CN113486433A - Method for calculating energy consumption shortage number of net zero energy consumption building and filling system - Google Patents
Method for calculating energy consumption shortage number of net zero energy consumption building and filling system Download PDFInfo
- Publication number
- CN113486433A CN113486433A CN202110816756.3A CN202110816756A CN113486433A CN 113486433 A CN113486433 A CN 113486433A CN 202110816756 A CN202110816756 A CN 202110816756A CN 113486433 A CN113486433 A CN 113486433A
- Authority
- CN
- China
- Prior art keywords
- energy consumption
- data
- deficit
- training
- missing
- 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
- 238000005265 energy consumption Methods 0.000 title claims abstract description 163
- 238000000034 method Methods 0.000 title claims abstract description 24
- 238000004364 calculation method Methods 0.000 claims abstract description 36
- 238000007476 Maximum Likelihood Methods 0.000 claims abstract description 22
- 230000006735 deficit Effects 0.000 claims abstract description 22
- 238000012549 training Methods 0.000 claims description 38
- 238000012795 verification Methods 0.000 claims description 29
- 238000012360 testing method Methods 0.000 claims description 19
- 230000000694 effects Effects 0.000 claims description 12
- 230000007812 deficiency Effects 0.000 claims description 10
- 238000004146 energy storage Methods 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 7
- 238000004378 air conditioning Methods 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 6
- 238000012544 monitoring process Methods 0.000 claims description 5
- 230000007717 exclusion Effects 0.000 claims description 4
- 238000004220 aggregation Methods 0.000 claims description 3
- 230000002776 aggregation Effects 0.000 claims description 3
- QVFWZNCVPCJQOP-UHFFFAOYSA-N chloralodol Chemical compound CC(O)(C)CC(C)OC(O)C(Cl)(Cl)Cl QVFWZNCVPCJQOP-UHFFFAOYSA-N 0.000 claims description 3
- 238000002790 cross-validation Methods 0.000 claims description 3
- 230000004927 fusion Effects 0.000 claims description 3
- 238000013461 design Methods 0.000 description 8
- 238000010276 construction Methods 0.000 description 7
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 230000003111 delayed effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 239000006185 dispersion Substances 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000002411 adverse Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000001816 cooling Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 238000013526 transfer learning Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/13—Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/06—Power analysis or power optimisation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- General Physics & Mathematics (AREA)
- Computer Hardware Design (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Civil Engineering (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Architecture (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Structural Engineering (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application relates to an energy consumption vacancy number calculation method and an energy consumption vacancy number filling system for a net zero energy consumption building, wherein the energy consumption vacancy number calculation method comprises the following steps: acquiring energy consumption data, establishing a database of fixed information and internal and external data, marking the data, extracting features, extracting a maximum likelihood value, forming a feature subset, predicting a missing value and filling the data; an energy consumption deficit fill system, comprising: data acquisition system, smart electric meter, computer, display device. According to the method for calculating the energy consumption shortage number, the net zero energy consumption building energy consumption data have strong time regularity, and the filled data follow the relevant regularity.
Description
Technical Field
The application relates to an energy consumption vacancy calculation method and an energy consumption vacancy filling system for filling missing data, in particular to a net zero energy consumption building, by using the technology of the Internet of things and a big data algorithm.
Background
In Net Zero Energy consumption buildings (NZEB), the Building design can fully adapt to local climatic features and site conditions, and the approximate balance of the supply side and the demand side of the Building Energy consumption is ensured through passive design and technical means, active technical measures and renewable resources inside and outside the site. Nowadays, internet of things (IoT) technology has been applied in large scale to the acquisition of building energy consumption data.
However, in the measurement process, due to reasons such as network disconnection, sensor failure, sensor battery exhaustion, manual operation errors and the like, the phenomenon of data loss often occurs in the energy consumption time sequence data. In addition, for some buildings with complex structures, a wired transmission mode is difficult to use in the construction of the Internet of things, wireless transmission can only be adopted, so that the data loss rate is even higher than 10%, and the total data loss rate is higher by considering other factors such as sensors, manpower and the like. Due to the loss of the energy consumption monitoring data, not only can the historical matching relationship between the energy consumption supply side and the energy consumption demand side of the building be measured and verified, but also the prediction precision of the future building productivity and load can be adversely affected, and the building energy-saving target of balanced supply and demand and zero net energy consumption can not be achieved.
The building energy consumption has particularity, the energy consumption data has strong time regularity, and the filled data must follow the relevant regularity. Firstly, building energy consumption and load demand are closely related to human activities and work and rest time; secondly, the photovoltaic energy yield in renewable energy is related to the sunlight intensity, irradiance, conversion efficiency, weather, climate and other factors.
In the prior art, a plurality of methods exist for filling missing values of time series data, and a plurality of scholars deduce a series of algorithms to fill the missing data through algorithms such as statistics, artificial intelligence and the like in order to ensure the integrity of the data, such as linear, polynomial, random sampling, latin hypercube sampling and the like.
The method gives a solution from the mathematical perspective, but does not consider the regularity and the particularity of the energy consumption time sequence data problem, and does not have a method with strong enough computing capability to accelerate the operation process and realize real-time data filling.
Disclosure of Invention
The technical problem to be solved by the application is how to avoid calculating the energy consumption shortage number of the net zero energy consumption building, the problems of regularity and particularity and data inaccuracy of an energy consumption time sequence data problem are not considered, and how to enhance the calculation capacity and accelerate the calculation process, and the problems of calculating and filling the energy consumption shortage number real-time data of the net zero energy consumption building are realized.
In order to solve the above technical problem, according to an aspect of the present application, there is provided a method for calculating an energy shortage number for a net zero energy consumption building, the method comprising the steps of:
first, energy consumption data is collected. Acquiring supply side energy consumption data and demand side energy consumption data of a net zero energy consumption building, wherein the supply side energy consumption data comprises energy consumption data such as photovoltaic, ground source heat pump, energy storage and the like, and the demand side energy consumption data comprises energy consumption data such as illumination, sockets, air conditioning equipment and the like;
and, a database of fixed information and internal and external data is established. The fixed information includes construction information including the year of construction, construction structure, construction function, longitude and latitude where the construction is located, underground and ground floor number, architectural design drawing, material and thermal conductivity of door and window, energy consumption of architectural design unit, etc., and equipment information including the year of production, design age, latest maintenance time, equipment model, rated power, rated cooling load, rated frequency, etc. The internal and external data comprise personnel activities and weather data, the personnel activities comprise working days or weekends, working hours and off-duty hours, holiday time, fixed activity people number, large activity arrangement and the like, and the weather data comprise historical temperature, humidity, irradiance, wind power, wind speed, wind direction, dry bulb temperature, wet bulb temperature and the like;
then, marking data, marking the acquired energy consumption data, distinguishing the data into normal data and missing data, selecting a time period of the normal data/the data which are not missing as a training set and a verification set, and selecting a time period of the missing data as a test set; performing feature extraction on energy consumption time sequence data in the training set, the verification set and the test set to correspondingly form a training feature set, a verification feature set and a test feature set, wherein the features comprise: time features, statistical features, timing features, cross term features. Time characteristics including year, month, day, hour, minute, holiday, week, season, etc., statistical characteristics including moving average, median, 25% quantile, 75% quantile, maximum, minimum, standard deviation, skewness, kurtosis, etc., timing characteristics including sum of squares, dispersion, sum of absolute values, delayed autocorrelation coefficient, approximate entropy, number of the elements greater than the mean, number of the elements less than the mean, position of the first occurrence of the maximum, position of the first occurrence of the minimum, etc., cross term characteristics, second-order cross term characteristics generated by a factorizer, wherein the training and verification characteristic set is extracted without being put back by adopting a self-service aggregation (Bagging) mode;
then, using the training feature set, the verification feature set and the test feature set as input items, applying a Hidden Markov Model (HMM) to extract a maximum likelihood value of the energy consumption time series data, and calculating a maximum likelihood Distance through a Euclidean Distance (Euclidean Distance), a cosine, a Minkowski Distance (Minkowski Distance), and the like; searching a training and verification feature subset which is most matched with the test feature set by comparing the maximum likelihood distance; on the training Feature subset and the verification Feature subset, a lightweight Gradient Boosting Model (LightGBM) is used for training through single-Side Gradient Sampling (GOSS) and mutual Exclusion Feature Bundling (EFB), data fusion (blending) is completed through the trained Model, a missing number filling Model for missing value prediction is obtained, and missing data is filled.
According to the embodiment of the application, the maximum likelihood value obtained by calculating corresponding energy consumption time sequence data by the hidden Markov model can be stored, and only once calculation is needed when new data is acquired in the future.
According to the embodiment of the application, on each training and verification feature subset, when a lightweight gradient lifting model is applied to training, a small part of data samples are selected to be operated on the subset.
According to the embodiment of the application, fine adjustment can be performed in the missing value prediction step, the latest missing number filling model of the lightweight gradient lifting model is stored, and fine adjustment of the model is achieved by applying incremental learning and cross validation methods to newly added data.
Preferably, the energy consumption timing data takes as input the most recent 20-30 time series window.
Preferably, the specified acquisition time interval may be 5 minutes, since the amplitude of the energy consumption data fluctuations is not large.
According to another aspect of the present application, there is provided an energy deficit filling system for a net zero energy consumption building, comprising: the system comprises a data acquisition system, a data acquisition unit and a sensor, wherein the data acquisition system is arranged in a net zero energy consumption building and is used for acquiring energy consumption data of the supply side of the building such as photovoltaic, ground source heat pump, energy storage and the like in real time; the intelligent electric meter is installed in a net zero energy consumption building, and energy consumption data of demand sides of the buildings such as lighting, sockets and air conditioning equipment are collected; the computer is used for calculating the energy consumption missing data number of the supply side and the demand side by using the energy consumption missing number calculation method for the net zero energy consumption building and filling corresponding missing data; and the display device is used for displaying the energy consumption monitoring data of the supply side and the energy consumption monitoring data of the demand side and displaying an energy consumption balance curve graph.
According to the embodiment of the application, the data collector of the energy consumption shortage filling system can be a single chip microcomputer, for example, Raspberry Pi 4B can be used; the sensor can adopt RS485 serial port protocol; the computer may implement a wireless connectivity network.
According to the embodiment of the application, the energy consumption shortage filling system is in wired connection as far as possible, if the structure is complex and wired cannot be installed, a wireless gateway is adopted to achieve data uploading, and the wireless gateway adopts Lora.
Because the energy consumption vacancy number calculation method and the energy consumption vacancy number filling system are used for the net zero energy consumption building, the energy consumption vacancy number calculation method comprises the following steps: acquiring energy consumption data, establishing a fixed information and internal and external data database, marking data, extracting features, extracting a maximum likelihood value, forming a feature subset, predicting a missing value and outputting an energy consumption value. The method comprises the steps of filling and converting an energy consumption missing value into a prediction problem, using a data non-missing time period as a training set and a verification set, and using a data missing time period as a test set; different from uncertainty of future weather and indoor personnel activity conditions in the prediction problem, information in a historical time period is relatively determined, and energy consumption data have strong time regularity; because the energy consumption time sequence data has certain autocorrelation, the hidden Markov can be applied to realize the clustering of the time sequence data, and the matching of the data from the test set to the training set is completed; the lightweight gradient lifting model has the advantages that the lightweight gradient lifting model is bound on a large data set through unilateral gradient sampling and mutual exclusion characteristics, and the lightweight gradient lifting model has high-speed and high-precision computing capability, so that the following beneficial effects can be realized:
and when the energy consumption shortage number of the net zero energy consumption building is calculated, the regularity and the particularity of the energy consumption time sequence data problem are considered, the calculation capacity is enhanced, the calculation process is accelerated, and the energy consumption shortage number real-time data calculation and filling of the net zero energy consumption building are realized.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments will be briefly introduced below, and it is apparent that the drawings in the following description only relate to some embodiments of the present application and are not limiting on the present application.
FIG. 1 is a flow chart of an implementation of a method for energy deficit calculation for net zero energy buildings according to an embodiment of the application;
FIG. 2 is a schematic diagram of an energy deficit filling system for a net zero energy consumption building, according to an embodiment of the application.
Description of reference numerals:
1, building with net zero energy consumption; 11 a photovoltaic panel; 12 a ground source heat pump; 13 an energy storage device; 21 air conditioning equipment; 31 a gateway; 41 supply side energy consumption profile; 42 missing number phenomenon; 43 demand side energy consumption curve; 44 energy consumption balance curve; 51 a display device; 101 energy consumption data; 102 fixing the information; 103 internal and external data; 201 marking data; 301, extracting features; 401 maximum likelihood value; 501 training and verifying a feature set subset; 601 missing value prediction; 701 is filled with data.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the drawings of the embodiments of the present application. It should be apparent that the described embodiments are only some of the embodiments of the present application, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the described embodiments of the application without any inventive step, are within the scope of protection of the application.
Unless defined otherwise, technical or scientific terms used herein shall have the ordinary meaning as understood by one of ordinary skill in the art to which this application belongs. The use of "first," "second," and similar terms in the description and claims of this patent application do not denote any order, quantity, or importance, but rather the terms are used to distinguish one element from another. Also, the use of the terms "a" or "an" and the like do not denote a limitation of quantity, but rather denote the presence of at least one.
Fig. 1 is a flowchart of an implementation of an energy consumption deficit calculation method for a net zero energy consumption building according to an embodiment of the present application.
As shown in fig. 1, the energy consumption deficiency calculation method for the net zero energy consumption building according to the embodiment of the present application includes the steps of collecting energy consumption data 101, establishing a fixed information 102 and internal and external data 103 database, marking data 201, feature extraction 301, extracting a maximum likelihood value 401, forming a training and verification feature subset 501, predicting a deficiency value 601, and filling data 701.
The energy consumption data 101 is acquired from the energy consumption data of the net zero energy consumption building, and the energy consumption data to be acquired includes the energy consumption data of the supply side and the energy consumption data of the demand side. The energy consumption data of the supply side comprises energy consumption data of photovoltaic, ground source heat pump, energy storage and the like; the energy consumption data of the demand side comprises energy consumption data of lighting, sockets, air conditioning equipment and the like. Each data subentry needs to obtain real-time energy consumption data through subentry measurement, and the time interval is preferably 5 minutes.
A database of fixed information 102 and internal and external data 103 is established. The fixed information includes building information and device information. The building information comprises construction year, building structure, building function, longitude and latitude of the building, underground and ground floor number, building design drawing, material and heat conductivity of door and window, energy consumption of building design unit, and the like; the equipment information includes the year of production, the design service life, the latest maintenance time, the equipment model, the rated power, the rated cold load, the rated frequency, and the like. The internal and external data includes personnel activity and weather data. The personnel activities comprise working days or weekends, working and working hours, holiday time, fixed activity personnel number, large-scale activity arrangement and the like; weather data includes historical temperature, humidity, irradiance, wind speed, wind direction, dry and wet bulb temperatures, and the like.
The marking data 202 is a marking of the collected energy consumption data. Normal data and missing data are distinguished and marked, normal data (namely, data which are not missing) time periods are selected as training sets and verification sets, and missing data time periods are selected as test sets.
The feature extraction 301 is to perform feature extraction on the energy consumption time sequence data in the training set, the verification set, and the test set to form a training feature set, a verification feature set, and a test feature set correspondingly. The characteristics comprise time characteristics, statistical characteristics, time sequence characteristics and cross item characteristics. Wherein, the time characteristics comprise year, month, day, hour, minute, whether to be holiday, week, season, etc.; statistical characteristics including moving average, median, 25% quantile, 75% quantile, maximum, minimum, standard deviation, skewness, kurtosis, etc.; the time sequence characteristics comprise a square sum, dispersion, a sum of absolute values, a delayed autocorrelation coefficient, approximate entropy, a number larger than the mean value, a number smaller than the mean value, a position where the maximum value appears for the first time, a position where the minimum value appears for the first time and the like; cross term features, second order cross term features generated by a factorization machine.
To avoid the phenomenon of negative transfer learning, the model obtained on the training and verification feature subset is weaker than the overall model. The method adopts a self-help aggregation mode to extract the training and verification feature set without putting back.
The extraction maximum likelihood value 401 is a maximum likelihood value of energy consumption time sequence data extracted by applying a hidden markov model with a training feature set, a verification feature set and a test feature set as input items. The maximum likelihood distance is calculated by the euclidean distance, the cosine, the minkowski distance, etc.
The feature subset is formed by comparing the maximum likelihood distance to find the training and verification feature subset 501 that best matches the test feature set.
The missing value prediction 601 is that a lightweight gradient lifting model is applied to a training feature subset and a verification feature subset to train through unilateral gradient sampling and mutual exclusion feature binding, and the model obtained through training is applied to complete data fusion to obtain a missing value prediction missing number filling model.
Finally, the padding data 701 is completed.
According to the embodiment of the application, the maximum likelihood value obtained by calculating corresponding energy consumption time sequence data by the hidden Markov model can be stored, and only once calculation is needed when new data is acquired in the future.
According to the embodiment of the application, when the lightweight gradient lifting model is applied to training on each training and verification feature subset, a small part of data samples are selected to be operated on the subset, and the calculation efficiency and the model robustness can be improved.
According to the embodiment of the application, the missing number filling model of the latest lightweight gradient lifting model can be stored, the model can be finely adjusted by applying incremental learning and cross validation methods to newly added data, the precision can be guaranteed, the calculation speed can be improved, and real-time data calculation and filling can be better realized.
Preferably, the energy consumption time series data may take as input the most recent 20-30 time series window.
Still preferably, the time interval for collecting the energy consumption data may be about 5 minutes.
FIG. 2 is a schematic diagram of an energy deficit filling system for a net zero energy consumption building, according to an embodiment of the application.
As shown in fig. 2, the energy shortage filling system for the net zero energy consumption building according to the embodiment of the present application includes a data acquisition system (not shown), a smart meter (not shown), a computer (not shown) and a display device 51.
Install the data acquisition system in net zero energy consumption building, including data collection station and sensor, gather the energy consumption data of the supply side of buildings such as photovoltaic, ground source heat pump, energy storage in real time, for example: photovoltaic panel 11, ground source heat pump 12, energy storage equipment 13.
The intelligent electric meter installed in the net zero energy consumption building collects energy consumption data of demand sides of buildings such as lighting, sockets and air conditioning equipment 21.
And the computer is used for calculating the energy consumption missing data number of the supply side and the demand side and filling corresponding missing data.
The display device 51 is used for displaying the supply-side energy consumption curve 41, the missing phenomenon 42 and the demand-side energy consumption curve 43 and displaying the energy consumption balance curve 44.
According to the embodiment of the application, the data collector of the energy consumption shortage filling system can be a single chip microcomputer, for example, Raspberry Pi 4B can be used; the sensor can adopt RS485 serial port protocol; a computer capable of wireless connection to a network may be used.
Preferably, the energy consumption shortage filling system adopts wired connection as far as possible, if the structure is complex and wired cannot be installed, a wireless gateway is adopted to upload data, and the wireless gateway adopts Lora.
Because the energy consumption vacancy number calculation method and the energy consumption vacancy number filling system are used for the net zero energy consumption building, the energy consumption vacancy number calculation method comprises the following steps: collecting energy consumption data, establishing a fixed information and internal and external data database, marking data, extracting features, extracting a maximum likelihood value, forming a feature subset, predicting a missing value, finely adjusting and outputting an energy consumption value. The method comprises the steps of filling and converting an energy consumption missing value into a prediction problem, using a data non-missing time period as a training set and a verification set, and using a data missing time period as a test set; different from uncertainty of future weather and indoor personnel activity conditions in the prediction problem, information in a historical time period is relatively determined, and energy consumption data have strong time regularity; the energy consumption time sequence data has certain autocorrelation, so that the data can be matched from a test set to a training set; and a fast and high-precision calculation method on a large data set is used for calculation, so the following beneficial effects can be realized:
and when the energy consumption shortage number of the net zero energy consumption building is calculated, the regularity and the particularity of the energy consumption time sequence data problem are considered, the calculation capacity is enhanced, the calculation process is accelerated, and the energy consumption shortage number real-time data calculation and filling of the net zero energy consumption building are realized.
The above are exemplary embodiments of the present application only, and are not intended to limit the scope of the present application, which is defined by the appended claims.
Claims (16)
1. A method of energy consumption deficit calculation for a net zero energy consumption building, the method comprising the steps of:
acquiring energy consumption data, and acquiring supply side energy consumption data and demand side energy consumption data of the net zero energy consumption building;
establishing a database of fixed information and internal and external data,
the fixed information comprises building information and equipment information, and the internal and external data comprises personnel activity and weather data;
marking data, namely marking the acquired energy consumption data, distinguishing normal data and missing data, selecting a time period of normal data/non-missing data as a training set and a verification set, and selecting a time period of missing data as a test set;
extracting characteristics, namely extracting the characteristics of the energy consumption time sequence data in the training set, the verification set and the test set to correspondingly form a training characteristic set, a verification characteristic set and a test characteristic set;
extracting a maximum likelihood value, and calculating the maximum likelihood value and the maximum likelihood distance by taking the training feature set, the verification feature set and the test feature set as input items;
forming a training and verification feature subset, and searching the training and verification feature subset which is most matched with the test feature set by comparing the maximum likelihood distance;
missing value prediction, namely acquiring a missing number filling model of the missing value prediction on the training feature subset and the verification feature subset; and
and filling data, and filling the missing data.
2. The energy consumption deficit calculation method according to claim 1, wherein the feature extraction is non-replacement extraction of the training feature set and the verification feature set in a self-help aggregation manner.
3. The energy consumption deficit calculation method according to claim 1, wherein the feature extracted features include: time features, statistical features, timing features, cross term features.
4. The energy consumption deficit calculation method according to claim 1, wherein the extracting the maximum likelihood value applies a hidden markov model to extract the maximum likelihood value of the energy consumption time series data, and the maximum likelihood distance is calculated by an euclidean distance, a cosine distance, a minkowski distance, or the like.
5. The energy consumption deficiency calculation method according to claim 1, wherein the deficiency value prediction is trained by using a lightweight gradient lifting model through unilateral gradient sampling and mutual exclusion feature bundling, and the model obtained through training is used for completing data fusion to obtain a deficiency filling model for deficiency value prediction.
6. The energy consumption deficit calculation method according to claim 4, wherein the maximum likelihood value obtained by calculating the corresponding energy consumption time series data by the hidden Markov model is stored and only needs to be calculated once every time new data is acquired later.
7. The energy consumption deficit calculation method according to claim 5, wherein, in the deficit value prediction step, when training is performed on the training feature subset and the verification feature subset by applying a lightweight gradient boosting model, a small number of data samples are selected and operated on the subsets.
8. The energy consumption deficiency calculation method according to claim 1, wherein the deficiency value prediction step further comprises:
and fine tuning, namely saving the missing number filling model of the latest lightweight gradient lifting model, and implementing the fine tuning of the model by applying incremental learning and cross validation methods to newly added data.
9. The energy consumption deficit calculation method according to claim 1, wherein the energy consumption time series data takes as input the most recent 20-30 time series window.
10. The energy consumption deficit calculation method according to claim 1, wherein the time interval for collecting energy consumption data is 5 minutes.
11. An energy deficit fill system for a net zero energy building, comprising:
a data acquisition system, comprising: the system comprises a data acquisition unit and a sensor, wherein the data acquisition unit is used for acquiring energy consumption data of the supply side of buildings such as photovoltaic buildings, ground source heat pumps, energy storage buildings and the like in real time;
the intelligent electric meter is used for collecting energy consumption data of demand sides of buildings such as lighting, sockets and air conditioning equipment, and is installed in the net zero energy consumption building;
a computer for calculating the energy consumption missing data number of the supply side and the demand side and filling corresponding missing data,
wherein the method of calculating the supply-side and demand-side energy consumption deficiency data number comprises the method of calculating the energy consumption deficiency number according to any one of claims 1 to 10; and
and the display device displays the energy consumption monitoring data of the supply side and the energy consumption monitoring data of the demand side and displays an energy consumption balance curve graph.
12. The energy consumption vacancy filling system of claim 11 wherein the data collector is a single-chip computer Raspberry Pi 4B.
13. The energy consumption deficit filling system according to claim 11, wherein the sensor employs an RS485 serial protocol.
14. The energy deficit filling system according to claim 11, wherein the energy deficit filling system employs a wired connection.
15. The energy deficit fill system according to claim 11, further comprising a wireless gateway, the energy deficit fill system employing wireless connectivity, the wireless gateway employing Lora.
16. The energy deficit filling system according to claim 11, wherein the computer is capable of implementing a wireless connection network.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011631189 | 2020-12-31 | ||
| CN2020116311896 | 2020-12-31 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113486433A true CN113486433A (en) | 2021-10-08 |
Family
ID=77942518
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110816756.3A Pending CN113486433A (en) | 2020-12-31 | 2021-07-20 | Method for calculating energy consumption shortage number of net zero energy consumption building and filling system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113486433A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117852928A (en) * | 2024-03-08 | 2024-04-09 | 国网北京市电力公司 | Near zero energy consumption building load prediction method, device, equipment and medium |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108615096A (en) * | 2018-05-10 | 2018-10-02 | 平安科技（深圳）有限公司 | Server, the processing method of Financial Time Series and storage medium |
| CN109858700A (en) * | 2019-02-01 | 2019-06-07 | 华北水利水电大学 | BP neural network heating system energy consumption prediction technique based on similar screening sample |
| CN109961177A (en) * | 2019-03-11 | 2019-07-02 | 浙江工业大学 | A kind of general water cooled central air conditioner energy consumption prediction technique based on shot and long term memory Recognition with Recurrent Neural Network |
| 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 |
| CN110597799A (en) * | 2019-09-17 | 2019-12-20 | 上海仪电（集团）有限公司中央研究院 | Automatic filling method, system and equipment for missing value of time sequence data |
| CN111046027A (en) * | 2019-11-25 | 2020-04-21 | 北京百度网讯科技有限公司 | Missing value filling method and device for time series data |
| CN111080009A (en) * | 2019-12-13 | 2020-04-28 | 北京瑞莱智慧科技有限公司 | Time series-based data prediction and completion method, device, medium, and apparatus |
| CN111476422A (en) * | 2020-04-10 | 2020-07-31 | 北京石油化工学院 | L ightGBM building cold load prediction method based on machine learning framework |
| CN112070317A (en) * | 2020-09-17 | 2020-12-11 | 四川长虹电器股份有限公司 | Hotel air conditioner energy consumption prediction method |
| CN112085125A (en) * | 2020-09-29 | 2020-12-15 | 西安交通大学 | Missing value filling method based on linear self-learning network, storage medium and system |
| CN112101755A (en) * | 2020-09-02 | 2020-12-18 | 江苏中设集团股份有限公司 | Service area energy consumption estimation method and system based on multi-factor comprehensive consideration |
-
2021
- 2021-07-20 CN CN202110816756.3A patent/CN113486433A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108615096A (en) * | 2018-05-10 | 2018-10-02 | 平安科技（深圳）有限公司 | Server, the processing method of Financial Time Series and storage medium |
| CN109858700A (en) * | 2019-02-01 | 2019-06-07 | 华北水利水电大学 | BP neural network heating system energy consumption prediction technique based on similar screening sample |
| 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 |
| CN109961177A (en) * | 2019-03-11 | 2019-07-02 | 浙江工业大学 | A kind of general water cooled central air conditioner energy consumption prediction technique based on shot and long term memory Recognition with Recurrent Neural Network |
| CN110597799A (en) * | 2019-09-17 | 2019-12-20 | 上海仪电（集团）有限公司中央研究院 | Automatic filling method, system and equipment for missing value of time sequence data |
| CN111046027A (en) * | 2019-11-25 | 2020-04-21 | 北京百度网讯科技有限公司 | Missing value filling method and device for time series data |
| CN111080009A (en) * | 2019-12-13 | 2020-04-28 | 北京瑞莱智慧科技有限公司 | Time series-based data prediction and completion method, device, medium, and apparatus |
| CN111476422A (en) * | 2020-04-10 | 2020-07-31 | 北京石油化工学院 | L ightGBM building cold load prediction method based on machine learning framework |
| CN112101755A (en) * | 2020-09-02 | 2020-12-18 | 江苏中设集团股份有限公司 | Service area energy consumption estimation method and system based on multi-factor comprehensive consideration |
| CN112070317A (en) * | 2020-09-17 | 2020-12-11 | 四川长虹电器股份有限公司 | Hotel air conditioner energy consumption prediction method |
| CN112085125A (en) * | 2020-09-29 | 2020-12-15 | 西安交通大学 | Missing value filling method based on linear self-learning network, storage medium and system |
Non-Patent Citations (3)
| Title |
|---|
| AOWABIN RAHMAN等: ""Predicting electricity consumption for commercial and residential buildings using deep recurrent neural networks"", 《APPLIED ENERGY》 * |
| 唐舟进 等: ""一种基于相关分析的局域最小二乘支持向量机小尺度网络流量预测算法"", 《物理学报》 * |
| 高山 等: ""全矢HMM在轴承剩余寿命预测中的应用"", 《机械设计与制造》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117852928A (en) * | 2024-03-08 | 2024-04-09 | 国网北京市电力公司 | Near zero energy consumption building load prediction method, device, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102705957B (en) | Method and system for predicting hourly cooling load of central air-conditioner in office building on line | |
| EP3614055A1 (en) | Heat supply regulation method and apparatus | |
| CN102779228B (en) | Method and system for online prediction on cooling load of central air conditioner in marketplace buildings | |
| CN107133702B (en) | Full-field power prediction method for wind power plant | |
| CN105373971A (en) | Method of building energy efficiency management on the basis of big data | |
| CN105260800A (en) | Photovoltaic module temperature prediction method and device | |
| CN105091225A (en) | Method and system for offline calculating indoor air conditioner starting and temperature standard exceeding | |
| CN104930589B (en) | Cell heating control system and method based on distributed sensor | |
| CN110298104A (en) | It is a kind of that energy saving building design information processing system and method are carried out using digital simulation | |
| CN102841592A (en) | Greenhouse control system based on wireless interconnection technology | |
| CN109508743A (en) | A kind of restorative procedure of the building energy consumption monitoring abnormal data based on KNN innovatory algorithm | |
| CN105429184B (en) | The overall process quantitative estimation method that a kind of wind-powered electricity generation priority scheduling is evaluated | |
| CN113486433A (en) | Method for calculating energy consumption shortage number of net zero energy consumption building and filling system | |
| US20200088433A1 (en) | Thermo Efficiency Measurement System | |
| CN108614173B (en) | Low-voltage side comprehensive energy consumption analysis method and low-voltage side comprehensive energy consumption analysis system | |
| CN103234583A (en) | Real-time monitoring system with light measuring and wind measuring functions | |
| TWI671996B (en) | Method for judging the orientation of a solar power module | |
| Chen et al. | Research on energy-saving optimization model based on building energy consumption data | |
| Wang | Research and design of intelligent monitoring system for greenhouse based on Internet of Things | |
| CN107133372B (en) | Mobile communication base station energy consumption benchmarking scene evaluation method and device | |
| CN105653804A (en) | Simulation analysis device and method for building energy consumption | |
| CN204155277U (en) | Based on the information continuous collecting apparatus of settlement indoor and outdoor surroundings | |
| CN206584209U (en) | A kind of information acquisition system for cotton storage storehouse | |
| CN202535372U (en) | Network management system of real-time wind resource of ten million kilowatt level wind power base | |
| Yang et al. | A forecasting method of air conditioning energy consumption based on extreme learning machine algorithm |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20211008 |