CN115423191A - Room temperature model prediction control method based on neural network and constrained by room heat load - Google Patents

Room temperature model prediction control method based on neural network and constrained by room heat load Download PDF

Info

Publication number
CN115423191A
CN115423191A CN202211077279.4A CN202211077279A CN115423191A CN 115423191 A CN115423191 A CN 115423191A CN 202211077279 A CN202211077279 A CN 202211077279A CN 115423191 A CN115423191 A CN 115423191A
Authority
CN
China
Prior art keywords
module
prediction
value
room temperature
time
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202211077279.4A
Other languages
Chinese (zh)
Inventor
王海超
薄盛
吴小舟
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.)
Dalian University of Technology
Original Assignee
Dalian 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 Dalian University of Technology filed Critical Dalian University of Technology
Priority to CN202211077279.4A priority Critical patent/CN115423191A/en
Publication of CN115423191A publication Critical patent/CN115423191A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • 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
    • F24D19/1012—Arrangement or mounting of control or safety devices for water heating systems for central heating by regulating the speed of a pump
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • Economics (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • General Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Tourism & Hospitality (AREA)
  • Operations Research (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Game Theory and Decision Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Quality & Reliability (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Development Economics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Thermal Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • Combustion & Propulsion (AREA)
  • Mechanical Engineering (AREA)
  • Steam Or Hot-Water Central Heating Systems (AREA)

Abstract

The invention belongs to the technical field of central heating room temperature control, deep learning and automation, and provides a room temperature model prediction control method based on a neural network and constrained by room heat load. The data input module converts the weather forecast data format and inputs the weather forecast data format into the heat load interval prediction module, the indoor temperature prediction module and the radiator module respectively; inputting a predicted value of the heat load interval, a predicted value of the indoor temperature and the heat dissipation capacity of the radiator into a rolling optimization module; the rolling optimization module searches a global optimal hot water flow sequence under the constraint of a predicted value of a thermal load interval, a first hot water flow value is transmitted to the flow frequency conversion module, and the output water pump rotating speed is input to the water pump controller module; the room temperature real value module returns the indoor temperature real value to the indoor temperature prediction module, and the calculation of the indoor temperature predicted value at the next moment is continued. The invention effectively achieves the purposes of reducing indoor temperature fluctuation and saving energy consumption, improves the response speed and has economic benefit and environmental benefit.

Description

Room temperature model prediction control method based on neural network and constrained by room heat load
Technical Field
The invention relates to the technical field of central heating room temperature control, deep learning and automation, in particular to a room temperature model prediction control method based on a neural network and constrained by room heat load.
Background
The main mode of winter heating in northern China is central heating, the area of urban central heating is in a rapid growth trend along with the promotion of urbanization, the continuous growth of the area of heating can cause larger energy consumption and higher carbon emission of heating, and the hot user side is an object which needs to be focused on realizing the energy conservation of a heating system. The model predictive control has good applicability to a heating system with large inertia and strong time lag, so that the problem of the optimal control of the indoor temperature can be effectively solved by utilizing the model predictive control theory.
With the development of intelligent algorithms and the proposal of neural networks, black box models which utilize a large amount of data to learn and predict complex mechanisms are also widely applied to the field of indoor temperature prediction, and the neural networks can learn the complex nonlinear relation among multiple input and output systems according to a large amount of data and perform rapid prediction, so that the neural network method can be used in the prediction control of the room temperature model to well describe the room temperature change process and solve the speed problem of room temperature dynamic regulation and control. At present, most of room temperature model prediction control utilizes a discretized state space equation to predict the indoor temperature, but the method has low calculation speed and cannot well meet the requirement of dynamic regulation and control of the indoor temperature. Meanwhile, the current room temperature model prediction control method lacks guidance of heat load, and has the problems that the optimal control sequence solving range is too large and the room temperature control effect needs to be improved. Aiming at the situation, the invention provides a method for adding room heat load constraint in the traditional room temperature model predictive control method to reduce the solving range of the optimal control sequence, and simultaneously, a method for adding a neural network in the room temperature model predictive control method to solve the speed problem of the dynamic regulation and control of the room temperature.
Disclosure of Invention
In view of the above background, a room temperature model predictive control method based on a neural network and constrained by room thermal load is proposed. The method can provide heat load guidance for room temperature model prediction control, reduce the solving range of an optimal control sequence, improve the room temperature dynamic regulation speed, reduce indoor temperature fluctuation, save energy consumption, improve the control effect and have both economic benefit and environmental benefit.
The technical scheme of the invention is as follows: a room temperature model prediction control method based on a neural network and constrained by room heat load is characterized in that the room temperature model prediction control method based on the neural network and constrained by the room heat load comprises a data input module 1, a heat load interval prediction module 2, an indoor temperature prediction module 3, a radiator module 4, a rolling optimization module 5, a flow and rotating speed conversion module 6, a water pump controller module 7 and a room temperature true value module 8; the data input module 1 carries out format conversion on the weather forecast data, and a weather forecast data set L after the format conversion is respectively input to the thermal load interval prediction module 2, the indoor temperature prediction module 3 and the radiator module 4; the predicted value of the heat load interval output by the heat load interval prediction module 2, the predicted value of the indoor temperature output by the indoor temperature prediction module 3 and the heat dissipation capacity of the radiator output by the radiator module 4 are input to the rolling optimization module 5; the rolling optimization module 5 searches for a globally optimal hot water flow sequence under the constraint of a predicted value of a thermal load interval, transmits a first hot water flow value to the flow frequency conversion module 6, and then inputs the water pump rotating speed output by the flow frequency conversion module 6 to the water pump controller module 7; the real room temperature value module 8 returns the real room temperature value to the indoor temperature prediction module 3, and continues to calculate the predicted room temperature value at the next moment.
The meteorological forecast data set output by the k moment data input module (1) is recorded as L k ，
L k ＝{L(k+1)，L(k+2)，...，L(k+n)，...L(k+T)} (1)
Wherein L is k A weather forecast data set indicating the (k + 1) to (k + T) times acquired at the k time, T being arbitrarily set by the operator. L (k + n) represents weather forecast data at time (k + n) acquired at time kA value of n is [1, T ]]Any integer in between.
The thermal load interval prediction module 2 comprises a thermal load prediction model and an error prediction model;
a thermal load prediction model, a meteorological forecast data set L obtained by inputting k time k Point prediction is carried out on the heat load in the subsequent prediction time domain T to obtain the heat load prediction values from (k + 1) to (k + T) time, and the heat load prediction values are arranged according to the ascending order of time to obtain a heat load prediction value time sequence set A k The number of the bits, as noted,
A k ＝F NN1 [L k ]＝{Q(k+1)，Q(k+2)，...，Q(k+n)，...Q(k+T)} (2)
wherein, F NN1 Forecasting a neural network model for a thermal load point, inputting a meteorological forecast data set L acquired at k moments k ，A k A set of predicted values of thermal load time series, Q, at times (k + 1) to (k + T) obtained using a neural network model for predicting thermal load points at time k k (k + n) represents a predicted value of the thermal load at the time (k + n) predicted at the time k, and n is [1, T ]]Any integer therebetween.
The error prediction model is used for predicting the error condition of each prediction point from the (k + 1) to (k + T) moments of the thermal load prediction model, Q k The error prediction value of (k + n) is noted,
ε k (k+n)＝|F NN2 [L(k+n)，Q(k+n)]| (3)
wherein, F NN2 For the error prediction neural network model, weather forecast data L (k + n) at time (k + n) acquired for time k and a thermal load prediction value Q (k + n) are input. It should be noted that n passes through [1, T ]]To obtain A k The error condition of each predicted point.
The predicted value of the heat load interval consists of a heat load predicted value and an error predicted value which are recorded,
I k (k+n)＝[Q(k+n)-ε(k+n)，Q(k+n)+ε(k+n)] (4)
wherein, I k (k + n) is a thermal load section at time (k + n) predicted at time k, Q (k + n) - ε (k + n) is a lower limit of the prediction of the thermal load section, and Q (k + n) + ε (k + n) is an upper limit of the prediction of the thermal load sectionIt is clear that n runs through [1, T ]]The predicted value of the thermal load section at each time of (k + 1) to (k + T) can be obtained.
At the moment k, the indoor temperature prediction module (3) is specifically,
x(k+n+1)＝F NN3 [L(k+n)，u(k+n)，x(k+n)] (5)
wherein, F NN3 For the indoor temperature prediction neural network model, x (k + n + 1) is the predicted room temperature value at the moment of (k + n + 1), L (k + n), u (k + n) and x (k + n) are respectively the weather forecast data value at the moment of (k + n), the drawn hot water flow value of (k + n) and the predicted room temperature value of (k + n), and n is spread over [0, T-1 ]]The whole number in (1).
The radiator module (4) obtains the heat dissipation capacity of the radiator by inputting the hot water flow and the temperature difference of supply and return water;
Q rad ＝F NN4 [u(k)，Δt] (6)
wherein Q is rad For the heat dissipation of the heat sink, F NN4 The prediction model is a neural network prediction model of the heat dissipating capacity of the radiator, u is the flow rate of hot water, delta T is the temperature difference between water supply and return of the radiator, and the delta T is measured by a temperature sensor on a water supply and return pipe of the radiator and does not change in each prediction time domain T;
the rolling optimization module (5) is added into a heat load interval, the allowable flow of the equipment and the room temperature allowable value set by a user to restrict the hot water flow passing through the radiator, the restriction condition is,
Q(k+n)-ε(k+n)≤Q rad ≤Q(k+n)+ε(k+n) (7)
u min ≤u(k)≤u max (8)
x min ≤x(k)≤x max (9)
wherein u is min Allowing minimum flow for the device, u max Maximum flow allowed for the device, x min Minimum allowable value of room temperature, x, set for user max Setting a maximum allowable value of the room temperature for a user, wherein Q (k + n) -epsilon (k + n) is the lower limit of the interval I output by the thermal load interval prediction module (2) at the moment k, and Q (k + n) + epsilon (k + n) is the upper limit of the interval I output by the thermal load interval prediction module (2) at the moment k;
the objective function of the rolling optimization module (5) is,
Figure BDA0003831688930000041
j is an objective function of minimizing room temperature fluctuation, and T is a prediction time domain of model prediction control;
the rolling optimization module 5 adopts a genetic algorithm with elite reservation to solve the optimal hot water flow sequence flowing through the radiator.
Figure BDA0003831688930000042
The first hot water flow value u * (k + 1) to the traffic frequency conversion module 6,
the flow rate and rotation speed conversion module 6 is used for realizing the corresponding relation between the rotation speed of the water pump and the flow rate of the hot water,
Figure BDA0003831688930000043
wherein, a 1 ，a 2 S is a parameter to be identified, and the data required for identification is obtained from historical operating data; n is a radical of an alkyl radical 0 And H 0 For selecting the rated speed of the plant and the lift at the rated speed, H p Is the pressure head difference at the inlet and outlet of the pipeline, n * (k + 1) is a hot water flow value u * (k + 1) corresponding water pump speed.
Output water pump rotating speed n of flow frequency conversion module 6 * (k + 1) to the water pump controller module 7.
In the (k + 1) - (k + 2) time, the water pump controller module 7 controls the water pump to keep the rotating speed n * (k + 1). Therefore, the solving and the control of the frequencies of the (k + 1) - (k + 2) water pumps are completed, and the aim of controlling the room temperature is further fulfilled.
And (k + 2) returning the real room temperature value acquired by the room temperature sensor to the indoor temperature prediction module 3 by the real room temperature value module 8, solving and controlling the frequency of the water pump within the time from (k + 2) to (k + 2), and so on until the example room stops working.
The invention has the beneficial effects that: the invention provides a room temperature model predictive control method based on a neural network and constrained by room heat load, which takes the predicted value of a room heat load interval based on the neural network as constraint, not only provides guidance for model predictive control, but also reduces the solving range of the optimal hot water flow sequence flowing through a radiator. Meanwhile, on the basis of the neural network, a heat load point prediction neural network model, an error prediction neural network model, an indoor temperature prediction neural network model and a radiator heat dissipation capacity neural network prediction model are constructed, the problem that the response time of a discretized state space equation is long is solved, the speed of dynamically regulating and controlling the indoor temperature is greatly improved, and the rapid dynamic regulation of the indoor temperature is realized. Meanwhile, a room temperature model prediction control method based on a neural network and constrained by room heat load is adopted to maintain stable room temperature, so that the purposes of reducing indoor temperature fluctuation and saving energy consumption can be effectively achieved, and economic benefits and environmental benefits are achieved.
Drawings
FIG. 1 is a block diagram of a room temperature model predictive control method based on a neural network and constrained by room thermal load.
In the figure: 1-a data input module; 2-a thermal load interval prediction module; 3-an indoor temperature prediction module; 4-a radiator module; 5-a rolling optimization module; 6-flow and rotation speed conversion module; 7-a water pump controller module; 8-real room temperature value module.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
A room temperature model prediction control method based on a neural network and constrained by room heat load comprises a data input module 1, a heat load interval prediction module 2, an indoor temperature prediction module 3, a radiator module 4, a rolling optimization module 5, a flow and rotating speed conversion module 6, a water pump controller module 7 and a room temperature true value module 8.
The data input module 1 carries out format conversion on the weather forecast data, and a weather forecast data set L after the format conversion is respectively input to the thermal load interval prediction module 2, the indoor temperature prediction module 3 and the radiator module 4; the predicted value of the heat load interval output by the heat load interval prediction module 2, the predicted value of the indoor temperature output by the indoor temperature prediction module 3 and the heat dissipation capacity of the radiator output by the radiator module 4 are input to the rolling optimization module 5; the rolling optimization module 5 searches a globally optimal hot water flow sequence under the constraint of a predicted value of a thermal load interval, transmits a first hot water flow value to the flow frequency conversion module 6, and inputs the rotating speed of the water pump output by the flow frequency conversion module 6 to the water pump controller module 7; the real room temperature value module 8 returns the real room temperature value to the indoor temperature prediction module 3, and continues to calculate the predicted room temperature value at the next moment.
For convenience of description and explanation, the sampling period of a room temperature model prediction control method based on a neural network and constrained by room heat load is set to be 5min, the prediction time domain is set to be 30min, the control time domain is set to be 30min, and the working time of an example room is 8:00-22:00.
the room at 8 points starts to operate, and the meteorological forecast data set output by the data input module (1) at 8 points is recorded as L 8 ，
L 8 ＝{L(8：05)，L(8：10)，L(8：15)，...L(8：30)} (13)
Wherein L is 8 Representation 8:00 acquired 8:05 to 8:30, of weather forecast data sets. L (8.
The thermal load interval prediction module 2 comprises a thermal load prediction model and an error prediction model;
thermal load prediction model, by input 8:00 acquired weather forecast dataset L 8 And performing point prediction on the heat load in the subsequent prediction time domain within 30min to obtain 8:05 to 8:30 and arranging the predicted heat load values according to the ascending order of time to obtain a time sequence set A of the predicted heat load values 8 The number of the bits, as noted,
A 8 ＝F NN1 [L 8 ]＝{Q(8：05)，Q(8：10)，Q(8：15)，...Q(8：30)} (14)
wherein, F NN1 For the prediction neural network model for the thermal load point, the input is 8:00 acquired weather forecast dataset L 8 The neural network needs to be trained in advance, and the training method is not described in detail since it does not belong to the present invention. A. The 8 Is 8:00 obtained using the thermal load point prediction neural network model 8:05 to 8: the thermal load prediction value time-series set of 30, Q (8.
The error prediction model is used to predict the thermal load prediction model 8:05 to 8: and 30, error condition of each prediction point, 8:00 using the error prediction model output 8:05-8: the error prediction value of the 30 thermal load value is,
Figure BDA0003831688930000061
wherein, F NN2 For the error prediction neural network model, the input quantity is the meteorological forecast data and the thermal load point prediction value at the moment corresponding to the error, the neural network needs to be trained in advance, and the training method is not described in detail because the training method does not belong to the invention.
The predicted value of the heat load interval consists of a heat load predicted value and an error predicted value which are recorded,
Figure BDA0003831688930000071
wherein, I 8 (8：05)-I 8 (8: 00 predicted 8:05-8: thermal load interval at time 30.
The output of the indoor temperature prediction module (3) is 8:05-8: a predicted value of the room temperature of 30,
Figure BDA0003831688930000072
wherein, F NN3 For the indoor temperature prediction neural network model, the input quantities are weather forecast data at the previous moment, a hot water flow value and a room temperature prediction value at the previous moment, and x (8: 00 real value at room temperature, the neural network needs to be trained in advance, and the training method is not described in detail because the training method does not belong to the invention.
The radiator module (4) obtains the heat dissipating capacity of the radiator by inputting the hot water flow and the temperature difference of the supply and return water,
Figure BDA0003831688930000073
wherein Q is rad For the heat dissipation of the heat sink, F NN4 The neural network prediction model for the heat dissipating capacity of the radiator inputs the hot water flow and the temperature difference between supplied water and returned water at the last moment, and the training is required to be carried out in advance through the network. Delta t is the temperature difference between the water supply and the water return of the radiator, and is measured by a temperature sensor on a water supply and return pipe of the radiator, and the delta t is not changed within 30 min;
the rolling optimization module (5) is added into a heat load interval, the allowable flow of the equipment and the room temperature allowable value set by a user to restrict the hot water flow passing through the radiator, the restriction condition is,
Figure BDA0003831688930000074
Figure BDA0003831688930000081
Figure BDA0003831688930000082
wherein u is min Allowing minimum flow for the device, u max Maximum flow allowed for the device, x min Minimum allowable value of room temperature, x, set for user max Setting a maximum allowable value of room temperature for a user;
the objective function of the rolling optimization module (5) is,
Figure BDA0003831688930000083
wherein J is an objective function for minimizing room temperature fluctuation;
the rolling optimization module 5 adopts a genetic algorithm with elite reservation to solve the optimal hot water flow sequence flowing through the radiator to obtain the optimal hot water flow sequence,
Figure BDA0003831688930000084
the first hot water flow value u * (8) to the traffic frequency conversion module 6,
the flow and rotating speed conversion module (6) is used for realizing the corresponding relation between the rotating speed of the water pump and the flow of hot water,
Figure BDA0003831688930000085
wherein, a 1 ，a 2 S is a parameter to be identified, and the data required for identification is obtained from historical operating data; n is a radical of an alkyl radical 0 And H 0 For selecting the rated speed of the plant and the lift at the rated speed, H p Is the pressure head difference at the inlet and outlet of the pipeline, n * (8 * (8.
Output water pump rotating speed n of flow frequency conversion module 6 * (8).
8:05-8: the water pump controller module 7 controls the water pump to keep during 10 hoursSpeed n * (8:05). Thus, finishing the steps of 8:05-8: and 10, solving and controlling the frequency of the water pump, thereby achieving the aim of controlling the room temperature.
8: and 10, returning the real room temperature value acquired by the room temperature sensor to the indoor temperature prediction module 3 by the real room temperature value module 8, and performing 8:10-8:15 solving and controlling the frequency of the water pump, and so on, completing the control of the water pump and the room temperature until 22:00 example the room stops working.

Claims (2)

1. A room temperature model prediction control method based on a neural network and constrained by room heat load is characterized by comprising a data input module (1), a heat load interval prediction module (2), an indoor temperature prediction module (3), a radiator module (4), a rolling optimization module (5), a flow and rotating speed conversion module (6), a water pump controller module (7) and a room temperature true value module (8); the data input module (1) converts the format of the weather forecast data, and the weather forecast data set L after format conversion is respectively input into the heat load interval prediction module (2), the indoor temperature prediction module (3) and the radiator module (4); the predicted value of the heat load interval output by the heat load interval prediction module (2), the predicted value of the indoor temperature output by the indoor temperature prediction module (3) and the heat dissipation capacity of the radiator output by the radiator module (4) are input to the rolling optimization module (5) together; the rolling optimization module (5) searches for a globally optimal hot water flow sequence under the constraint of a predicted value of a thermal load interval, transmits a first hot water flow value to the flow frequency conversion module (6), and inputs the rotating speed of the water pump output by the flow frequency conversion module (6) to the water pump controller module (7); the real room temperature value module (8) returns the real room temperature value to the indoor temperature prediction module (3) and continues to calculate the predicted room temperature value at the next moment;
the meteorological forecast data set output by the k moment data input module (1) is recorded as L k ，
L k ＝{L(k+1),L(k+2),…,L(k+n),…L(k+T)}(1)
Wherein L is k Weather forecast representing time (k + 1) to (k + T) acquired at time kReporting a data set, wherein T is set according to time requirements; l (k + n) represents a weather forecast data value at the time (k + n) acquired at the time k, and n is [1, T ]]Any integer in between;
the thermal load interval prediction module (2) comprises a thermal load prediction model and an error prediction model;
thermal load prediction model by inputting a weather forecast data set L obtained at time k k Point prediction is carried out on the heat load in the subsequent prediction time domain T to obtain the heat load prediction values from (k + 1) to (k + T) time, and the heat load prediction values are arranged according to the ascending order of time to obtain a heat load prediction value time sequence set A k The number of the bits, as noted,
A k ＝F NN1 [L k ]＝{Q(k+1),Q(k+2),…,Q(k+n),…Q(k+T)} (2)
wherein, F NN1 Forecasting a neural network model for a thermal load point, inputting a meteorological forecast data set L acquired at a moment k k ，A k A set of predicted values of thermal load time series, Q, at times (k + 1) to (k + T) obtained using a neural network model for predicting thermal load points at time k k (k + n) represents a predicted value of the thermal load at the time (k + n) predicted at the time k, and n is [1, T ]]Any integer in between;
the error prediction model is used for predicting the error condition of each prediction point from the (k + 1) to (k + T) time of the heat load prediction neural network model, and Q is k The error prediction value of (k + n) is noted,
ε k (k+n)＝|F NN2 [L(k+n),Q(k+n)]| (3)
wherein, F NN2 Inputting weather forecast data L (k + n) and a heat load predicted value Q (k + n) at the (k + n) moment acquired for the k moment for an error prediction neural network model; wherein n passes through [1, T ]]To obtain A k The error condition of each predicted point is determined;
the predicted value of the heat load interval consists of a heat load predicted value and an error predicted value which are recorded,
I k (k+n)＝[Q(k+n)-ε(k+n),Q(k+n)+ε(k+n)] (4)
wherein, I k (k + n) is a thermal load section at time (k + n) predicted at time k, Q (k + n) -epsilon (k + n) is a lower limit of the prediction of the thermal load section, and Q(k + n) + ε (k + n) is the upper limit of the thermal load interval prediction; wherein n is represented by [1, T ]]The predicted value of the thermal load interval at each time from (k + 1) to (k + T) is obtained;
at the moment k, the indoor temperature prediction module (3) is specifically,
x(k+n+1)＝F NN3 [L(k+n)，u(k+n)，x(k+n)] (5)
wherein, F NN3 For the indoor temperature prediction neural network model, x (k + n + 1) is the predicted room temperature value at the moment of (k + n + 1), L (k + n), u (k + n) and x (k + n) are the weather forecast data value at the moment of (k + n), the drawn hot water flow value of (k + n) and the predicted room temperature value of (k + n), respectively, and n is spread over [0, T-1 ]]The integer of (1);
the radiator module (4) obtains the heat dissipation capacity of the radiator by inputting the hot water flow and the temperature difference of supply and return water;
Q rad ＝F NN4 [u(k)，Δt] (6)
wherein Q is rad For heat dissipation of the radiator, F NN4 The method comprises the steps that a neural network prediction model of the heat dissipation capacity of a radiator is adopted, u is the flow of hot water, delta T is the temperature difference between supply water and return water of the radiator, and the delta T is measured by a temperature sensor on a supply water return pipe of the radiator and does not change in each prediction time domain T;
the rolling optimization module (5) is added into a heat load interval, the allowable flow of the equipment and the room temperature allowable value set by a user to restrict the hot water flow passing through the radiator, the restriction condition is,
Q(k+n)-ε(k+n)≤Q rad ≤Q(k+n)+ε(k+n) (7)
u min ≤u(k)≤u max (8)
x min ≤x(k)≤x max (9)
wherein u is min Allowing minimum flow for the device, u max Maximum flow allowed for the device, x min Minimum allowable value of room temperature, x, set for user max Setting a maximum allowable value of the room temperature for a user, wherein Q (k + n) -epsilon (k + n) is the lower limit of the interval I output by the thermal load interval prediction module (2) at the moment k, and Q (k + n) + epsilon (k + n) is the upper limit of the interval I output by the thermal load interval prediction module (2) at the moment k;
the objective function of the rolling optimization module (5) is,
Figure FDA0003831688920000031
wherein J is an objective function of minimizing room temperature fluctuation, and T is a prediction time domain of model prediction control;
the flow and rotating speed conversion module (6) is used for realizing the corresponding relation between the rotating speed of the water pump and the flow of the hot water,
Figure FDA0003831688920000032
wherein, a 1 ，a 2 S is a parameter to be identified, and the data required for identification is obtained from historical operating data; n is 0 And H 0 For selecting the rated speed of the plant and the lift at the rated speed, H p Is the pressure head difference of the inlet and the outlet of the pipeline, and n is the rotating speed of the water pump corresponding to the flow u.
2. The room temperature model predictive control method based on the neural network and constrained by the room heat load as claimed in claim 1, wherein the solution method of the rolling optimization module (5) adopts a genetic algorithm with elite reservation to solve the optimal frequency sequence of the variable frequency water pump.
CN202211077279.4A 2022-09-05 2022-09-05 Room temperature model prediction control method based on neural network and constrained by room heat load Pending CN115423191A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211077279.4A CN115423191A (en) 2022-09-05 2022-09-05 Room temperature model prediction control method based on neural network and constrained by room heat load

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211077279.4A CN115423191A (en) 2022-09-05 2022-09-05 Room temperature model prediction control method based on neural network and constrained by room heat load

Publications (1)

Publication Number Publication Date
CN115423191A true CN115423191A (en) 2022-12-02

Family

ID=84201499

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211077279.4A Pending CN115423191A (en) 2022-09-05 2022-09-05 Room temperature model prediction control method based on neural network and constrained by room heat load

Country Status (1)

Country Link
CN (1) CN115423191A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116307074A (en) * 2023-02-01 2023-06-23 中国建筑科学研究院有限公司 Method for acquiring real thermal data, system and method for constructing neural network model of thermal data, and method for predicting thermal load
CN117580345A (en) * 2024-01-19 2024-02-20 广州豪特节能环保科技股份有限公司 Cloud computing-based centralized control method and system for indirect evaporative cooling equipment

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116307074A (en) * 2023-02-01 2023-06-23 中国建筑科学研究院有限公司 Method for acquiring real thermal data, system and method for constructing neural network model of thermal data, and method for predicting thermal load
CN117580345A (en) * 2024-01-19 2024-02-20 广州豪特节能环保科技股份有限公司 Cloud computing-based centralized control method and system for indirect evaporative cooling equipment
CN117580345B (en) * 2024-01-19 2024-04-19 广州豪特节能环保科技股份有限公司 Cloud computing-based centralized control method and system for indirect evaporative cooling equipment

Similar Documents

Publication Publication Date Title
CN115423191A (en) Room temperature model prediction control method based on neural network and constrained by room heat load
CN111256294B (en) Model prediction-based optimization control method for combined operation of water chilling unit
CN110288164B (en) Predictive control method for building air-conditioning refrigeration station system
CN113091123B (en) Building unit heat supply system regulation and control method based on digital twin model
CN111473407B (en) Model-based centralized heating system on-demand accurate regulation and control method
CN108990383B (en) Predictive control method for air conditioning system of data center
CN104359195B (en) Based on the central air-conditioning freezing water controling method that dynamic response end total load changes
CN111550861B (en) Self-adaptive optimization control method, system and device for heat pump and electric heat storage equipment
CN107023966B (en) Method for optimizing set value of outlet water temperature of cooling water of air conditioner of subway station
CN105864016B (en) A kind of more water pump distributing system variable water volume runing adjustment methods of open type
CN104698843A (en) Model prediction control based energy saving control method of data center
Li et al. Rule-based fuzzy control method for static pressure reset using improved Mamdani model in VAV systems
CN101655272A (en) Energy-saving control management system of network central air conditioner and method thereof
CN110110416B (en) Distributed energy source cold network cold supply optimization method based on graph theory
CN113446661B (en) Intelligent and efficient heat supply network operation adjusting method
CN109917656B (en) Circulating cooling water minimum pressure difference energy-saving control system and method based on process medium multi-temperature target
WO2023083235A1 (en) Optimization method and system for performing prediction on power distribution network in distributed manner
CN109114662A (en) A kind of heating control method and system of the electric heating energy storage equipment based on multiple agent
Tianyi et al. An optimal differential pressure reset strategy based on the most unfavorable thermodynamic loop on-line identification for a variable water flow air conditioning system
CN110486896B (en) Cascade air conditioning system optimization control method based on water chilling unit energy consumption model
CN111473408A (en) Control method of heat supply control system based on climate compensation
CN111523210A (en) Prediction analysis method and system for temperature rise and drop process of urban central heating system
CN113375220B (en) Heat exchange station multi-mode regulation and control method based on load prediction
CN114358604A (en) Urban centralized heating system time-sharing and zone-dividing regulation and control method considering building heat storage characteristics
CN115614812A (en) Hydraulic balance management method based on Internet of things and edge computing technology

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