CN106570561A - System and method for predicting insulator surface non-soluble deposit density - Google Patents
System and method for predicting insulator surface non-soluble deposit density Download PDFInfo
- Publication number
- CN106570561A CN106570561A CN201610983705.9A CN201610983705A CN106570561A CN 106570561 A CN106570561 A CN 106570561A CN 201610983705 A CN201610983705 A CN 201610983705A CN 106570561 A CN106570561 A CN 106570561A
- Authority
- CN
- China
- Prior art keywords
- neural network
- nsdd
- prediction
- data
- insulator
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 239000012212 insulator Substances 0.000 title claims abstract description 89
- 238000000034 method Methods 0.000 title claims abstract description 23
- 238000013528 artificial neural network Methods 0.000 claims abstract description 137
- 238000012360 testing method Methods 0.000 claims abstract description 7
- 238000004380 ashing Methods 0.000 claims description 23
- 230000005540 biological transmission Effects 0.000 claims description 18
- 238000012549 training Methods 0.000 claims description 17
- 239000013049 sediment Substances 0.000 claims description 14
- 238000003062 neural network model Methods 0.000 claims description 11
- 230000009466 transformation Effects 0.000 claims description 11
- 230000002087 whitening effect Effects 0.000 claims description 11
- 238000009825 accumulation Methods 0.000 claims description 10
- 238000009499 grossing Methods 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 7
- 230000009467 reduction Effects 0.000 claims description 5
- 238000001556 precipitation Methods 0.000 claims description 2
- 230000006870 function Effects 0.000 description 9
- 238000010586 diagram Methods 0.000 description 7
- 210000002569 neuron Anatomy 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 230000007547 defect Effects 0.000 description 3
- 238000012546 transfer Methods 0.000 description 3
- 239000000284 extract Substances 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000011109 contamination Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000003344 environmental pollutant Substances 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000009422 external insulation Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 238000012806 monitoring device Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 231100000719 pollutant Toxicity 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Insulators (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a system and a method for predicting the insulator surface non-soluble deposit density. The system comprises an original data acquisition unit, a series gray neural network prediction unit, a parallel gray neural network prediction unit, an embedded gray neural network prediction unit, a NSDD (Non Soluble Deposit Density) predicted value output unit and an NSDD early warning unit. Insulator surface non-soluble deposit density data and local meteorological data are inputted into the series gray neural network prediction unit, the parallel gray neural network prediction unit and the embedded gray neural network prediction unit, and prediction is performed on an insulator NSDD value through the three prediction units; then judgment is performed on the prediction accuracy of the three prediction unit by using a test sample, output of the prediction unit with high prediction accuracy is enabled to act as an insulator NSDD predicted value; and an early warning is given out through the NSDD early warning unit according to whether predicted values outputted by two or more of the prediction units reach a preset graded early warning threshold or not.
Description
Technical Field
The invention belongs to the technical field of external insulation of power systems, and particularly relates to a system and a method for predicting the density of insoluble sediments on the surface of an insulator.
Background
Due to the accumulation of pollutants on the surface of the insulator under normal working voltage, pollution flashover accidents easily occur in severe weather such as overcast and rainy days, heavy fog days and the like, and the serious threat is formed to the safe and stable operation of a power system. It is very necessary to predict the pollution degree of the insulator on the power transmission line so as to prevent pollution flashover accidents in time. The degree of contamination of the surface of the insulator is generally evaluated using a Non-Soluble Deposit Density (NSDD), abbreviated as ash Density.
The gray model is widely applied to the aspect of insulator surface pollution degree prediction because of the advantages of less sample data required by modeling, no need of considering distribution rules and variation trends, simplicity in modeling and convenience in operation, but the gray system is poor in information processing capacity due to the lack of self-learning, self-organization and self-adaption capacity, and cannot independently complete a prediction task.
Disclosure of Invention
Aiming at the defects or the improvement requirements of the prior art, the invention provides a system and a method for predicting the density of insoluble sediments on the surface of an insulator, and aims to provide a method for predicting the density of insoluble sediments which can be widely applied to insulators of any type by combining a gray model and a neural network and provide a pollution flashover early warning function.
In order to achieve the above object, according to an aspect of the present invention, there is provided a system for predicting density of insoluble deposits on a surface of an insulator, including a raw data acquisition unit, a series-type gray neural network prediction unit, a parallel-type gray neural network prediction unit, an embedded type gray neural network prediction unit, and an NSDD prediction value output unit;
the system comprises an original data acquisition unit, a data acquisition unit and a data acquisition unit, wherein the original data acquisition unit is used for acquiring NSDD data and meteorological data of an insulator on a power transmission line; the series connection type grey neural network prediction unit, the parallel connection type grey neural network prediction unit and the embedded type grey neural network prediction unit are respectively used for predicting the NSDD on the power transmission line according to the NSDD data and the meteorological data of the insulator on the power transmission line; correspondingly, the series-connection type grey neural network prediction unit outputs a first prediction value, the parallel-connection type grey neural network prediction unit outputs a second prediction value, and the embedded type grey neural network prediction unit outputs a third prediction value;
and the NSDD predicted value output unit is used for performing parallel comparison on the three predicted values and the sample data, selecting a predicted unit with the highest prediction accuracy from the three predicted units according to a comparison result, and taking the predicted value output by the predicted unit as the insulator NSDD predicted value.
Preferably, the insulator surface insoluble sediment density prediction system further comprises an NSDD early warning unit;
the NSDD early warning unit is used for generating an early warning signal according to the first predicted value, the second predicted value, the third predicted value and a preset early warning threshold value; specifically, when two or more of the first predicted value, the second predicted value, and the third predicted value reach the warning threshold, the warning signal is generated.
Preferably, in the insulator surface insoluble deposit density prediction system, the tandem type gray neural network prediction unit includes a first GM (1,1) model, a second GM (1,1) model, a third GM (1,1) model, and a neural network, which are arranged in parallel;
wherein, the GM (1,1) model is one of the gray models and is a first order differential equation only containing univariates; the neural network comprises an input layer, a hidden layer and an output layer, and the transfer function of the neural network adopts a Sigmoid function;
the input interfaces of the first GM (1,1) model, the second GM (1,1) model and the third GM (1,1) model are all connected with an original data acquisition unit; the input end of the neural network is connected with the output interfaces of the first GM (1,1) model, the second GM (1,1) model and the third GM (1,1) model;
performing insulator NSDD prediction on a first GM (1,1) model, a second GM (1,1) model and a third GM (1,1) model according to insulator NSDD data on a power transmission line respectively to obtain three groups of ashing prediction results; and performing NSDD prediction on the insulator by the neural network according to the three groups of ashing prediction results to obtain a first prediction value.
Preferably, in the insulator surface insoluble sediment density prediction system, the parallel gray neural network prediction unit includes a GM (1,1) model, a neural network, and a combined prediction network; wherein the GM (1,1) model is connected with the neural network in parallel;
the input ends of the GM (1,1) model and the neural network are connected with an original data acquisition unit; the input end of the combined prediction network is connected with the output end of the GM (1,1) model and the output end of the neural network;
performing insulator NSDD prediction on the GM (1,1) model and the neural network according to insulator NSDD data and meteorological data on the power transmission line respectively; and the combined prediction network weights the prediction results of the GM (1,1) model and the neural network to obtain a second prediction value.
Preferably, in the system for predicting the density of insoluble deposits on the surface of an insulator, the embedded gray neural network prediction unit comprises a gray layer, a neural network and a whitening layer which are sequentially connected in series;
the graying layer is used for performing accumulation transformation and smoothing on insulator NSDD data and meteorological data on an original power transmission line; the neural network is used for conducting insulator NSDD prediction according to the accumulated and smoothed data, and the whitening layer is used for conducting subtraction transformation reduction processing on the output data of the neural network to obtain a third prediction value.
According to another aspect of the present invention, based on the system for predicting the density of insoluble deposits on the surface of an insulator, a method for predicting the density of insoluble deposits on the surface of an insulator is provided, which includes the following steps:
(1) establishing three GM (1,1) models with different sequence lengths according to the acquired original insulator NSDD data; taking the predicted NSDD values of the three GM (1,1) models as input quantities, taking measured NSDD as output quantities, and inputting the measured NSDD into a neural network for training to obtain the optimal weight and threshold of the neural network; performing insulator NSDD prediction by using the trained neural network to obtain a first prediction value;
wherein, measuring NSDD refers to collected NSDD data;
(2) insulator NSDD prediction is carried out through a grey model and a neural network model respectively, and two initial prediction data are obtained; determining weighting coefficients of the two initial prediction data according to the test sample; weighting the two initial prediction data according to the weight coefficient to obtain a second prediction value;
in the step, for a time node k for predicting the NSDD value, taking (k-10) measured NSDD data of 1 to (k-10) nodes as training samples, and taking 10 measured NSDD data values of (k-10) to k as test samples;
(3) training a neural network, and performing ashing treatment on the NSDD data of the original insulator; inputting the data subjected to ashing treatment into a trained neural network, and performing NSDD prediction on the insulator by using the trained neural network; whitening the output data of the neural network to obtain a third predicted value; wherein, the ashing treatment comprises accumulated variation and smoothing treatment; the whitening processing refers to an accumulation subtraction transformation;
(4) and performing parallel matching on the first predicted value, the second predicted value and the third predicted value with a test sample, selecting a prediction unit with the highest prediction accuracy from the three prediction units according to a comparison result, and taking the predicted value output by the prediction unit as a final insulator NSDD predicted value.
Preferably, the method for predicting the density of insoluble deposits on the surface of the insulator, in step (3), comprises the following substeps:
(3.1) training the neural network by adopting meteorological data, original insulator NSDD data before a time node m and ashing data of 10 time nodes before the time node m to obtain an optimal weight and a threshold; constructing a trained neural network according to the optimal weight and the threshold;
wherein m is a time node for training and predicting an NSDD data value, and input data adopted by the training neural network comprise meteorological data, original NSDD data of time nodes from 1 to (m-1), and ashing data obtained by ashing the original NSDD data of 10 time nodes from (m-10) to m; the meteorological data comprises wind speed, precipitation and relative humidity; the output data adopted by the training neural network is ashing data of the NSDD data of the original insulators with 10 time nodes (m-9) - (m +1) in total after ashing treatment;
(3.2) carrying out ashing treatment on the original insulator NSDD data of the mth node; putting the ashed data into a trained neural network to obtain an initial predicted value; and whitening the initial predicted value to obtain a second predicted value.
Preferably, the method for predicting the density of insoluble deposits on the surface of the insulator further comprises the step (5):
(5) and comparing the first predicted value, the second predicted value and the third predicted value with a preset early warning threshold value respectively, and generating an early warning signal when two or more of the first predicted value, the second predicted value and the third predicted value reach the early warning threshold value.
In general, compared with the prior art, the above technical solution contemplated by the present invention can achieve the following beneficial effects:
(1) according to the insulator surface insoluble sediment density prediction system and method provided by the invention, a gray model and a neural network model are organically fused, so that the system has the unique method that the gray system is modeled by small sample data, and has the advantage that the neural network model has self-adaptive capacity to nonlinear and non-precise rules; different information of the system can be obtained from different angles and different models through combination, the purposes of improving the prediction precision and increasing the stability and the result reliability are achieved, the combined model has stronger robustness to the change of a data structure, and the defect that a single prediction method is inaccurate is effectively overcome;
(2) the invention provides a system and a method for predicting the density of insoluble sediments on the surface of an insulator, wherein a series gray neural network prediction unit is adopted to connect a neural network and a gray model in a system in a series mode, the output of the gray model is used as the input of the neural network, the system and the method can be used for fault-tolerant analysis and prediction of a complex system, the training time of the neural network is greatly reduced, and the problem that the prediction of a single neural network is easy to fall into a local minimum value is effectively solved;
(3) according to the insulator surface insoluble sediment density prediction system and method provided by the invention, the grey model and the neural network are combined in a weighting mode to construct the parallel grey neural network prediction unit, so that the defect that a single model is easy to lose information is overcome, the randomness is reduced, and the effect of improving the prediction precision is achieved;
(4) according to the insulator surface insoluble sediment density prediction system and method provided by the invention, the embedded grey neural network prediction unit weakens the randomness of original data by setting the grey layer, is easy to approximate to a nonlinear excitation function of a neural network, greatly shortens the network learning time, and accelerates the convergence process while improving the prediction precision;
(5) the insulator surface insoluble sediment density prediction system and the method provided by the invention use the gray system to assist in constructing the neural network, because the information structure of the gray system comprises the deterministic information and the uncertain information, the deterministic information in the gray system is used to assist in constructing the neural network, the deterministic information is used to guide the structure of the neural network, and the learning algorithm of the neural network is improved;
(6) according to the insulator surface insoluble sediment density prediction system and method provided by the invention, the neural network can effectively enhance a grey system; because the gray system has empty set (time zone without information) in the information time zone, only approximate and incompletely determined gray differential equation can be established, but the gray differential equation is difficult to be directly used in practical application and needs to be analyzed; the invention constructs a neural network to analyze the gray parameters of the gray differential equation, extracts a sample from the known data of a gray system to train the neural network, extracts the analyzed gray differential equation parameters from the neural network when the neural network converges, obtains the determined differential equation and realizes accurate prediction.
Drawings
FIG. 1 is a schematic diagram of a system for predicting the density of insoluble deposits on the surface of an insulator according to an embodiment;
FIG. 2 is a schematic diagram of the prediction flow of the GM (1,1) model in the example;
FIG. 3 is a schematic diagram of a neural network model structure in an embodiment;
FIG. 4 is a schematic diagram of a neural network prediction flow in an embodiment;
FIG. 5 is a schematic structural diagram of a tandem gray neural network model in an embodiment;
FIG. 6 is a schematic structural diagram of a parallel gray neural network model in an embodiment;
fig. 7 is a schematic structural diagram of an embedded gray neural network model in the embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
The insulator surface insoluble sediment density prediction system provided by the embodiment of the invention comprises an original data acquisition unit, a series-connection type grey neural network prediction unit, a parallel-connection type grey neural network prediction unit, an embedded type grey neural network prediction unit, an NSDD prediction value output unit and an NSDD early warning unit, wherein the original data acquisition unit is connected with the series-connection type grey neural network prediction unit;
the system comprises an original data acquisition unit, a data acquisition unit and a data acquisition unit, wherein the original data acquisition unit is used for acquiring NSDD data and meteorological data of an insulator on a power transmission line; the series connection type grey neural network prediction unit, the parallel connection type grey neural network prediction unit and the embedded type grey neural network prediction unit respectively predict the insulator NSDD according to the insulator NSDD data and meteorological data on the power transmission line, and correspondingly obtain a first prediction value, a second prediction value and a third prediction value; in the embodiment, on-line monitoring equipment or other on-line monitoring devices for power transmission and transformation density of the optical sensor are used for collecting NSDD data of the insulator on the power transmission line; the meteorological data is data collected around the insulator to be predicted;
the NSDD predicted value output unit combines the three predicted values with sample data, selects a predicted unit with the highest predicted accuracy from the three predicted units, and takes the predicted value output by the predicted unit as the NSDD predicted value of the insulator;
the NSDD early warning unit generates an early warning signal according to the first predicted value, the second predicted value, the third predicted value and a preset grading early warning threshold value; specifically, when two or more of the first predicted value, the second predicted value, and the third predicted value reach the grading early warning threshold, the early warning signal is generated.
The prediction flow of the GM (1,1) model used in this embodiment is illustrated in fig. 2, which is as follows:
first, input the original data sequence X(0)＝[x(0)(1),x(0)(2),…,x(0)(n)]；
Then judging the original data sequence X(0)Whether the modeling condition is met or not; if not, carrying out data smoothing treatment, if yes, carrying out accumulation treatment on the original data to obtain an accumulation sequence X(1)；x(1)＝[x(1)(1),x(1)(2),…,x(1)(n)](ii) a Wherein
Then construct on X(1)First order linear differential equation ofSolving by a least square method to obtain parameters a and b;
wherein,
thereby obtaining a predicted value of X (1),
obtaining the predicted value of X (0) through accumulation reduction,
the structure of the neural network model used in this embodiment is as illustrated in fig. 3, and includes an input layer, a hidden layer, and an output layer; wijIs the weight of the input layer to the hidden layer, θjThreshold for hidden layer neurons, VijFor the implicit layer to output layer connection weights,is the threshold of the output layer; connection weight W of neural networkij、VijAnd a threshold value thetaj、Obtained through training;
input to neurons in the hidden layerWherein XiThe input quantity of each input node of the neural network is obtained;
the transfer function of the neural network adopts a Sigmoid function f (x) 1/(1+ e-x)；
Output of hidden layer
Input to output layer neurons
Output of neurons in the output layer, i.e. predictive values of neural networks
The flow of prediction by the neural network is shown in fig. 4, and includes the following steps:
firstly, constructing a basic structure of a neural network: setting the number of nodes of an input layer, a hidden layer and an output layer of the neural network, inputting an initial weight and a threshold of the hidden layer, and inputting an initial weight and a threshold of the output layer;
then selecting a sample from the sample library to train the neural network; in the training process, the weight and the threshold of the hidden layer and the weight and the threshold of the output layer are continuously adjusted according to the error and the output of the hidden layer and the error and the output of the output layer;
and (4) finishing the training of the neural network after the neural network trained by all samples meets the requirements of learning times and error conditions.
In the embodiment, the structure of the adopted tandem gray neural network prediction unit is shown in fig. 5; comprising 3 GM (1,1) models, GM1、GM2And GM3And a neural network type 3 × 7 × 1;
the neural network comprises 3 input layer nodes, 7 hidden layer nodes and 1 output layer node; 3 GM (1,1) models1，GM2，GM3Is the input quantity of the neural network; the transfer function from the output layer to the hidden layer and from the hidden layer to the output layer is a Sigmoid function, and the hidden layer is arranged according to the Kolmogorov theorem; the output of the neural network is the first predicted value.
The steps of performing NSDD prediction by using the tandem gray neural network prediction unit are as follows:
(1.1) establishing 3 GM (1,1) models by using NSDD data acquired by a data acquisition module, wherein the sequence lengths of the 3 GM (1,1) models are respectively 10, 8 and 6; namely, the data of the (k-20) th to (k-10), (k-18) to (k-10) th time nodes and (k-16) to (k-10) th time nodes; wherein k refers to a time node for insulator NSDD prediction;
(1.2) respectively predicting by using the 3 GM (1,1) models to obtain 3 groups of NSDD prediction data, wherein each group comprises 10 NSDD prediction data values of (k-10) -k time nodes;
(1.3) the common NSDD predicted value of the 3 GM (1,1) models is 16 NSDD data from (k-16) -k, the 16 NSDD predicted data values are used as the input quantity of the neural network, the actually acquired original insulator NSDD data is used as the output quantity of the neural network to train the neural network, and the optimal weight value and the threshold value of the neural network are obtained; in the embodiment, the neural network structure adopted in the step is 3 × 7 × 1 type;
and (1.4) predicting the NSDD value of the insulator at the future moment after the k time node by adopting the trained neural network to obtain a first predicted value.
In the embodiment, the structure of the parallel gray neural network prediction unit is shown in fig. 6, and includes a GM (1,1) model, a neural network, and a combined prediction network; the GM (1,1) model is connected with the neural network in parallel, the output of the GM model and the neural network is used as the input of the combined prediction network, and the output of the combined prediction network is a second predicted value; the steps of adopting the parallel gray neural network prediction unit to carry out NSDD prediction are as follows:
(2.1) respectively predicting by utilizing a gray prediction model GM (1,1) model and a neural network model to obtain an initial prediction NSDD value y1And y2；
(2.2) Using the check value Y(0)(t) subtracting the initial predicted NSDD values y, respectively1And y2To obtain a prediction error e1And e2；
According to prediction errorCalculating to obtain a weight coefficient omega of the GM (1,1) model1And weight coefficient omega of neural network model2；ω1+ω2＝1；
(2.3) according to yc＝ω1y1+ω2y2And obtaining a second predicted value.
Wherein the weight coefficient omega1And a weight coefficient ω2The method comprises the following steps:
according to the test sequence Y(0)(t) obtaining e1、e2And ecIs composed of
Second predicted value y of insulator NSDDcHas a variance of
For Var (e)c) Obtaining a minimum value
Due to, ω2＝1-ω1Therefore, the weight coefficient ω1And a weight coefficient ω2Respectively as follows:
wherein, Var (e)1)＝σ11，Var(e2)＝σ22，cov(e1,e2)＝σ12。
The structure of the embedded gray neural network prediction unit adopted in this embodiment is shown in fig. 7, and a gray layer, a neural network, and a whitening layer are sequentially connected in series; the graying layer carries out accumulation transformation and smoothing processing on the original data, and the whitening layer carries out accumulation reduction transformation reduction processing on the output data of the neural network; wherein, the structure of the neural network is 4 multiplied by 9 multiplied by 1 type; the input layer of the neural network is meteorological data and NSDD data; the NSDD data is accumulated data of 10 time nodes from the (k-10) th node to the k-th node; the hidden layer is set according to the Kolmogorov theorem; the output layer is predicted NSDD data, and the value of the predicted NSDD data is accumulated data of 10 time nodes from (k-9) th to (k +1) th;
the steps of adopting the embedded grey neural network prediction unit to carry out NSDD prediction are as follows:
(3.1) for the original data sequence X(0)＝(x(0)(1),x(0)(2),…,x(0)(n)) performing an accumulative transformation to obtain a 1-AGO sequence X(1)＝(x(1)(1),x(1)(2),…,x(1)(n))；
Wherein x is(0)(k)≥0,k＝1,2,…,n；
(3.2) performing three-point smoothing treatment on the 1-AGO sequence;
for a node with k 2,3, …, n-1:
the smoothing formula is
Two endpoints of the pair k 1 and k n:
the smoothing formula is
(3.3) predicting according to the result of the ashing treatment through a neural network;
(3.4) output sequence X to neural network(3)＝(x(3)(1),x(3)(2),…,x(3)(n)) performing a subtraction transformation,
obtaining a third predicted value X(4)＝(x(4)(1),x(4)(2),…,x(4)(n)), wherein x(3)(k)≥0,k＝1,2,…,n；
x(3)(k)＝x(2)(k)-x(2)(k-1)。
In this embodiment, the NSDD early warning unit is adopted to perform early warning prediction according to the predicted values of the units; the early warning unit sets A, B, C, D a total of 4 early warning levels.
Wherein, when the NSDD predicted value output by two of the three prediction units of the system reaches the value rho of the insulator NSDD value possibly causing pollution flashoverF95%, i.e. 95% ρFThe system sends out A-level early warning; when the NSDD predicted value of two prediction models in the three prediction models of the system reaches the value rho of the insulator NSDD value possibly causing pollution flashoverF90%, i.e. 90% pFThe system sends out a B-level early warning; when the NSDD predicted value of two prediction models in the three prediction models of the system reaches the value rho of the insulator NSDD value possibly causing pollution flashoverF85%, i.e. 85% ρFThe system sends out a C-level early warning; when the NSDD predicted value of two prediction models in the three prediction models of the system reaches the value rho of the insulator NSDD value possibly causing pollution flashoverF80%, i.e. 80% pFAnd the system sends out a D-level early warning.
The NSDD pre-warning unit compares the NSDD pre-warning value with the NSDD value of the insulator in the pollution flashover to generate pre-warning information for processing by operating personnel, and the function of effectively preventing the pollution flashover accident of the power transmission line in time can be achieved.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (8)
1. The insulator surface insoluble sediment density prediction system is characterized by comprising an original data acquisition unit, a series connection type grey neural network prediction unit, a parallel connection type grey neural network prediction unit, an embedded type grey neural network prediction unit and an NSDD prediction value output unit;
the original data acquisition unit is used for acquiring NSDD data and meteorological data of the insulator on the power transmission line; the series connection type grey neural network prediction unit, the parallel connection type grey neural network prediction unit and the embedded type grey neural network prediction unit are used for predicting the NSDD on the power transmission line according to NSDD data and meteorological data on the power transmission line respectively to obtain three prediction values;
and the NSDD predicted value output unit is used for performing parallel comparison on the three predicted values and the sample data, selecting a predicted unit with the highest predicted accuracy from the three predicted units according to a comparison result, and taking the predicted value output by the predicted unit as the NSDD predicted value of the insulator.
2. The insulator surface insoluble deposit density prediction system of claim 1, further comprising an NSDD pre-warning unit;
the NSDD early warning unit is used for generating an early warning signal according to the three predicted values and a preset early warning threshold value; specifically, when two or three of the three predicted values reach an early warning threshold, an early warning signal is generated.
3. The insulator surface insoluble deposit density prediction system according to claim 1 or 2, wherein the series-type gray neural network prediction unit comprises a first GM (1,1) model, a second GM (1,1) model, a third GM (1,1) model, and a neural network in parallel;
the input interfaces of the first GM (1,1) model, the second GM (1,1) model and the third GM (1,1) model are all connected with the original data acquisition unit; the input end of the neural network is connected with the output interfaces of the first GM (1,1) model, the second GM (1,1) model and the third GM (1,1) model;
the first GM (1,1) model, the second GM (1,1) model and the third GM (1,1) model respectively carry out insulator NSDD prediction according to insulator NSDD data on the power transmission line to obtain three groups of ashing prediction results; and performing NSDD prediction on the insulator by the neural network according to the three groups of ashing prediction results to obtain a first prediction value.
4. The insulator surface insoluble deposit density prediction system according to claim 1 or 2, wherein the parallel type gray neural network prediction unit comprises a GM (1,1) model, a neural network, and a combined prediction network;
the GM (1,1) model is connected with a neural network in parallel; the input ends of the GM (1,1) model and the neural network are connected with an original data acquisition unit; the input end of the combined prediction network is connected with the output end of the GM (1,1) model and the output end of the neural network;
the GM (1,1) model and the neural network are used for respectively predicting the insulator NSDD according to the insulator NSDD data and the meteorological data on the power transmission line; and the combined prediction network is used for weighting the prediction results of the GM (1,1) model and the neural network to obtain a second prediction value.
5. The insulator surface insoluble deposit density prediction system according to claim 1 or 2, wherein the embedded gray neural network prediction unit comprises a gray layer, a neural network and a white layer connected in series in sequence;
the graying layer is used for performing accumulation transformation and smoothing on insulator NSDD data and meteorological data on the power transmission line; the neural network is used for conducting insulator NSDD prediction according to the data after the smoothing is accumulated, and the whitening layer is used for conducting subtraction transformation reduction processing on the output data of the neural network to obtain a third prediction value.
6. An insulator surface insoluble deposit density prediction method based on the insulator surface insoluble deposit density prediction system according to any one of claims 1 to 5, characterized by comprising the following steps:
(1) establishing three GM (1,1) models with different sequence lengths according to the acquired original insulator NSDD data; taking the predicted NSDD values output by the three GM (1,1) models as input quantities, and taking measured NSDD data as output quantities to carry out neural network training; performing insulator NSDD prediction by using the trained neural network to obtain a first prediction value;
wherein, measuring NSDD refers to collected NSDD data;
(2) insulator NSDD prediction is carried out through a grey model and a neural network model respectively, and two initial prediction data are obtained; determining weight coefficients of the two initial prediction data according to the test sample; weighting the two initial prediction data according to the weight coefficient to obtain a second prediction value;
(3) training a neural network, and performing ashing treatment on measured NSDD data; using the data after ashing as input data, and adopting a trained neural network to predict NSDD; whitening the output data of the neural network to obtain a third predicted value; the ashing treatment comprises accumulated variation and smoothing treatment; the whitening processing is accumulation subtraction transformation;
(4) and performing parallel comparison on the first predicted value, the second predicted value and the third predicted value with a test sample, and selecting one with the highest prediction accuracy from the three predicted values as an insulator NSDD predicted value according to a comparison result.
7. The insulator surface insoluble deposit density prediction system according to claim 6, wherein said step (3) comprises the sub-steps of:
(3.1) training the neural network by adopting meteorological data, original insulator NSDD data before a time node m and ashing data of 10 time nodes before the time node m to obtain an optimal weight and a threshold; constructing a trained neural network according to the optimal weight and the threshold;
wherein m is a time node for training and predicting an NSDD data value, and input data adopted by the training neural network comprise meteorological data, original NSDD data of time nodes from 1 to (m-1), and ashing data obtained by ashing the original NSDD data of 10 time nodes from (m-10) to m; the meteorological data comprises wind speed, precipitation and relative humidity; the output data adopted by the training neural network is ashing data of the NSDD data of the original insulators with 10 time nodes (m-9) - (m +1) in total after ashing treatment;
(3.2) carrying out ashing treatment on the original insulator NSDD data of the mth node; putting the ashed data into a trained neural network to obtain an initial predicted value; and whitening the initial predicted value to obtain a second predicted value.
8. The insulator surface insoluble deposit density prediction system according to claim 6, further comprising step (5):
(5) and comparing the first predicted value, the second predicted value and the third predicted value with a preset early warning threshold value respectively, and generating an early warning signal when two or three of the first predicted value, the second predicted value and the third predicted value reach the early warning threshold value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610983705.9A CN106570561B (en) | 2016-11-08 | 2016-11-08 | A kind of insoluble sediment density forecasting system of insulator surface and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610983705.9A CN106570561B (en) | 2016-11-08 | 2016-11-08 | A kind of insoluble sediment density forecasting system of insulator surface and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106570561A true CN106570561A (en) | 2017-04-19 |
| CN106570561B CN106570561B (en) | 2019-04-12 |
Family
ID=58540708
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610983705.9A Active CN106570561B (en) | 2016-11-08 | 2016-11-08 | A kind of insoluble sediment density forecasting system of insulator surface and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106570561B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113841150A (en) * | 2019-05-16 | 2021-12-24 | 大金工业株式会社 | Learning model generation method, program, storage medium, and learning completion model |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102590677A (en) * | 2012-02-28 | 2012-07-18 | 浙江省电力试验研究院 | Analyzing and processing method for test data of manual pollution flashover of insulator |
| CN104502410A (en) * | 2013-07-21 | 2015-04-08 | 国家电网公司 | Prediction method for insulator equivalent salt deposit density and non-soluble deposit density by least squares support vector machine and genetic algorithm |
| CN104992055A (en) * | 2015-06-16 | 2015-10-21 | 哈尔滨工业大学 | Pollution-flashover-caused trip probability calculation method of overhead lines in sand and dust environment |
-
2016
- 2016-11-08 CN CN201610983705.9A patent/CN106570561B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102590677A (en) * | 2012-02-28 | 2012-07-18 | 浙江省电力试验研究院 | Analyzing and processing method for test data of manual pollution flashover of insulator |
| CN104502410A (en) * | 2013-07-21 | 2015-04-08 | 国家电网公司 | Prediction method for insulator equivalent salt deposit density and non-soluble deposit density by least squares support vector machine and genetic algorithm |
| CN104992055A (en) * | 2015-06-16 | 2015-10-21 | 哈尔滨工业大学 | Pollution-flashover-caused trip probability calculation method of overhead lines in sand and dust environment |
Non-Patent Citations (1)
| Title |
|---|
| 张寒，文习山，丁辉: "用神经网络预测基于气象因素的绝缘子等值附盐密度", 《高压电器》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113841150A (en) * | 2019-05-16 | 2021-12-24 | 大金工业株式会社 | Learning model generation method, program, storage medium, and learning completion model |
| CN113841150B (en) * | 2019-05-16 | 2024-04-09 | 大金工业株式会社 | Learning model generation method, article evaluation determination and output method, and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106570561B (en) | 2019-04-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Su et al. | Concrete cracks detection using convolutional neuralnetwork based on transfer learning | |
| US5465320A (en) | Method of automated learning, an apparatus therefor, and a system incorporating such an apparatus | |
| CN111028939B (en) | Multigroup intelligent diagnosis system based on deep learning | |
| CN108090629B (en) | Load prediction method and system based on nonlinear autoregressive neural network | |
| CN106453293A (en) | Network security situation prediction method based on improved BPNN (back propagation neural network) | |
| Hu et al. | A review on empirical methods of pavement performance modeling | |
| CN113609955A (en) | Three-phase inverter parameter identification method and system based on deep learning and digital twinning | |
| CN110530876B (en) | Insulator pollution degree development prediction method based on long-term and short-term memory neural network | |
| CN111967308A (en) | Online road surface unevenness identification method and system | |
| CN106997373A (en) | A kind of link prediction method based on depth confidence network | |
| CN110991776A (en) | Method and system for realizing water level prediction based on GRU network | |
| CN106815635A (en) | A kind of forecasting system and method for insulator surface equivalent salt deposit density | |
| Liu et al. | Deep transfer learning-based vehicle classification by asphalt pavement vibration | |
| CN111402249B (en) | Image evolution analysis method based on deep learning | |
| CN114048546A (en) | Graph convolution network and unsupervised domain self-adaptive prediction method for residual service life of aircraft engine | |
| CN117392091A (en) | Flocculation effect image detection method, flocculation effect image detection system, electronic equipment and medium | |
| Kim et al. | Untangling the contribution of input parameters to an artificial intelligence PM2. 5 forecast model using the layer-wise relevance propagation method | |
| CN116451322A (en) | Bayesian optimization-based LSTM deep learning network mechanical prediction method | |
| CN106570561B (en) | A kind of insoluble sediment density forecasting system of insulator surface and method | |
| CN117218375B (en) | Priori knowledge and data driven based environment visibility prediction method and device | |
| Di et al. | An automatic and integrated self-diagnosing system for the silting disease of drainage pipelines based on SSAE-TSNE and MS-LSTM | |
| KR101585545B1 (en) | A method of Wavelet-based autoregressive fuzzy modeling for forecasting algal blooms | |
| Ehteram et al. | An advanced deep learning model for predicting water quality index | |
| CN115359197A (en) | Geological curved surface reconstruction method based on spatial autocorrelation neural network | |
| CN115510763A (en) | Air pollutant concentration prediction method and system based on data-driven exploration |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |