CN114970719B - Internet of things operation index prediction method based on improved SVR model - Google Patents
Internet of things operation index prediction method based on improved SVR model Download PDFInfo
- Publication number
- CN114970719B CN114970719B CN202210584561.5A CN202210584561A CN114970719B CN 114970719 B CN114970719 B CN 114970719B CN 202210584561 A CN202210584561 A CN 202210584561A CN 114970719 B CN114970719 B CN 114970719B
- Authority
- CN
- China
- Prior art keywords
- operation index
- svr model
- model
- svr
- data
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 32
- 238000005457 optimization Methods 0.000 claims abstract description 10
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 239000002245 particle Substances 0.000 claims description 19
- 238000012360 testing method Methods 0.000 claims description 18
- 238000012549 training Methods 0.000 claims description 18
- 238000002790 cross-validation Methods 0.000 claims description 9
- 238000012795 verification Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 8
- 230000006870 function Effects 0.000 claims description 6
- 238000010200 validation analysis Methods 0.000 claims description 6
- 238000006073 displacement reaction Methods 0.000 claims description 3
- 239000000463 material Substances 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 2
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 238000007637 random forest analysis Methods 0.000 description 20
- 230000000694 effects Effects 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 238000010845 search algorithm Methods 0.000 description 3
- 238000010219 correlation analysis Methods 0.000 description 2
- 230000002068 genetic effect Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012163 sequencing technique Methods 0.000 description 2
- 238000012706 support-vector machine Methods 0.000 description 2
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000005684 electric field Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000010921 in-depth analysis Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000002203 pretreatment Methods 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an improved SVR model-based method for predicting an operation index of an Internet of things, which comprises the following steps: preprocessing climate data and the operation index data of the Internet of things; performing feature selection on the preprocessed climate data, and eliminating redundant features; establishing an RF-SVR model, and optimizing parameters of the RF-SVR model by using an intelligent optimization algorithm PSO to obtain the RF-PSO-SVR model; and outputting the prediction result of the operation index of the Internet of things through the RF-PSO-SVR model. The method improves the reliability of feature selection, is beneficial to improving the prediction precision of a subsequent prediction model, aims at that the SVR is only used to realize the accurate prediction of the local Internet of things operation index data, and optimizes the SVR model after feature selection by using a PSO algorithm, thereby improving the model fitting goodness, reducing the mean square error and effectively improving the prediction precision of the Internet of things operation index data.
Description
Technical Field
The invention belongs to the field of prediction of an Internet of things operation index of an electric power system, and particularly relates to an Internet of things operation index prediction method based on an improved SVR model.
Background
The internet technology in China is gradually penetrated from the original combination with the consumption service field to the combination with the industrial field, and the traditional industrial field is in an important window period of digital transformation and upgrading. In the face of massive data on the user side of the power system, the prior art research lacks in-depth analysis on local Internet of things operation index influence factors and influence intensities, and an effective prediction method is difficult to form.
The prediction model between the combined operation index and the climate factors can be analyzed and established by developing the research of a machine learning algorithm. The machine learning prediction method comprises two important research directions of data preprocessing and model establishment. The research of the data preprocessing direction is focused on a feature selection part, and common feature selection methods include a gray correlation analysis method, a genetic algorithm, a Random Forest (RF) and the like. The gray correlation analysis method is suitable for occasions with strong subjectivity, and influences the specificity of the feature selection result to a certain extent. While genetic algorithms require multiple iterations, they are far more computationally complex than RF. While more efficient, the RF algorithm alone ignores the contribution of features to model fit. Accordingly, researchers have proposed a method of calculating feature importance using RF and determining feature dimensions using a sequence forward selection (Sequential Forward Selection, SFS) algorithm based on classification accuracy. The K-fold cross-validation and search algorithm may perform well in further enhancing the reliability of the training set feature selection results. The things-to-things operation index belongs to continuous variables, so a regression model needs to be established. In the regression model, linear regression is most widely used, but has a great limitation in dealing with nonlinear problems. In the electric field, many variables are nonlinear. Therefore, a model needs to be built by means of an algorithm with higher generalization capability, the support vector machine regression (Support Vector Regression, SVR) maps data to a high-dimensional feature space, the optimization problem is converted into a convex quadratic programming problem, and the modeling of nonlinear data can be realized. However, accurate prediction of local internet of things operation index data cannot be achieved by using only SVR.
Therefore, a new solution is needed to solve this problem.
Disclosure of Invention
The invention aims to: in order to overcome the defects in the prior art, the method for predicting the combined operation index based on the improved SVR model is simple to use and can effectively improve the prediction accuracy.
The technical scheme is as follows: in order to achieve the above object, the present invention provides an improved SVR model-based prediction method for an operation index of an internet of things, comprising the steps of:
S1: preprocessing climate data and the operation index data of the Internet of things;
S2: performing feature selection on the preprocessed climate data, and eliminating redundant features;
S3: according to the pre-processed material-to-material operation index data and the weather data after feature selection, a weather and material-to-material operation index prediction (RF-SVR) model is established, and parameters of the RF-SVR model are optimized by using an intelligent optimization algorithm PSO to obtain an RF-PSO-SVR model;
S4: and outputting the prediction result of the operation index of the Internet of things through the RF-PSO-SVR model.
Further, the pretreatment method in the step S1 is as follows:
A1: randomly dividing a training set and a testing set according to a proportion by using a climate and Internet of things operation index data set;
A2: and respectively carrying out standardization processing on the training set and the testing set to eliminate the influence of dimension.
Further, the step S2 specifically includes:
b1: selecting training set data to perform feature selection;
B2: traversing the set number of RF trees using a grid search;
B3: determining the number of RF trees using K-fold cross-validation with minimum validation error as a criterion;
B4: the RF after optimizing the parameter is utilized to sort the feature importance degree;
B5: for importance-ranked features, dimension selection is performed using a goodness-of-fit sequence forward selection (Sequential Forward Selection, SFS) algorithm. The purpose of dimension selection is to remove redundancy and interference characteristics, improve the accuracy class of the SVR model and reduce the calculation complexity of the SVR model.
Further, the step B3 specifically includes:
the L-fold cross validation divides the data set into K groups, circulates K times, selects a group of data which is not repeated before as a validation set in each circulation, and uses the rest K-1 groups as training sets; after the cycle is finished, K parameter values and K verification errors corresponding to the K parameter values can be obtained, the average error of the K errors is calculated and used as a new verification error, and finally, the parameter value corresponding to the minimum verification error, namely the number of RF trees, is selected to be brought into the RF feature selection model.
Further, in the step B5, the method for performing dimension selection by using a sequence forward selection algorithm based on goodness of fit includes:
and adding the features into the SVR models one by one from the empty set according to the importance degree, and selecting the feature dimension which enables the SVR models to be fitted with the highest importance degree.
Further, in the step S3, the process of optimizing the parameters of the weather and the combined operation index prediction model by using the intelligent optimization algorithm PSO is as follows:
C1: setting parameter information such as iteration times, population numbers and the like of a particle swarm algorithm;
C2: initializing a penalty parameter (C), a loss function (epsilon) and a kernel coefficient (gamma) of the SVR model, initializing the position and the speed of particles in a particle swarm algorithm, and then calculating the fitness (R 2) of the SVR model of a training set as fitness;
And C3: continuously iterating, updating the position and the speed of the particles, calculating the corresponding fitness, and recording C, epsilon and gamma corresponding to the global optimal fitness;
And C4: and C3, carrying the optimal parameters C, epsilon and gamma obtained in the step C3 into a test set SVR model for prediction.
Further, in the step S3, two indexes of R 2 and mean square error (Mean Square Error, MSE) are used to test SVR model performance.
Further, the position and velocity of the particles in steps C2 and C3 are achieved by the following formula:
for particles i=1, 2,3,., N, the speed calculation formula is:
vi(t+1)=vi(t)+c1*rand()*(pbesti(t)-xi(t))+c2*rand()*(gbesti(t)-xi(t))
Wherein c 1 and c 2 are learning factors, pbest i is an optimal position obtained by a single search, gbest i is an optimal position obtained by a global search, and rand () represents a number between randomly selected intervals (0, 1);
the position calculation formula is:
xi(t+1)=xi(t)+vi(t)。
Further, the SVR model in the step C4 is implemented by the following formula:
Where α, α * are lagrange multipliers, k (x i,xj) represents a kernel function, and b is a displacement.
The method is used for predicting local Internet of things operation indexes of the metering equipment on the client side of the intelligent power grid under different climatic conditions. Firstly, characteristic selection is carried out on the climate data so as to realize improvement of the model performance and simultaneously reduce the calculation complexity, and then an optimization algorithm is used for further optimizing the model performance. During feature selection, a grid search algorithm and K-fold cross validation are used for random forest (RandomForest, RF) parameter selection. For the computed RF feature importance, a sequence forward selection (Sequential Forward Selection, SFS) algorithm based on the goodness of fit is used for dimension selection. After feature selection, a support vector machine regression (Support Vector Regression, SVR) model is optimized using a Particle Swarm Optimization (PSO) algorithm.
The beneficial effects are that: compared with the prior art, the invention has the following advantages:
1. According to the invention, the RF parameters are traversed by using a grid search algorithm, the RF parameters are determined by K-fold cross validation, and for the calculated feature importance, the feature dimension is selected by using an SFS algorithm based on the fitting goodness, so that the reliability of feature selection is improved, and the prediction precision of a subsequent prediction model is improved.
2. Aiming at the fact that accurate prediction of local Internet of things operation index data cannot be achieved only by using SVR, the SVR model after feature selection is optimized by using a PSO algorithm, so that model fitting goodness is improved, mean square error is reduced, and prediction accuracy of the Internet of things operation index data is effectively improved.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 shows the goodness of fit of SVR models corresponding to features of different dimensions;
FIG. 3 is a graph of the predictive effects of different algorithmic models.
Detailed Description
The present application is further illustrated in the accompanying drawings and detailed description which are to be understood as being merely illustrative of the application and not limiting of its scope, and various modifications of the application, which are equivalent to those skilled in the art upon reading the application, will fall within the scope of the application as defined in the appended claims.
The embodiment provides an improved SVR model-based method for predicting an operation index of an Internet of things, as shown in FIG. 1, comprising the following steps:
S1: preprocessing climate data and the operation index data of the Internet of things;
the climate variable data selected in the embodiment are data of highest temperature, lowest temperature, average temperature, relative humidity, rainfall, wind speed and air pressure of a certain area in Jiangsu province for three years continuously; the continuous three-year acquisition success rate data are used as the index variable of the combined operation.
The pretreatment process in this embodiment is as follows:
S1-1: randomly dividing climate data and an Internet of things operation index data set according to the proportion of 80% of a training set and 20% of a testing set;
S1-2: and respectively carrying out standardization treatment on the training set and the testing set to eliminate the influence of dimension.
S2: the method specifically comprises the following steps of:
S2-1: selecting training set data to perform feature selection;
s2-2: the set number of RF trees is traversed using a grid search.
S2-3: the number of preset trees is traversed through gridding search: 20. 40, 60, 80, 100, 120, 140, 160, 180, 200, 300, 500.
S2-4: determining the number of RF trees using K-fold cross-validation with minimum validation error as a criterion;
In this example, the dataset was divided into 5 groups using 5-fold cross-validation, and cycled 5 times. One set of data that was not repeated previously was selected as the validation set for each cycle, and the remaining 4 sets were used as the training set. After the cycle is completed, 5 parameter values and 5 verification errors corresponding to the parameter values can be obtained. The average of these 5 errors is calculated and used as the new verification error. Finally, the parameter value 160 corresponding to the minimum verification error is selected to be brought into the RF feature selection model.
S2-5: the RF after optimizing the parameter is utilized to sort the feature importance degree;
The feature importance ranking results in this embodiment are shown in table 1:
TABLE 1 training set feature importance
The importance of the highest temperature, the lowest temperature, the average temperature, the relative humidity, the air pressure, the rainfall and the average wind speed are respectively as follows: 0.45573, 0.25744, 0.21525, 0.02632, 0.02465, 0.01339, 0.00722.
S2-6: and carrying out dimension selection by using an SFS algorithm based on the goodness of fit.
In combination with the feature importance descending order sequencing result of table 1, features are added to the SVR model one by one and the corresponding fitting goodness is calculated, and the result is shown in fig. 2. And eliminating the average wind speed characteristic according to the model fitting goodness and the characteristic sequencing result shown in the figure 2 and the table 1.
S3: establishing a weather and combined operation index prediction (RF-SVR) model according to the preprocessed combined operation index data and the weather data after feature selection, and optimizing parameters of the RF-SVR model by using an intelligent optimization algorithm PSO to obtain an RF-PSO-SVR model;
In this embodiment, the steps specifically include the following procedures:
S3-1: setting parameter information such as iteration times, population numbers and the like of a particle swarm algorithm;
The PSO algorithm parameter setting conditions are as follows: iterating for 20 times, wherein the population size is 20, the weight factor is 0.4, and the learning factor is 2.
S3-2: initializing penalty parameters (C), loss functions (epsilon) and kernel coefficients (gamma) of the SVR model, initializing positions and speeds of particles in a particle swarm algorithm, and then calculating R 2 of a training set SVR model as fitness;
Initialized c= 2.08346, epsilon= 0.23507, gamma= 0.31808, the initial velocity of the particles is 0.8, and the initial position is a random number. The final calculated R 2 for the training set SVR model was 0.28.
S3-3: continuously iterating, updating the position and the speed of the particles, calculating the corresponding fitness, and recording C, epsilon and gamma corresponding to the global optimal fitness;
for particles i=1, 2,3,., N, the speed calculation formula is:
vi(t+1)=vi(t)+c1*rand()*(pbesti(t)-xi(t))+c2*rand()*(gbesti(t)-xi(t))
Wherein c 1 and c 2 are learning factors, pbest i is an optimal position obtained by single search, and gbest i is an optimal position obtained by global search. rand () represents the number between randomly chosen intervals (0, 1).
The position calculation formula is:
xi(t+1)=xi(t)+vi(t)
After 20 iterations are completed, the optimal parameters are c=7.27886351, epsilon= 0.29907942 gamma= 0.13297306.
S3-4: carrying the optimal parameters into a test set SVR model for prediction;
the SVR model is implemented by the following formula:
Where α, α * are lagrange multipliers, k (x i,xj) represents a kernel function, and b is a displacement.
S3-5: model performance was checked using two indicators, R 2 and mean square error (Mean Square Error, MSE). The SVR model test set R 2 corresponding to the optimal parameters is 0.79849, and the test set MSE is 0.20061.
S4: and outputting the prediction result of the combined operation index through the SVR model after optimizing the parameters.
In order to verify the prediction effect of the RF-PSO-SVR model provided by the present invention, the RF-PSO-SVR model, the SVR model before feature selection, and the RF-SVR model after feature selection were compared with the same data, and specific comparison data are shown in table 2 and fig. 3.
TABLE 2 comparison of different model Performance
As can be seen from table 2, the SVR model test set R 2 after feature selection increased by about 4.2% and the test set MSE decreased by about 4.2% compared to before feature selection. The interpretation of the weather factor features on the daily acquisition success rate is enhanced. Interference features are effectively removed after feature selection, and prediction accuracy is improved while a model is simplified. The prediction accuracy of the SVR model after feature selection optimized by the PSO algorithm is further improved, the test set R 2 is improved by about 9.9%, and the MSE of the test set is reduced by about 9.9%.
Fig. 3 further shows the prediction effect of the SVR, RF-PSO-SVR model by plotting a scatter plot with the true value as the abscissa and the predicted value as the ordinate. The model data points directly subjected to regression in fig. 3 (a) are most scattered, the distribution of the data points is converged after feature selection in fig. 3 (b), the model data points after PSO optimization in fig. 3 (c) are more concentrated near y=x, and the fitting effect is better. The three graphs (a), (b) and (c) of fig. 3 verify the variation of the test set R 2 and the test set MSE under different models. Further demonstrating the superiority of the RF-PSO-SVR algorithm in improving model performance.
Claims (8)
1. An improved SVR model-based method for predicting an operation index of an Internet of things is characterized by comprising the following steps:
S1: preprocessing climate data and the operation index data of the Internet of things;
s2: performing feature selection on the preprocessed climate data;
S3: establishing an RF-SVR model according to the preprocessed material-to-material operation index data and the climate data after feature selection, and optimizing parameters of the RF-SVR model by using an intelligent optimization algorithm PSO to obtain the RF-PSO-SVR model;
s4: outputting a prediction result of the operation index of the combined object through the RF-PSO-SVR model;
in the step S3, the process of optimizing the parameters of the climate and the combined operation index prediction model by using the intelligent optimization algorithm PSO is as follows:
c1: setting parameter information of a particle swarm algorithm;
c2: initializing a penalty parameter C, a loss function epsilon and a kernel coefficient gamma of the SVR model, initializing the position and the speed of particles in a particle swarm algorithm, and then calculating a fitness R 2 of the SVR model of a training set as a fitness;
And C3: continuously iterating, updating the position and the speed of the particles, calculating the corresponding fitness, and recording C, epsilon and gamma corresponding to the global optimal fitness;
And C4: and C3, carrying the optimal parameters C, epsilon and gamma obtained in the step C3 into a test set SVR model for prediction.
2. The method for predicting the combined operation index based on the improved SVR model according to claim 1, wherein the preprocessing method in step S1 is as follows:
A1: randomly dividing a training set and a testing set according to a proportion by using a climate and Internet of things operation index data set;
A2: and respectively carrying out standardization processing on the training set and the testing set to eliminate the influence of dimension.
3. The method for predicting the combined operation index based on the improved SVR model as claimed in claim 2, wherein the step S2 specifically comprises:
b1: selecting training set data to perform feature selection;
B2: traversing the set number of RF trees using a grid search;
B3: determining the number of RF trees using K-fold cross-validation with minimum validation error as a criterion;
B4: the RF after optimizing the parameter is utilized to sort the feature importance degree;
b5: and for the features with the ordered importance degrees, carrying out dimension selection by using a sequence forward selection algorithm based on the goodness of fit.
4. The method for predicting the combined operation index based on the improved SVR model as claimed in claim 3, wherein said step B3 specifically comprises:
Dividing a data set into K groups by K-fold cross validation, circulating K times, selecting a group of data which is not repeated before as a validation set in each circulation, and using the rest K-1 groups as training sets; after the cycle is finished, K parameter values and K verification errors corresponding to the K parameter values can be obtained, the average error of the K errors is calculated and used as a new verification error, and finally, the parameter value corresponding to the minimum verification error, namely the number of RF trees, is selected to be brought into the RF feature selection model.
5. The method for predicting the combined operation index based on the improved SVR model as claimed in claim 3, wherein the method for performing dimension selection in step B5 by using the sequence forward selection algorithm based on goodness of fit is as follows:
and adding the features into the SVR models one by one from the empty set according to the importance degree, and selecting the feature dimension which enables the SVR models to be fitted with the highest importance degree.
6. The method for predicting the combined operation index based on the improved SVR model according to claim 1, wherein in said step S3, the SVR model performance is checked by using two indexes of R 2 and mean square error.
7. The method for predicting the combined operation index based on the improved SVR model according to claim 1, wherein the positions and the velocities of the particles in the steps C2 and C3 are achieved by the following formula:
for particles i=1, 2,3,., N, the speed calculation formula is:
vi(t+1)=vi(t)+c1*rand()*(pbesti(t)-xi(t))+c2*rand()*(gbesti(t)-xi(t))
Wherein c 1 and c 2 are learning factors, pbest i is an optimal position obtained by a single search, gbest i is an optimal position obtained by a global search, and rand () represents a number between randomly selected intervals (0, 1);
the position calculation formula is:
xi(t+1)=xi(t)+vi(t)。
8. The method for predicting the combined operation index based on the improved SVR model according to claim 1, wherein the SVR model in the step C4 is implemented by the following formula:
Where α, α * are lagrange multipliers, k (x i,xj) represents a kernel function, and b is a displacement.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210584561.5A CN114970719B (en) | 2022-05-27 | 2022-05-27 | Internet of things operation index prediction method based on improved SVR model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210584561.5A CN114970719B (en) | 2022-05-27 | 2022-05-27 | Internet of things operation index prediction method based on improved SVR model |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114970719A CN114970719A (en) | 2022-08-30 |
CN114970719B true CN114970719B (en) | 2024-04-26 |
Family
ID=82956187
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210584561.5A Active CN114970719B (en) | 2022-05-27 | 2022-05-27 | Internet of things operation index prediction method based on improved SVR model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114970719B (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102419584A (en) * | 2011-11-30 | 2012-04-18 | 锦州华冠环境科技实业公司 | Method of estimating and evaluating emission law of pollution source by using internet of things and internet-of-things controller |
CN109190979A (en) * | 2018-09-03 | 2019-01-11 | 深圳市智物联网络有限公司 | A kind of industry internet of things data analysis method, system and relevant device |
CN113887571A (en) * | 2021-09-10 | 2022-01-04 | 上海工业自动化仪表研究院有限公司 | Electronic equipment fault prediction method for improving SVR algorithm |
-
2022
- 2022-05-27 CN CN202210584561.5A patent/CN114970719B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102419584A (en) * | 2011-11-30 | 2012-04-18 | 锦州华冠环境科技实业公司 | Method of estimating and evaluating emission law of pollution source by using internet of things and internet-of-things controller |
CN109190979A (en) * | 2018-09-03 | 2019-01-11 | 深圳市智物联网络有限公司 | A kind of industry internet of things data analysis method, system and relevant device |
CN113887571A (en) * | 2021-09-10 | 2022-01-04 | 上海工业自动化仪表研究院有限公司 | Electronic equipment fault prediction method for improving SVR algorithm |
Also Published As
Publication number | Publication date |
---|---|
CN114970719A (en) | 2022-08-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111738512B (en) | Short-term power load prediction method based on CNN-IPSO-GRU hybrid model | |
CN111860982A (en) | Wind power plant short-term wind power prediction method based on VMD-FCM-GRU | |
CN111069328A (en) | Isothermal extrusion process parameter optimization method based on particle swarm optimization | |
CN111832839B (en) | Energy consumption prediction method based on sufficient incremental learning | |
CN112884236B (en) | Short-term load prediction method and system based on VDM decomposition and LSTM improvement | |
CN116937579B (en) | Wind power interval prediction considering space-time correlation and interpretable method thereof | |
CN112884012A (en) | Building energy consumption prediction method based on support vector machine principle | |
CN115995810A (en) | Wind power prediction method and system considering weather fluctuation self-adaptive matching | |
CN111815026A (en) | Multi-energy system load prediction method based on feature clustering | |
CN111709454B (en) | Multi-wind-field output clustering evaluation method based on optimal copula model | |
CN115759415A (en) | Power consumption demand prediction method based on LSTM-SVR | |
CN115115090A (en) | Wind power short-term prediction method based on improved LSTM-CNN | |
CN112288157A (en) | Wind power plant power prediction method based on fuzzy clustering and deep reinforcement learning | |
CN115374710A (en) | Wind speed prediction method based on improved gull optimization algorithm and multi-core extreme learning machine | |
CN116826745B (en) | Layered and partitioned short-term load prediction method and system in power system background | |
CN114117852A (en) | Regional heat load rolling prediction method based on finite difference working domain division | |
CN114240687A (en) | Energy hosting efficiency analysis method suitable for comprehensive energy system | |
CN113627674A (en) | Distributed photovoltaic power station output prediction method and device and storage medium | |
CN114970719B (en) | Internet of things operation index prediction method based on improved SVR model | |
CN110276478B (en) | Short-term wind power prediction method based on segmented ant colony algorithm optimization SVM | |
CN114912195B (en) | Aerodynamic sequence optimization method for commercial vehicle | |
CN116307250A (en) | Short-term load prediction method and system based on typical daily feature selection | |
CN111950696A (en) | Short-term power load prediction method based on dimension reduction and improved neural network | |
CN112149052A (en) | Daily load curve clustering method based on PLR-DTW | |
CN113449466B (en) | Solar radiation prediction method and system for optimizing RELM based on PCA and chaos GWO |
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 |