CN108895532B - Area heating energy-saving control method based on random distribution control algorithm - Google Patents

Area heating energy-saving control method based on random distribution control algorithm Download PDF

Info

Publication number
CN108895532B
CN108895532B CN201810365407.2A CN201810365407A CN108895532B CN 108895532 B CN108895532 B CN 108895532B CN 201810365407 A CN201810365407 A CN 201810365407A CN 108895532 B CN108895532 B CN 108895532B
Authority
CN
China
Prior art keywords
temperature
function
probability distribution
weight
basis function
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
Application number
CN201810365407.2A
Other languages
Chinese (zh)
Other versions
CN108895532A (en
Inventor
王丽艳
韩东升
任密峰
乔铁柱
阎高伟
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Taiyuan University of Technology
Original Assignee
Taiyuan University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Taiyuan University of Technology filed Critical Taiyuan University of Technology
Priority to CN201810365407.2A priority Critical patent/CN108895532B/en
Publication of CN108895532A publication Critical patent/CN108895532A/en
Application granted granted Critical
Publication of CN108895532B publication Critical patent/CN108895532B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24D—DOMESTIC- OR SPACE-HEATING SYSTEMS, e.g. CENTRAL HEATING SYSTEMS; DOMESTIC HOT-WATER SUPPLY SYSTEMS; ELEMENTS OR COMPONENTS THEREFOR
    • F24D19/00—Details
    • F24D19/10—Arrangement or mounting of control or safety devices
    • F24D19/1006—Arrangement or mounting of control or safety devices for water heating systems
    • F24D19/1009—Arrangement or mounting of control or safety devices for water heating systems for central heating
    • F24D19/1048—Counting of energy consumption
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24D—DOMESTIC- OR SPACE-HEATING SYSTEMS, e.g. CENTRAL HEATING SYSTEMS; DOMESTIC HOT-WATER SUPPLY SYSTEMS; ELEMENTS OR COMPONENTS THEREFOR
    • F24D2220/00—Components of central heating installations excluding heat sources
    • F24D2220/04—Sensors
    • F24D2220/042—Temperature sensors

Abstract

The invention belongs to the technical field of heat supply control, and particularly relates to a random-based heat supply control methodA regional heat supply energy-saving control method of a machine distribution control algorithm. The heat supply control method comprises the following steps: step S100: acquiring user temperature data of a plurality of points in an area; step S200: estimating the temperature probability distribution of a heat supply area; step S300: establishing a basic function representation model of a temperature probability distribution density function; step S400: establishing input variables and leadsn-a non-linear prediction model between 1 weight vector to predict the temperature probability distribution at the next moment; step S500: carrying out feedback correction on the predicted output information in combination with the temperature probability distribution; step S600: and calculating the output of the randomly distributed control quantity, and outputting and controlling a frequency converter to drive a water pump to provide heat for the region. The core of the control method is to process the probability information of temperature distribution, determine the control quantity by using the random distribution control theory, improve the balance degree of district heating to the maximum extent and reduce the energy consumption.

Description

Area heating energy-saving control method based on random distribution control algorithm
Technical Field
The invention belongs to the technical field of heat supply control, and particularly relates to a district heat supply energy-saving control method based on a random distribution control algorithm.
Background
At present, most of cities or regions adopt a centralized heating mode, but the system is a system with random interference factors due to different building structures in the regions, different distances from heat sources, larger resistance difference of a heating pipe network and influence of environment and other factors, and various interferences have non-Gaussian characteristics. The temperature balance of a heat supply area is difficult to control by adopting a conventional control method, the temperature of part of the area is high frequently, the temperature of some areas is too low frequently, and if the temperature and the flow of a heat source are simply increased, energy waste can be caused. There is therefore a need for a control method that can control the temperature distribution relatively evenly throughout a heating area, or according to a set temperature distribution.
The existing control method generally adopts PID and Smith-PID to control and regulate the temperature of a certain point, which causes large fluctuation of the heat supply network.
Disclosure of Invention
The invention provides a district heating energy-saving control method based on a random distribution control algorithm, aiming at realizing balanced heating of a district or heating according to given temperature distribution information.
The invention adopts the following technical scheme: a district heating energy-saving control method based on a random distribution control algorithm comprises the following steps:
s100: the method comprises the steps of obtaining Q user temperature data, water supply and return data of a heat supply pipe network and output control information at different positions in an area, and storing the obtained temperature information as a historical data set.
The historical data set includes Q pieces of thermal user temperature data T ═ T in a time period in the heat supply area1,T2,…TQ]And the temperature T of the primary water supply at the corresponding time pointG1Primary net return water temperature TH1Water supply temperature T of secondary networkG2Secondary net return water temperature TH2And the ambient temperature TE。
S200: and estimating the temperature probability distribution of the heat supply area according to the acquired information, comprising the following steps.
S201: selecting different positions (x) in the heat supply area from the data obtained in the step S100j,yj) J is 1,2, …, Q is Q, and M groups of different temperature data T in one day are provided for hot usersjAnd corresponding heating system control output ui(k) I is 1,2, … M, and are combined to form a temperature value matrix
Figure DEST_PATH_IMAGE002
S202: using a temperature matrix
Figure DEST_PATH_IMAGE004
Constructing temperature fields at different moments according to different time points, and performing mathematical approximation processing on the whole temperature field data to obtain a probability density function gamma (T, u) of the temperature Ti(k))：
Figure DEST_PATH_IMAGE006
Wherein N (T) represents the number of temperature values less than T in the temperature field of the heat supply area, Ti,maxAnd Ti,minRepresenting the upper and lower temperature bounds in the temperature field at time i.
S300: the method for representing the known temperature probability distribution density function by using the basis function representation model comprises the following steps.
S301: selecting a basis function, adopting a Gaussian RBF network as the basis function, and obtaining the following expression:
Figure DEST_PATH_IMAGE008
wherein T is the collected temperature information; mu.si,σiThe center value and width of the function for the ith network node.
S302: determining the weighted expression of the basis function, and expressing the probability density function of the temperature distribution at the moment k by using the form of the weighted sum of the basis functions according to the RBF network approximation principle, wherein the expression is as follows:
γ(T,u(k))＝C(T)V(k)+Bn(T)wn(k)+e0(T,k) (3)
wherein C (T) ═ B1(T),B2(T),…,Bn-1(T)]，V(k)＝[w1(k),w2(k),…,wn-1(k)]T，wn(k) Is the weight corresponding to the nth basis function, e0(T, k) is the error of the approximation to the region temperature probability distribution density function.
S303: determining the weight of each basis function, the nth weight ωn(k) The nonlinear function h (v (k)) of the weight vector v (k) can be expressed as:
Figure DEST_PATH_IMAGE010
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE012
Figure DEST_PATH_IMAGE014
ignoring the approximation error, combining equations (3) and (4) yields:
Figure DEST_PATH_IMAGE016
both sides left-hand multiplication by [ CT(T) Bn(T)]TAnd in the interval [ TminTmax]Is integrated, when the matrix is
Figure DEST_PATH_IMAGE018
When not singularity, can be transformed to obtain:
Figure DEST_PATH_IMAGE020
the weight V (k) of each basis function of the temperature probability distribution density function is obtained by the above equation (6).
In addition, the selection of the corresponding basis function is not limited to the above method, and another basis function is provided below to represent the known temperature probability distribution density function, and the specific steps are as follows:
s301: the selection of the basis function, because the B-spline model can estimate any distribution curve and can represent the random distribution of the variable in the form of discrete weight, the B-spline model is used as the basis function for representing the temperature probability distribution, and the second-order B-spline basis function is as follows:
Figure DEST_PATH_IMAGE022
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE024
h (x) is the Heviside function and s ═ j-3.
S302: determining a weighted expression of the basis function, and describing the probability distribution of the collected random temperature information by using a B-spline-based estimation model, wherein the expression is as follows:
γ(T,u(k))＝C(T)V(k)+Bn(T)wn(k)+e0(T,k) (8)
wherein C (T) ═ B1(T),B2(T),…,Bn-1(T)]，V(k)＝[w1(k),w2(k),…,wn-1(k)]T，wn(k) For the nth basis function corresponding to the weight, e0(T, k) is the error of the approximation to the region temperature probability distribution density function.
S303: determining the weight of the basis function, and the formula Error! The corresponding weight calculation formula of Reference source not found is:
Figure DEST_PATH_IMAGE026
wherein, ci＝(xi-xi-3) N is the total number of data collected, Bi(xl) Is a basis function.
Using the above formula, the different basis functions of each basis function can be determined, i.e., its basis function vector V (k) can be determined.
The selection of the basis functions is not limited to the above-mentioned method, and various choices of the basis functions are available, and the present invention is not repeated herein, and the scope of the claims of the present invention should be covered without departing from the spirit of the present invention.
S400: establishing a nonlinear prediction model between the input variable and the first n-1 weight vectors, and predicting the temperature probability distribution at the next moment.
S401: selecting input variables, and adding the primary network water supply temperature T obtained in the step S100G1Primary net return water temperature TH1Water supply temperature T of secondary networkG2Secondary net return water temperature TH2Ambient temperature TECurrent control output u (k) ═ u1(k),u2(k),…,um(k)]And the first n-1 weight vectors V (k) of the temperature probability distribution at the current moment are merged as input and are recorded as: x ═ TG1,TH1,TG2,TH2,TE,u(k),V(k)]L×(4+n+m)。
S402: selecting a prediction model, selecting a random weight neural network, and expressing the network model as follows:
Figure DEST_PATH_IMAGE028
wherein, ω isj＝[ωj1,ωj2,…,ωjm]TConnecting the input weights of the jth hidden cell for m input nodes, βj＝[βj1,βj2,…,βj(n-1)]TConnecting the output weights of the output nodes for the jth hidden layer, bjIs the bias of the jth implicit cell.
S403: training the model, randomly giving a group of input layer weights and biases, and training the model by using L groups of collected historical data, wherein an objective function is as follows:
Figure DEST_PATH_IMAGE030
s404: obtaining optimal output weight value by solving generalized inverse of H matrix
Figure DEST_PATH_IMAGE032
The formula is as follows:
Figure DEST_PATH_IMAGE034
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE036
Figure DEST_PATH_IMAGE038
s405: establishing a relation between the basis function weight and the input variable:
Figure DEST_PATH_IMAGE040
s406: and predicting the temperature probability distribution density function at the next moment:
γm(T,k+1)＝C(T)Vm(k+1)+Bn(T)wn(k+1) (14)。
in addition, the selection of the prediction model is not limited to the above method, and another model is selected for description, and the specific steps are as follows:
s402: selecting a prediction model, considering the correlation among factors influencing regional heat supply, and selecting an LS _ SVM nonlinear model, wherein the model is expressed as:
V(k+1)＝WTΦ(X(k))+Β (15)
wherein W is a diagonal weight matrix, and V (k +1) ═ V1(k+1),…,vn(k+1)]TAnd (3) representing the weight value of prediction at the moment of k +1, B is a deviation vector, and the nonlinear mapping phi (X (k)) converts input data into a high-dimensional space.
S403: training the model, and training the model by using the collected L groups of historical data, wherein the target function is as follows:
Figure DEST_PATH_IMAGE042
wherein, B ═ B1… bn]T，
Figure DEST_PATH_IMAGE044
Respectively representing a weight and a deviation, EiIs the fitting error.
S404: optimizing the optimization problem by constructing a Lagrange function to obtain the weight of the model:
Figure DEST_PATH_IMAGE046
s405: establishing a relation between the basis function weight and the input variable:
V(k+1)＝WTΦ(X(k))+Β (18)。
s406: and predicting the temperature probability distribution density function at the next moment:
γm(T,k+1)＝C(T)Vm(k+1)+Bn(T)wn(k+1) (19)。
the choice of the prediction model is not limited to the above method, and the present invention is not repeated herein, and the scope of the claims of the present invention should be covered without departing from the spirit of the present invention.
S500: the feedback correction is performed using the prediction output information in combination with the temperature probability distribution, including the following steps.
S501: determining the error between the actual output temperature probability distribution density at the kth moment and the predicted output temperature probability distribution density as follows:
e(T,k)＝γ(T,k)-γm(T,k) (20)。
s502: and performing feedback correction on the predicted output temperature probability distribution density by using the error, wherein the corrected predicted output is as follows:
γJ(T,k+1)＝γm(T,k+1)+ηe(T,k) (21)；
wherein η is a correction coefficient, and the value range is 0 & lt η & lt 1.
S600: the heating output control is adjusted in real time according to the temperature correction, the random distribution control quantity output of the heating system is calculated, the following method is adopted,
s601: calculating the difference between the expected temperature and the predicted temperature at the moment k +1 as:
eJ(T,k+1)＝γg(T,k+1)-γJ(T,k+1) (22)。
s602: constructing a performance indicator function that minimizes temperature variation:
J1(u(k))＝∫(γ(T,u(k))-γg(T,u(k))2dT (23)。
s603: considering the energy-saving effect of the whole heating system, the performance index function minimizing the energy is as follows:
J2(u(k))＝u(k)TRu(k) (24)。
s604: and (3) optimizing and solving, namely solving the control output which simultaneously satisfies the performance indexes of the equations (23) and (24) by using an optimization algorithm, namely:
argminJ(u(k))＝(J1(u(k)),J2(u(k))) (25)。
compared with the prior art, the invention utilizes the heat supply control method of random distribution control theory to make control decision, so as to realize balanced heat supply to the area or heat supply according to the given temperature distribution information. The feedback quantity adopted by the invention is the temperature distribution information of the whole area, and the temperature distribution of the whole heat supply area is controlled by a random distribution control algorithm, so that the waste of energy is reduced.
Drawings
FIG. 1 is a block diagram of the system;
FIG. 2 System embodiment;
FIG. 3 illustrates a temperature data collection and storage process;
FIG. 4 illustrates a temperature field at a time within a heating area;
figure 5 temperature probability distribution for different control outputs of the heating system.
Detailed Description
The method and the system for controlling the energy conservation of the district heating based on the random distribution control algorithm can be realized by adopting a modeling process and a real-time process in the implementation mode:
the modeling process mainly comprises the following steps:
s100: the method comprises the steps of obtaining a plurality of user temperature data of different positions in an area, heat supply and return water data of a heat supply pipe network and output control information, and storing the obtained temperature information as a historical data set. The historical data set includes Q pieces of user temperature data T ═ T [ T ] for a period of time (24 hours) within the heat supply region1,T2,…TQ]And the temperature T of the primary water supply at the corresponding time pointG1Primary net return water temperature TH1Water supply temperature T of secondary networkG2Secondary net return water temperature TH2And the ambient temperature TE。
Establishing the historical data set can be realized by the following steps:
historical data needs to collect information of indoor and outdoor temperatures, supply and return water temperatures of a primary network and supply and return water temperatures of a secondary network, a plurality of heat users at different positions away from a heat supply source are selected in a heat supply area, a plurality of representative time periods within 24 hours are selected, and indoor and outdoor temperature values of the heat users at the same time interval in a certain time period are collected by using a temperature sensor and stored. Meanwhile, a detection device is installed at a water supply and return interface of the heat exchange between the heat source and the heat user for two times, and corresponding water supply and return temperatures are collected and stored, as shown in the attached drawing 3.
S200: estimating the temperature probability distribution of the heat supply area according to the acquired information, wherein the estimation of the temperature probability distribution of the heat supply area is realized by adopting the following mode:
selecting different positions (x) in the heat supply areaj,yj) J-1, 2, … Q total Q different temperature data T of M groups in one day for hot usersjAnd corresponding heating system control output ui(k) I is 1,2, … M, and are combined to form a temperature value matrix
Figure DEST_PATH_IMAGE048
Using a temperature matrix
Figure DEST_PATH_IMAGE050
The temperature field was constructed at different time points as shown in fig. 4. The data of the whole temperature field is processed by mathematical approximation to obtain a probability density function gamma (T, u) of the temperature Ti(k))：
Figure DEST_PATH_IMAGE052
Wherein N (T) represents the number of temperature values less than T in the temperature field of the heat supply area, Ti,maxAnd Ti,minRepresenting the upper and lower temperature bounds in the temperature field at time i. The temperature field can be converted into a probability distribution of different temperatures T of the region by the above formula, as shown in fig. 5.
S300: the known temperature probability distribution is represented by a basis function, and the method is realized by the following steps:
s301: selecting a basis function, adopting a Gaussian RBF network as the basis function, and obtaining the following expression:
Figure DEST_PATH_IMAGE054
wherein T is the collected temperature information; mu.si,σiThe center value and width of the function for the ith network node.
S302: determining the weighted expression of the basis function, and expressing the probability density function of the temperature distribution at the moment k by using the form of the weighted sum of the basis functions according to the RBF network approximation principle, wherein the expression is as follows:
γ(T,u(k))＝C(T)V(k)+Bn(T)wn(k)+e0(T,k) (3)
wherein C (T) ═ B1(T),B2(T),…,Bn-1(T)]，V(k)＝[w1(k),w2(k),…,wn-1(k)]T，wn(k) Is the weight corresponding to the nth basis function, e0(T, k) is the error of the approximation to the region temperature probability distribution density function.
S303: determining the weight of each basis function, the nth weight ωn(k) The nonlinear function h (v (k)) of the weight vector v (k) can be expressed as:
Figure DEST_PATH_IMAGE056
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE058
Figure DEST_PATH_IMAGE060
ignoring the approximation error, combining equations (3) and (4) yields:
Figure DEST_PATH_IMAGE062
both sides left-hand multiplication by [ CT(T) Bn(T)]TAnd in the interval [ TminTmax]Is integrated, when the matrix is
Figure DEST_PATH_IMAGE064
When not singularity, can be transformed to obtain:
Figure DEST_PATH_IMAGE066
the weight V (k) of each basis function of the temperature probability distribution density function is obtained by the above equation (6).
In addition, the selection of the corresponding basis function is not limited to the above method, and another basis function is provided below to represent the known temperature probability distribution density function, and the specific steps are as follows:
s301: the selection of the basis function, because the B-spline model can estimate any distribution curve and can represent the random distribution of the variable in the form of discrete weight, the B-spline model is used as the basis function for representing the temperature probability distribution, and the second-order B-spline basis function is as follows:
Figure DEST_PATH_IMAGE068
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE070
h (x) is the Heviside function and s ═ j-3.
S302: determining a weighted expression of the basis function, and describing the probability distribution of the collected random temperature information by using a B-spline-based estimation model, wherein the expression is as follows:
γ(T,u(k))＝C(T)V(k)+Bn(T)wn(k)+e0(T,k) (8)
wherein C (T) ═ B1(T),B2(T),…,Bn-1(T)]，V(k)＝[w1(k),w2(k),…,wn-1(k)]T，wn(k) For the nth basis function corresponding to the weight, e0(T, k) is the error of the approximation to the region temperature probability distribution density function.
S303: determining the weight of the basis function, and the formula Error! The corresponding weight calculation formula of Reference source not found is:
Figure DEST_PATH_IMAGE072
wherein, ci＝(xi-xi-3) N is the total number of data collected, Bi(xl) Is a basis function.
Using the above formula, the different basis functions of each basis function can be determined, i.e., its basis function vector V (k) can be determined.
The selection of the basis functions is not limited to the above-mentioned method, and various choices of the basis functions are available, and the present invention is not repeated herein, and the scope of the claims of the present invention should be covered without departing from the spirit of the present invention.
S400: establishing a nonlinear model between the input variable and the first n-1 weight vectors, and establishing a relation between the input variable and the temperature probability distribution density, wherein the method is realized by the following steps:
s401: selecting input variables, and adding the primary network water supply temperature T obtained in the step S100G1Primary net return water temperature TH1Water supply temperature T of secondary networkG2Secondary net return water temperature TH2Ambient temperature TECurrent control output u (k) ═ u1(k),u2(k),…,um(k)]And the first n-1 weight vectors V (k) of the temperature probability distribution at the current moment are merged as input and are recorded as: x ═ TG1,TH1,TG2,TH2,TE,u(k),V(k)]L×(4+n+m)。
S402: selecting a prediction model, selecting a random weight neural network, and expressing the network model as follows:
Figure DEST_PATH_IMAGE074
wherein, ω isj＝[ωj1,ωj2,…,ωjm]TConnecting the input weights of the jth hidden cell for m input nodes, βj＝[βj1,βj2,…,βj(n-1)]TConnecting output nodes for jth hidden layerOutput weight of the point, bjIs the bias of the jth implicit cell.
S403: training the model, randomly giving a group of input layer weights and biases, and training the model by using L groups of collected historical data, wherein an objective function is as follows:
Figure DEST_PATH_IMAGE076
s404: obtaining optimal output weight value by solving generalized inverse of H matrix
Figure DEST_PATH_IMAGE078
The formula is as follows:
Figure DEST_PATH_IMAGE080
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE082
Figure DEST_PATH_IMAGE084
s405: establishing a relation between the basis function weight and the input variable:
Figure DEST_PATH_IMAGE086
s406: and predicting the temperature probability distribution density function at the next moment:
γm(T,k+1)＝C(T)Vm(k+1)+Bn(T)wn(k+1) (14)。
in addition, the selection of the prediction model is not limited to the above method, and another model is selected for description, and the specific steps are as follows:
s402: selecting a prediction model, considering the correlation among factors influencing regional heat supply, selecting an LS _ SVM nonlinear model, wherein a network model is expressed as:
V(k+1)＝WTΦ(X(k))+Β (15)
wherein W is a diagonal weight matrix, and V (k +1) ═ V1(k+1),…,vn(k+1)]TAnd (3) representing the weight value of prediction at the moment of k +1, B is a deviation vector, and the nonlinear mapping phi (X (k)) converts input data into a high-dimensional space.
S403: training the model, and training the model by using the collected L groups of historical data, wherein the target function is as follows:
Figure DEST_PATH_IMAGE088
wherein, B ═ B1… bn]T，
Figure DEST_PATH_IMAGE090
Respectively representing a weight and a deviation, EiIs the fitting error.
S404: optimizing the optimization problem by constructing a Lagrange function to obtain the weight of the model:
Figure DEST_PATH_IMAGE092
s405: establishing a relation between the basis function weight and the input variable:
V(k+1)＝WTΦ(X(k))+Β (18)。
s406: and predicting the temperature probability distribution density function at the next moment:
γm(T,k+1)＝C(T)Vm(k+1)+Bn(T)wn(k+1) (19)。
the choice of the prediction model is not limited to the above method, and the present invention is not repeated herein, and the scope of the claims of the present invention should be covered without departing from the spirit of the present invention.
S500: after the temperature probability distribution at the next moment is obtained, feedback correction is carried out by utilizing the prediction output information and combining the temperature probability distribution, and real-time feedback correction is carried out, wherein the specific implementation mode is as follows:
determining the error between the actual output temperature probability distribution density at the kth moment and the predicted output temperature probability distribution density as follows:
e(T,k)＝γ(T,k)-γm(T,k) (20)。
and performing feedback correction on the predicted output temperature probability distribution density by using the error, wherein the corrected predicted output is as follows:
γJ(T,k+1)＝γm(T,k+1)+ηe(T,k) (21)；
wherein η is a correction coefficient, and the value range is 0 & lt η & lt 1.
S600: and (3) adjusting the heat supply output control in real time according to the temperature correction at the k +1 moment, wherein the specific implementation mode is as follows:
calculating the difference between the expected temperature and the predicted temperature at the moment k +1 as:
eJ(T,k+1)＝γg(T,k+1)-γJ(T,k+1) (22)。
constructing a performance indicator function that minimizes temperature variation:
J1(u(k))＝∫(γ(T,u(k))-γg(T,u(k))2dT (23)。
considering the energy-saving effect of the whole heating system, the performance index function minimizing the energy is as follows:
J2(u(k))＝u(k)TRu(k) (24)。
and (3) optimizing and solving, namely solving the control output which simultaneously satisfies the performance indexes of the equations (23) and (24) by using an optimization algorithm, namely:
arg min J(u(k))＝(J1(u(k)),J2(u(k))) (25)。
and determining the control output of the heating system by using an optimization algorithm by taking the difference value between the expected temperature distribution and the temperature distribution at the (k +1) th moment as input. Constructing a performance index which simultaneously meets the minimum temperature difference of a heat user and the minimum energy consumption of a heat supply system, and optimizing the performance index by utilizing an optimization algorithm to obtain the final system output control quantity u*(k)。
The overall process of the district heating energy-saving control method based on the random distribution control algorithm is shown in the attached figure 1, and the error feedback adjustment process according to the probability distribution of the predicted temperature and the probability distribution of the expected temperature in the real-time process is shown in the attached figure 2. The invention estimates the temperature probability distribution of the heat supply area by a random distribution algorithm, establishes the relation between the heat supply input and the temperature probability distribution, realizes the timely adjustment of the frequency converter of the heat supply system according to the real-time change of the temperature, ensures that the heat supply effect is more definite, not only meets the expected temperature requirement of a heat user, but also effectively reduces the unnecessary energy loss, realizes the purpose of energy-saving control, and has important practical value.
The area heating energy-saving control based on the random distribution control algorithm can be realized through the specific implementation mode. The present invention is not limited to the above-described embodiments, and any modifications or partial substitutions without departing from the spirit and scope of the present invention should be covered in the claims of the present invention.

Claims (10)

1. A district heating energy-saving control method based on a random distribution control algorithm is characterized in that: comprises the following steps of (a) carrying out,
s100: acquiring Q user temperature data, heat supply and return water data of a heat supply pipe network and output control information in different positions in an area, and storing the acquired temperature information as a historical data set;
s200: estimating the temperature probability distribution of the heat supply area according to the acquired information;
s300: representing the known temperature probability distribution density function by using a basic function representation model;
s400: establishing a nonlinear prediction model between the input variable and the first n-1 weight vectors, and predicting the temperature probability distribution at the next moment;
s500: performing feedback correction by using the predicted output information in combination with the temperature probability distribution;
s600: and (4) adjusting the heat supply output control in real time according to the temperature correction, and calculating the random distribution control quantity output of the heat supply system.
2. The random distribution control based algorithm of claim 1The regional heating energy-saving control method is characterized by comprising the following steps: in step S100, the historical data set includes Q pieces of user temperature data T ═ T in a time period in the collected heat supply region1,T2,…TQ]And the temperature T of the primary water supply at the corresponding time pointG1Primary net return water temperature TH1Water supply temperature T of secondary networkG2Secondary net return water temperature TH2And the ambient temperature TE。
3. The district heating energy-saving control method based on the random distribution control algorithm according to claim 2, characterized in that: the step S200 takes the following approach,
s201: selecting different positions (x) in the heat supply area from the data obtained in the step S100j,yj) J-1, 2, … Q total Q different temperature data T of M groups in one day for hot usersjAnd corresponding heating system control output ui(k) I is 1,2, … M, and are combined to form a temperature value matrix
Figure FDA0002427959140000011
S202: using a temperature matrix
Figure FDA0002427959140000012
Constructing temperature fields at different moments according to different time points, and performing mathematical approximation processing on the whole temperature field data to obtain a probability density function gamma (T, u) of the temperature Ti(k))：
Figure FDA0002427959140000013
Wherein N (T) represents the number of temperature values less than T in the temperature field of the heat supply area, Ti,maxAnd Ti,minRepresenting the upper and lower temperature bounds in the temperature field at time i.
4. The district heating energy-saving control method based on the random distribution control algorithm according to claim 3, characterized in that: the step S300 takes the following method,
s301: selecting a basis function;
s302: determining a weighted representation of the basis functions;
s303: a weight value for each basis function is determined.
5. The district heating energy-saving control method based on the random distribution control algorithm according to claim 4, characterized in that: in the step S300, the step of,
s301: selecting a basis function, adopting a Gaussian RBF network as the basis function, and obtaining the following expression:
Figure FDA0002427959140000014
wherein T is the collected temperature information; mu.si,σiThe center value and width of the function for the ith network node.
S302: determining the weighted expression of the basis function, and expressing the probability density function of the temperature distribution at the moment k by using the form of the weighted sum of the basis functions according to the RBF network approximation principle, wherein the expression is as follows:
γ(T,u(k))＝C(T)V(k)+Bn(T)wn(k)+e0(T,k) (3)
wherein C (T) ═ B1(T),B2(T),…,Bn-1(T)]，V(k)＝[w1(k),w2(k),…,wn-1(k)]T，wn(k) Is the weight corresponding to the nth basis function, e0(T, k) is the error of the approximation to the region temperature probability distribution density function.
S303: determining the weight of each basis function, the nth weight ωn(k) The nonlinear function h (v (k)) of the weight vector v (k) can be expressed as:
Figure FDA0002427959140000021
wherein the content of the first and second substances,
Figure FDA0002427959140000022
Figure FDA0002427959140000023
ignoring the approximation error, combining equations (3) and (4) yields:
Figure FDA0002427959140000024
both sides left-hand multiplication by [ CT(T) Bn(T)]TAnd in the interval [ TminTmax]Is integrated, when the matrix is
Figure FDA0002427959140000025
When not singularity, can be transformed to obtain:
Figure FDA0002427959140000026
the weight V (k) of each basis function of the temperature probability distribution density function is obtained by the above equation (6).
6. The district heating energy-saving control method based on the random distribution control algorithm according to claim 5, characterized in that:
s301: and selecting a basis function, wherein a B-spline model is used as the basis function for representing the temperature probability distribution, and the second-order B-spline basis function is in the following form:
Figure FDA0002427959140000031
wherein the content of the first and second substances,
Figure FDA0002427959140000032
h (x) is a Heviside function and s ═ j-3;
s302: determining a weighted expression of the basis function, and describing the probability distribution of the collected random temperature information by using a B-spline-based estimation model, wherein the expression is as follows:
γ(T,u(k))＝C(T)V(k)+Bn(T)wn(k)+e0(T,k) (8)
wherein C (T) ═ B1(T),B2(T),…,Bn-1(T)]，V(k)＝[w1(k),w2(k),…,wn-1(k)]T，wn(k) For the nth basis function corresponding to the weight, e0(T, k) is the error of the approximation of the region temperature probability distribution density function;
s303: determining the weight value of the basis function, wherein the weight calculation formula corresponding to the formula (7) is as follows:
Figure FDA0002427959140000033
wherein, ci＝(xi-xi-3) N is the total number of data collected, Bi(xl) Is a basis function;
using the above formula, different basis functions for each basis function can be determined, and the vector of basis functions V (k) can be determined.
7. The district heating energy-saving control method based on the random distribution control algorithm according to claim 5 or 6, characterized in that: the step S400 takes the following method,
s401: selecting input variables, and adding the primary network water supply temperature T obtained in the step S100G1Primary net return water temperature TH1Water supply temperature T of secondary networkG2Secondary net return water temperature TH2Ambient temperature TECurrent control output u (k) ═ u1(k),u2(k),…,um(k)]And the first n-1 weight vectors V (k) of the temperature probability distribution at the current moment are merged as input and are recorded as: x ═ TG1,TH1,TG2,TH2,TE,u(k),V(k)]L×(4+n+m)；
S402: selecting a prediction model, selecting a random weight neural network, and expressing the network model as follows:
Figure FDA0002427959140000034
wherein, ω isj＝[ωj1,ωj2,…,ωjm]TConnecting the input weights of the jth hidden cell for m input nodes, βj＝[βj1,βj2,…,βj(n-1)]TConnecting the output weights of the output nodes for the jth hidden layer, bjIs the bias of the jth hidden cell;
s403: training the model, randomly giving a group of input layer weights and biases, and training the model by using L groups of collected historical data, wherein an objective function is as follows:
Figure FDA0002427959140000041
s404: obtaining optimal output weight value by solving generalized inverse of H matrix
Figure FDA0002427959140000042
The formula is as follows:
Figure FDA0002427959140000043
wherein the content of the first and second substances,
Figure FDA0002427959140000044
Figure FDA0002427959140000045
s405: establishing a relation between the basis function weight and the input variable:
Figure FDA0002427959140000046
s406: and predicting the temperature probability distribution density function at the next moment:
γm(T,k+1)＝C(T)Vm(k+1)+Bn(T)wn(k+1) (14)。
8. the district heating energy-saving control method based on the random distribution control algorithm according to claim 5 or 6, characterized in that: the step S400 takes the following method,
s401: selecting input variables, and adding the primary network water supply temperature T obtained in the step S100G1Primary net return water temperature TH1Water supply temperature T of secondary networkG2Secondary net return water temperature TH2Ambient temperature TECurrent control output u (k) ═ u1(k),u2(k),…,um(k)]And the first n-1 weight vectors V (k) of the temperature probability distribution at the current moment are merged as input and are recorded as: x ═ TG1,TH1,TG2,TH2,TE,u(k),V(k)]L×(4+n+m)；
S402: selecting a prediction model, selecting an LS _ SVM nonlinear model, and expressing a network model as follows:
V(k+1)＝WTΦ(X(k))+Β (15)
wherein W is a diagonal weight matrix, and V (k +1) ═ V1(k+1),…,vn(k+1)]TRepresenting the weight value predicted at the moment of k +1, wherein B is a deviation vector, and the nonlinear mapping phi (X (k)) converts input data into a high-dimensional space;
s403: training the model, and training the model by using the collected L groups of historical data, wherein the target function is as follows:
Figure FDA0002427959140000051
wherein, B ═ B1…bn]T，
Figure FDA0002427959140000052
Respectively representing a weight and a deviation, EiIs the fitting error;
s404: optimizing by constructing a Lagrange function to obtain the weight of the model:
Figure FDA0002427959140000053
s405: establishing a relation between the basis function weight and the input variable:
V(k+1)＝WTΦ(X(k))+Β (18)；
s406: and predicting the temperature probability distribution density function at the next moment:
γm(T,k+1)＝C(T)Vm(k+1)+Bn(T)wn(k+1) (19)。
9. the district heating energy-saving control method based on the random distribution control algorithm according to claim 8, characterized in that: the step S500 described takes the following method,
s501: determining the error between the actual output temperature probability distribution density at the kth moment and the predicted output temperature probability distribution density as follows:
e(T,k)＝γ(T,k)-γm(T,k) (20)；
s502: and performing feedback correction on the predicted output temperature probability distribution density by using the error, wherein the corrected predicted output is as follows:
γJ(T,k+1)＝γm(T,k+1)+ηe(T,k) (21)；
wherein η is a correction coefficient, and the value range is 0 & lt η & lt 1.
10. The district heating energy-saving control method based on the random distribution control algorithm according to claim 9, characterized in that: the step S600 described takes the following approach,
s601: calculating the difference between the expected temperature and the predicted temperature at the moment k +1 as:
eJ(T,k+1)＝γg(T,k+1)-γJ(T,k+1) (22)；
s602: constructing a performance indicator function that minimizes temperature variation:
J1(u(k))＝∫(γ(T,u(k))-γg(T,u(k))2dT (23)；
s603: considering the energy-saving effect of the whole heating system, the performance index function minimizing the energy is as follows:
J2(u(k))＝u(k)TRu(k) (24)；
s604: and (3) optimizing and solving, namely solving the control output which simultaneously satisfies the performance indexes of the equations (23) and (24) by using an optimization algorithm, namely:
argminJ(u(k))＝(J1(u(k)),J2(u(k))) (25)。
CN201810365407.2A 2018-04-20 2018-04-20 Area heating energy-saving control method based on random distribution control algorithm Active CN108895532B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810365407.2A CN108895532B (en) 2018-04-20 2018-04-20 Area heating energy-saving control method based on random distribution control algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810365407.2A CN108895532B (en) 2018-04-20 2018-04-20 Area heating energy-saving control method based on random distribution control algorithm

Publications (2)

Publication Number Publication Date
CN108895532A CN108895532A (en) 2018-11-27
CN108895532B true CN108895532B (en) 2020-06-30

Family

ID=64342326

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810365407.2A Active CN108895532B (en) 2018-04-20 2018-04-20 Area heating energy-saving control method based on random distribution control algorithm

Country Status (1)

Country Link
CN (1) CN108895532B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109695174B (en) * 2018-12-28 2021-09-07 东北大学 PDF shape prediction method and system for fiber length distribution in pulping process
CN110956275B (en) * 2019-11-27 2021-04-02 支付宝(杭州)信息技术有限公司 Risk prediction and risk prediction model training method and device and electronic equipment

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102519086A (en) * 2012-01-16 2012-06-27 江苏大学 Zigbee-based central heating regulation and control system of public building
KR20160074279A (en) * 2014-12-18 2016-06-28 주식회사 부창이엔지 Boiler Having a Main-Controller For Control Heating Temperature of Room

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20170211862A1 (en) * 2016-01-25 2017-07-27 Sharp Kabushiki Kaisha Dual temperature heat pump system

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102519086A (en) * 2012-01-16 2012-06-27 江苏大学 Zigbee-based central heating regulation and control system of public building
KR20160074279A (en) * 2014-12-18 2016-06-28 주식회사 부창이엔지 Boiler Having a Main-Controller For Control Heating Temperature of Room

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
小区集中供热系统水力平衡调节节能潜力分析;刘兰斌等;《建筑科学》;20140630;第30卷(第6期);全文 *

Also Published As

Publication number Publication date
CN108895532A (en) 2018-11-27

Similar Documents

Publication Publication Date Title
CN108256672B (en) Whole-network balance adjusting method for central heating system
CN110984062B (en) Simulation scheduling method for large-scale reservoir group of watershed main and branch flows
Li et al. Predicting hourly cooling load in the building: A comparison of support vector machine and different artificial neural networks
Li et al. Assessment of linear emulators in lightweight Bayesian calibration of dynamic building energy models for parameter estimation and performance prediction
Asadi et al. Multi-objective optimization for building retrofit: A model using genetic algorithm and artificial neural network and an application
CN111365828A (en) Model prediction control method for realizing energy-saving temperature control of data center by combining machine learning
WO2021196997A1 (en) Method for predicting and controlling water level of series of water conveyance canals on basis of fuzzy neural network
Yi et al. Site-specific optimal energy form generation based on hierarchical geometry relation
CN109409614A (en) A kind of Methods of electric load forecasting based on BR neural network
JP6104235B2 (en) How to predict building energy consumption
CN111561732B (en) Heat exchange station heat supply adjusting method and system based on artificial intelligence
CN113112077A (en) HVAC control system based on multi-step prediction deep reinforcement learning algorithm
CN105910169A (en) Urban heating system heating network regulating method and system based on mechanism model prediction control
Ceballos-Fuentealba et al. A simulation and optimisation methodology for choosing energy efficiency measures in non-residential buildings
CN108895532B (en) Area heating energy-saving control method based on random distribution control algorithm
CN112232980A (en) Regulation and control method for heat pump unit of regional energy heat supply system
KR101119983B1 (en) Method for predicting water demand using Group Method of Data Handling algorithm
CN108266792A (en) It is a kind of to build the quick calculation method for representing temperature
CN104598765A (en) Building energy consumption prediction method based on elastic adaptive neural network
CN110110416A (en) A kind of cold net cooling supply optimization method of distributed energy based on graph theory
CN114484557A (en) Building group heat supply load regulation and control method based on target energy consumption management and control
Moon et al. Prediction models and control algorithms for predictive applications of setback temperature in cooling systems
CN111461404A (en) Short-term load and hydropower prediction method based on neural network prediction interval
CN111626003B (en) Heating system heat load layered prediction method, storage medium and prediction equipment
CN110097929A (en) A kind of blast furnace molten iron silicon content on-line prediction method

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