00003390198800032.png,HDA00003390198800042.png"\; state="\{\{state\}\}">w</figure-callout>}}_{j_{}}^{{}_1}{\hat{x}}_{k^{\&prime;}}(t^{\&prime;}+p)$ (20)

The input layer number equals the number of multilayer neural network, and network does not have hidden layer, and a linearity output unit is arranged, and is used for determining in final predicting the outcome the weights of each component correspondence

J is natural number, can calculate by the contrary method of Moore-Penrose of asking formula (20).

With

As power distribution cabinet health index function, the time series that generates with collection capacity is input, calculates the power distribution cabinet health index and predicts the outcome;

Figure 7 shows that empirical mode and neural network mixture model structure, in the phase one,

Be the pattern resolving cell; I is interpolating unit, in the subordinate phase

Be DRNN structural design unit, in the phase III,

It is Dynamical Recurrent Neural Networks.

Step 4, data processor are sent to the power distribution cabinet health index of predicting in the industrial computer and store, and are sent to the remote dispatching terminal by wireless communication module, so that the maintenance personal in time overhauls.Power distribution cabinet health index forecast model predict the outcome with measured value more as shown in Figure 8, in the power distribution cabinet predictive index forecast model of setting up, the error of power distribution cabinet predictive index is controlled within ± 6% basically.

Claims (4)

1. electric health index prediction unit of power distribution cabinet, it is characterized in that: comprise current sensor, voltage sensor, frequency sensor, insulation tester, data acquisition chip, central processing unit, industrial computer and wireless communication module, wherein, the output terminal of current sensor, the output terminal of voltage sensor, the output terminal of frequency sensor and the output terminal of insulation tester are connected four road input ends of data acquisition chip respectively, the output terminal of data acquisition chip connects the input end of central processing unit, and the two-way output terminal of central processing unit connects the input end of input end of industrial control machine and wireless communication module respectively.

2. adopt the electric health index prediction unit of the described power distribution cabinet of claim 1 to carry out forecast method, it is characterized in that, may further comprise the steps:

Step 1, five parameters of insulating coefficient, operating voltage, dropout current, frequency, open and close times of gathering power distribution cabinet, and above-mentioned five parameters are sent in the data acquisition chip;

Step 2, data acquisition chip carry out analog to digital conversion to five parameters of gathering, and five parameters after the analog to digital conversion are sent in the data processor;

Step 3, data processor are predicted the electric health index of power distribution cabinet according to five parameters of gathering;

Step 3-1, employing empirical mode decomposition method are carried out resolution process to the time series that five parameters of insulating coefficient, operating voltage, dropout current, frequency, open and close times by power distribution cabinet constitute;

Step 3-2, employing recurrent neural network method are predicted decomposing the back time series, and based on the phase space reconfiguration theory, time series after decomposing is embedded in the phase space, employing mutual information method is determined the time delay in the phase space, adopt embedding dimension algorithm to determine the correlation dimension of phase space, and then obtain a plurality of different anticipation components;

Step 3-3, the Dynamical Recurrent Neural Networks structure is set, the a plurality of different anticipation component input Dynamical Recurrent Neural Networks that step 3-2 obtains is trained, and adopt the method for linear combination that all the different components after training are carried out linear combination, namely obtain the electric health index anticipation function of power distribution cabinet;

Step 4, data processor are sent to the electric health index of predicting of power distribution cabinet in the industrial computer and store, and are sent to the remote dispatching terminal by wireless communication module, so that the maintenance personal in time overhauls.

3. the electric health index prediction unit of employing power distribution cabinet according to claim 2 carries out forecast method, it is characterized in that: the described employing empirical mode decomposition of step 3-1 method is carried out resolution process to the time series that five parameters of insulating coefficient, operating voltage, dropout current, frequency, open and close times by power distribution cabinet constitute, and concrete steps are as follows:

Step 3-1-1, determine the Local Extremum that time series signal is all, and adopt the cubic spline interpolation curve that all local maximum points are coupled together to form the coenvelope line;

Step 3-1-2, adopt the cubic spline interpolation curve that all local minizing points are coupled together to form the lower envelope line, determine all data points of envelope envelope according to envelope up and down;

Step 3-1-3, the mean value of envelope is designated as m up and down ₁, obtain difference h ₁:

h ₁=z(t)-m ₁ (1)

Wherein, z (t)=[z ₁(t), z ₂(t) ..., z ₅(t)]; z ₁(t) data of the 1st input quantity in expression t five parameters of gathering constantly; z ₂(t) data of the 2nd input quantity in expression t five parameters of gathering constantly; z ₅(t) data of the 5th input quantity in expression t five parameters of gathering constantly;

If h ₁Maximal value h _MaxWith minimum value h _MinDifference less than a predefined value ε, ε≤0.01

|h _max-h _min|≤ε （2）

H then ₁A mode component for z (t); Otherwise, execution in step 3-1-4;

Step 3-1-4, with h ₁As another primary data, repeating step 3-1-1 recirculates k time, until h to step 3-1-3 _k=h _K-1-m _kMiddle h _kSatisfy the condition of formula (2), note c ₁=h _k, c then ₁First mode component for signal z (t);

Step 3-1-5, with c ₁From z (t), separate, obtain z (t) remainder r ₁:

r ₁=z(t)-c ₁ （3）

With r ₁To step 3-1-4, n recirculates as another primary data repeating step 3-1-1 _tInferior, obtain the n of signal z (t) _tIndividual mode component;

When the remainder of z (t) was a monotonic quantity, circulation finished, and signal z (t) is decomposed into n _tIndividual mode component

With a residual amount Sum, that is:

$z\left(t\right)=\underset{i_1=1}{\overset{n_t}{\mathrm{\&Sigma;}}}{c}_{i_1}+r_{n_t}$ （4）

Wherein, component

Represent the characteristic signal of the different time that comprises among the original signal z (t), established x (t), and order

X (t) has represented the central tendency of signal z (t).

4. the electric health index prediction unit of employing power distribution cabinet according to claim 2 carries out forecast method, it is characterized in that: step 3-2 is described based on the phase space reconfiguration theory, time series after decomposing is embedded in the phase space, time series x (t) after being about to decompose is embedded into phase space X (t)={ x (t), x (t-τ) ..., x (t-(m-1) τ) } in, line number is the number of parameters of gathering in the described phase space, and columns is correlation dimension;

Wherein, the dynamics state of etching system during X (t) expression t, τ is time delay, τ=1,2 ..., n ₁, n ₁Be natural number; M is the dimension of embedded space;

Described employing recurrent neural network method predicts decomposing the back time series, determines to certainly exist function F () in the phase space and make time series state after the time-delay

And satisfy between the current state X (t):

$\hat{x}(t+\mathrm{\&tau;})=F\left(X\left(t\right)\right)$ （14）

F () is the electric health index anticipation function of power distribution cabinet to be sought;

And adopt interpolation method that sequence is expanded, obtain a plurality of different anticipation components:

$\hat{x}(t\&prime;+(\mathrm{\&lambda;}+1)p)=F(x\&prime;(t\&prime;),x\&prime;(t\&prime;-\mathrm{\&tau;}),...,x\&prime;(t\&prime;-(m-1)\mathrm{\&tau;}))$ （16）

Wherein, λ is the interpolation multiple, and t ' is the moment after the interpolation, p is the prediction step number, x ' (t ') is the time series after the interpolation, x ' (t '-τ) be the time series that postpones the τ time after the interpolation, x ' (t '-(m-1) τ) be the time series of delay (m-1) τ time after the interpolation.

Images