A kind of rural power grids level of intelligence evaluation method neural network based
Technical field
The present invention relates to a kind of integrated evaluating methods, more particularly, to a kind of rural power grids intelligence neural network based
Level evaluation method.
Background technique
Agricultural civilization is modern civilization, the foundation of urban civilization, is cultural root arteries and veins and spiritual home.In modernization construction
Obvious to all, faith missing, culture sand are impacted in the contemporary society of rapid development, industrialization, urbanization caused by traditional culture
Desert allows modern not know where is returned in township, looks for less than " happiness ".The beautiful rural area of creation is exactly in this social economy and culture
The new issue proposed under historical background, and in beautiful rural forming process, electric energy is indispensable link, is rural area
It is that rural environment is promoted, use can experience the important step promoted under electric energy substitutes to an important ring for beautiful rural transformation.
On the basis of pushing forward rural power " transformation, reform and same price " comprehensively, " new countryside, new electric power, new service " is vigorously implemented
Rural power development strategy, accelerates building and upgrading of rural power grids paces, implements " being powered in every family " engineering, it is electrified to carry out new rural village
Construction, starting upgrade of rural power grids upgrade engineering, are made that positive contribution for the fast development of rural economy society.
By studying rural power grids overall merit, the assessment such as new technology, new energy is focused on, is conducive to
The construction of power grid itself, and be conducive to the development of entire power industry, national economy sustainable health development and society
It can harmonious stable development.
Summary of the invention
In view of this, in view of the deficiencies of the prior art, it is an object of the present invention to provide a kind of rural areas neural network based
Power grid level of intelligence evaluation method can accurately make scientific and reasonable analysis to rural power grids intelligent level.
In order to achieve the above objectives, the invention adopts the following technical scheme:
A kind of rural power grids level of intelligence evaluation method neural network based, comprising the following steps:
(1) index system of rural power grids overall merit, including first class index and corresponding with first class index two are constructed
Grade index;
(2) weight is calculated using entropy assessment;
(3) weight is calculated using analytic hierarchy process (AHP);
(4) weight is calculated using neural network;
(5) result and step for the analytic hierarchy process (AHP) weight for obtaining entropy assessment weight that step (2) obtains, step (3)
(4) method of weighting obtained compares, and verifies the feasibility of neural network.
Further, the first class index include power supply quality, power supply capacity, grid equipment, operation of power networks, informationization,
Automation, interactive, DG/ microgrid access, new technology and clean and environmental protection.
Further, the power supply quality in first class index includes rate of qualified voltage, power supply reliability and rate of qualified voltage etc.
Two-level index;
Power supply capacity in first class index includes capacity load ratio of network, substation's load factor qualification rate, main transformer " N-1 " qualification
Rate, middle pressure and low-voltage circuit load factor qualification rate, middle pressure and low-voltage circuit power supply capacity deficiency ratio, medium-voltage line can turn to supply
Rate qualification rate and 10kV outgoing line interval utilization rate;
Grid equipment in first class index includes on-load voltage regulation main transformer ratio, low-voltage distribution equipment one kind equipment ratio, opens
Close oil-free rate, energy-saving ratio-change, middle pressure and low-voltage circuit one kind equipment ratio, line insulation rate, the line conductor of matching is cut
Face standardized rate;
Operation of power networks in first class index includes medium-voltage line and low-voltage circuit high load rate route specific gravity, high load rate
Main transformer and distribution transforming specific gravity, three-phase imbalance platform area accounting;
Informationization in first class index includes substation and power supply station's optical fiber coverage rate, communication line percentage reserve, has letter
Company's ratio at county level, the power distribution station ratio with communication function of breathization Horizontal communication;
The corresponding two-level index of automation in first class index includes the unattended rate of electric substation, dispatch automated system
Realize that the route ratio of distribution automation, substation have five distant function ratios, substation is idle automatic throwing in practical, county town
Cutting apparatus coverage rate, 100kVA and above customer charge monitoring ratio, relay equipment popularize installation rate, platform area and user's low pressure
Aftercurrent protecting equipment operational percentage;
Interactive corresponding two-level index in first class index includes that successfully repair rate, self-help charging rate, electricity consumption are complained
Rate, the functionization of 95598 customer service systems and intelligent electric meter utility ratio;
It includes generation of electricity by new energy permeability, distributed generation resource that DG/ microgrid in first class index, which accesses corresponding two-level index,
Utilization rate and new energy installation rate, distributed generation resource Capacity Margin, distributed generation resource rate of qualified voltage, electric car low ebb fill
Electric rate, electric car peak discharge rate;
The corresponding two-level index of new technology in first class index includes realizing that the control of whole network voltage reactive comprehensive, development are set
Standby on-line monitoring and repair based on condition of component, development 10kV power distribution live-wire, development harmonic synthesis improvement and power distribution network self-healing control;
The corresponding two-level index of environmental index in first class index include clean energy resource Generation Rate, polluted gas discharge amount,
Controllable load occupation ratio.
Further, it in the step (2), is calculated using entropy assessment as follows:
(1) original index data matrix is constructed
X in matrixijFor the evaluation of estimate of j-th of project under i-th of index, wherein i≤m, j≤n, m are the total of evaluation object
Quantity, n are the total quantity of the project under i-th of evaluation object;
(2) two-level index carries out unison quantization under each scale, calculates the specific gravity p of the i-th scheme index value under jth item indexij
(3) the entropy e of jth item index is calculatedj
In formula: k > 0, ln are natural logrithm;
(4) objective weight for calculating jth item index, for given j, xijThe smaller then e of othernessjIt is bigger;Work as xijEntirely
When portion is equal, ej=emax=1, at this point for the comparison of scheme, index xjAlmost without effect;When each scheme index value difference is got over
When big, ejSmaller, this index is bigger for project plan comparison role;Objective weight value vjWith ejIt is inversely proportional, so jth
The weight of item index can be by 1-ejTo measure;The objective weight of each index acquired by entropy assessment is
Further, in the step (3), the method for calculating weight using analytic hierarchy process (AHP) is as follows:
(1) it is filled in by expert as follows for a certain application form for assessing object:
B in upper tableijReflect significance level of the index j relative to index i, wherein m be evaluation index number, i, j≤
m;
(2) judgment matrix is established according to application form
(3) the maximum eigenvalue λ of judgment matrix B is soughtmaxAnd its corresponding feature vector v
Bv=λmaxv
(4) each index weights are sought
(5) consistency check
Consistency ratio CR is at random
In formula, CI is coincident indicator;RI is Aver-age Random Consistency Index;
Coincident indicator CI can be determined according to following formula
In formula, m is judgment matrix order, works as m=1, and 2 when is not required to examine.
Further, in the step (4), the method for calculating two-level index weight using neural network is as follows:
A. neural network is a kind of Multi-layered Feedforward Networks by Back Propagation Algorithm training, is most widely used at present
One of general neural network model.BP network can learn and store a large amount of input-output mode map relationship, without thing
It is preceding to disclose the math equation for describing this mapping relations.Its learning rules be using steepest descent method, by backpropagation come
The constantly weight and threshold value of adjustment network, utilizes the error sum of squares minimum of network.Neural network model is divided into input layer
(input), hidden layer (hide layer) and output layer (output layer).
B. neural network three layer model defines
Input layer: X=xi, i=1,2 ..., n
Wherein xiFor the index raw value of input.
Transforming function transformation function f (x) is continuous, guidable unipolarity Sigmoid function:
F ' (x)=f (x) [1-f (x)]
Hidden layer: yj=f (netj), j=1,2 ..., m
netj=∑nvijxi, j=1,2 ..., m
Output layer:
Wherein, yjThe output of the linear combiner of input signal, wjk、vjkFor weighted value, OkFor output signal.
C. algorithm and step
Hidden layer:
Output layer:
One error signal is respectively defined to hidden layer and output layer, is had
With
Again by netj=∑nvijxi,Adjust weighted value:
With
ByIn summary method obtains:
The training step for concluding neural network model is as follows:
(1) initial value is assigned to all weights.Randomly initial value is accompanied by the threshold value of whole weights and neuron.
(2) training sample output and input is given.
(3) each layer reality output is calculated.
(4) weight is corrected.According to the deviation of output calculated value and actual value, weight is reversely successively adjusted from output layer, directly
Deviation E is fallen within setting value by correcting each weight to input layer.
(5) reach deviation precision or cycle-index requirement, output is as a result, otherwise, return step 2.
The beneficial effects of the present invention are:
The invention proposes the multinomial novel indexes about rural power grids intelligent level, including grid automation, information
Change, support technology and the indexs such as clean, more comprehensively predicts and cover the developing direction of rural power grids from now on, energy
It is enough that there is certain reference in the following power grid construction timing;Secondly, proposing mind by the research to single evaluation method
It is applied in evaluation through network algorithm, avoids entirely subjective or completely objective extreme end value, utilize the excellent of neural network
Gesture constructs the Rating Model based on first class index, is evaluated finally by the smart grid level in a certain area, verifies
The science and reasonability of neural network method provide strong branch to the problems such as investment decision person's lookup electrical network weak link
Support, while being of great significance to the operation of Intelligent rural network.
Other features and advantages of the present invention will be illustrated in the following description, also, partly becomes from specification
It obtains it is clear that understand through the implementation of the invention.The objectives and other advantages of the invention can be by written explanation
Specifically noted structure is achieved and obtained in book, claims and attached drawing.
Detailed description of the invention
Fig. 1 is flow diagram of the invention;
Fig. 2 is the index system of rural area smart grid Integrated Assessment On The Level in the present invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
Attached drawing, the technical solution of the embodiment of the present invention is clearly and completely described.Obviously, described embodiment is this hair
Bright a part of the embodiment, instead of all the embodiments.Based on described the embodiment of the present invention, the common skill in this field
Art personnel every other embodiment obtained, shall fall within the protection scope of the present invention.
As shown in Figure 1, a kind of rural power grids level of intelligence evaluation method neural network based, comprising the following steps:
(1) index system of rural power grids overall merit, including first class index and corresponding with first class index two are constructed
Grade index;
(2) weight is calculated using entropy assessment;
(3) weight is calculated using analytic hierarchy process (AHP);
(4) weight is calculated using neural network;
(5) result and step for the analytic hierarchy process (AHP) weight for obtaining entropy assessment weight that step (2) obtains, step (3)
(4) method of weighting obtained compares, and verifies the feasibility of neural network.
As shown in Fig. 2, the first class index includes power supply quality, power supply capacity, grid equipment, operation of power networks, information
Change, automation, interactive, DG/ microgrid access, new technology and environmental index.
Power supply quality in first class index includes the two-level index such as rate of qualified voltage, power supply reliability and rate of qualified voltage;
Power supply capacity in first class index includes capacity load ratio of network, substation's load factor qualification rate, main transformer " N-1 " qualification
Rate, middle pressure and low-voltage circuit load factor qualification rate, middle pressure and low-voltage circuit power supply capacity deficiency ratio, medium-voltage line can turn to supply
Rate qualification rate and 10kV outgoing line interval utilization rate;
Grid equipment in first class index includes on-load voltage regulation main transformer ratio, low-voltage distribution equipment one kind equipment ratio, opens
Close oil-free rate, energy-saving ratio-change, middle pressure and low-voltage circuit one kind equipment ratio, line insulation rate, the line conductor of matching is cut
Face standardized rate;
Operation of power networks in first class index includes medium-voltage line and low-voltage circuit high load rate route specific gravity, high load rate
Main transformer and distribution transforming specific gravity, three-phase imbalance platform area accounting;
Informationization in first class index includes substation and power supply station's optical fiber coverage rate, communication line percentage reserve, has letter
Company's ratio at county level, the power distribution station ratio with communication function of breathization Horizontal communication;
The corresponding two-level index of automation in first class index includes the unattended rate of electric substation, dispatch automated system
Realize that the route ratio of distribution automation, substation have five distant function ratios, substation is idle automatic throwing in practical, county town
Cutting apparatus coverage rate, 100kVA and above customer charge monitoring ratio, relay equipment popularize installation rate, platform area and user's low pressure
Aftercurrent protecting equipment operational percentage;
Interactive corresponding two-level index in first class index includes that successfully repair rate, self-help charging rate, electricity consumption are complained
Rate, the functionization of 95598 customer service systems and intelligent electric meter utility ratio;
It includes generation of electricity by new energy permeability, distributed generation resource that DG/ microgrid in first class index, which accesses corresponding two-level index,
Utilization rate and new energy installation rate, distributed generation resource Capacity Margin, distributed generation resource rate of qualified voltage, electric car low ebb fill
Electric rate, electric car peak discharge rate;
The corresponding two-level index of new technology in first class index includes realizing that the control of whole network voltage reactive comprehensive, development are set
Standby on-line monitoring and repair based on condition of component, development 10kV power distribution live-wire, development harmonic synthesis improvement and power distribution network self-healing control;
The corresponding two-level index of environmental index in first class index include clean energy resource Generation Rate, polluted gas discharge amount,
Controllable load occupation ratio.
In the step (2), the method for calculating the entropy assessment weight of each two-level index using entropy assessment is as follows:
(1) original index data matrix is constructed
X in matrixijFor the evaluation of estimate of j-th of project under i-th of index, wherein i≤m, j≤n, m are the sum of index
Amount, n are the total quantity of the project under i-th of index;
(2) unison quantization is carried out under each scale, calculates the specific gravity p of the i-th scheme index value under jth item indexij
(3) the entropy e of jth item index is calculatedj
In formula: k > 0, ln are natural logrithm;
(4) objective weight for calculating jth item index, for given j, xijThe smaller then e of othernessjIt is bigger;Work as xijEntirely
When portion is equal, ej=emax=1, at this point for the comparison of scheme, index xjAlmost without effect;When each scheme index value difference is got over
When big, ejSmaller, this index is bigger for project plan comparison role;Objective weight value vjWith ejIt is inversely proportional, so jth
The weight of item index can be by 1-ejTo measure;The objective weight of each index acquired by entropy assessment is
As shown in table 1, it is two-level index entropy assessment weight analysis table:
Table 1:
In the step (3), the method for the analytic hierarchy process (AHP) weight of each two-level index is calculated using analytic hierarchy process (AHP)
It is as follows:
1) judgment matrix is established according to application form, wherein bjiSignificance level of the index j relative to index i is reflected, wherein
M is the number of evaluation index, i, j≤m;
2) the maximum eigenvalue λ of judgment matrix B is soughtmaxAnd its corresponding feature vector v
Bv=λmaxv
3) weight is sought
As shown in table 2, it is two-level index analytic hierarchy process (AHP) weight analysis table:
In the step (4), the method for calculating weight using neural network is as follows:
A. neural network is a kind of Multi-layered Feedforward Networks by Back Propagation Algorithm training, is most widely used at present
One of general neural network model.BP network can learn and store a large amount of input-output mode map relationship, without thing
It is preceding to disclose the math equation for describing this mapping relations.Its learning rules be using steepest descent method, by backpropagation come
The constantly weight and threshold value of adjustment network, utilizes the error sum of squares minimum of network.Neural network model is divided into input layer
(input), hidden layer (hide layer) and output layer (output layer).
B. neural network three layer model defines
Input layer: X=xi, i=1,2 ..., n
Wherein xiFor the index raw value of input.
Transforming function transformation function f (x) is continuous, guidable unipolarity Sigmoid function:
F ' (x)=f (x) [1-f (x)]
Hidden layer: yj=f (netj), j=1,2 ..., m
netj=∑nvijxi, j=1,2 ..., m
Output layer:
Wherein, yjThe output of the linear combiner of input signal, wjk、vjkFor weighted value, OkFor output signal.
C. algorithm and step
Hidden layer:
Output layer:
One error signal is respectively defined to hidden layer and output layer, is had
With
Again by netj=∑nvijxi,Adjust weighted value:
With
ByIn summary method obtains:
As shown in table 3, it is two-level index neural network weight analysis table:
It is each two-level index enabling legislation weight contrast table such as table 4:
Table 4:
As shown in table 4, this patent, can be more by proposing that neural network algorithm applies the new strategy in power grid evaluation
Clearly find out, the weights of evaluation are more scientific, compared to single subjective estimate method and objective evaluation, have higher
Science, by the comparison to weighted value, the value that neural network algorithm obtains mostly among chromatographic assays and entropy assessment,
Do not occur against scientific extremum, by verifying, method can be used in evaluation procedure.
It is the final scoring of each first class index such as table 5:
Table 5
First class index |
Scoring |
First class index |
Scoring |
Power supply quality |
96.64 |
Power supply capacity |
70.77 |
Grid equipment |
35.05 |
The level of IT application |
58.5 |
Automatization level |
69.65 |
It is interactive |
80.83 |
Distribution access |
54.97 |
Clean and environmental protection |
52.94 |
As shown in table 5, the scoring of rural power grids constitution element is obtained, can clearly find weak link at this stage.
The invention proposes the multinomial novel index about rural area intelligent level, including grid automation, informationization,
It support technology and the indexs such as cleans, more comprehensively predicts and cover the developing direction of rural power grids from now on, Neng Gou
The following power grid construction timing has certain reference;Secondly, proposing nerve net by the research to single evaluation method
Network algorithm is applied in evaluation, and entirely subjective or objective extreme end value completely is avoided, using the advantage of neural network,
The Rating Model based on first class index is constructed, is evaluated finally by the smart grid level in a certain area, verifying nerve
The science and reasonability of network method provide strong support to the problems such as investment decision person's lookup electrical network weak link, together
When be of great significance to the operation of Intelligent rural network.
Finally, it is stated that the above examples are only used to illustrate the technical scheme of the present invention and are not limiting, this field is common
Other modifications or equivalent replacement that technical staff makes technical solution of the present invention, without departing from the technology of the present invention side
The spirit and scope of case, are intended to be within the scope of the claims of the invention.