CN112419092A - Line loss prediction method based on particle swarm optimization extreme learning machine - Google Patents
Line loss prediction method based on particle swarm optimization extreme learning machine Download PDFInfo
- Publication number
- CN112419092A CN112419092A CN202011350048.7A CN202011350048A CN112419092A CN 112419092 A CN112419092 A CN 112419092A CN 202011350048 A CN202011350048 A CN 202011350048A CN 112419092 A CN112419092 A CN 112419092A
- Authority
- CN
- China
- Prior art keywords
- learning machine
- line loss
- extreme learning
- particle swarm
- particle
- 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.)
- Pending
Links
- 239000002245 particle Substances 0.000 title claims abstract description 96
- 238000005457 optimization Methods 0.000 title claims abstract description 49
- 238000000034 method Methods 0.000 title claims abstract description 43
- 210000002569 neuron Anatomy 0.000 claims abstract description 22
- 230000008569 process Effects 0.000 claims abstract description 13
- 238000012549 training Methods 0.000 claims abstract description 13
- 238000004422 calculation algorithm Methods 0.000 claims description 33
- 230000003247 decreasing effect Effects 0.000 claims description 8
- 230000015654 memory Effects 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 8
- 238000003860 storage Methods 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 4
- 230000001133 acceleration Effects 0.000 claims description 3
- 239000000203 mixture Substances 0.000 claims 3
- 239000002994 raw material Substances 0.000 claims 3
- 230000006870 function Effects 0.000 description 12
- 238000010586 diagram Methods 0.000 description 11
- 238000004590 computer program Methods 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 5
- 238000012360 testing method Methods 0.000 description 5
- 239000011159 matrix material Substances 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000005284 excitation Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 241001106462 Ulmus Species 0.000 description 1
- 241000949731 Ulmus davidiana Species 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- 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
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Economics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Molecular Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a line loss prediction method based on a particle swarm optimization extreme learning machine, which comprises the steps of obtaining daily power supply; inputting the daily power supply amount into a pre-trained extreme learning machine to obtain line loss; in the training process, a particle swarm optimization is adopted to optimize the initial hidden layer neuron threshold of the extreme learning machine and the connection weight between the input layer and the hidden layer. The initial hidden layer neuron threshold value of the extreme learning machine and the connection weight between the input layer and the hidden layer are optimized by adopting the particle swarm optimization, so that the precision and the stability of the extreme learning machine are improved, and the line loss prediction precision is improved.
Description
Technical Field
The invention relates to a line loss prediction method based on a particle swarm optimization extreme learning machine, and belongs to the field of line loss prediction.
Background
With the steady development of economy, the continuous improvement of living standard and the increasing aggravation of global energy crisis conditions, the economic requirement on the operation of a power grid is continuously improved. The line loss and the line loss rate are very important comprehensive indexes, and the large-scale energy-saving loss-reducing scheme is embodied in various aspects of the power grid and plays a positive role in improving the economical efficiency of the power grid operation and reducing the energy waste. The line loss value refers to the energy loss emitted in the form of heat energy, namely the effective power consumed by the resistance and the conductance, and is short for the electric energy loss of the power grid. The line loss rate is an important index capable of comprehensively reflecting the comprehensive management level and the operation level of regional electric power, and a power grid company needs to further strengthen line loss management by means of science and technology, so that the purpose of reducing the line loss rate and improving comprehensive strength is achieved. The line loss is analyzed and researched by using an advanced technology, so that the loss of energy generated in transmission is reduced as much as possible, and the method is one of the operation management targets of power enterprises.
An Extreme Learning Machine (ELM) is a single hidden layer feedforward neural network proposed by south american studios in singapore. In recent years, the extreme learning machine has attracted attention of many experts and scholars at home and abroad due to its excellent performance and simple structure, and is applied to the field of line loss prediction. However, the prediction accuracy of the extreme learning machine is greatly influenced due to the random generation of the connection weight between the ELM input layer and the hidden layer and the threshold of the neuron of the hidden layer. Therefore, a method for providing line loss prediction accuracy is urgently needed.
Disclosure of Invention
The invention provides a line loss prediction method based on a particle swarm optimization extreme learning machine, which solves the problems disclosed in the background technology.
In order to solve the technical problems, the technical scheme adopted by the invention is as follows:
a line loss prediction method based on a particle swarm optimization extreme learning machine comprises the following steps,
acquiring daily power supply amount;
inputting the daily power supply amount into a pre-trained extreme learning machine to obtain line loss;
in the training process, a particle swarm optimization is adopted to optimize the initial hidden layer neuron threshold of the extreme learning machine and the connection weight between the input layer and the hidden layer.
And acquiring daily power supply amount, and performing normalization processing on the daily power supply amount.
In the particle swarm optimization, an initial hidden layer neuron threshold value, a connection weight between an input layer and a hidden layer are used as particles, and a root mean square error output by an extreme learning machine is used as particle fitness.
In the particle swarm optimization, the particle velocity is updated by adopting the inertia weight with linear decrement.
The linear decreasing inertia weight is adopted to update the particle speed, the specific formula is,
wherein,for k iterations corresponding to the particle velocity, wkFor the inertial weights corresponding to the k iterations,particle velocity corresponding to k-1 iterations, c1、c2As an acceleration factor, r1、r2Is a value range of [0,1]The random number of (a) is set,respectively iterating the optimal position of the particle extremum and the optimal position of the particle swarm extremum for k times,the particle i position is iterated k-1 times.
The formula of the inertial weight is as follows,
wk=wmax-(wmax-wmin)k/K
wherein, wmax、wminThe maximum value and the minimum value of the inertia weight are respectively, and K is the maximum iteration number of the particle swarm algorithm.
A line loss prediction system based on a particle swarm optimization extreme learning machine comprises,
an acquisition module: acquiring daily power supply amount;
a prediction module: inputting the daily power supply amount into a pre-trained extreme learning machine to obtain line loss;
in the training process, a particle swarm optimization is adopted to optimize the initial hidden layer neuron threshold of the extreme learning machine and the connection weight between the input layer and the hidden layer.
In the particle swarm optimization, an initial hidden layer neuron threshold value, a connection weight between an input layer and a hidden layer are used as particles, a root mean square error output by an extreme learning machine is used as particle fitness, and the particle speed is updated by adopting linearly decreasing inertial weight.
A computer readable storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by a computing device, cause the computing device to perform a method of line loss prediction based on a particle swarm optimization extreme learning machine.
A computing device comprising one or more processors, one or more memories, and one or more programs stored in the one or more memories and configured to be executed by the one or more processors, the one or more programs including instructions for performing a particle swarm optimization extreme learning machine-based line loss prediction method.
The invention achieves the following beneficial effects: the initial hidden layer neuron threshold value of the extreme learning machine and the connection weight between the input layer and the hidden layer are optimized by adopting the particle swarm optimization, so that the precision and the stability of the extreme learning machine are improved, and the line loss prediction precision is improved.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is an ELM network;
FIG. 3 is a graph of sample numbers for a 10kV Ulmus Davidiana Branch No. 4 pole transformer;
FIG. 4 is a graph of predicted correlation error results;
FIG. 5 is a graph comparing the prediction results of PSO-ELM and EIM for a 10kV ELM 7-tun branch line No. 4 rod transformer;
FIG. 6 is an iteration diagram of a three-door Song line 48# three-six-Tun transformer sample;
FIG. 7 is a graph of error results associated with the prediction of a three-gate Song line 48# three-six-Tung transformer;
FIG. 8 is a graph comparing the predicted results of PSO-ELM and EIM for a three-door Song line 48# three-six-Tun transformer;
FIG. 9 is a graph of sample numbers of transformers from 5 teams on a long ridge;
FIG. 10 is a graph of long ridge sub-5 team transformer prediction correlation error results;
FIG. 11 is a graph comparing the prediction results of PSO-ELM and EIM of 5 long-ridge sub-team transformers;
FIG. 12 is a plot of number of samples of inert forest in a common 01512 station area;
FIG. 13 is a graph of predicted correlation error results;
FIG. 14 is a comparison graph of PSO-ELM and EIM prediction results of inert forest in public 01512 station area;
FIG. 15 is a graph of sample numbers for transformers on No. 131 poles of a 10kV Yongle line Yongle trunk;
FIG. 16 is a diagram of the results of the prediction related errors of the transformer on the No. 131 pole of the 10kV Yongle line Yongle stem;
FIG. 17 is a graph comparing the prediction results of PSO-ELM and EIM of transformer on pole 131 of 10kV Yongle line Yongle stem.
Detailed Description
The invention is further described below with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby.
As shown in fig. 1, a line loss prediction method based on a particle swarm optimization extreme learning machine includes the following steps:
An Extreme Learning Machine (ELM) is a single hidden layer feedforward neural network learning algorithm, and the ELM is characterized in that a threshold value and a connection weight value are not required to be changed after initialization, the weight of an output layer is directly obtained by a generalized inverse method, and the weight of a traditional BP neural network needs to be continuously adjusted by error feedback, so that the training learning speed of the ELM is greatly improved compared with that of the BP. The ELM assigns the connection weight and the threshold value arbitrarily, model parameters are not required to be changed in the training process, and the weight of the output layer can be obtained by a least square method only by setting the number of hidden layer neurons. But the ELM connection weight value and the threshold value are randomly generated, so that the accuracy of the extreme learning machine is greatly influenced.
As shown in fig. 2, the whole network model of the ELM network is divided into three layers, i.e., an input layer, a hidden layer, and an output layer. The input layer realizes the function of accepting input variables of an external environment, the hidden layer is mainly used for realizing the functions of calculation, identification and the like, and the output layer is used for outputting results. When the hidden layer mapping function of the single hidden layer feedforward neural network meets infinite and differentiable requirements, the connection weight of the input layer and the hidden layer and the threshold value of the hidden layer can be randomly set, and the hidden layer mapping function is not required to be adjusted after being set. When a suitable network structure is chosen, the neural network can fit any continuous function without error. And the connection weight between the hidden layer and the output layer is determined at one time in a way of solving an equation set without iterative adjustment, so that overfitting of training data is avoided, and the model prediction speed is improved.
The calculation of ELM is implemented as follows:
taking L hidden layer neurons, wherein the excitation function of the L hidden layer neurons is g (·) which is a nonlinear function, and the excitation function is particularly Sigmoid;
1) random initialization weight threshold (a)i,bi),ai、biRespectively are the connection weight between the input layer and the hidden layer and the neuron threshold of the hidden layer;
2) calculating hidden layer node output matrix H ═ g (a)i,bi,xi);
Wherein x isi′For the input of ELM, i' e [1, n];
3) Calculating the output weight from the hidden layer to the output layer;
4) and when the calculated output error is less than or equal to an arbitrary constant epsilon, the ELM finishes training.
The predicted quality was evaluated using the following 3 indices, the calculation formulas for which are Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Mean absolute percentage error of expected value (MAPE), respectively, in order:
The Particle Swarm Optimization (PSO) simulates the migration and clustering behaviors of the bird group, and the flight process of the particles is used as the individual searching process. Each particle corresponds to a fitness value and a speed that affects its flight direction, distance; each particle in the search space is searched according to the current optimal particle, and the particle completes iteration by tracking two extreme values.
Aiming at the problems that invalid hidden nodes may appear on initial weight values and threshold values generated randomly by the ELM and the generalization capability is insufficient, a Particle Swarm Optimization (PSO) is introduced to optimize the initial input weight values and the threshold values of the ELM, so that the defect of random selection of the initial input weight values and the threshold values of an extreme learning machine is overcome. The PSO-ELM model for detecting the line loss of the transformer area is established by adopting the connection weight and the threshold in the PSO optimized ELM algorithm, and the prediction effect of the PSO-ELM algorithm and the ELM algorithm on data with different complexity degrees is compared.
The utility model comprises the following steps: in the particle swarm optimization, an initial hidden layer neuron threshold value, a connection weight between an input layer and a hidden layer are used as particles, a root mean square error output by an extreme learning machine is used as particle fitness, and the particle speed is updated by adopting linearly decreasing inertial weight.
The particle velocity is updated by adopting the linearly decreasing inertial weight, the local optimization capability in the later iteration stage of the algorithm is enhanced, and the specific formula is as follows:
wherein,for k iterations corresponding to the particle velocity, wkFor the inertial weights corresponding to the k iterations,particle velocity corresponding to k-1 iterations, c1、c2As an acceleration factor, r1、r2Is a value range of [0,1]The random number of (a) is set,respectively iterating the optimal position of the particle extremum and the optimal position of the particle swarm extremum for k times,the particle i position is iterated k-1 times.
The inertial weight formula is:
wk=wmax-(wmax-wmin)k/K
wherein, wmax、wminThe maximum value and the minimum value of the inertia weight are respectively, and K is the maximum iteration number of the particle swarm algorithm.
The specific process is as follows:
1) and (4) preprocessing data. Given learning samples (input vector and desired output vector) are input and subjected to normalization processing.
2) Particle swarm initialization (input weight and threshold of ELM); and (3) constructing an optimized new model of the PSO-ELM (determining the number of neurons of an input layer, a hidden layer and an output layer).
3) The algorithm parameters are initialized, as well as the velocity and position of the particles.
4) And the local optimization capability at the later iteration stage of the algorithm is enhanced by adopting the linearly decreasing inertial weight in combination with the characteristic of the inertial weight.
5) And taking the root mean square error of the ELM output as the fitness of the PSO. Calculating and evaluating an adaptive value of each particle, training an ELM model by using parameter information contained in the current position of the particle, and predicting a test sample.
6) Marking the current optimal position of the particle individuals and the optimal position of the particle swarm.
7) And adjusting the speed and the position of each particle according to the individual extreme value and the group extreme value of the particle, and calculating the corresponding fitness value of the current particle. And stopping the optimization iteration process when the maximum iteration times or the optimal fitness is reached, and finally obtaining the optimal particle position.
8) And (3) the optimal particle position obtained by executing the steps is the corresponding optimal input weight and hidden layer threshold in the extreme learning machine, the obtained weight and hidden layer threshold are substituted into an ELM to remold the network for modeling prediction, and an output weight matrix is calculated to obtain a prediction result.
To further illustrate the effect of the above method, the following tests were performed:
the data items selected in the test comprise the power supply amount in the day, the power consumption in the day and the line loss rate of the transformer area. The PSO-ELM is used for verifying PSO-ELM established in the text, and line loss prediction and research data are truly acquired by five station files of a 7-station branch line No. 4 pole transformer of a 10kV ELM tree, a 10kV Yongle line Yongle stem No. 131 pole post transformer, a three-gate Song line No. 48# three-six-station transformer, a 5-team long-ridge sub-transformer and a public 01512 station zone Sunlin. The collection date is 1/2018 to 14/2020/7 once a day. And dividing the line loss data of the five transformer areas into a training set and a testing set according to the proportion of 5:1 because the total data amount of each transformer area after the abnormal values are removed is inconsistent.
The current day power supply and the current day power consumption are used as input parameters of the ELM, the output is the line loss rate, and the line loss data of five districts, namely a 7-branch line 4-pole transformer of a 10kV ELM, a 10kV Yongle line Yongle stem 131-pole-mounted transformer, a three-door Song line 48# three-six-branch line transformer, a Changling sub 5-team transformer and a public 01512 mountain forest are subjected to prediction simulation.
FIG. 3 is a graph showing the number of samples of a 7-tap 4-bar transformer 921 for a 10kV elm tree; FIG. 4 is a graph of predicted correlation error results, which can be obtained by predicting results for 153 samples, and the correlation error value is substantially below 0.1; fig. 5 is a comparison graph of the prediction results of the PSO-ELM and the ELM, and it can be seen that the prediction accuracy of the ELM algorithm is 85.658%, the accuracy of the line loss prediction using the PSO-ELM algorithm reaches 96.592%, and the accuracy is improved by more than 10%. The prediction precision of the PSO-ELM optimization algorithm is greatly improved compared with that of the ELM algorithm.
Fig. 6 is a graph showing the number of samples of 918 three-gate sony line 48# three-six-bin transformers; FIG. 7 is a graph of predicted correlation error results, which can be obtained by predicting results for 153 samples, and the correlation error value is substantially below 0.1; fig. 8 is a comparison graph of the prediction results of the PSO-ELM and the ELM, and it can be seen that the prediction accuracy of the ELM algorithm is 41.958%, the accuracy of the line loss prediction using the PSO-ELM algorithm reaches 96.097%, and the accuracy is improved by more than 50%. The prediction precision of the PSO-ELM optimization algorithm is greatly improved compared with that of the ELM algorithm.
FIG. 9 is a graph showing 640 sample numbers of transformer areas of 5 teams of a long ridge; FIG. 10 is a graph of predicted correlation error results, wherein the correlation error values of the transformer regions of the long ridge sub-5 team are substantially below 0.02 as can be obtained from the results of prediction on 108 samples; fig. 11 is a comparison graph of the prediction results of the PSO-ELM and the ELM, and it can be seen that the prediction accuracy of the ELM algorithm is 67.266%, the accuracy of the line loss prediction using the PSO-ELM algorithm reaches 99.715%, and the accuracy is improved by more than 30%. The prediction precision of the PSO-ELM optimization algorithm is greatly improved compared with that of the ELM algorithm.
FIG. 12 is a graph showing 706 sample numbers of inert forest zones in a public 01512 zone; fig. 13 is a prediction correlation error result graph, and it can be obtained from the results of 117 samples that the correlation error value of the stand area of an inert forest of the public 01512 stand area is substantially below 0.03; fig. 14 is a comparison graph of the PSO-ELM and ELM prediction results of the inert forest district of the public 01512 district, and it can be seen that the prediction accuracy of the ELM algorithm is 88.579%, the accuracy of the line loss prediction performed by the PSO-ELM algorithm reaches 91.169%, and the accuracy is improved. The prediction precision of the PSO-ELM optimization algorithm is greatly improved compared with that of the ELM algorithm.
Fig. 15 is a graph of 119 samples in the transformer area on the 131 # pole of the 10kV permanent magnet wire permanent magnet pole; fig. 16 is a diagram of the result of prediction correlation error, which can be obtained by predicting the results of 21 samples, and the correlation error value of the station area is small; fig. 17 is a comparison graph of prediction results of PSO-ELM and ELM of transformer area on pole 131 of 10kV permanent magnet wire permanent magnet pole, which shows that the prediction accuracy of the ELM algorithm is 72.005%, the accuracy of line loss prediction using the PSO-ELM algorithm reaches 99.929%, and the accuracy is improved by more than 20%. The prediction precision of the PSO-ELM optimization algorithm is greatly improved compared with that of the ELM algorithm.
TABLE 1 error results comparison table
The data of the five regions are contrastively analyzed, so that the prediction result of the ELM prediction model is not stable and is greatly influenced by the number of samples. The line loss prediction model established by the ELM has certain limitations, is easy to fall into local optimization, and has strong randomness of prediction results and over-objective parameter setting, thus leading to poor prediction effect.
According to the table 1, the PSO-ELM algorithm predicts that the pole transformer MAE of the No. 4 branch 7-tree branch of the ELM is 0.2069, the MAPE is 0.055 and the RMSE is 0.2135, the prediction precision can reach 96.592%, and the prediction precision is superior to the index value predicted by the ELM model. The MAE of a public 01512 station inert forest predicted by the PSO-ELM algorithm is 0.0199, the MAPE is 0.0107 and the RMSE is 0.0248, the prediction precision can reach 91.169%, and the prediction precision is superior to the index value predicted by the ELM model; the MAE of the transformer area on the pole column of No. 131 of the 10kV Yongle line Yonglegan predicted by the PSO-ELM algorithm is 0.0029, the MAPE is 0.00063 and the RMSE is 0.0093, the prediction precision can reach 99.929 percent, and the prediction precision is superior to the index value predicted by the ELM model. It can be seen that whether the large sample, the medium sample or the small sample can explain the applicability and the effectiveness of the line loss prediction model established by the PSO-ELM.
In experimental tests, the line loss rate prediction based on the PSO-ELM basically consumes about 15s or even lower time, and can meet the requirement of rapidity of power line loss rate prediction.
According to the method, the particle swarm optimization is adopted to optimize the initial hidden layer neuron threshold value of the extreme learning machine and the connection weight between the input layer and the hidden layer, so that the precision and the stability of the extreme learning machine are improved, and the line loss prediction precision is improved.
A line loss prediction system based on a particle swarm optimization extreme learning machine comprises,
an acquisition module: acquiring daily power supply amount;
a prediction module: inputting the daily power supply amount into a pre-trained extreme learning machine to obtain line loss;
in the training process, a particle swarm optimization is adopted to optimize the initial hidden layer neuron threshold of the extreme learning machine and the connection weight between the input layer and the hidden layer.
In the particle swarm optimization, an initial hidden layer neuron threshold value, a connection weight between an input layer and a hidden layer are used as particles, a root mean square error output by an extreme learning machine is used as particle fitness, and the particle speed is updated by adopting linearly decreasing inertial weight.
A computer readable storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by a computing device, cause the computing device to perform a method of line loss prediction based on a particle swarm optimization extreme learning machine.
A computing device comprising one or more processors, one or more memories, and one or more programs stored in the one or more memories and configured to be executed by the one or more processors, the one or more programs including instructions for performing a particle swarm optimization extreme learning machine-based line loss prediction method.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The present invention is not limited to the above embodiments, and any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention are included in the scope of the claims of the present invention which are filed as the application.
Claims (10)
1. A line loss prediction method based on a particle swarm optimization extreme learning machine is characterized by comprising the following steps: comprises the steps of (a) preparing a mixture of a plurality of raw materials,
acquiring daily power supply amount;
inputting the daily power supply amount into a pre-trained extreme learning machine to obtain line loss;
in the training process, a particle swarm optimization is adopted to optimize the initial hidden layer neuron threshold of the extreme learning machine and the connection weight between the input layer and the hidden layer.
2. The line loss prediction method based on the particle swarm optimization extreme learning machine according to claim 1, characterized in that: and acquiring daily power supply amount, and performing normalization processing on the daily power supply amount.
3. The line loss prediction method based on the particle swarm optimization extreme learning machine according to claim 1, characterized in that: in the particle swarm optimization, an initial hidden layer neuron threshold value, a connection weight between an input layer and a hidden layer are used as particles, and a root mean square error output by an extreme learning machine is used as particle fitness.
4. The line loss prediction method based on the particle swarm optimization extreme learning machine according to claim 3, wherein the line loss prediction method comprises the following steps: in the particle swarm optimization, the particle velocity is updated by adopting the inertia weight with linear decrement.
5. The line loss prediction method based on the particle swarm optimization extreme learning machine according to claim 4, wherein the line loss prediction method comprises the following steps: the linear decreasing inertia weight is adopted to update the particle speed, the specific formula is,
wherein,for k iterations corresponding to the particle velocity, wkFor the inertial weights corresponding to the k iterations,particle velocity corresponding to k-1 iterations, c1、c2As an acceleration factor, r1、r2Is a value range of [0,1]The random number of (a) is set,respectively iterating the optimal position of the particle extremum and the optimal position of the particle swarm extremum for k times,the particle i position is iterated k-1 times.
6. The line loss prediction method based on the particle swarm optimization extreme learning machine according to claim 5, wherein the line loss prediction method comprises the following steps: the formula of the inertial weight is as follows,
wk=wmax-(wmax-wmin)k/K
wherein, wmax、wminThe maximum value and the minimum value of the inertia weight are respectively, and K is the maximum iteration number of the particle swarm algorithm.
7. The utility model provides a line loss prediction system based on limit learning is optimized to particle swarm, its characterized in that: comprises the steps of (a) preparing a mixture of a plurality of raw materials,
an acquisition module: acquiring daily power supply amount;
a prediction module: inputting the daily power supply amount into a pre-trained extreme learning machine to obtain line loss;
in the training process, a particle swarm optimization is adopted to optimize the initial hidden layer neuron threshold of the extreme learning machine and the connection weight between the input layer and the hidden layer.
8. The line loss prediction system based on the particle swarm optimization extreme learning machine according to claim 7, wherein: in the particle swarm optimization, an initial hidden layer neuron threshold value, a connection weight between an input layer and a hidden layer are used as particles, a root mean square error output by an extreme learning machine is used as particle fitness, and the particle speed is updated by adopting linearly decreasing inertial weight.
9. A computer readable storage medium storing one or more programs, characterized in that: the one or more programs include instructions that, when executed by a computing device, cause the computing device to perform any of the methods of claims 1-6.
10. A computing device, characterized by: comprises the steps of (a) preparing a mixture of a plurality of raw materials,
one or more processors, one or more memories, and one or more programs stored in the one or more memories and configured to be executed by the one or more processors, the one or more programs including instructions for performing any of the methods of claims 1-6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011350048.7A CN112419092A (en) | 2020-11-26 | 2020-11-26 | Line loss prediction method based on particle swarm optimization extreme learning machine |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011350048.7A CN112419092A (en) | 2020-11-26 | 2020-11-26 | Line loss prediction method based on particle swarm optimization extreme learning machine |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112419092A true CN112419092A (en) | 2021-02-26 |
Family
ID=74842569
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011350048.7A Pending CN112419092A (en) | 2020-11-26 | 2020-11-26 | Line loss prediction method based on particle swarm optimization extreme learning machine |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112419092A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114172541A (en) * | 2021-12-02 | 2022-03-11 | 广东电网有限责任公司 | Energy consumption balance control method, device and system for power line communication network |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103729695A (en) * | 2014-01-06 | 2014-04-16 | 国家电网公司 | Short-term power load forecasting method based on particle swarm and BP neural network |
CN104635080A (en) * | 2015-01-29 | 2015-05-20 | 国网河南省电力公司鹤壁供电公司 | Method for predicating line loss rate of power grid |
US20180240018A1 (en) * | 2016-05-19 | 2018-08-23 | Jiangnan University | Improved extreme learning machine method based on artificial bee colony optimization |
CN108694467A (en) * | 2018-05-22 | 2018-10-23 | 中国电力科学研究院有限公司 | A kind of method and system that Line Loss of Distribution Network System rate is predicted |
CN110163410A (en) * | 2019-04-08 | 2019-08-23 | 国网信通亿力科技有限责任公司 | It is a kind of based on neural network-time series line loss power predicating method |
-
2020
- 2020-11-26 CN CN202011350048.7A patent/CN112419092A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103729695A (en) * | 2014-01-06 | 2014-04-16 | 国家电网公司 | Short-term power load forecasting method based on particle swarm and BP neural network |
CN104635080A (en) * | 2015-01-29 | 2015-05-20 | 国网河南省电力公司鹤壁供电公司 | Method for predicating line loss rate of power grid |
US20180240018A1 (en) * | 2016-05-19 | 2018-08-23 | Jiangnan University | Improved extreme learning machine method based on artificial bee colony optimization |
CN108694467A (en) * | 2018-05-22 | 2018-10-23 | 中国电力科学研究院有限公司 | A kind of method and system that Line Loss of Distribution Network System rate is predicted |
CN110163410A (en) * | 2019-04-08 | 2019-08-23 | 国网信通亿力科技有限责任公司 | It is a kind of based on neural network-time series line loss power predicating method |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114172541A (en) * | 2021-12-02 | 2022-03-11 | 广东电网有限责任公司 | Energy consumption balance control method, device and system for power line communication network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Shamshirband et al. | A survey of deep learning techniques: application in wind and solar energy resources | |
Sun et al. | Research and application of parallel normal cloud mutation shuffled frog leaping algorithm in cascade reservoirs optimal operation | |
Fei et al. | Research on data mining algorithm based on neural network and particle swarm optimization | |
Lu et al. | An adaptive neural architecture search design for collaborative edge-cloud computing | |
CN112419092A (en) | Line loss prediction method based on particle swarm optimization extreme learning machine | |
Zhu | Research on adaptive combined wind speed prediction for each season based on improved gray relational analysis | |
Sun et al. | Improved chimpanzee algorithm based on CEEMDAN combination to optimize ELM short-term wind speed prediction | |
Xiaojuan et al. | Short-time wind speed prediction for wind farm based on improved neural network | |
CN111310974A (en) | Short-term water demand prediction method based on GA-ELM | |
Li et al. | Prediction of short-term traffic flow based on PSO-optimized chaotic BP neural network | |
Xiong et al. | Prediction method of icing thickness of transmission line based on MEAO | |
Kehe et al. | Research of wind power prediction model based on RBF neural network | |
Wang | A study on the stock market prediction based on genetic neural network | |
Min et al. | Research on mid-long term load forecasting based on combination forecasting mode | |
Chen et al. | Based on a prediction method for improving WOA-Elman air quality prediction | |
Qu et al. | A new neural network with genetic algorithm in searching nonlinear function extremum | |
Sun et al. | Prediction of rural residents' tourism demand based on back propagation neural network | |
Kaliraj | Intrusion Detection Using Whale Optimization Based Weighted Extreme Learning Machine in Applied Nonlinear Analysis | |
Wang et al. | The application of photovoltaic power prediction technology | |
Niu et al. | Forecasting residential electricity based on FOAGMNN | |
Li et al. | An artificial intelligence-based fusion method for wind power prediction | |
Zhou | Construction of genetic algorithm model for fitness program optimization of middle school students | |
Li et al. | A hybrid algorithm combining auto-encoder network with sparse Bayesian regression optimized by artificial bee colony for short-term wind power forecasting | |
Sun et al. | Quarterly electricity consumption prediction based on time series decomposition method and gray model | |
Ren et al. | C4. 5 Data Algorithm for Industrialized Process |
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 |