A method and system for predicting building cooling and heating loads based on dynamic time warping and multiple models

Abstract

The present invention discloses a method and system for predicting building cold and hot loads based on dynamic time warping and multiple models, including: constructing a temperature dynamic model of heat medium in a building's heating pipeline, a static heat balance model at a node, and a radiator power model, and combining them to obtain a heat load prediction model for the building; constructing several energy consumption models, and obtaining a cold load prediction model for the building based on the several energy consumption models; using hierarchical clustering based on dynamic time warping to obtain the correlation between model data sequences, and merging the model data sequences to obtain a model data sequence to be input; inputting the model data sequence to be input into the cold and hot load prediction model to obtain cold and hot load prediction results; wherein the several energy consumption models include: a chiller energy consumption model, a refrigeration water pump energy consumption model, a cooling water pump energy consumption model, a cooling tower energy consumption model, and a terminal air conditioning energy consumption model. The present invention achieves the purpose of accurately predicting the cold and hot loads of a building.

Description

Technical Field

[0001] The present invention relates to the technical field of building equipment control, and in particular to a method and system for predicting building cooling and heating loads based on dynamic time warping and multiple models. Background Art

[0002] As the energy consumption intensity of public buildings increases, energy conservation in public buildings has become an important part of building energy conservation and carbon reduction. Building load forecasting plays an important role in building operation stages such as building energy management, energy operation and building equipment control strategy to improve building energy efficiency and tap the potential of building energy conservation.

[0003] The reason for the increase in energy intensity of public buildings lies in the continuous increase in energy intensity per unit area of ​​public buildings and the gradually increasing terminal energy demand of various public buildings (such as air conditioning, lighting, etc.). "Passive priority to reduce demand, active optimization to improve efficiency" is the basic support for the reduction of building energy demand, and it is also the hub for the development of building energy conservation. One of the key technologies to achieve "active optimization to improve efficiency" is the prediction research of terminal energy demand of public buildings, among which the prediction of cooling demand of air conditioning (i.e. building cooling load prediction) is the main one.

[0004] Existing technologies mainly focus on two aspects when implementing "active optimization to improve efficiency" to solve the problem of high energy consumption in public buildings. One is to use better building envelope design during building construction, and the other is to build various advanced intelligent energy-saving control technologies to realize intelligent buildings. Building load forecasting can be used to evaluate energy-saving potential in building design and renovation, and can also be used as a basis for control strategies of intelligent buildings. Therefore, accurate and effective building load forecasting can not only guide building design, but also optimize the scheduling strategy of building energy equipment, thereby developing the energy-saving potential of buildings.

[0005] With the progress of society, more and more public buildings are equipped with heating systems during construction, so the building heat load forecast has become an important part of the public building terminal energy demand forecast. However, most of the existing public building load forecasting methods only forecast the building cooling load, while there is a lack of effective and accurate forecasting models for the building heat load forecast. Therefore, the forecast results obtained in the forecast of public building terminal energy demand are inaccurate and incomplete, which is not conducive to achieving "active optimization to improve efficiency".

[0006] In addition, when forecasting building load, it is necessary to rely on building data in different time series, and there are certain differences between these building data. The existing building load forecasting methods do not exclude data with large differences, which in turn affects the accuracy of the load forecasting results. Summary of the invention

[0007] In order to overcome the shortcomings of the prior art, the present invention provides a method and system for predicting building cooling and heating loads based on dynamic time warping and multiple models, which are used to solve the technical problems that the existing building load prediction methods lack effective building thermal load prediction means and the prediction accuracy is poor due to large data differences, thereby achieving the purpose of accurately predicting the building cooling and heating loads.

[0008] To solve the above problems, the technical solution adopted by the present invention is as follows:

[0009] A method for predicting building cooling and heating loads based on dynamic time warping and multiple models includes the following steps:

[0010] Construct a dynamic temperature model of the heat medium in the building's heating pipes, a static heat balance model at the nodes, and a radiator power model, and combine them to obtain a heat load prediction model for the building;

[0011] Constructing several energy consumption models, and obtaining a cooling load prediction model of the building based on the several energy consumption models;

[0012] Hierarchical clustering based on dynamic time warping is used to obtain the correlation between model data sequences, and the model data sequences are merged to obtain the model data sequence to be input;

[0013] Inputting the model data sequence to be input into the cold and hot load prediction model to obtain cold and hot load prediction results;

[0014] The energy consumption models include: a chiller energy consumption model, a refrigeration water pump energy consumption model, a cooling water pump energy consumption model, a cooling tower energy consumption model and a terminal air conditioning energy consumption model.

[0015] As a preferred embodiment of the present invention, when constructing a temperature dynamic model of a heat medium, it includes:

[0016] According to the heat medium density, the cross-sectional area of ​​the pipeline, the length of the pipeline, and the heat medium mass flow rate of the pipeline, the pipeline The second coefficient related to the transmission delay , as shown in Formula 1:

[0017] (1);

[0018] In the formula, is the time interval, For pipeline The cross-sectional area of For pipeline Length, For pipeline The heat medium mass flow rate, is the heat medium density, is the ceiling rounding function;

[0019] According to arrive Time flows through the pipeline The total mass of the heat medium in the pipeline The first coefficient related to the transmission delay , as shown in Formula 2 and Formula 3:

[0020] (2);

[0021] (3);

[0022] In the formula, for arrive Time flows through the pipeline Total mass of heat medium;

[0023] According to the first coefficient and the second coefficient Get pipeline Heat loss coefficient .

[0024] As a preferred embodiment of the present invention, after obtaining the pipeline Heat loss coefficient After, including:

[0025] According to the pipeline The first coefficient related to the transmission delay , the second coefficient And pipeline Heat loss coefficient , get the transmission delay of the heat medium and transmission heat loss ;

[0026] According to the transmission delay of heat medium and transmission heat loss , construct the temperature dynamic model of the heat medium, as shown in Formula 7:

[0027] (7);

[0028] In the formula, It is the temperature dynamic model of the heat medium.

[0029] As a preferred embodiment of the present invention, when constructing a static heat balance model at a node, it includes:

[0030] According to the heat medium mass flow rate of the pipeline, the heat medium temperature at the nodes in the water supply and return network is obtained;

[0031] Based on the specific heat capacity of the heat medium and the heat medium temperature at the nodes in the water supply and return networks, the heat power and heat load power provided by the heat source at the node are obtained, as shown in Formula 9 and Formula 10:

[0032] (9);

[0033] (10);

[0034] In the formula, for Time Node The heat power provided by the heat source at for Time Node The heat load power at is the specific heat capacity of the heat medium, is the set of source nodes in the heating pipeline, is the set of load nodes in the heating pipeline, and for Nodes in the water supply and return network at all times The heat medium temperature at , Outflow and inflow nodes respectively A collection of pipelines.

[0035] As a preferred embodiment of the present invention, when constructing a radiator power model, it includes:

[0036] According to the indoor and outdoor temperature, equivalent thermal resistance and heat load power of the building, the equivalent heat capacity of the building is obtained. , and obtain the node load of the heating pipeline within a period of time, as shown in formula 12:

[0037] (12);

[0038] In the formula, for The outdoor temperature of the building at the moment, for The indoor temperature of the building at the moment, for The heat load power at the moment, is the interval time, is a natural constant, is the set of load nodes in the heating pipeline, is the equivalent resistance of the building, For interval time Nodes of internal heating pipes load;

[0039] According to the node load and node The radiator parameters of the building are used to obtain the radiator power model, as shown in Formula 13:

[0040] (13);

[0041] In the formula, and For Node The first and second radiator parameters of the building, and for Nodes in the water supply and return network at all times The heat medium temperature at is the heat sink power model.

[0042] As a preferred embodiment of the present invention, when constructing several energy consumption models, it includes:

[0043] Construct the energy consumption model of the chiller according to the chiller load rate and the chiller cooling water outlet temperature;

[0044] According to the frequency of refrigeration water pump and cooling water pump, energy consumption model of refrigeration water pump and cooling water pump is constructed respectively;

[0045] Construct a cooling tower energy consumption model based on the cooling tower fan frequency;

[0046] The terminal air-conditioning energy consumption model is constructed based on the cooling capacity of the air-conditioning terminal and the energy efficiency ratio of the terminal system.

[0047] As a preferred embodiment of the present invention, when obtaining the correlation between model data sequences, it includes:

[0048] Obtain the first and second model data sequences, as shown in Formula 21 and Formula 22:

[0049] (twenty one);

[0050] (twenty two);

[0051] In the formula, is the first cooling and heating load prediction model data sequence, is the second cooling and heating load prediction model data sequence, , The first and second model data sequences , middle , The value of the moment;

[0052] The shortest dynamic regularization path of the first and second model data sequences is obtained, as shown in Formula 23:

[0053] (twenty three);

[0054] In the formula, The form is , the shortest dynamic integration path from Start to Finish.

[0055] As a preferred implementation mode of the present invention, after obtaining the shortest dynamic regularization path, it includes:

[0056] The similarity between the first and second model data sequences under the shortest dynamic regularization path is obtained, as shown in Formula 24:

[0057] (twenty four);

[0058] In the formula, is the similarity, and are the first and second model data sequences under the shortest dynamic regularization path.

[0059] As a preferred embodiment of the present invention, when merging the model data sequences, it includes:

[0060] The first and second model data sequences are treated as separate clusters, denoted as and ;

[0061] The similarity between the first and second model data sequences under the shortest dynamic regularization path is expressed as a matrix Represents that the matrix Elements in ;

[0062] The first and second model data sequences with similarity greater than the threshold are merged to obtain a new model data sequence. ;

[0063] Update the matrix according to the new model data sequence , , update the current number of clusters ;

[0064] The merging of model data sequences and the updating of the matrix and the current number of clusters are repeated until all model data sequences with similarities greater than a threshold are merged to obtain a model data sequence to be input.

[0065] A building cooling and heating load prediction system based on dynamic time warping and multiple models, comprising:

[0066] The first model building unit is used to build a temperature dynamic model of the heat medium in the heating pipeline of the building, a static heat balance model at the node, and a radiator power model, and combine them to obtain a heat load prediction model of the building;

[0067] The second model building unit is used to build a plurality of energy consumption models, and obtain a cooling load prediction model of the building based on the plurality of energy consumption models;

[0068] Data clustering unit: used to obtain the correlation between model data sequences by hierarchical clustering based on dynamic time warping, and merge the model data sequences to obtain the model data sequence to be input;

[0069] Load prediction unit: used for inputting the model data sequence to be input into the cold and hot load prediction model to obtain cold and hot load prediction results;

[0070] The energy consumption models include: a chiller energy consumption model, a refrigeration water pump energy consumption model, a cooling water pump energy consumption model, a cooling tower energy consumption model and a terminal air conditioning energy consumption model.

[0071] Compared with the prior art, the present invention has the following beneficial effects:

[0072] (1) The present invention fully reflects the transmission delay and transmission heat loss of the heat medium through the temperature dynamic model of the heat medium, and combines the heat power and heat load power provided by the heat source at the node obtained by the static heat balance model at the node, and combines the radiator power model, so as to fully consider the influence of transmission delay, transmission heat loss and radiator on heat load, and realize accurate prediction of heat load of public buildings;

[0073] (2) When constructing the temperature dynamic model of the heat medium, the present invention comprehensively considers multiple influencing factors (such as heat medium density, cross-sectional area of ​​the pipeline, length of the pipeline, mass flow rate of heat medium in the pipeline, etc.), accurately obtains the first coefficient and the second coefficient related to the pipeline and the transmission delay, and obtains the accurate heat loss coefficient of the pipeline based on the first coefficient and the second coefficient, which lays a foundation for the subsequent accurate prediction of the transmission delay and transmission heat loss of the heat medium;

[0074] (3) The heat medium temperature dynamic model provided by the present invention reflects the transmission delay and transmission heat loss of the heat medium. It fully combines the first and second coefficients and the heat loss coefficient obtained in the above steps with other influencing factors (such as pipeline ambient temperature, the heat medium temperature at the inlet and outlet of the water supply pipeline at a certain moment, etc.), so it can effectively reflect the transmission delay and transmission heat loss generated by public buildings in the actual heating process.

[0075] (4) The static heat balance model at the node provided by the present invention achieves accurate prediction of the heat power and heat load power provided by the heat source at the node by obtaining the temperature of the heat medium at the node in the water supply and return network and combining it with the specific heat capacity of the heat medium;

[0076] (5) The radiator power model provided by the present invention first obtains the equivalent heat capacity of the building, then obtains the node load of the heating pipeline within a period of time based on the equivalent heat capacity, and finally obtains the radiator power model by combining the radiator parameters of the building at the node. The model can fully reflect the impact of the radiator on the heat load of public buildings during the actual heating process;

[0077] (6) The main energy-consuming equipment of the existing public building air-conditioning system includes: chillers, cooling water pumps, refrigeration water pumps, cooling tower fans and terminal air conditioners. However, the existing building cooling load prediction model often only considers one or several of them, so the obtained cooling load prediction results are inaccurate and incomplete. The cooling load prediction model provided by the present invention fully considers the influence of the main energy-consuming equipment of the air-conditioning system, so that the obtained cooling load prediction results are both accurate and comprehensive.

[0078] (7) The present invention obtains the shortest dynamic regularization path between model data sequences, obtains the similarity between model data sequences under the shortest dynamic regularization path, and merges the model data sequences with similarity greater than a threshold to generate a model data sequence to be input, thereby eliminating the influence of model data sequences with large differences and improving the accuracy of cold and hot load prediction results.

[0079] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1 It is a step diagram of the building cooling and heating load prediction method based on dynamic time warping and multiple models provided by the present invention. DETAILED DESCRIPTION

[0081] The building cooling and heating load prediction method based on dynamic time warping and multiple models provided by the present invention is as follows: Figure 1 As shown, the following steps are included:

[0082] Step S1: construct a temperature dynamic model of the heat medium in the heating pipeline of the building, a static heat balance model at the node, and a radiator power model, and combine them to obtain a heat load prediction model of the building;

[0083] Step S2: constructing a plurality of energy consumption models, and obtaining a cooling load prediction model of the building based on the plurality of energy consumption models;

[0084] Step S3: using hierarchical clustering based on dynamic time warping to obtain the correlation between model data sequences, and merging the model data sequences to obtain the model data sequence to be input;

[0085] Step S4: inputting the model data sequence to be input into the cooling load prediction model and the heating load prediction model to obtain the cooling load prediction result and the heating load prediction result;

[0086] Among them, several energy consumption models include: chiller energy consumption model, refrigeration water pump energy consumption model, cooling water pump energy consumption model, cooling tower energy consumption model and terminal air conditioning energy consumption model.

[0087] Specifically, the present invention fully reflects the transmission delay and transmission heat loss of the heat medium through the temperature dynamic model of the heat medium, and combines the heat power and heat load power provided by the heat source at the node obtained by the static heat balance model at the node, as well as the radiator power model, thereby fully considering the transmission delay, transmission heat loss and the influence of the radiator on the heat load, and realizing accurate prediction of the heat load of public buildings.

[0088] In the above step S1, when constructing the temperature dynamic model of the heat medium, it includes:

[0089] According to the heat medium density, the cross-sectional area of ​​the pipeline, the length of the pipeline, and the heat medium mass flow rate of the pipeline, the pipeline The second coefficient related to the transmission delay , as shown in Formula 1:

[0090] (1);

[0091] In the formula, is the time interval in units of ; For pipeline The cross-sectional area in units of ; For pipeline The length of ; For pipeline The heat medium mass flow rate is in ; is the heat medium density, in units of , is the ceiling rounding function;

[0092] According to arrive Time flows through the pipeline The total mass of the heat medium in the pipeline The first coefficient related to the transmission delay , as shown in Formula 2 and Formula 3:

[0093] (2);

[0094] (3);

[0095] In the formula, for arrive Time flows through the pipeline The total mass of the heat medium is ;

[0096] According to the first coefficient and the second coefficient Get pipeline Heat loss coefficient , as shown in Formula 4:

[0097] (4);

[0098] In the formula, is the overall heat loss coefficient of the pipeline, in units of ; is the specific heat capacity of the heat medium.

[0099] Specifically, when constructing the temperature dynamic model of the heat medium, the present invention comprehensively considers multiple influencing factors (such as heat medium density, the cross-sectional area of ​​the pipeline, the length of the pipeline, the heat medium mass flow rate of the pipeline, etc.), accurately obtains the first coefficient and the second coefficient related to the pipeline and the transmission delay, and obtains the accurate heat loss coefficient of the pipeline based on the first coefficient and the second coefficient, which lays a foundation for the subsequent accurate prediction of the transmission delay and transmission heat loss of the heat medium.

[0100] Furthermore, after obtaining the pipeline Heat loss coefficient After, including:

[0101] According to the pipeline The first coefficient related to the transmission delay , the second coefficient And pipeline Heat loss coefficient , the transmission delay and transmission heat loss of the heat medium are obtained, as shown in Formula 5 and Formula 6:

[0102] (5);

[0103] (6);

[0104] In the formula, A collection of heating pipes for buildings. For pipeline The heat loss coefficient, for The pipeline ambient temperature at the moment, in °C; for The temperature of the heat medium at the inlet of the water supply pipe at the moment, in °C; for The heat medium temperature at the inlet of the return water pipe at the moment, in °C; When only transmission delay is considered without considering heat loss The heat medium temperature at the outlet of the water supply pipe at the moment, in °C; When only transmission delay is considered without considering heat loss The heat medium temperature at the outlet of the return water pipe at the moment, in °C; is the transmission delay of the heat medium, It is the heat loss of heat medium transmission;

[0105] According to the transmission delay and transmission heat loss of the heat medium, a temperature dynamic model of the heat medium is constructed, as shown in Formula 7:

[0106] (7);

[0107] In the formula, It is the temperature dynamic model of the heat medium.

[0108] Specifically, the heat medium temperature dynamic model provided by the present invention reflects the transmission delay and transmission heat loss of the heat medium, which fully combines the first and second coefficients and the heat loss coefficient obtained in the above steps with other influencing factors (such as pipeline ambient temperature, the heat medium temperature at the inlet and outlet of the water supply pipeline at a certain moment, etc.), so it can effectively reflect the transmission delay and transmission heat loss generated by public buildings in the actual heating process.

[0109] In the above step S1, when constructing the static thermal balance model at the node, it includes:

[0110] According to the heat medium mass flow rate of the pipeline, the heat medium temperature at the nodes in the water supply and return network is obtained, as shown in Formula 8:

[0111] (8);

[0112] In the formula, , Outflow and inflow nodes respectively A collection of pipelines, is the set of intersection nodes in the heating pipeline, and for Water supply and return pipes at all times The transmission heat loss, and for Nodes in the water supply and return network at all times The heat medium temperature at , in °C; for The temperature of the heat medium at the inlet of the water supply pipe at the moment, in °C; for The heat medium temperature at the inlet of the return water pipe at the moment, in °C;

[0113] Based on the specific heat capacity of the heat medium and the heat medium temperature at the nodes in the water supply and return networks, the heat power and heat load power provided by the heat source at the node are obtained, as shown in Formula 9 and Formula 10:

[0114] (9);

[0115] (10);

[0116] In the formula, for Time Node The heat power provided by the heat source at ; for Time Node The heat load power at ; is the specific heat capacity of the heat medium, is the set of source nodes in the heating pipeline, It is the collection of load nodes in the heating pipeline.

[0117] Specifically, the static heat balance model at the node provided by the present invention achieves accurate prediction of the heat power and heat load power provided by the heat source at the node by obtaining the heat medium temperature at the node in the water supply and return network and combining the specific heat capacity of the heat medium.

[0118] In the above step S1, when constructing the heat sink power model, it includes:

[0119] According to the indoor and outdoor temperature, equivalent thermal resistance and heat load power of the building, the equivalent heat capacity of the building is obtained, as shown in formula 11:

[0120] (11);

[0121] In the formula, is the equivalent heat capacity of the building, in units of ; is the equivalent resistance of the building, in units of ; is the outdoor temperature of the building, in ; is the indoor temperature of the building, in ; is the heat load power, in units of ;

[0122] Based on the equivalent heat capacity of the building, the node load of the heating pipeline within a period of time is obtained, as shown in Formula 12:

[0123] (12);

[0124] In the formula, for The outdoor temperature of the building at the time, in ; for The indoor temperature of the building at the time, in ; for The heat load power at the moment, in units of ; is the interval time, is a natural constant, is the set of load nodes in the heating pipeline, For interval time Nodes of internal heating pipes load;

[0125] According to the node load and node The radiator parameters of the building are used to obtain the radiator power model, as shown in Formula 13:

[0126] (13);

[0127] In the formula, and For Node The first and second radiator parameters of the building, and for Nodes in the water supply and return network at all times The heat medium temperature at ; is the heat sink power model.

[0128] Specifically, the radiator power model provided by the present invention first obtains the equivalent heat capacity of the building, and then obtains the node load of the heating pipeline within a time period based on the equivalent heat capacity, and finally combines the radiator parameters of the building at the node to obtain the radiator power model. This model can fully reflect the impact of the radiator on the heat load of public buildings during the actual heating process.

[0129] Specifically, the heat load prediction model is a combination model of a temperature dynamic model of a heat medium, a static heat balance model at a node, and a radiator power model.

[0130] In the above step S2, when constructing the chiller energy consumption model, it includes:

[0131] According to the chiller load rate and chiller cooling water outlet temperature, the chiller energy consumption model is constructed, as shown in Formula 14:

[0132] (14);

[0133] In the formula, is the outlet temperature of cooling water of the chiller, in ; is the water unit load rate, in units of ; to is the energy consumption model parameter of the chiller, is the energy consumption of the chiller.

[0134] In the above step S2, when constructing the energy consumption model of the chilled water pump and the energy consumption model of the cooling water pump, it includes:

[0135] According to the chilled water pump frequency and the cooling water pump frequency, the chilled water pump energy consumption model and the cooling water pump energy consumption model are constructed respectively, as shown in Formula 15 and Formula 16:

[0136] (15);

[0137] (16);

[0138] In the formula, is the chilled water pump frequency, in ; is the cooling water pump frequency, in ; to is the chilled water pump energy consumption model parameter, to is the cooling water pump energy consumption model parameter, is the energy consumption of the chilled water pump, is the energy consumption of cooling water pump.

[0139] In the above step S2, when constructing the cooling tower energy consumption model and the terminal air conditioning energy consumption model, it includes:

[0140] The cooling tower energy consumption model is constructed according to the cooling tower fan frequency, as shown in Formula 17:

[0141] (17);

[0142] In the formula, is the cooling tower fan frequency, in , to is the cooling tower energy consumption model parameter, is the energy consumption of cooling tower;

[0143] The terminal air conditioning energy consumption model is constructed according to the cooling capacity of the air conditioning terminal and the energy efficiency ratio of the terminal system, as shown in Formula 18 and Formula 19:

[0144] (18);

[0145] (19);

[0146] In the formula, is the cooling capacity of the air conditioner terminal, in units of ; is the energy efficiency ratio of the terminal system; , are the mixing ratios of fresh air and return air, is the energy consumption of terminal air conditioning.

[0147] Specifically, the parameters to Based on the historical actual chiller cooling water outlet temperature and historical actual water unit load rate, the least squares method was used for data fitting and identification.

[0148] parameter to Based on the actual historical chilled water pump frequency, the least square method was used to fit the data and identify it.

[0149] parameter to Based on the historical actual cooling water pump frequency, the least square method was used for data fitting and identification.

[0150] parameter to Based on the historical actual cooling tower fan frequency, the least squares method was used for data fitting and identification.

[0151] In the above step S2, when obtaining the cooling load prediction model of the building, it includes:

[0152] Based on the energy consumption model of the chiller, the energy consumption model of the chilled water pump, the energy consumption model of the cooling water pump, the energy consumption model of the cooling tower, and the energy consumption model of the terminal air conditioner, the cooling load prediction model is obtained, as shown in Formula 20:

[0153] (20);

[0154] In the formula, For the Energy consumption of chillers, For the Energy consumption of chilled water pumps, For the Energy consumption of cooling water pumps, For the Energy consumption of cooling tower For the Energy consumption of terminal air conditioner, , , , , They are the number of chillers, chilled water pumps, cooling water pumps, cooling towers and terminal air conditioners in operation. is the overall energy consumption of the air conditioning system.

[0155] Specifically, the main energy-consuming equipment of the existing public building air-conditioning system includes: chillers, cooling water pumps, refrigeration water pumps, cooling tower fans and terminal air conditioners. However, the cooling load prediction model of existing buildings often only considers one or several of them, so the obtained cooling load prediction results are inaccurate and incomplete. The cooling load prediction model provided by the present invention fully considers the influence of the main energy-consuming equipment of the air-conditioning system, so that the obtained cooling load prediction results are both accurate and comprehensive.

[0156] In the above step S3, when obtaining the correlation between the model data sequences, it includes:

[0157] Obtain the first and second model data sequences, as shown in Formula 21 and Formula 22:

[0158] (twenty one);

[0159] (twenty two);

[0160] In the formula, is the first heating load prediction model data sequence or the first cooling load prediction model data sequence, is a second heating load prediction model data sequence or a second cooling load prediction model data sequence, is the first model data sequence middle The value of the moment, The second model data series middle The value of the moment;

[0161] The shortest dynamic regularization path of the first and second model data sequences is obtained, as shown in Formula 23:

[0162] (twenty three);

[0163] In the formula, The form is , the shortest dynamic integration path from Start to Finish;

[0164] Among them, the direction of the shortest dynamically regularized path must be from the first point in the lower left corner to the last point in the upper right corner; the midpoint of the shortest dynamically regularized path can only be matched with adjacent points; the shortest dynamically regularized path can only be carried out in chronological order.

[0165] Specifically, when obtaining the shortest dynamic regularized path of the first and second model data sequences, it includes: obtaining the shortest dynamic regularized path of the first heat load prediction model data sequence and the second heat load prediction model data sequence; obtaining the shortest dynamic regularized path of the first cooling load prediction model data sequence and the second cooling load prediction model data sequence.

[0166] Furthermore, after obtaining the shortest dynamic regularization path, it includes:

[0167] The similarity between the first and second model data sequences under the shortest dynamic regularization path is obtained, as shown in Formula 24:

[0168] (twenty four);

[0169] In the formula, is the similarity, and are the first and second model data sequences under the shortest dynamic regularization path.

[0170] In the above step S3, when merging the model data sequences, it includes:

[0171] The first and second model data sequences are treated as separate clusters, denoted as and ;

[0172] The similarity between the first and second model data sequences under the shortest dynamic regularization path is expressed as a matrix Represents that the matrix Elements in ;

[0173] The first model data sequence and the second model data sequence whose similarity is greater than a threshold are merged to obtain a new model data sequence. ;

[0174] According to the new model data sequence, the number of the following clusters is moved forward by one, and the matrix is ​​deleted. No. Row and Column, update matrix , , update the current number of clusters ;

[0175] The merging of model data sequences and the updating of the matrix and the current number of clusters are repeated until all model data sequences with similarities greater than a threshold are merged to obtain a model data sequence to be input.

[0176] Specifically, when the present invention predicts the cold and hot loads, the model data sequences used are model data under different time series, and there are certain differences between the model data under these different time series. If the model data sequences with large differences are directly used for the cold and hot load predictions, it will affect the accuracy of the cold and hot load prediction results. The present invention obtains the shortest dynamic regularization path between the model data sequences, obtains the similarity between the model data sequences under the shortest dynamic regularization path, and merges the model data sequences with similarities greater than a threshold to generate a model data sequence to be input, thereby eliminating the influence of the model data sequences with large differences and improving the accuracy of the cold and hot load prediction results.

[0177] The building cooling and heating load prediction system based on dynamic time warping and multiple models provided by the present invention includes: a first model construction unit, a second model construction unit, a data clustering unit, and a load prediction unit.

[0178] The first model building unit is used to build a temperature dynamic model of the heat medium in the heating pipeline of the building, a static heat balance model at the node, and a radiator power model, and combine them to obtain a heat load prediction model of the building;

[0179] The second model building unit is used to build a plurality of energy consumption models, and obtain a cooling load prediction model of the building based on the plurality of energy consumption models;

[0180] Data clustering unit: used to obtain the correlation between model data sequences by hierarchical clustering based on dynamic time warping, and merge the model data sequences to obtain the model data sequence to be input;

[0181] Load forecasting unit: used to input the model data sequence to be input into the cooling load forecasting model and the heating load forecasting model to obtain the cooling load forecasting result and the heating load forecasting result;

[0182] Among them, several energy consumption models include: chiller energy consumption model, refrigeration water pump energy consumption model, cooling water pump energy consumption model, cooling tower energy consumption model and terminal air conditioning energy consumption model.

[0183] The above-mentioned embodiments are only preferred embodiments of the present invention and cannot be used to limit the scope of protection of the present invention. Any non-substantial changes and substitutions made by technicians in this field on the basis of the present invention shall fall within the scope of protection required by the present invention.

Claims

1. A building cooling and heating load prediction method based on dynamic time warping and multiple models, characterized in that: The following steps are involved: Construct a dynamic temperature model of the heat medium in the building's heating pipes, a static heat balance model at the nodes, and a radiator power model, and combine them to obtain a heat load prediction model for the building; Constructing several energy consumption models, and obtaining a cooling load prediction model of the building based on the several energy consumption models; Hierarchical clustering based on dynamic time warping is used to obtain the correlation between model data sequences, and the model data sequences are merged to obtain the model data sequence to be input; Inputting the model data sequence to be input into the cold and hot load prediction model to obtain cold and hot load prediction results; The energy consumption models include: a chiller energy consumption model, a refrigeration water pump energy consumption model, a cooling water pump energy consumption model, a cooling tower energy consumption model and a terminal air conditioning energy consumption model; When constructing the temperature dynamic model of the heat medium, it includes: According to the heat medium density, the cross-sectional area of ​​the pipeline, the length of the pipeline, and the heat medium mass flow rate of the pipeline, the second coefficient related to the pipeline k and the transmission delay is obtained. As shown in Formula 1: Where Δy is the time interval, is the cross-sectional area of ​​the pipe k, is the length of the pipeline k, is the heat medium mass flow rate in pipeline k, ρ e is the heat medium density, is the ceiling function, is the second coefficient related to the transmission delay of pipeline k; According to The total mass of heat medium flowing through pipe k at time y, and the first coefficient related to pipe k and transmission delay is obtained As shown in Formula 2 and Formula 3: In the formula, for The total mass of heat medium flowing through pipe k at time y is: is the first coefficient related to the transmission delay of pipeline k; According to the first coefficient and the second coefficient Get the heat loss coefficient of pipeline k In order to obtain the heat loss coefficient of the pipeline k After that, it also includes: According to the first coefficient of pipeline k related to the transmission delay The second coefficient and the heat loss coefficient of the pipeline k Get the transmission delay C and transmission heat loss R of the heat medium; According to the transmission delay C and transmission heat loss R of the heat medium, the temperature dynamic model of the heat medium is constructed, as shown in Formula 7: D = -CR (7); In the formula, D is the temperature dynamic model of the heat medium; When constructing a static heat balance model at a node, include: According to the heat medium mass flow rate of the pipeline, the heat medium temperature at the nodes in the water supply and return network is obtained; Based on the specific heat capacity of the heat medium and the heat medium temperature at the nodes in the water supply and return networks, the heat power and heat load power provided by the heat source at the node are obtained, as shown in Formula 9 and Formula 10: In the formula, is the thermal power provided by the heat source at node l at time y, is the heat load power at node l at time y, v e is the specific heat capacity of the heat medium, Θ dm is the set of source nodes in the heating pipeline, Θ am is the set of load nodes in the heating pipeline, and are the heat medium temperatures at node l in the water supply and return networks at time y, are the sets of pipes flowing out of and into node l respectively; When building a heat sink power model, include: According to the indoor and outdoor temperature, equivalent thermal resistance and heat load power of the building, the equivalent heat capacity V of the building is obtained, and the node load of the heating pipeline in a period of time is obtained, as shown in Formula 12: In the formula, is the outdoor temperature of the building at time y-1, is the indoor temperature of the building at time y-1, is the heat load power at time y-1, Δy is the interval time, e is the natural constant, Θ am is the set of load nodes in the heating pipeline, R d is the equivalent resistance of the building, is the node l load of the heating pipeline at time y; According to the node load of the heating pipeline in a period of time and the radiator parameters of the building at node l, the radiator power model is obtained, as shown in formula 13: In the formula, and are the first and second radiator parameters of the building at node l, and are the heat medium temperatures at node l in the water supply and return networks at time y, is the heat load power at node l at time y; When building several energy consumption models, including: Construct the energy consumption model of the chiller according to the chiller load rate and the chiller cooling water outlet temperature; According to the frequency of refrigeration water pump and cooling water pump, energy consumption model of refrigeration water pump and cooling water pump is constructed respectively; Construct a cooling tower energy consumption model based on the cooling tower fan frequency; The terminal air-conditioning energy consumption model is constructed based on the cooling capacity of the air-conditioning terminal and the energy efficiency ratio of the terminal system.

2. The method for predicting building cooling and heating loads based on dynamic time warping and multiple models according to claim 1 is characterized in that: When obtaining the correlation between model data series, include: Obtain the first and second model data sequences, as shown in Formula 21 and Formula 22: Where Q1 is the first cooling and heating load prediction model data sequence, Q2 is the second cooling and heating load prediction model data sequence, are the first and second model data sequences Q1 and Q2 respectively. m ,t z The value of the moment; The shortest dynamic regularization path of the first and second model data sequences is obtained, as shown in Formula 23: In the formula, e j The form is (o,k), and the shortest dynamic integration path is Start to Finish.

3. The method for predicting building cooling and heating loads based on dynamic time warping and multiple models according to claim 2 is characterized in that: After obtaining the shortest dynamically regularized path, it includes: The similarity between the first and second model data sequences under the shortest dynamic regularization path is obtained, as shown in Formula 24: In the formula, is the similarity, and They are respectively the first and second model data sequences under the shortest dynamic regularization path.

4. The method for predicting building cooling and heating loads based on dynamic time warping and multiple models according to claim 3 is characterized in that: When merging model data sequences, include: The first and second model data sequences are treated as separate clusters, denoted as V o ={c o } and V k ={c k }; The similarity between the first and second model data sequences under the shortest dynamic regularization path is represented by a matrix Z, where the element z in the matrix Z is ok =F(c o ,c k ); The first and second model data sequences with similarity greater than the threshold are merged to obtain a new model data sequence, V o ′=V o +V k ; Update the matrix z according to the new model data sequence ok =F(c o ,c k ), k = 1, 2, ..., w-1, update the current number of clusters w′ = w-1; The merging of model data sequences and the updating of the matrix and the current number of clusters are repeated until all model data sequences with similarities greater than a threshold are merged to obtain a model data sequence to be input.

5. A building cooling and heating load prediction system based on dynamic time warping and multiple models, characterized in that: include: The first model building unit is used to build a temperature dynamic model of the heat medium in the heating pipeline of the building, a static heat balance model at the node, and a radiator power model, and combine them to obtain a heat load prediction model of the building; The second model building unit is used to build a plurality of energy consumption models, and obtain a cooling load prediction model of the building based on the plurality of energy consumption models; Data clustering unit: used to obtain the correlation between model data sequences by hierarchical clustering based on dynamic time warping, and merge the model data sequences to obtain the model data sequence to be input; Load prediction unit: used for inputting the model data sequence to be input into the cold and hot load prediction model to obtain cold and hot load prediction results; The energy consumption models include: a chiller energy consumption model, a refrigeration water pump energy consumption model, a cooling water pump energy consumption model, a cooling tower energy consumption model and a terminal air conditioning energy consumption model; When constructing the temperature dynamic model of the heat medium, it includes: According to the heat medium density, the cross-sectional area of ​​the pipeline, the length of the pipeline, and the heat medium mass flow rate of the pipeline, the second coefficient related to the pipeline k and the transmission delay is obtained. As shown in Formula 1: Where Δy is the time interval, is the cross-sectional area of ​​the pipe k, is the length of the pipeline k, is the heat medium mass flow rate in pipeline k, ρ e is the heat medium density, is the ceiling function, is the second coefficient related to the transmission delay of pipeline k; According to The total mass of heat medium flowing through pipe k at time y, and the first coefficient related to pipe k and transmission delay is obtained As shown in Formula 2 and Formula 3: In the formula, for The total mass of heat medium flowing through pipe k at time y is: is the first coefficient related to the transmission delay of pipeline k; According to the first coefficient and the second coefficient Get the heat loss coefficient of pipeline k In order to obtain the heat loss coefficient of the pipeline k After that, it also includes: According to the first coefficient of pipeline k related to the transmission delay The second coefficient and the heat loss coefficient of the pipeline k Get the transmission delay C and transmission heat loss R of the heat medium; According to the transmission delay C and transmission heat loss R of the heat medium, the temperature dynamic model of the heat medium is constructed, as shown in Formula 7: D = -CR (7); In the formula, D is the temperature dynamic model of the heat medium; When constructing a static heat balance model at a node, include: According to the heat medium mass flow rate of the pipeline, the heat medium temperature at the nodes in the water supply and return network is obtained; Based on the specific heat capacity of the heat medium and the heat medium temperature at the nodes in the water supply and return networks, the heat power and heat load power provided by the heat source at the node are obtained, as shown in Formula 9 and Formula 10: In the formula, is the thermal power provided by the heat source at node l at time y, is the heat load power at node l at time y, v e is the specific heat capacity of the heat medium, Θ dm is the set of source nodes in the heating pipeline, Θ am is the set of load nodes in the heating pipeline, and are the heat medium temperatures at node l in the water supply and return networks at time y, are the sets of pipes flowing out of and into node l respectively; When building a heat sink power model, include: According to the indoor and outdoor temperature, equivalent thermal resistance and heat load power of the building, the equivalent heat capacity V of the building is obtained, and the node load of the heating pipeline in a period of time is obtained, as shown in Formula 12: In the formula, is the outdoor temperature of the building at time y-1, is the indoor temperature of the building at time y-1, is the heat load power at time y-1, Δy is the interval time, e is the natural constant, Θ am is the set of load nodes in the heating pipeline, R d is the equivalent resistance of the building, is the node l load of the heating pipeline at time y; According to the node load of the heating pipeline in a period of time and the radiator parameters of the building at node l, the radiator power model is obtained, as shown in formula 13: In the formula, and are the first and second radiator parameters of the building at node l, and are the heat medium temperatures at node l in the water supply and return networks at time y, is the heat load power at node l at time y; When building several energy consumption models, including: Construct the energy consumption model of the chiller according to the chiller load rate and the chiller cooling water outlet temperature; According to the frequency of refrigeration water pump and cooling water pump, energy consumption model of refrigeration water pump and cooling water pump is constructed respectively; Construct a cooling tower energy consumption model based on the cooling tower fan frequency; The terminal air-conditioning energy consumption model is constructed based on the cooling capacity of the air-conditioning terminal and the energy efficiency ratio of the terminal system.

Images