CN105225020A - A kind of running status Forecasting Methodology based on BP neural network algorithm and system - Google Patents

Abstract

This application discloses a kind of running status Forecasting Methodology based on BP neural network algorithm and system, the method and system are applied to power information system, predict with the running status of the equipment to be predicted to power information system.Be specially multiple equipment indexes of the equipment to be predicted first determining power information system; Be that equipment index carries out repeatedly value according to historical data; Utilize this repeatedly value to preset BP neural network model train, obtain running status forecast model; The prediction running status of equipment to be predicted is obtained according to the physical device index of equipment to be predicted and running status forecast model.Operation maintenance personnel can, determining that according to prediction running status equipment to be predicted needs to overhaul in time during maintenance, thus can be avoided causing harmful effect to the normal operation of network system.

Description

A kind of running status Forecasting Methodology based on BP neural network algorithm and system

Technical field

The application relates to technical field of electric power, more particularly, relates to a kind of running status Forecasting Methodology based on BP neural network algorithm and system.

Background technology

Along with the development of electric system, power information system is progressively dissolved into the various aspects of power grid enterprises' production and operation, progressively becomes ingredient indispensable in production and operation link, becomes the key factor ensureing enterprise safety operation.Current present situation is that power information system quantity is many, platform is various, causes maintenance work heavy.Studying effective infosystem operation management method in industry always, to improve power information system operation management automatization level, but be no matter the installation of robotization operation and maintenance tools or fault self-recovery mechanism etc., all can not predict the running status of system, could can only find when breaking down, thus cause maintenance not in time, cause harmful effect to the normal operation of network system.

Summary of the invention

In view of this, the application provides a kind of running status Forecasting Methodology based on BP neural network algorithm and system, for predicting the running status of power information system, to enable operation maintenance personnel overhaul in time when power information system will break down, avoid causing harmful effect to the normal operation of network system.

To achieve these goals, the existing scheme proposed is as follows:

Based on a running status Forecasting Methodology for BP neural network algorithm, be applied to power information system, comprise step:

Determine multiple equipment indexes of the equipment to be predicted of described power information system;

Be that any described equipment index carries out repeatedly value according to historical data;

Utilize the described repeatedly value of described multiple equipment index to train the BP neural network model preset, obtain running status forecast model;

The physical device index of described equipment to be predicted is inputted described running status forecast model, obtains the prediction running status of described equipment to be predicted.

Optionally, described equipment to be predicted comprises the supply unit of described power information system;

Described multiple equipment index comprises redundancy condition, runs the time limit and aging conditions, outward appearance, alarm indicator state and familial product.

Optionally, described equipment to be predicted comprises the database of described power information system;

Described multiple equipment index comprises database performance, database table space, DataBase combining number, database backup system and database journal space.

Optionally, described prediction running status comprises severe exception status, abnormality, idea state or normal condition.

Based on a running status prognoses system for BP neural network algorithm, be applied to power information system, comprise:

Equipment index determination module, for determining multiple equipment indexes of the equipment to be predicted of described power information system;

Assignment module, for being that any described equipment index carries out repeatedly value according to historical data;

Training module, for utilizing described in described multiple equipment index repeatedly value to train the BP neural network model preset, obtains running status forecast model;

Prediction module, for the physical device index of described equipment to be predicted is inputted described running status forecast model, obtains the prediction running status of described equipment to be predicted.

Optionally, described equipment to be predicted comprises the supply unit of described power information system;

Described multiple equipment index comprises redundancy condition, runs the time limit and aging conditions, outward appearance, alarm indicator state and familial product.

Optionally, described equipment to be predicted comprises the database of described power information system;

Described multiple equipment index comprises database performance, database table space, DataBase combining number, database backup system and database journal space.

Optionally, described prediction running status comprises severe exception status, abnormality, idea state or normal condition.

As can be seen from above-mentioned technical scheme, this application discloses a kind of running status Forecasting Methodology based on BP neural network algorithm and system, the method and system are applied to power information system, predict with the running status of the equipment to be predicted to power information system.Be specially multiple equipment indexes of the equipment to be predicted first determining power information system; Be that equipment index carries out repeatedly value according to historical data; Utilize this repeatedly value to preset BP neural network model train, obtain running status forecast model; The prediction running status of equipment to be predicted is obtained according to the physical device index of equipment to be predicted and running status forecast model.Operation maintenance personnel can, determining that according to prediction running status equipment to be predicted needs to overhaul in time during maintenance, thus can be avoided causing harmful effect to the normal operation of network system.

Accompanying drawing explanation

In order to be illustrated more clearly in the embodiment of the present application or technical scheme of the prior art, be briefly described to the accompanying drawing used required in embodiment or description of the prior art below, apparently, accompanying drawing in the following describes is only some embodiments of the application, for those of ordinary skill in the art, under the prerequisite not paying creative work, other accompanying drawing can also be obtained according to these accompanying drawings.

The flow chart of steps of a kind of running status Forecasting Methodology based on BP neural network algorithm that Fig. 1 provides for the embodiment of the present application;

A kind of simple network BP algorithm model that Fig. 2 provides for the application;

A kind of frequency of training that Fig. 3 provides for the application and relationship by objective (RBO) figure;

The structured flowchart of a kind of running status prognoses system based on BP neural network algorithm that Fig. 4 provides for another embodiment of the application.

Embodiment

Below in conjunction with the accompanying drawing in the embodiment of the present application, be clearly and completely described the technical scheme in the embodiment of the present application, obviously, described embodiment is only some embodiments of the present application, instead of whole embodiments.Based on the embodiment in the application, those of ordinary skill in the art are not making the every other embodiment obtained under creative work prerequisite, all belong to the scope of the application's protection.

Embodiment one

The flow chart of steps of a kind of running status Forecasting Methodology based on BP neural network algorithm that Fig. 1 provides for the embodiment of the present application.

BP (BackPropagation) network is proposed by the scientist group headed by Rumelhart and McCelland for 1986, being a kind of Multi-layered Feedforward Networks by Back Propagation Algorithm training, is one of current most widely used neural network model.BP network can learn and store a large amount of input-output mode map relations, and without the need to disclosing the math equation describing this mapping relations in advance.Its learning rules use method of steepest descent, constantly adjusted the weights and threshold of network, make the error sum of squares of network minimum by backpropagation.Fig. 2 shows a kind of BP neural network model of simple network.

Suppose there be P training sample, existing P inputoutput pair (I _p, T _p), p=1,2 ... P.Wherein, input vector is: I _p=(i _p1..., i _pm) ^t, target output vector is T _p=(t _p1..., t _pn) ^t, network output vector (in theory):

O _p＝(o _p1,...,o _pn) ^T(1)

Note w _ijfor from the jth of input vector (j=1 ..., m) individual to output vector i-th (i=1 ..., the n) weight of individual component.Usual theoretical value and actual value have certain error, e-learning then refer to constantly with compare, and according to minimum principle amendment parameter w _ij, error sum of squares is reached minimum:

$min\underset{i=1}{\overset{n}{\Σ}}{(t_{pi}-o_{pi})}^2(p=1,...,P)---\left(2\right)$

Delta learning principle:

Note Δ w _ijrepresent recursion index word once, then have:

$w_{\mathrm{ij}}+\mathrm{\Δ}{w}_{\mathrm{ij}}\⇒w_{\mathrm{ij}}$

${\mathrm{\Δw}}_{ij}=\underset{p=1}{\overset{P}{\Σ}}\mathrm{\η}(t_{pi}-o_{pi})i_{pj}=\underset{p=1}{\overset{P}{\Σ}}{\mathrm{\η\δ}}_{pi}{i}_{pj}---\left(4\right)$

δ _pi＝t _pi-o _pi(5)

η becomes learning efficiency.

Note: from (1) formula, i-th neuronic output is:

i _pm=-1, w _im=(i-th neuronic threshold value) (6)

It is special in f is linear function,

$o_{pi}=a(\underset{j=1}{\overset{m}{\Σ}}{w}_{ij}{i}_{pj})+b---\left(7\right)$

According to above-described neural network, if wherein each neuron is linear, getting training quota is:

$E=\underset{p=1}{\overset{P}{\Σ}}{E}_p---\left(8\right)$

$E_p=\frac{1}{2}\underset{i=1}{\overset{n}{\Σ}}{(t_{pi}-o_{pi})}^2---\left(9\right)$

Time, ask the gradient steepest descent method of the minimum value of E to be exactly Delta learning rules.

As shown in Figure 1, the running status Forecasting Methodology that the present embodiment provides is applicable to power information system, specifically comprises step:

S101: the multiple equipment indexes determining equipment to be predicted.

Namely from needing to choose multiple equipment index that can reflect the operation conditions of this equipment the equipment relevant device index of prediction.For the power module of power information system, choose the following equipment index of power module:

Redundancy condition: refer to power module whether redundancy, with or without configuring redundancy power supply

Run the time limit and aging conditions: refer to the time that power supply puts into operation and ageing equipment situation

Outward appearance: whether power supply outward appearance is clean and with or without laying dust

Alarm indicator state: power alarm pilot lamp has no alarm situation

Familial product: same model and with the open hidden danger defect of batch products, announce accessory life situations.

Power supply status amount evaluation content and score value are in table 1, and power supply status evaluation is carried out in the mode quantized, and full marks are 50 points, and the importance according to evaluation content distributes score value.

The fractional value that associated specialist provides because power module has alarm to be equal to single supply equipment, to play two status values equal, mark is all 15 points, and remaining is all 5 points, is artificial distribution.

Evaluation content	Quantity of state	Mark
			Redundancy condition	Power module whether redundancy, with or without configuring redundancy power supply	15
Run the time limit and aging conditions	The time that power supply puts into operation and ageing equipment situation	10
			Visual condition	Whether power supply outward appearance is clean and with or without laying dust	5
Alarm indicator state	Power alarm pilot lamp has no alarm situation	15
			Familial product	Same model and with the open hidden danger defect of batch products, announce accessory life situations	5

Table 1 power supply status amount evaluation content and mark

S102: be that arbitrary equipment index carries out repeatedly value according to historical data.

Suppose i _p1, i _p2, i _p3, i _p4, i _p5represent the redundancy condition of power supply respectively, run the quantity of state evaluation score value of the time limit and aging conditions, visual condition, alarm indicator situation and familial product situation, each value value 20 times, adjustable according to actual needs, the size of value is the numerical value after the normalized drawn in repair based on condition of component process.

S103: utilize equipment index to train the BP neural network model preset.

From above, from repair based on condition of component process, obtain nearest 20 values, i.e. I _p1=(i _p1,1..., i _p1,20) ^t, in hidden layer, neuronic number is 4, and the input function of hidden layer is transig, and the activation function of output layer is %trsnsig, and training function is Gradient Descent function, i.e. above-described standard learning algorithm:

$P=\left[\begin{array}{c}{i}_{p_{1,1}},\mathrm{...},i_{p_{1,20}}\\ i_{p_{2,1}},\mathrm{...},i_{p_{2,20}}\\ .\\ .\\ .\\ i_{p_{5,1}},\mathrm{...},i_{p_{5,20}}\end{array}\right]=\left[\begin{array}{cccccccccccccccccccc}0.33& 0.67& 0.67& 0.67& 0.67& 0.33& 0.67& 0.67& 0.33& 0.67& 1.00& 0.67& 0.33& 0.00& 0.00& 0.00& 0.33& 0.33& 0.67& 1.00\\ 0.50& 0.50& 1.00& 1.00& 1.00& 1.00& 1.00& 1.00& 1.00& 1.00& 0.50& 0.50& 0.50& 0.50& 0.00& 0.00& 0.00& 0.50& 0.50& 1.00\\ 0.00& 0.00& 0.00& 0.00& 0.00& 0.00& 0.00& 0.00& 0.00& 0.50& 0.50& 0.50& 0.50& 0.50& 0.50& 0.50& 1.00& 1.00& 1.00& 1.00\\ 0.33& 0.67& 0.67& 0.67& 0.67& 0.33& 0.67& 0.67& 0.33& 0.67& 1.00& 0.67& 0.33& 0.00& 0.00& 0.00& 0.33& 0.33& 0.67& 1.00\\ 0.00& 0.00& 0.00& 0.00& 0.00& 0.00& 0.00& 0.00& 0.00& 0.50& 0.50& 0.50& 0.50& 0.50& 0.50& 0.50& 1.00& 1.00& 1.00& 1.00\end{array}\right]$

T＝[0.950.950.950.950.950.950.950.950.950.950.950.950.950.950.950.950.950.950.950.95]

net＝newff(([01；01；01；01；01]),[5,4,1],{‘tansig’,‘logsig’},’traingd’)；

net.trainParam.epochs＝15000；

net.trainParam.goal＝0.01；

LP.lr＝0.1；

Net＝train(net,P,T)；

From Fig. 3, we can find out that the error that training sample can reach with target fractional 0.95 is 0.01 through 184 training, and error can be passed through to increase to run step number and improve default error precision industry and reduce further.

Meanwhile, we can obtain each neuronic weighted value, and wherein input layer is to hidden layer weights V=net.iw{1,1};

$V=\left[\begin{array}{ccccc}2.4868& 1.4480& 0.6377& -0.5209& -2.4425\\ -1.7060& -0.2762& 1.8524& 2.7631& -0.9336\\ 0.4958& -2.2344& 1.9574& 1.9346& 1.4533\\ -0.0847& 1.9427& 1.4398& -0.5423& -2.9625\\ 2.0452& -0.2890& -1.6920& 2.0575& -1.8874\end{array}\right]$

Hidden layer to output layer weights W=net.lw{2,1},

W＝[-1.4433-1.46290.5040-1.1062-1.4625]

S104: the prediction running status obtaining equipment to be predicted.

In actual maintenance process, the equipment index can treating predict device carries out value, and substitutes into above-mentioned running status forecast model, is predicted the outcome.

Situation is patrolled and examined when what provide a power supply, comprise redundancy condition, run the time limit and aging conditions, visual condition, alarm indicator situation and familial product situation, can predict whether this power module can reach the standard of 0.95 point above according to the weighted value drawn, thus arrange corresponding service work.

The power module performance evaluation table that the corresponding relation of mark and evaluation content is shown in Table 2.

Evaluation content	Kilter	Normal condition	Attention state	Abnormality	Severe exception status
						Target output value	0.98-1	0.95-0.98	0.90-0.95	0.85-0.90	≤0.85

Table 2

When overhauling power module, by the maintenance classification shown in table 3 and repertory.

Maintenance classification	Overhauling project
		Category-A overhauls	A.1 integral device is changed
Category-B overhauls	B.1 port is changed B.2 power supply and is changed B.3 port and change B.4 diagnostic test
		C class is overhauled	C.1 C.2 traditional performance test is cleaned, checks, is keeped in repair
D class is overhauled	D.1 D.2 live testing keeps in repair, maintains D.3 maintainer's specialty inspection tour

Table 3

Table 3 gives maintenance classification and corresponding overhauling project, and the condition selecting that can run according to equipment selects corresponding Strategies of Maintenance:

" severe exception status " Strategies of Maintenance: Whole Equipment state is severe conditions, maintenance suggestion is for performing category-A maintenance.

" abnormality " Strategies of Maintenance: Whole Equipment state is abnormality, maintenance suggestion is for performing category-B maintenance.

" attention state " Strategies of Maintenance: Whole Equipment state is attention state, maintenance suggestion is for performing the maintenance of C class.

" normal condition " Strategies of Maintenance: Whole Equipment state is normal condition, maintenance suggestion is for performing the maintenance of D class.

As can be seen from technique scheme, present embodiments provide a kind of running status Forecasting Methodology based on BP neural network algorithm, the method is applied to power information system, predicts with the running status of the equipment to be predicted to power information system.Be specially multiple equipment indexes of the equipment to be predicted first determining power information system; Be that equipment index carries out repeatedly value according to historical data; Utilize this repeatedly value to preset BP neural network model train, obtain running status forecast model; The prediction running status of equipment to be predicted is obtained according to the physical device index of equipment to be predicted and running status forecast model.Operation maintenance personnel can, determining that according to prediction running status equipment to be predicted needs to overhaul in time during maintenance, thus can be avoided causing harmful effect to the normal operation of network system.

At present, database positioning service work progressively realizes robotization, the robotization repair based on condition of component index that can complete multiple capacity class index and performance class index on the database of O&M is disposed, and has possessed discovery in advance and the data analysis capabilities of capacity and partial properties problem.Database positioning primary evaluation content and score value as shown in table 4.

Evaluation content	Quantity of state	Mark
			Database performance	Session number, things number per second, Database lock quantity, deadlock quantity etc.	15
Database table space	Table space state, utilization rate, use space, surplus ratio etc.	15
			DataBase combining number	User's call-rate, linking number, connection utilization factor	10
Database backup system	Can database data back up in time, and can filing space remove in time	5
			Database journal space	Whether database journal space is full	5

Table 4

The same, suppose i _p1, i _p2, i _p3, i _p4, i _p5distinguish the evaluation score of the state variables such as the performance of representation database, table space, linking number, standby system and log space situation, each state variable evaluation score value 20 times, adjustable according to actual needs, the size of value is the numerical value after the normalized drawn in repair based on condition of component process.Also can obtain optimal training sequence according to Delta learning rules, thus judge that whether database positioning is abnormal, thus select corresponding Strategies of Maintenance according to database health status.

Embodiment two

The structured flowchart of a kind of running status prognoses system based on BP neural network algorithm that Fig. 4 provides for the embodiment of the present application.

As shown in Figure 4, the running status prognoses system that the present embodiment provides is applicable to power information system, specifically comprises equipment index determination module 10, assignment module 20, training module 30 and prediction module 40.

Equipment index determination module 10 is for determining multiple equipment indexes of equipment to be predicted.

Namely from needing to choose multiple equipment index that can reflect the operation conditions of this equipment the equipment relevant device index of prediction.For the power module of power information system, choose the following equipment index of power module:

Redundancy condition: refer to power module whether redundancy, with or without configuring redundancy power supply

Run the time limit and aging conditions: refer to the time that power supply puts into operation and ageing equipment situation

Outward appearance: whether power supply outward appearance is clean and with or without laying dust

Alarm indicator state: power alarm pilot lamp has no alarm situation

Familial product: same model and with the open hidden danger defect of batch products, announce accessory life situations.

Power supply status amount evaluation content and score value are shown in it is the table 1 of a upper embodiment, and power supply status evaluation is carried out in the mode quantized, and full marks are 50 points, and the importance according to evaluation content distributes score value.

The fractional value that associated specialist provides because power module has alarm to be equal to single supply equipment, to play two status values equal, mark is all 15 points, and remaining is all 5 points, is artificial distribution.

Assignment module 20 is for being that arbitrary equipment index carries out repeatedly value according to historical data.

Suppose i _p1, i _p2, i _p3, i _p4, i _p5represent the redundancy condition of power supply respectively, run the quantity of state evaluation score value of the time limit and aging conditions, visual condition, alarm indicator situation and familial product situation, each value value 20 times, adjustable according to actual needs, the size of value is the numerical value after the normalized drawn in repair based on condition of component process.

Training module 30 is trained the BP neural network model preset for utilizing equipment index.

From above, from repair based on condition of component process, obtain nearest 20 values, i.e. Ip1=(ip ₁, 1 ..., ip1,2 ₀) ^t, in hidden layer, neuronic number is 4, and the input function of hidden layer is transig, and the activation function of output layer is %trsnsig, and training function is Gradient Descent function, i.e. above-described standard learning algorithm:

$P=\left[\begin{array}{c}{i}_{p_{1,1}},\mathrm{...},i_{p_{1,20}}\\ i_{p_{2,1}},\mathrm{...},i_{p_{2,20}}\\ .\\ .\\ .\\ i_{p_{5,1}},\mathrm{...},i_{p_{5,20}}\end{array}\right]=\left[\begin{array}{cccccccccccccccccccc}0.33& 0.67& 0.67& 0.67& 0.67& 0.33& 0.67& 0.67& 0.33& 0.67& 1.00& 0.67& 0.33& 0.00& 0.00& 0.00& 0.33& 0.33& 0.67& 1.00\\ 0.50& 0.50& 1.00& 1.00& 1.00& 1.00& 1.00& 1.00& 1.00& 1.00& 0.50& 0.50& 0.50& 0.50& 0.00& 0.00& 0.00& 0.50& 0.50& 1.00\\ 0.00& 0.00& 0.00& 0.00& 0.00& 0.00& 0.00& 0.00& 0.00& 0.50& 0.50& 0.50& 0.50& 0.50& 0.50& 0.50& 1.00& 1.00& 1.00& 1.00\\ 0.33& 0.67& 0.67& 0.67& 0.67& 0.33& 0.67& 0.67& 0.33& 0.67& 1.00& 0.67& 0.33& 0.00& 0.00& 0.00& 0.33& 0.33& 0.67& 1.00\\ 0.00& 0.00& 0.00& 0.00& 0.00& 0.00& 0.00& 0.00& 0.00& 0.50& 0.50& 0.50& 0.50& 0.50& 0.50& 0.50& 1.00& 1.00& 1.00& 1.00\end{array}\right]$

T＝[0.950.950.950.950.950.950.950.950.950.95０.950.950.950.950.950.950.950.950.950.95]

net＝newff(([01；01；01；01；01]),[5,4,1],{‘tansig’,‘logsig’},’traingd’)；

net.trainParam.epochs＝15000；

net.trainParam.goal＝0.01；

LP.lr＝0.1；

Net＝train(net,P,T)；

From Fig. 3, we can find out that the error that training sample can reach with target fractional 0.95 is 0.01 through 184 training, and error can be passed through to increase to run step number and improve default error precision industry and reduce further.

Meanwhile, we can obtain each neuronic weighted value, and wherein input layer is to hidden layer weights V=net.iw{1,1};

$V=\left[\begin{array}{ccccc}2.4868& 1.4480& 0.6377& -0.5209& -2.4425\\ -1.7060& -0.2762& 1.8524& 2.7631& -0.9336\\ 0.4958& -2.2344& 1.9574& 1.9346& 1.4533\\ -0.0847& 1.9427& 1.4398& -0.5423& -2.9625\\ 2.0452& -0.2890& -1.6920& 2.0575& -1.8874\end{array}\right]$

Hidden layer to output layer weights W=net.lw{2,1},

W＝[-1.4433-1.46290.5040-1.1062-1.4625]

Prediction module 40 is for obtaining the prediction running status of equipment to be predicted.

In actual maintenance process, the equipment index can treating predict device carries out value, and substitutes into above-mentioned running status forecast model, is predicted the outcome.

Situation is patrolled and examined when what provide a power supply, comprise redundancy condition, run the time limit and aging conditions, visual condition, alarm indicator situation and familial product situation, can predict whether this power module can reach the standard of 0.95 point above according to the weighted value drawn, thus arrange corresponding service work.

The corresponding relation of mark and evaluation content is shown in the power module performance evaluation table shown in table 2 of a upper embodiment.

When overhauling power module, by a upper embodiment table 3 shown in maintenance classification and repertory.

Table 3 gives maintenance classification and corresponding overhauling project, and the condition selecting that can run according to equipment selects corresponding Strategies of Maintenance:

" severe exception status " Strategies of Maintenance: Whole Equipment state is severe conditions, maintenance suggestion is for performing category-A maintenance.

" abnormality " Strategies of Maintenance: Whole Equipment state is abnormality, maintenance suggestion is for performing category-B maintenance.

" attention state " Strategies of Maintenance: Whole Equipment state is attention state, maintenance suggestion is for performing the maintenance of C class.

" normal condition " Strategies of Maintenance: Whole Equipment state is normal condition, maintenance suggestion is for performing the maintenance of D class.

As can be seen from technique scheme, present embodiments provide a kind of running status prognoses system based on BP neural network algorithm, this system is applied to power information system, predicts with the running status of the equipment to be predicted to power information system.Be specially multiple equipment indexes of the equipment to be predicted first determining power information system; Be that equipment index carries out repeatedly value according to historical data; Utilize this repeatedly value to preset BP neural network model train, obtain running status forecast model; The prediction running status of equipment to be predicted is obtained according to the physical device index of equipment to be predicted and running status forecast model.Operation maintenance personnel can, determining that according to prediction running status equipment to be predicted needs to overhaul in time during maintenance, thus can be avoided causing harmful effect to the normal operation of network system.

In this instructions, each embodiment adopts the mode of going forward one by one to describe, and what each embodiment stressed is the difference with other embodiments, between each embodiment identical similar portion mutually see.To the above-mentioned explanation of the disclosed embodiments, professional and technical personnel in the field are realized or uses the application.To be apparent for those skilled in the art to the multiple amendment of these embodiments, General Principle as defined herein when not departing from the spirit or scope of the application, can realize in other embodiments.Therefore, the application can not be restricted to these embodiments shown in this article, but will meet the widest scope consistent with principle disclosed herein and features of novelty.

Claims (8)

1., based on a running status Forecasting Methodology for BP neural network algorithm, be applied to power information system, it is characterized in that, comprise step:

Determine multiple equipment indexes of the equipment to be predicted of described power information system;

According to historical data for equipment index described in each carries out repeatedly value;

Utilize the described repeatedly value of described multiple equipment index to train the BP neural network model preset, obtain running status forecast model;

The physical device index of described equipment to be predicted is inputted described running status forecast model, obtains the prediction running status of described equipment to be predicted.

2. running status Forecasting Methodology as claimed in claim 1, it is characterized in that, described equipment to be predicted comprises the supply unit of described power information system;

Described multiple equipment index comprises redundancy condition, runs the time limit and aging conditions, outward appearance, alarm indicator state and familial product.

3. running status Forecasting Methodology as claimed in claim 1, it is characterized in that, described equipment to be predicted comprises the database of described power information system;

Described multiple equipment index comprises database performance, database table space, DataBase combining number, database backup system and database journal space.

4. the running status Forecasting Methodology as described in any one of claims 1 to 3, is characterized in that, described prediction running status comprises severe exception status, abnormality, idea state or normal condition.

5., based on a running status prognoses system for BP neural network algorithm, be applied to power information system, it is characterized in that, comprising:

Equipment index determination module, for determining multiple equipment indexes of the equipment to be predicted of described power information system;

Assignment module, for being that any described equipment index carries out repeatedly value according to historical data;

Training module, for utilizing described in described multiple equipment index repeatedly value to train the BP neural network model preset, obtains running status forecast model;

Prediction module, for the physical device index of described equipment to be predicted is inputted described running status forecast model, obtains the prediction running status of described equipment to be predicted.

6. running status prognoses system as claimed in claim 5, it is characterized in that, described equipment to be predicted comprises the supply unit of described power information system;

Described multiple equipment index comprises redundancy condition, runs the time limit and aging conditions, outward appearance, alarm indicator state and familial product.

7. running status prognoses system as claimed in claim 5, it is characterized in that, described equipment to be predicted comprises the database of described power information system;

Described multiple equipment index comprises database performance, database table space, DataBase combining number, database backup system and database journal space.

8. the running status prognoses system as described in any one of claim 5 ~ 7, is characterized in that, described prediction running status comprises severe exception status, abnormality, idea state or normal condition.