CN112365082A - Public energy consumption prediction method based on machine learning - Google Patents
Public energy consumption prediction method based on machine learning Download PDFInfo
- Publication number
- CN112365082A CN112365082A CN202011344904.8A CN202011344904A CN112365082A CN 112365082 A CN112365082 A CN 112365082A CN 202011344904 A CN202011344904 A CN 202011344904A CN 112365082 A CN112365082 A CN 112365082A
- Authority
- CN
- China
- Prior art keywords
- energy consumption
- data
- machine learning
- consumption prediction
- training
- 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
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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- G06N3/02—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
-
- 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/08—Construction
Abstract
The invention discloses a public energy consumption prediction method based on machine learning, and belongs to the field of energy consumption prediction. The invention comprises the following steps: s101, collecting data; s102, data preprocessing, including: A. removing outliers by adopting an MAD algorithm; B. replacing missing values; C. carrying out variable reduction by adopting a PCA algorithm; s103, predictive modeling; the calculation was performed using a DNN deep neural network, using the collected data to use DNN with more hidden layers in the R software tool in the Keras library. The invention overcomes the current situation that the public energy consumption prediction in the prior art still lacks effective means, and aims to provide a public energy consumption prediction method based on machine learning, mainly solves the problem of how to integrate a big data platform and machine learning into an intelligent system for managing energy, and the efficiency of a public department is an important component of the concept of a smart city.
Description
Technical Field
The invention relates to the technical field of energy consumption prediction, in particular to a public energy consumption prediction method based on machine learning.
Background
In the context of smart cities, how to realize accurate energy consumption of public facilities is an important problem to be solved urgently, because large public buildings are main energy consumers, especially public large buildings with high use frequency such as education, sanitation and governments, and an accurate energy consumption prediction model can effectively provide decision basis for energy consumption supervision and energy saving optimization. However, recent developments in machine learning have not fully exploited the big data environment in this area.
Through retrieval, the Chinese patent application number: 201811519685.5, the name of invention creation is: the application discloses a method and a device for predicting energy consumption, which comprises the following steps: inputting the acquired first data into an ARIMA model which is trained in advance, and inputting the acquired second data into an SVM model which is trained in advance; determining a first energy consumption predicted value based on an ARIMA model, and determining a second energy consumption predicted value based on an SVM model; and determining a target energy consumption predicted value of the first predicted time according to the first energy consumption predicted value and the second energy consumption predicted value. When energy consumption prediction is carried out, a first energy consumption predicted value is determined based on an ARIMA model, and a second energy consumption predicted value is determined based on an SVM model. And determining a target energy consumption predicted value by combining the first energy consumption predicted value determined by the ARIMA model and the second energy consumption predicted value determined by the SVM model. And the determined target energy consumption predicted value is more accurate by utilizing the advantages of the ARIMA model and the SVM model. There is still room for further optimization.
Disclosure of Invention
1. Technical problem to be solved by the invention
The invention aims to overcome the current situation that the public energy consumption prediction in the prior art still lacks effective means, and provides a public energy consumption prediction method based on machine learning, mainly aiming at solving the problem of how to integrate a big data platform and the machine learning into an intelligent system for managing energy, wherein the efficiency of a public department is an important component of a smart city concept.
2. Technical scheme
In order to achieve the purpose, the technical scheme provided by the invention is as follows:
the invention discloses a public energy consumption prediction method based on machine learning, which comprises the following steps:
s101, collecting data;
s102, data preprocessing, including:
A. removing outliers by adopting an MAD algorithm;
B. replacing missing values;
C. carrying out variable reduction by adopting a PCA algorithm;
s103, predictive modeling; calculating by using a DNN deep neural network, using the collected data to use DNN together with more hidden layers in an R software tool in a Keras library, and evaluating the accuracy average error percentage of all DNN models by using symmetry; dividing the training sample into a training sub sample and a testing sub sample, carrying out k times of iterative training on the training sample by the neural network, and testing on the testing sub sample; if the test is erroneous, the process is repeated with a sub-sample reduction or a maximum number of iterations of 100000, and if the error starts to increase, the process stops.
Further, the data collection in step S101 includes data collection from the following three types of procedures: a. the method comprises the following steps of (1) transfer construction, wherein vitality, geographic space and static occupational attributes of each public are obtained from an EMIS information system; b. an SCADA (supervisory control and data acquisition) automatic reading energy consumption sensor is used in the internet of things network to collect energy consumption data and dynamic occupancy data; c. environmental data in the network is collected, including air temperature, wind speed, air pressure.
Further, the outlier elimination in step S102 specifically includes the following processes:
a1, calculating the median X of all elementsmedian；
A2, calculating the absolute deviation of all elements from the median value: bias ═ Xi-Xmedian|；
A3, obtaining the median MAD of absolute deviation as biasmedian；
A4, determining the parameter n, then all the data can be adjusted as follows:
further, the replacing the missing value in step S102 specifically includes the following processes:
b1, solving all order difference quotient formulas of n sample samples without missing values;
b2, establishing an interpolation polynomial f (x) by a simultaneous difference quotient formula;
and B3, substituting the attribute point x corresponding to the sample containing the missing value into the interpolation polynomial f (x) to obtain an approximate value.
Further, the reduction of the variables in step S102 specifically includes the following processes:
c1, performing a line on one feature, averaging each feature, and subtracting the average value of each feature from the original data to obtain new centralized data;
c2, solving a feature covariance matrix;
c3, solving an eigenvalue and an eigenvector according to the covariance matrix;
c4, arranging the eigenvalues in descending order, correspondingly giving eigenvectors, selecting principal components, and solving a projection matrix; selecting the more accurate data of the part with the front sequence for projection matrix;
c5, solving the data after dimension reduction according to the projection matrix;
where all preprocessing steps are performed with the clustering, the missing values in each input attribute have been replaced, as derived from the average of the remaining values in that attribute.
Further, the specific process of predictive modeling in step S103 is as follows:
using an artificial neural network having an input layer, a hidden layer and an output layer, for a hidden layer, the basic computation comprises a summation function: inputting the weighted input of the layer unit; and activating a function: calculating an output of the hidden layer by using a linear function or a non-linear function; the calculation is as follows:
where yc is the output of the computation, xi is the element X of the input vector, wi is the element of the weight vector W, the value of the weight is initially determined randomly from the interval [ -1,1], and then the error term is adjusted, and n is the hidden node number layer therein;
using the collected data to use DNN with the 2 nd, 3 rd and 4 th hidden layers in an R software tool in a Keras library, wherein the weight is adjusted to be better generalized through a dropping method, the dropping rate is 0.1, the hidden number optimizes units in each hidden layer through a cross validation program, and the number of the hidden layers is selected as a number in random intervals; the activation function is applied to each hidden layer and each output layer, the optimization algorithm of each hidden layer is an Adam algorithm, the learning rate is 0.001, and the training is carried out for 200 times at most;
the average error percentage of accuracy for all DNN models was evaluated by using symmetry:
wherein y istIs the target actual output, and ycIs the output of the calculation, and n is the number of observations in the sample;
the training sample is further divided into training and testing subsamples so as to determine the training time program of the DNN through cross validation; in the process, the neural network carries out iterative training k times on a training sample, tests on a testing subsample, and if the test is wrong, the subsample is reduced by repeating the process, or the maximum iterative times are 100000, and if the error starts to increase, the process is stopped.
3. Advantageous effects
Compared with the prior art, the technical scheme provided by the invention has the following beneficial effects:
(1) according to the public energy consumption prediction method based on machine learning, the big data platform and the machine learning are integrally applied to the energy management system, the energy efficiency of public management can be effectively improved, the adopted prediction model integrates big data collection and predicts the energy consumption of each energy in a public building, the modeling method of the prediction model is simple, and the energy consumption prediction is more accurate based on the machine learning.
(2) The model established by the public energy consumption prediction method based on machine learning can be realized in an intelligent system which is proposed and named as MERIDA, the system integrates big data collection and energy consumption models of each energy in a public building, and realizes the synergistic effect of the big data collection and the energy consumption models to form a management platform, so that the energy efficiency environment of a public department in big data is improved, and the energy consumption and the cost are reduced.
(3) According to the public energy consumption prediction method based on machine learning, the intelligent public buildings are connected as a part of the intelligent city, and the digital transformation energy management can improve the energy efficiency of public management, higher service quality and healthier environment.
Drawings
FIG. 1 is a schematic diagram of a DNN model with three hidden layers according to the present invention.
Detailed Description
For a further understanding of the invention, reference should be made to the following detailed description taken in conjunction with the accompanying drawings.
In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, and are only for convenience of description and simplicity of description, but do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus, should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
The present invention will be further described with reference to the following examples.
Example 1
The public energy consumption prediction method based on machine learning comprises the following steps:
s101, collecting data;
a big data collection method in a framework comprises data collection from the following three types of programs: a. the method comprises the following steps of (1) transfer construction, wherein vitality, geographic space and static occupational attributes of each public are obtained from an EMIS information system; b. an SCADA (supervisory control and data acquisition) automatic reading energy consumption sensor is used in the internet of things network to collect energy consumption data and dynamic occupancy data; c. environmental data in the network is collected, including air temperature, wind speed, air pressure, etc.
The above-described procedure performs national level public departments within each building to create large data sets with high number, variety and speed of data sets stored in the cloud.
S102, data preprocessing, including:
A. removing outliers by adopting an MAD algorithm; the method specifically comprises the following steps:
a1, calculating the median X of all elementsmedian；
A2, calculating the absolute deviation of all elements from the median value: bias ═ Xi-Xmedian|；
A3, obtaining the median MAD of absolute deviation as biasmedian；
A4, determining the parameter n, then all the data can be adjusted as follows:
B. replacing missing values; the method specifically comprises the following steps:
b1, solving all order difference quotient formulas of n sample samples without missing values;
b2, establishing an interpolation polynomial f (x) by a simultaneous difference quotient formula;
and B3, substituting the attribute point x corresponding to the sample containing the missing value into the interpolation polynomial f (x) to obtain an approximate value.
C. Carrying out variable reduction by adopting a PCA algorithm; the method specifically comprises the following steps:
c1, performing a line on one feature, averaging each feature, and subtracting the average value of each feature from the original data to obtain new centralized data;
c2, solving a feature covariance matrix;
c3, solving an eigenvalue and an eigenvector according to the covariance matrix;
c4, arranging the eigenvalues in descending order, correspondingly giving eigenvectors, selecting principal components, and solving a projection matrix; selecting the more accurate data of the part with the front sequence for projection matrix;
c5, solving the data after dimension reduction according to the projection matrix;
where all preprocessing steps are performed with the clustering, the missing values in each input attribute have been replaced, as derived from the average of the remaining values in that attribute.
S103, predictive modeling; calculating by using a DNN deep neural network, using the collected data to use DNN together with more hidden layers in an R software tool in a Keras library, and evaluating the accuracy average error percentage of all DNN models by using symmetry; dividing the training sample into a training sub sample and a testing sub sample, carrying out k times of iterative training on the training sample by the neural network, and testing on the testing sub sample; if the test is erroneous, the process is repeated with a sub-sample reduction or a maximum number of iterations of 100000, and if the error starts to increase, the process stops.
The specific process is as follows:
with an artificial neural network, usually having one input layer, hidden layer and output layer, DNN can add many hidden layers, which is more suitable for analyzing the intrinsic regularity of a sample under a large data set. For a hidden layer, the basic computation includes a summation function: inputting the weighted input of the layer unit; and activating a function: calculating an output of the hidden layer by using a linear function or a non-linear function; the calculation is as follows:
where yc is the output of the computation, xi is the element X of the input vector, wi is the element of the weight vector W, the value of the weight is initially determined randomly from the interval [ -1,1] and then the error term is adjusted, and n is the hidden node number layer therein.
Using the collected data to use DNN with more hidden layers such as 2 nd, 3 rd, 4 th hidden layers in R software tools in a Keras library, the weight being adjusted to be generalized better by a miss rate of 0.1, the number of hidden layers being selected as a number in a random interval by optimizing the units in each hidden layer by a cross-validation procedure; the activation function is applied to each hidden layer and each output layer, the optimization algorithm of each hidden layer is an Adam algorithm, the learning rate is 0.001, and the training is carried out for 200 times at most; the DNN model with three hidden layers is shown in detail in fig. 1.
The average error percentage of accuracy for all DNN models was evaluated by using symmetry:
wherein y istIs the target actual output, and ycIs the output of the calculation, and n is the number of observations in the sample;
the training sample is further divided into training and testing subsamples so as to determine the training time program of the DNN through cross validation; in the process, the neural network carries out iterative training k times on a training sample, tests on a testing subsample, and if the test is wrong, the subsample is reduced by repeating the process, or the maximum iterative times are 100000, and if the error starts to increase, the process is stopped.
The present embodiment integrates big data platforms and machine learning into an intelligent system for managing energy. Deep neural networks are used to create a specific predictive model of energy consumption for department buildings. The established model can be realized in an intelligent system which is proposed and named as MERIDA, the system integrates big data collection and energy consumption models of each energy source in a forecast public building, and realizes the synergistic effect of the big data collection and the forecast public building to form a management platform, so that the energy efficiency environment of a public department in big data is improved, and the energy consumption and the cost are reduced. In the embodiment, the intelligent public buildings are connected as a part of the intelligent city, and the digital transformation energy management can improve the energy efficiency, higher service quality and healthier environment of public management.
For the specific case below, the EMIS system used with real data sets from energy management information in a location initially contained 17000 multiple public building physics with large variables, environmental attributes and their energy consumption. The set of physical attributes includes building, heating, cooling and energy data, weather, geospatial and occupational data attributes describing environmental factors. The input space consists of 82 attributes and energy consumption data for the following energy sources: electricity, natural gas and heat, water and carbon dioxide emissions during 2008 to 2019.
The variable in this model is the energy consumption Q1HNDREF representing the Specific Energy Consumption (SEC) (expressed in kWh/(m2. a)) as the building area command for the energy building consumed per square meter of heating. It also includes the duration of the indoor (19 ℃) and outdoor temperatures during the heating season and the heating season. The data also normalizes the minimum and maximum values of each attribute by the distance between them. After a pre-processing stage, 575 samples of public buildings were selected for creating machine learning value. The data set was randomly divided into training and testing subsets such that 70% of the data was used for training and 30% was a machine learning method used to verify that all three results retained samples were tested. The training subset is additionally divided into training data (80% of the training set) for training DNN, while the remaining training set (20%) is used to optimize the architecture, parameters and learning time program in network cross-validation.
The present invention and its embodiments have been described above schematically, without limitation, and what is shown in the drawings is only one of the embodiments of the present invention, and the actual structure is not limited thereto. Therefore, if the person skilled in the art receives the teaching, without departing from the spirit of the invention, the person skilled in the art shall not inventively design the similar structural modes and embodiments to the technical solution, but shall fall within the scope of the invention.
Claims (6)
1. A public energy consumption prediction method based on machine learning is characterized by comprising the following steps:
s101, collecting data;
s102, data preprocessing, including:
A. removing outliers by adopting an MAD algorithm;
B. replacing missing values;
C. carrying out variable reduction by adopting a PCA algorithm;
s103, predictive modeling; calculating by using a DNN deep neural network, using the collected data to use DNN together with more hidden layers in an R software tool in a Keras library, and evaluating the accuracy average error percentage of all DNN models by using symmetry; dividing the training sample into a training sub sample and a testing sub sample, carrying out k times of iterative training on the training sample by the neural network, and testing on the testing sub sample; if the test is erroneous, the process is repeated with a sub-sample reduction or a maximum number of iterations of 100000, and if the error starts to increase, the process stops.
2. The machine learning-based public energy consumption prediction method according to claim 1, characterized in that: data collection in step S101 includes data collection from the following three types of programs: a. the method comprises the following steps of (1) transfer construction, wherein vitality, geographic space and static occupational attributes of each public are obtained from an EMIS information system; b. an SCADA (supervisory control and data acquisition) automatic reading energy consumption sensor is used in the internet of things network to collect energy consumption data and dynamic occupancy data; c. environmental data in the network is collected, including air temperature, wind speed, air pressure.
3. The machine learning-based public energy consumption prediction method according to claim 1, characterized in that: the outlier elimination in step S102 specifically includes the following steps:
a1, calculating the median X of all elementsmedian；
A2, calculating the absolute deviation of all elements from the median value:
a3, obtaining the median of the absolute deviation
A4, determining the parameter n, then all the data can be adjusted as follows:
4. the machine learning-based public energy consumption prediction method according to claim 3, characterized in that: the replacing of the missing value in step S102 specifically includes the following steps:
b1, solving all order difference quotient formulas of n sample samples without missing values;
b2, establishing an interpolation polynomial f (x) by a simultaneous difference quotient formula;
and B3, substituting the attribute point x corresponding to the sample containing the missing value into the interpolation polynomial f (x) to obtain an approximate value.
5. The machine learning-based public energy consumption prediction method according to claim 4, characterized in that: the variable reduction in step S102 specifically includes the following steps:
c1, performing a line on one feature, averaging each feature, and subtracting the average value of each feature from the original data to obtain new centralized data;
c2, solving a feature covariance matrix;
c3, solving an eigenvalue and an eigenvector according to the covariance matrix;
c4, arranging the eigenvalues in descending order, correspondingly giving eigenvectors, selecting principal components, and solving a projection matrix; selecting the more accurate data of the part with the front sequence for projection matrix;
c5, solving the data after dimension reduction according to the projection matrix;
where all preprocessing steps are performed with the clustering, the missing values in each input attribute have been replaced, as derived from the average of the remaining values in that attribute.
6. The machine learning-based public energy consumption prediction method according to claim 1, characterized in that: the specific process of predictive modeling in step S103 is as follows:
using an artificial neural network having an input layer, a hidden layer and an output layer, for a hidden layer, the basic computation comprises a summation function: inputting the weighted input of the layer unit; and activating a function: calculating an output of the hidden layer by using a linear function or a non-linear function; the calculation is as follows:
where yc is the output of the computation, xi is the element X of the input vector, wi is the element of the weight vector W, the value of the weight is initially determined randomly from the interval [ -1,1], and then the error term is adjusted, and n is the hidden node number layer therein;
using the collected data to use DNN with the 2 nd, 3 rd and 4 th hidden layers in an R software tool in a Keras library, wherein the weight is adjusted to be better generalized through a dropping method, the dropping rate is 0.1, the hidden number optimizes units in each hidden layer through a cross validation program, and the number of the hidden layers is selected as a number in random intervals; the activation function is applied to each hidden layer and each output layer, the optimization algorithm of each hidden layer is an Adam algorithm, the learning rate is 0.001, and the training is carried out for 200 times at most;
the average error percentage of accuracy for all DNN models was evaluated by using symmetry:
wherein y istIs the target actual output, and ycIs the output of the calculation, and n is the number of observations in the sample;
the training sample is further divided into training and testing subsamples so as to determine the training time program of the DNN through cross validation; in the process, the neural network carries out iterative training k times on a training sample, tests on a testing subsample, and if the test is wrong, the subsample is reduced by repeating the process, or the maximum iterative times are 100000, and if the error starts to increase, the process is stopped.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011344904.8A CN112365082A (en) | 2020-11-25 | 2020-11-25 | Public energy consumption prediction method based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011344904.8A CN112365082A (en) | 2020-11-25 | 2020-11-25 | Public energy consumption prediction method based on machine learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112365082A true CN112365082A (en) | 2021-02-12 |
Family
ID=74534266
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011344904.8A Pending CN112365082A (en) | 2020-11-25 | 2020-11-25 | Public energy consumption prediction method based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112365082A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113114161A (en) * | 2021-03-26 | 2021-07-13 | 哈尔滨工业大学 | Electromechanical system signal filtering method for eliminating outliers by using minimum median method |
| CN114580876A (en) * | 2022-02-24 | 2022-06-03 | 山西省交通新技术发展有限公司 | Energy data acquisition, analysis, management and control system and method based on energy conservation and environmental protection |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106199174A (en) * | 2016-07-01 | 2016-12-07 | 广东技术师范学院 | Extruder energy consumption predicting abnormality method based on transfer learning |
| CN110046743A (en) * | 2019-03-06 | 2019-07-23 | 上海交通大学 | Energy Consumption of Public Buildings prediction technique and system based on GA-ANN |
| CN110222887A (en) * | 2019-05-27 | 2019-09-10 | 国网安徽省电力有限公司 | Prediction technique based on VMD and DNN and the application in short-term load forecasting |
| CN110287233A (en) * | 2019-06-18 | 2019-09-27 | 华北电力大学 | A kind of system exception method for early warning based on deep learning neural network |
-
2020
- 2020-11-25 CN CN202011344904.8A patent/CN112365082A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106199174A (en) * | 2016-07-01 | 2016-12-07 | 广东技术师范学院 | Extruder energy consumption predicting abnormality method based on transfer learning |
| CN110046743A (en) * | 2019-03-06 | 2019-07-23 | 上海交通大学 | Energy Consumption of Public Buildings prediction technique and system based on GA-ANN |
| CN110222887A (en) * | 2019-05-27 | 2019-09-10 | 国网安徽省电力有限公司 | Prediction technique based on VMD and DNN and the application in short-term load forecasting |
| CN110287233A (en) * | 2019-06-18 | 2019-09-27 | 华北电力大学 | A kind of system exception method for early warning based on deep learning neural network |
Non-Patent Citations (1)
| Title |
|---|
| 蒂罗•斯特鲁兹著，王鼎等译: "《数据拟合与不确定度-加权最小二乘拟合以及推广（第2版）》", 国防工业出版社, pages: 48 - 52 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113114161A (en) * | 2021-03-26 | 2021-07-13 | 哈尔滨工业大学 | Electromechanical system signal filtering method for eliminating outliers by using minimum median method |
| CN114580876A (en) * | 2022-02-24 | 2022-06-03 | 山西省交通新技术发展有限公司 | Energy data acquisition, analysis, management and control system and method based on energy conservation and environmental protection |
| CN114580876B (en) * | 2022-02-24 | 2023-02-10 | 山西省交通新技术发展有限公司 | Energy data acquisition, analysis, management and control system and method based on energy conservation and environmental protection |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Yu et al. | Extracting knowledge from building-related data—A data mining framework | |
| CN103853106B (en) | A kind of energy consumption Prediction Parameters optimization method of building energy supplied equipment | |
| Hobson et al. | Clustering and motif identification for occupancy-centric control of an air handling unit | |
| CN106951611A (en) | A kind of severe cold area energy-saving design in construction optimization method based on user's behavior | |
| CN104881735A (en) | System and method of smart power grid big data mining for supporting smart city operation management | |
| CN108399470B (en) | Indoor PM2.5 prediction method based on multi-example genetic neural network | |
| CN112365082A (en) | Public energy consumption prediction method based on machine learning | |
| Guo et al. | Prediction of heating and cooling loads based on light gradient boosting machine algorithms | |
| CN112149890A (en) | Comprehensive energy load prediction method and system based on user energy label | |
| CN111179108A (en) | Method and device for predicting power consumption | |
| CN114897241A (en) | Intelligent building energy efficiency supervision and prediction method based on digital twins | |
| Ahmad et al. | Efficient energy planning with decomposition-based evolutionary neural networks | |
| CN111582588B (en) | Building energy consumption prediction method based on triple convolution fusion GRU | |
| CN117332288A (en) | Building energy saving technology evaluation system and method | |
| CN115879652B (en) | Hierarchical collaborative planning method and device for energy network, electronic equipment and storage medium | |
| Khayatian et al. | Benchmarking HVAC controller performance with a digital twin | |
| CN117033923A (en) | Method and system for predicting crime quantity based on interpretable machine learning | |
| CN116662860A (en) | User portrait and classification method based on energy big data | |
| Yu | Mining hidden knowledge from measured data for improving building energy performance | |
| Testasecca et al. | Towards Digital Twins of buildings and smart energy networks: Current and future trends | |
| CN115013858B (en) | Heat supply secondary network household regulation and control method based on lag time identification | |
| CN110322113A (en) | Based on markovian Line Loss of Distribution Network System processing method | |
| Bortolini et al. | Digital Twins’ Applications for Building Energy Efficiency: A Review. Energies 2022, 15, 7002 | |
| CN116204779B (en) | Method, system and readable storage medium for judging operation state of energy storage salt cavern | |
| Jradi et al. | A Digital Twin platform for energy efficient and smart buildings applications |
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 |