CN112465198B - Double-agent assisted search energy optimization method for small sample data of park - Google Patents
Double-agent assisted search energy optimization method for small sample data of park Download PDFInfo
- Publication number
- CN112465198B CN112465198B CN202011279183.7A CN202011279183A CN112465198B CN 112465198 B CN112465198 B CN 112465198B CN 202011279183 A CN202011279183 A CN 202011279183A CN 112465198 B CN112465198 B CN 112465198B
- Authority
- CN
- China
- Prior art keywords
- heat
- energy
- rated
- formula
- lithium bromide
- 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
- 238000005457 optimization Methods 0.000 title claims abstract description 45
- 238000000034 method Methods 0.000 title claims abstract description 31
- 238000005070 sampling Methods 0.000 claims abstract description 21
- AMXOYNBUYSYVKV-UHFFFAOYSA-M lithium bromide Chemical compound [Li+].[Br-] AMXOYNBUYSYVKV-UHFFFAOYSA-M 0.000 claims description 70
- 239000003795 chemical substances by application Substances 0.000 claims description 34
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 claims description 28
- 230000005611 electricity Effects 0.000 claims description 24
- 238000005338 heat storage Methods 0.000 claims description 20
- 238000002485 combustion reaction Methods 0.000 claims description 18
- 230000006870 function Effects 0.000 claims description 18
- 239000007789 gas Substances 0.000 claims description 16
- 239000003345 natural gas Substances 0.000 claims description 14
- 238000005259 measurement Methods 0.000 claims description 11
- 230000008901 benefit Effects 0.000 claims description 10
- 238000010248 power generation Methods 0.000 claims description 8
- 239000002918 waste heat Substances 0.000 claims description 8
- 230000009977 dual effect Effects 0.000 claims description 7
- 238000012549 training Methods 0.000 claims description 7
- 230000020169 heat generation Effects 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 4
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 claims description 2
- 238000004146 energy storage Methods 0.000 claims description 2
- 238000004519 manufacturing process Methods 0.000 claims description 2
- 238000005312 nonlinear dynamic Methods 0.000 claims description 2
- 230000008569 process Effects 0.000 claims description 2
- 238000012546 transfer Methods 0.000 claims description 2
- 150000001875 compounds Chemical class 0.000 claims 1
- 238000011156 evaluation Methods 0.000 claims 1
- 238000013135 deep learning Methods 0.000 abstract description 4
- 230000010354 integration Effects 0.000 abstract description 4
- 230000000694 effects Effects 0.000 abstract description 3
- 230000002708 enhancing effect Effects 0.000 abstract description 2
- 238000010845 search algorithm Methods 0.000 description 3
- 238000009434 installation Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 239000000470 constituent Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/047—Probabilistic or stochastic 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/08—Probabilistic or stochastic CAD
-
- 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)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Evolutionary Computation (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Development Economics (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Finance (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Molecular Biology (AREA)
- General Business, Economics & Management (AREA)
- Computing Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Accounting & Taxation (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Game Theory and Decision Science (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Operations Research (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Probability & Statistics with Applications (AREA)
Abstract
The invention discloses a double-agent auxiliary search energy optimization method (DSEO-SSD for short) aiming at small sample data of a park. Aiming at the engineering problem of intelligent park multi-energy collaborative energy optimization, the method adopts a double-agent model combining global/local search to improve the deep learning optimization algorithm. In DSEO-SSD, dividing a small sample database of the intelligent park into a plurality of subsets, and enhancing the prediction precision of the agent model through ensemble learning; and simultaneously, a Kriging-based proxy model strategy and an RBF proxy model strategy are designed and are respectively used for global searching and local searching, and the Kriging-based proxy model strategy and the RBF proxy model strategy are combined into a high-precision proxy integration. If the problem is solved by using the traditional optimization algorithm of single agent assisted search, the problems of insufficient sampling data and possible local optimization are often faced, and when the method is applied to the intelligent park multi-energy collaborative energy optimization engineering problem, the effect is more obvious than that of the common intelligent optimization method.
Description
Technical Field
The invention relates to a small sample data optimization algorithm based on Kriging and RBF double-agent assisted local/global search for intelligent park multi-energy collaborative energy optimization.
Background
The intelligent park is one of the important constituent modules of intelligent city development, and is not only an industrial park but also a resident park at present, and one of the bottlenecks restricting the quick transformation is the problem of insufficient data acquisition caused by insufficient installation of a measurement terminal, insufficient measurement data, too long measurement time interval and the like in the early stage.
With the construction of the energy internet and the ubiquitous electric power internet of things, the problem of cooperative energy optimization including different energy forms such as cold, heat, electricity and gas is considered to be an important research subject at present. For the problem of collaborative energy optimization of an intelligent park under the condition of insufficient data, if a traditional optimization algorithm of single agent assisted search is used for solving the problem, the problem of local optimum is often faced.
Disclosure of Invention
In view of the above, the invention provides a double-agent assisted search energy optimization method for small sample data in a park, which aims to solve the following technical problems, technical schemes and technical effects:
the invention aims at the technical problems that: the problems of insufficient installation of the early-stage measurement terminal, insufficient measurement data, too long measurement time interval and the like exist in the existing park, and the problems of insufficient data acquisition, unqualified data prediction precision and large prediction error are caused. In this scenario, it is therefore difficult to make an optimal solution for the multi-energy collaborative energy optimization of the intelligent park. If the problem is solved by using the conventional optimization algorithm of single agent assisted search, the problem of local optimum is often faced.
The technical scheme of the invention is as follows: the invention discloses a double-agent assisted search energy optimization method for small sample data of a park, which is applied to multi-energy collaborative energy optimization under the condition that only small sample data of an intelligent park is acquired. Firstly, setting multiple optimization targets of intelligent park multi-energy collaborative scheduling, such as triple optimization targets of maximum cold, heat and electricity sales benefits, highest new energy utilization rate and minimum electricity consumption peak Gu Chazhi, and unifying the three targets into a single target through target weighting, thereby simplifying the complexity of a model and improving the optimization efficiency. Then, operational equations and inequality constraints are determined based on the capacity devices of the intelligent campus energy station and the load devices within the campus. And secondly, dividing a small sample data set of the capacity, the load energy and the like of the universal station collected in the park into a plurality of subsets, and training a plurality of models on the same sample set through ensemble learning to enhance the prediction accuracy of the models. Finally, two model management strategies, namely a Kriging agent model management strategy and an RBF model management strategy are designed and used for global searching and local searching respectively, and the two searching modes are skillfully combined to form a high-precision Kriging agent integration and RBF agent integration, so that an optimal solution can be found more quickly and accurately.
The invention has the technical effects that: 1) The small sample data set of the intelligent park is divided into a plurality of subsets, and a plurality of models are trained on the same sample set through ensemble learning, so that the prediction accuracy of the models can be enhanced, and the prediction error is effectively reduced under the condition of insufficient data; 2) The double-agent auxiliary search algorithm adopts the thought of combining local search and global search, so that the double-agent auxiliary search algorithm can fully search near the current optimal point, and the balance of convergence quality and convergence efficiency is obtained.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a dual agent assisted search optimization algorithm of the present invention;
FIG. 2 is a schematic illustration of a echelon sampling employed in the present invention;
FIG. 3 is a schematic diagram of the integrated learning step of the proxy model of the present invention.
Detailed Description
In order to make the above objects, features and advantages of the present invention more comprehensible, the following detailed description of the technical solution of the present invention refers to the accompanying drawings and specific embodiments. It should be noted that the described embodiments are only some embodiments of the present invention, and not all embodiments, and that all other embodiments obtained by persons skilled in the art without making creative efforts based on the embodiments in the present invention are within the protection scope of the present invention.
As shown in FIG. 1, the invention provides a double-agent assisted search energy optimization method (DSEO-SSD) for small sample data of a park, and adopts a global/local search combined double-agent model to improve a deep learning optimization algorithm for the engineering problem of intelligent park multi-energy collaborative energy optimization. The data set collected by the measurement terminal divides the original data set into a plurality of subsets through random sampling and echelon sampling, and the prediction precision of the model is enhanced through ensemble learning (namely, a plurality of models are trained on the same sample set); two brand new model management strategies are designed at the same time, namely a Kriging agent model strategy and a deep learning Radial Basis Function (RBF) agent model strategy are respectively used for global searching and local searching, and are skillfully combined to form a high-precision Kriging-RBF agent integration. The sampling thought is as follows: the random sampling and the echelon sampling method are combined, each sample unit of the random sampling is randomly extracted, the overall parameters can be estimated according to the probability theory sample statistics, and the sampling error can be calculated, so that the overall inference of the data samples is obtained; the echelon sampling can improve the sampling speed, can effectively avoid sampling abnormal data, and can ensure the integrity of sample data. First, a small sample data set S t And then sequentially extracting the data set with the size of K as a sample subset according to the size of the data sequence number and the sequence from the smaller sequence number to the larger sequence number.
The specific implementation steps are as follows:
step 1: three optimization targets of maximum benefits of selling cold, heat and electricity (time-sharing charging of cold, heat and electricity), highest renewable energy utilization rate (namely lowest discarding rate) and minimum electricity consumption peak Gu Chazhi of the intelligent park are set, and the three targets are unified into a single target through target weighting, so that complexity of a model is simplified, and optimization efficiency is improved. The related expression is as follows:
the multi-objective dynamic nonlinear optimization function based on the weight function unifies three objectives into a simplified expression of single-objective nonlinear dynamic optimization through objective weighting as (1), wherein A, B, C are respectively different objective functions, and alpha, beta and delta are respectively weighting coefficients of different objective functions, and the weighting coefficients are selected according to different application scenes.
minF＝αA+βB+δC (1)
In the application of the invention, different objective functions A, B, C respectively correspond to the maximum benefits of cold, heat and electricity sold by the universal station, the highest renewable energy utilization rate and the minimum electricity consumption peak Gu Chazhi, and the objective functions are respectively expressed in the formulas (2) - (4).
A＝max(f 1 +f 2 +f 3 +f g -f c ) (2)
Wherein f 1 For electricity utilization benefits, f 2 To use heat gain, f 3 To use the cold returns, f g F, for surplus electricity Internet surfing benefits c Is the running cost.
Wherein N is re In order to be able to use the renewable energy source,is a single some renewable energy source j.
Wherein L, H is the highest and lowest limit values of electric power, T is the total number of electric power measurement time intervals, P at time i i,j Representing the electrical power.
Step 2: operational equation and inequality constraints are determined based on the intelligent campus energy plant energy production facility and the campus load facility. In the application of the invention, the energy generating equipment of the universal station comprises a gas boiler, a lithium bromide machine, a heat pump and a heat storage tank, and the park load equipment comprises a transferable load, a charging load, an energy storage load and the like, and the related constraint conditions are as follows:
1) Electric balance constraint
E G (t)＝E R (t)+E d (t)+E s (t) (5)
E G (t)＝η g-e ·c 1 (t) (6)
E G (t)≤E G-rated (t) (7)
E in formula (5) G (t) represents the power generation amount of the internal combustion engine, E R (t) represents the electricity consumption of the station, E d (t) represents the load electricity consumption, E s And (t) represents the remaining internet power. Eta in the formula (6) g-e Representing the conversion of natural gas into electricity c 1 And (t) represents natural gas consumed by the internal combustion generator for power generation. Formula (7) is the capacity constraint of the internal combustion engine, E G-rated And (t) represents the rated power of the internal combustion generator.
2) Thermal equilibrium constraints
S bo H bo (t)+S li-h H li-h (t)+S hp H hp (t)+S sk-d H sk-d (t)＝H d (t)+S sk-c H sk-c (t) (8)
H in (8) bo (t)、H li-h (t)、H hp (t)、H sk-d (t)、H sk-c (t)、H d (t) respectively representing the heat generated by a gas boiler, the heat generated by a lithium bromide machine, the heat generated by a heat pump unit, the heat release of a heat storage tank, the heat charge of the heat storage tank and the heat load; s in (9) j Representing the running state of a certain unit equipment j, the subscript of which is different from the corresponding unit equipment, e.g. S bo 、S li-h 、S hp 、S sk-d 、S sk-c The method respectively represents the operation state of the gas boiler, the operation state of the lithium bromide machine, the operation state of the heat pump, the heat release state of the heat storage tank and the heat charging state of the heat storage tank.
A. Gas boiler
H bo (t)＝η g-bo ·c 2 (t) (10)
H bo (t)≤H bo-rated (t) (11)
Formula (10) is a heat generation expression of the gas boiler, wherein eta g-bo C is the efficiency of gas-fired boiler gas-to-heat conversion 2 (t) represents natural gas consumed by a gas boiler; formula (11) is the heat capacity constraint of the gas boiler, H bo-rated And (t) represents the rated power of the gas boiler.
B lithium bromide machine
H li-h (t)＝η g-hc ·η li-h ·c 1 (t) (12)
H li-h (t)≤H li-rated (t) (13)
In order to optimize the energy cascade utilization, in the scheme, only a waste heat utilization mode is considered for the lithium bromide unit, namely, the waste heat of the internal combustion engine is converted into cold or hot, and a direct-combustion natural gas mode, namely, the situation of directly combusting natural gas is not considered. Formula (12) is a heat generation expression of lithium bromide machine, wherein eta g-hc Represents the efficiency, eta of the air-to-waste heat generated by the internal combustion generating set during power generation li-h The efficiency of the lithium bromide machine in converting the waste heat of the internal combustion generator into heat is shown. c 1 And (t) represents natural gas consumed by the internal combustion generator for power generation. Formula (13) is the heat capacity constraint of the lithium bromide machine, H li-rated And (t) represents the rated power of lithium bromide.
C. Heat pump
H hp (t)＝η g-h ·c 3 (t) (14)
H hp (t)≤H hp-rated (t) (15)
Formula (14) is a heat generation expression of the heat pump unit, wherein eta g-h C is the efficiency of heat transfer of the heat pump unit 3 And (t) represents natural gas consumed by the heat pump. Equation (15) is a heat capacity constraint of the heat pump.
D. Heat storage tank
E sk-min ≤E sk (t)≤E sk-max (17)
S sk-c +S sk-d ≤1 (18)
Equation (16) represents the heat of the heat storage tank, where η c Represents the heat-filling efficiency eta f Indicating the heat release efficiency. Equation (17) is used for approximating the capacity of the heat storage tank, S sk-min 、S sk-max Representing the maximum and minimum values of the heat storage tank, respectively. Formula (18) represents that the heat storage tank can only release heat or charge heat per unit time, wherein S sk-d (t)、S sk-c (t) is a 2-ary variable.
3) Cold balance constraint
S cen C cen (t)+S li-c C li-c (t)＝C d (t) (19)
C in formula (19) cen (t)、C li-c (t)、C d (t) respectively representing the cold produced by a centrifugal machine and the cold produced by a lithium bromide machine and the size of the cold load; s in (20) j The running state of a certain unit device j is represented, and the subscript of the running state is different from the corresponding unit device.
A. Centrifugal machine
C cen (t)＝η e-c E R-cen (t) (21)
C cen (t)≤C cen-rated (t) (22)
Formula (21) is a cold-producing expression of the centrifuge, wherein eta e-c Indicating the refrigerating efficiency of the centrifugal machine, E R-cen And (t) represents the amount of electricity consumed by the centrifuge. Formula (22) restricts the capacity of the centrifuge, C cen-rated And (t) represents the rated heat capacity of the centrifuge.
B. Lithium bromide machine
C li-c (t)＝η li-c ·c 1 (t) (23)
C li-c (t)≤C li-rated (t) (24)
S li-c +S li-h ≤1 (25)
Formula (23) is the cold-producing expression of lithium bromide machine, eta li-c Indicating the efficiency with which the lithium bromide machine will refrigerate. Formula (24) is the cold capacity constraint of the lithium bromide machine, C li-rated And (t) represents the rated heat capacity of the lithium bromide machine. The formula (25) shows that the lithium bromide machine can only generate heat or cool, S li-c (t)、S li-h (t) is a 2-ary variable, i.e. when S li-c When (t) =1, the lithium bromide machine is producing cold, when S li-h When (t) =1, the lithium bromide machine generates heat.
Step 3: initializing data Z collected by a measurement terminal of an intelligent park by Latin hypercube sampling t ＝<L 1 t ,L 2 t ,…,L N t >The method comprises the steps of carrying out a first treatment on the surface of the The data preprocessing mode is used for improving the quality of data.
Step 4: reuse scheduling scheme function<h 1 t ,h 2 t ,…,h N t >Evaluating;
step 5: creation of a small sample database S from initial data t ＝{(L 1 t ,h 1 t ),(L 2 t ,h 1 t ),…,(L N t ,h N t ) And building a proxy model based on the above.
Step 6: and performing an optimizing process by taking a Kriging and RBF double-agent auxiliary search optimizing algorithm as an optimizer. And carrying out global and local search optimization under the condition that the termination condition of the algorithm is not met. The whole idea of the step is to divide the original data set into a plurality of subsets through random sampling and echelon sampling, and then to train a plurality of models on the same sample set through concentrated learning so as to enhance the prediction precision of the models. Wherein a schematic of the echelon sampling is shown in fig. 2 and the proxy model ensemble learning step for enhancing model prediction accuracy is shown in fig. 3.
The global optimizing step comprises the following steps: 1) According to the global model management strategy, in the small sample database S t M subsets; 2) Training M subsets using a Kriging proxy model; 3) An optimization algorithm is adopted to find the optimal value of each Kriging agent model; 4) Average value O of the optimal values of M subsets av The method comprises the steps of carrying out a first treatment on the surface of the 5) Judging if O av If the value is the current optimal value, repeating the global optimizing step, otherwise, entering local optimizing.
The local optimizing step is as follows: 1) According to local model management strategy, in small sample database S t N subsets; 2) Training the N subsets using an RBF proxy model; 3) An optimal value of each RBF agent model is found by adopting an optimization algorithm; 4) Find the weighted sum O of the optimal values of the N subsets ws The method comprises the steps of carrying out a first treatment on the surface of the 5) Judging if O ws If the value is the current optimal value, repeating the local optimizing step, otherwise, entering global optimizing.
The global model management strategy comprises the following management steps: 1) According to dimension W of decision space dim Determining the number M= ((W) of subsets to be divided dim +1)(W dim +2)/2100); 2) To maintain a balance between data set diversity and data set size, a small sample database S is run with a probability of 0.5 t Random sampling is performed for each point in (a).
The local model management strategy comprises the following management steps: 1) In a small sample database S with N data points t In the method, a echelon sampling method is adopted, and N-1 data are sampled each time, namely S j ＝{S t \L i J=1, 2, …, N }, where L i Is the jth data point; 2) Training N subsets using RBF proxy model and taking a weighted sum O of N models ws The expression is formula (26). Wherein Y is p Is the RBF predicted value finally obtained, Y pj Is the predicted value of the jth RBF model, lambda j Is the weight of the jth RBF model, defined as equation (27), where e j Is the root mean square error of the jth RBF model.
Y p (x)＝λ 1 Y p1 +λ 2 Y p2 +...+λ N Y pN (26)
Step 7: and evaluating the obtained current optimal solution by using a scheduling scheme function, and adding the obtained new sample points into a sample database.
Step 8: and when the algorithm does not reach the preset termination condition, continuing iteration, and when the algorithm reaches the preset termination condition, terminating the algorithm, and outputting an optimal solution of the intelligent park multi-energy collaborative energy optimization problem to obtain an energy optimization scheme of the universal energy station and park equipment.
Aiming at an intelligent park powered by a universal station, the invention solves the scheduling problem of multi-objective collaborative optimization of cold, hot and electric multi-energy comprehensive demand response under the condition of insufficient data sample collection quantity. The adopted double-agent assisted search algorithm is a new deep learning algorithm improved on the basis of single-agent assisted search. The algorithm adopts the thought of combining local search and global search, so that the algorithm can fully search near the current optimal point, and the balance of convergence quality and convergence efficiency is obtained.
The foregoing examples illustrate only a few embodiments of the invention and are described in detail herein without thereby limiting the scope of the invention. It should be noted that it will be apparent to those skilled in the art that several variations and modifications can be made without departing from the spirit of the invention, which are all within the scope of the invention. Accordingly, the scope of protection of the present invention is to be determined by the appended claims.
Claims (9)
1. The double-agent assisted search energy optimization method for small sample data of a park is characterized by comprising the following steps of:
step 1: setting three optimization targets of maximum cold, heat and electricity sales benefits, maximum renewable energy utilization rate and minimum electricity consumption peak Gu Chazhi of the intelligent park energy station, and unifying the three targets into a single target through target weighting;
step 2: determining constraint conditions of an operation equation and an inequality according to energy production equipment of a smart park energy station and park load equipment;
step 3: initializing data Z collected by a measurement terminal of an intelligent park by Latin hypercube sampling t ＝<L 1 t ,L 2 t ,…,L N t >Wherein L is j t Representing data Z t Is the j-th element of (2);
step 4: reuse scheduling scheme function<h 1 t ,h 2 t ,…,h N t >Performing an evaluation, wherein h j t A j-th element representing a scheduling scheme function;
step 5: creation of a small sample database S from initial data t ＝{(L 1 t ,h 1 t ),(L 2 t ,h 1 t ),…,(L N t ,h N t ) Establishing a proxy model on the basis;
step 6: performing a optimizing process by taking a Kriging and RBF double-agent auxiliary search optimizing algorithm as an optimizer, and performing global and local search optimizing under the condition that the termination condition of the algorithm is not achieved;
step 7: the obtained current optimal solution is evaluated by a scheduling scheme function, and the obtained new sample points are added into a sample database;
step 8: when the algorithm does not reach the preset termination condition, iteration is continued, and when the algorithm reaches the preset termination condition, the algorithm is terminated, and an optimal solution of the intelligent park multi-energy collaborative energy optimization problem is output, so that an energy optimization scheme of the universal energy station and park equipment is obtained;
the related expression in step 1 is as follows:
the multi-objective dynamic nonlinear optimization function based on the weight function unifies three objectives into a simplified expression of single-objective nonlinear dynamic optimization through objective weighting, wherein A, B, C is respectively different objective functions, alpha, beta and delta are respectively weighting coefficients of different objective functions, and the weighting coefficients are selected according to different application scenes;
minF＝αA+βB+δC (1)
different objective functions A, B, C respectively correspond to maximum benefits of cold, heat and electricity sold by the universal station, maximum renewable energy utilization rate and minimum electricity consumption peak Gu Chazhi, and the objective functions are expressed as (2) - (4);
A＝max(f 1 +f 2 +f 3 +f g -f c ) (2)
wherein f 1 For electricity utilization benefits, f 2 To use heat gain, f 3 To use the cold returns, f g F, for surplus electricity Internet surfing benefits c Is the running cost;
wherein N is re In order to be able to use the renewable energy source,is a single renewable energy source j;
wherein L, H is the highest and lowest limit values of electric power, T is the total number of electric power measurement time intervals, P at time i i,j Representing the electrical power.
2. The method of optimizing energy for a dual agent assisted search for small sample data on a campus of claim 1, wherein the energy generating facility of the flood station in step 2 comprises a gas boiler, lithium bromide machine, heat pump and heat storage tank, and the campus load facility comprises a transferable load, a charging load and an energy storage load.
3. The method of optimizing energy for a dual agent assisted search for small sample data on a campus of claim 2, wherein in step 2, the constraint of electrical balance:
E G (t)＝E R (t)+E d (t)+E s (t) (5)
E G (t)＝η g-e ·c 1 (t) (6)
E G (t)≤E G-rated (t) (7)
e in formula (5) G (t) represents the power generation amount of the internal combustion engine, E R (t) represents the electricity consumption of the station, E d (t) represents the load electricity consumption, E s (t) represents the remaining amount of power on the network; eta in the formula (6) g-e Representing the conversion of natural gas into electricity c 1 (t) represents natural gas consumed for power generation by the internal combustion generator; formula (7) is the capacity constraint of the internal combustion engine, E G-rated And (t) represents the rated power of the internal combustion generator.
4. The method of double agent assisted search energy optimization for small sample data on a campus of claim 2, wherein the thermal equilibrium constraint in step 2:
S bo H bo (t)+S li-h H li-h (t)+S hp H hp (t)+S sk-d H sk-d (t)＝H d (t)+S sk-c H sk-c (t)(8)
h in (8) bo (t)、H li-h (t)、H hp (t)、H sk-d (t)、H sk-c (t)、H d (t) respectively representing the heat generated by a gas boiler, the heat generated by a lithium bromide machine, the heat generated by a heat pump unit, the heat release of a heat storage tank, the heat charge of the heat storage tank and the heat load; s in (9) j Representing the running state of a certain unit equipment j, the subscript of which is different from the corresponding unit equipment, e.g. S bo 、S li-h 、S hp 、S sk-d 、S sk-c Respectively representing the operation state of the gas boiler, the operation state of the lithium bromide machine, the operation state of the heat pump, the heat release state of the heat storage tank and the heat charging state of the heat storage tank;
A. gas boiler
H bo (t)＝η g-bo ·c 2 (t) (10)
H bo (t)≤H bo-rated (t) (11)
Formula (10) is a heat generation expression of the gas boiler, wherein eta g-bo C is the efficiency of gas-fired boiler gas-to-heat conversion 2 (t) represents natural gas consumed by a gas boiler; formula (11) is the heat capacity constraint of the gas boiler, H bo-rated (t) represents the rated power of the gas boiler;
B. lithium bromide machine
H li-h (t)＝η g-hc ·η li-h ·c 1 (t) (12)
H li-h (t)≤H li-rated (t) (13)
In order to optimize the energy cascade utilization, the lithium bromide unit only considers the waste heat utilization mode, namely the condition of converting the waste heat of the internal combustion engine into cold or heat, and does not consider the direct-combustion natural gas mode, namely the condition of directly combusting natural gas; formula (12) is a heat generation expression of lithium bromide machine, wherein eta g-hc Represents the efficiency, eta of the air-to-waste heat generated by the internal combustion generating set during power generation li-h The efficiency of the lithium bromide machine for converting the waste heat of the internal combustion generator into heat is represented; c 1 (t) represents natural gas consumed for power generation by the internal combustion generator; formula (13) is the heat capacity constraint of the lithium bromide machine, H li-rated (t) represents the rated power of lithium bromide;
C. heat pump
H hp (t)＝η g-h ·c 3 (t) (14)
H hp (t)≤H hp-rated (t) (15)
Formula (14) is a heat generation expression of the heat pump unit, wherein eta g-h C is the efficiency of heat transfer of the heat pump unit 3 (t) represents natural gas consumed by the heat pump; formula (15) is a heat capacity constraint of the heat pump, H hp-rated (t) represents the rated power of the heat pump;
D. heat storage tank
E sk-min ≤E sk (t)≤E sk-max (17)
S sk-c +S sk-d ≤1 (18)
Equation (16) represents the heat of the heat storage tank, where η c Represents the heat-filling efficiency eta f Indicating the heat release efficiency; formula (17) restricts the capacity of the heat storage tank, S sk-min 、S sk-max Respectively representing the maximum value and the minimum value of the heat storage tank; formula (18) represents that the heat storage tank can only release heat or charge heat per unit time, wherein S sk-d (t)、S sk-c (t) is a 2-ary variable.
5. The method of optimizing energy for a dual agent assisted search for small sample data on a campus of claim 2, wherein the cold balance constraint in step 2:
S cen C cen (t)+S li-c C li-c (t)＝C d (t) (19)
c in formula (19) cen (t)、C li-c (t)、C d (t) respectively representing the cold produced by a centrifugal machine and the cold produced by a lithium bromide machine and the size of the cold load; s in (20) j The operation state of a certain unit device j is represented, and the subscript of the unit device j is different from the corresponding unit device;
A. centrifugal machine
C cen (t)＝η e-c E R-cen (t) (21)
C cen (t)≤C cen-rated (t) (22)
Formula (21) is a cold-producing expression of the centrifuge, wherein eta e-c Indicating the refrigerating efficiency of the centrifugal machine, E R-cen (t) represents the amount of electricity consumed by the centrifuge; formula (22) restricts the capacity of the centrifuge, C cen-rated (t) represents the rated heat capacity of the centrifuge;
B. lithium bromide machine
C li-c (t)＝η li-c ·c 1 (t) (23)
C li-c (t)≤C li-rated (t) (24)
S li-c +S li-h ≤1 (25)
Formula (23) is the cold-producing expression of lithium bromide machine, eta li-c Indicating the efficiency of the lithium bromide machine to refrigerate; formula (24) is the cold capacity constraint of the lithium bromide machine, C li-rated (t) represents the rated heat capacity of the lithium bromide machine; the formula (25) shows that the lithium bromide machine can only generate heat or cool, S li-c (t)、S li-h (t) is a 2-ary variable, i.e. when S li-c When (t) =1, the lithium bromide machine is producing cold, when S li-h When (t) =1, the lithium bromide machine generates heat.
6. The method for optimizing energy for a dual agent assisted search for small sample data on a campus of claim 1, wherein the global optimizing step in step 6 is: 1) According to the global model management strategy, in the small sample database S t M subsets; 2) Training M subsets using a Kriging proxy model; 3) An optimization algorithm is adopted to find the optimal value of each Kriging agent model; 4) Average value O of the optimal values of M subsets av The method comprises the steps of carrying out a first treatment on the surface of the 5) Judging if O av If the value is the current optimal value, repeating the global optimizing step, otherwise, entering local optimizing.
7. The method for optimizing energy for a dual agent assisted search for small sample data on a campus of claim 1, wherein the step of locally optimizing in step 6 is: 1) According to local model management strategy, in small sample database S t N subsets; 2) Training the N subsets using an RBF proxy model; 3) An optimal value of each RBF agent model is found by adopting an optimization algorithm; 4) Find the weighted sum O of the optimal values of the N subsets ws The method comprises the steps of carrying out a first treatment on the surface of the 5) Judging if O ws If the value is the current optimal value, repeating the local optimizing step, otherwise, entering global optimizing.
8. According to claimThe method for optimizing energy for double-agent assisted search for small sample data in a campus of claim 6, wherein the global model management policy comprises the following management steps: 1) According to dimension W of decision space dim Determining the number of subsets m= ((W) dim +1)(W dim +2)/2100); 2) To maintain a balance between data set diversity and data set size, a small sample database S is run with a probability of 0.5 t Random sampling is performed for each point in (a).
9. The method for optimizing energy for a dual agent assisted search for small sample data on a campus of claim 7, wherein the local model management policy is managed as follows: 1) In a small sample database S with N data points t In the method, a echelon sampling method is adopted, and N-1 data are sampled each time, namely S j ＝{S t \L i J=1, 2, …, N }, where L i Is the jth data point; 2) Training N subsets using RBF proxy model and taking a weighted sum O of N models ws The expression of the compound is shown as a formula (26); wherein Y is p Is the RBF predicted value finally obtained, Y pj Is the predicted value of the jth RBF model, lambda j Is the weight of the jth RBF model, defined as equation (27), where e j Is the root mean square error of the jth RBF model;
Y p (x)＝λ 1 Y p1 +λ 2 Y p2 +...+λ N Y pN (26)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011279183.7A CN112465198B (en) | 2020-11-16 | 2020-11-16 | Double-agent assisted search energy optimization method for small sample data of park |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011279183.7A CN112465198B (en) | 2020-11-16 | 2020-11-16 | Double-agent assisted search energy optimization method for small sample data of park |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112465198A CN112465198A (en) | 2021-03-09 |
| CN112465198B true CN112465198B (en) | 2023-07-28 |
Family
ID=74836912
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011279183.7A Active CN112465198B (en) | 2020-11-16 | 2020-11-16 | Double-agent assisted search energy optimization method for small sample data of park |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112465198B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111241751A (en) * | 2020-01-15 | 2020-06-05 | 深圳大学 | Wing profile optimization method and device based on agent assisted evolution algorithm |
| CN111914382A (en) * | 2019-05-07 | 2020-11-10 | 宁波大学 | Constraint evolution optimization method of atmospheric and vacuum pressure device based on proxy model |
-
2020
- 2020-11-16 CN CN202011279183.7A patent/CN112465198B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111914382A (en) * | 2019-05-07 | 2020-11-10 | 宁波大学 | Constraint evolution optimization method of atmospheric and vacuum pressure device based on proxy model |
| CN111241751A (en) * | 2020-01-15 | 2020-06-05 | 深圳大学 | Wing profile optimization method and device based on agent assisted evolution algorithm |
Non-Patent Citations (3)
| Title |
|---|
| 基于动态径向基函数代理模型的优化策略;彭磊;刘莉;龙腾;;机械工程学报(07);全文 * |
| 基于多级代理模型的优化算法;王琦;丁运亮;陈昊;;南京航空航天大学学报(04);全文 * |
| 模型替代方法在优化设计中的应用;刘万刚;宋述芳;樊维超;吕震宙;;中国工程机械学报(02);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112465198A (en) | 2021-03-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ge et al. | A hybrid model for short-term PV output forecasting based on PCA-GWO-GRNN | |
| Yan et al. | Stochastic multi-scenario optimization for a hybrid combined cooling, heating and power system considering multi-criteria | |
| CN103138256B (en) | A kind of new energy electric power reduction panorama analytic system and method | |
| CN106874630A (en) | Based on the regional power grid new energy development potential evaluation method that electricity is dissolved | |
| CN112749462A (en) | Method, system and equipment for optimizing current operation parameters of comprehensive energy system | |
| CN107666155B (en) | Markov model-based random stability analysis method for multi-energy complementary system | |
| CN112994115B (en) | New energy capacity configuration method based on WGAN scene simulation and time sequence production simulation | |
| CN110739726B (en) | Multi-type power supply capacity long-term planning method considering offshore wind power access | |
| CN112952807B (en) | Multi-objective optimization scheduling method considering wind power uncertainty and demand response | |
| CN110009141B (en) | Climbing event prediction method and system based on SDAE feature extraction and SVM classification model | |
| CN111598289A (en) | Distributed optimization method of integrated energy system considering LSTM photovoltaic output prediction | |
| CN113469412B (en) | Real-time operation strategy optimization method and system for comprehensive energy system | |
| CN109670694B (en) | Load prediction method for multi-energy supply system | |
| CN111404204A (en) | Optimized scheduling method for power system including wind power plant by using improved wolf pack algorithm | |
| CN112183841A (en) | Optimized dispatching method of micro-grid containing electric automobile based on simulated annealing algorithm | |
| CN117993611A (en) | Flexible heat source new energy consumption capability assessment method based on scene time sequence | |
| CN114037209A (en) | Comprehensive benefit analysis method and device for distributed photovoltaic access direct-current power distribution system | |
| CN112465198B (en) | Double-agent assisted search energy optimization method for small sample data of park | |
| CN113158547A (en) | Regional comprehensive energy system optimal configuration method considering economy and reliability | |
| CN109586309B (en) | Power distribution network reactive power optimization method based on big data free entropy theory and scene matching | |
| CN116128154A (en) | Energy optimal configuration method and device for agricultural park comprehensive energy system | |
| CN115912421A (en) | Power distribution network energy storage site selection constant-volume multi-objective optimization method and system | |
| CN114997460A (en) | Regional micro-energy network operation optimization method considering maximum consumption of renewable energy | |
| CN113361805A (en) | Power distribution network planning method and system | |
| CN117094613B (en) | Model construction method and device applied to comprehensive energy system |
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 |