Multi-objective optimization control method, training method and system of building radiation energy supply system

Abstract

The invention provides a multi-objective optimization control method, a training method and a system for a building radiation energy supply system, which can be applied to the field of temperature control technology and building energy supply system optimization control. The method comprises the steps of obtaining historical control sequence information, historical environmental factor information and disturbance parameter information of a building radiation energy supply system for a target building corresponding to a target time period in response to a received energy supply demand for the target time period; taking the lowest energy supply cost of the building radiation energy supply system and the highest indoor temperature control precision as target functions, processing historical control sequence information, historical environment factor information and disturbance parameter information of the building radiation energy supply system based on target constraint conditions, and generating target control sequence information of the building radiation energy supply system; and based on the target control sequence information, operating parameters of the building radiation energy supply system are controlled, so that the energy supply cost is the lowest while the target building indoor temperature reaches the target threshold value.

Description

Technical Field

[0001] The present disclosure relates to the field of temperature control technology, and in particular to a multi-objective optimization control method, a training method and a device for a building radiation energy supply system. Background Art

[0002] The embedded pipe energy supply system embeds heat-carrying fluid pipes inside the building's envelope structure and uses these pipes to store and transfer energy, thereby achieving control of the building's indoor environment. In related technologies, the control of the building's indoor environment is mainly achieved through rule-based control methods. However, since rule-based control methods rely on preset rules and schedules to operate building systems, they cannot adapt to rapidly changing environments or complex system requirements, resulting in low accuracy and efficiency in controlling the building's indoor temperature. Summary of the invention

[0003] In view of the above problems, the present disclosure provides a multi-objective optimization control method, training method, device and equipment for a building radiation energy supply system.

[0004] According to a first aspect of the present disclosure, a multi-objective optimization control method for a building radiant energy supply system is provided, comprising: in response to receiving an energy supply demand for a target time period, obtaining historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiant energy supply system for a target building corresponding to the target time period; taking the lowest energy supply cost of the building radiant energy supply system and the highest indoor temperature control accuracy as the objective function, based on objective constraints, processing the historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiant energy supply system to generate target control sequence information of the building radiant energy supply system; based on the target control sequence information, controlling the operating parameters of the building radiant energy supply system so that the indoor temperature of the target building reaches a target threshold while minimizing the energy supply cost; wherein the target threshold is determined according to user demand, and the user demand includes energy supply demand and temperature demand.

[0005] A second aspect of the present disclosure provides a training method for a target model of a building radiation energy supply system, characterized in that the target model includes three layers of long short-term memory network layers, two layers of random drop layers and one layer of full connection layer, and the method includes: obtaining a target feature target pair data set; using a first layer of long short-term memory network to process the target feature target pair data set, and outputting a sample first time series feature target pair sequence; using a first layer of random drop layer to process the sample first time series feature target pair sequence, and outputting a sample first time series feature target pair sequence; using a second layer of long short-term memory network layer to process the sample first time series feature parameter information Processing, output the second time series feature parameter information of the sample; using the second random discard layer to process the second time series feature parameter information of the sample, output the second target time series feature parameter information of the sample; using the third long short-term memory network layer to process the second target time series feature parameter information of the sample, output the third time series feature parameter information of the sample; using the fully connected layer to process the third time series feature parameter information of the sample, output indoor temperature prediction information and energy consumption information; according to the indoor temperature prediction result, energy consumption information, historical control sequence information, historical environmental factor information and disturbance parameter information, train the target model to obtain the trained target model.

[0006] A third aspect of the present disclosure provides a multi-objective optimization control device for a building radiation energy supply system, comprising: a first acquisition module, a first processing module and a control module.

[0007] The first acquisition module is used to obtain historical control sequence information, historical environmental factor information and disturbance parameter information of a building radiation energy supply system for a target building corresponding to the target period in response to receiving an energy supply demand for the target period.

[0008] The first processing module is used to process the historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiant energy supply system based on the target constraints, taking the lowest energy supply cost of the building radiant energy supply system and the highest indoor temperature control accuracy as the objective function, and generate the target control sequence information of the building radiant energy supply system.

[0009] The control module is used to control the operating parameters of the building radiation energy supply system based on the target control sequence information so that the indoor temperature of the target building reaches the target threshold while the energy supply cost is minimized; wherein the target threshold is determined according to user needs, and the user needs include temperature needs and energy supply needs.

[0010] A fourth aspect of the present disclosure provides a training device for a target model of a building radiation energy supply system, wherein the target model includes three long short-term memory network layers, two random dropout layers, and one fully connected layer, and the device includes: a second acquisition module, a second processing module, a third processing module, a fourth processing module, a fifth processing module, a sixth processing module, a seventh processing module, and a training module.

[0011] The second acquisition module is used to acquire a target feature target pair data set.

[0012] The second processing module is used to process the target feature target pair data set using the first layer of long short-term memory network, and output the sample first time series feature target pair sequence.

[0013] The third processing module is used to process the sample first time series feature target pair sequence by using the first random discard layer, and output the sample first target time series feature target pair sequence.

[0014] The fourth processing module is used to process the first target temporal characteristic parameter information of the sample by using the second long short-term memory network layer, and output the second temporal characteristic parameter information of the sample.

[0015] The fifth processing module is used to process the second time series characteristic parameter information of the sample by using the second random discarding layer, and output the second target time series characteristic parameter information of the sample.

[0016] The sixth processing module is used to process the second target time series feature parameter information of the sample using the third long short-term memory network layer, and output the third time series feature parameter information of the sample.

[0017] The seventh processing module is used to use the fully connected layer to process the third target time series characteristic parameter information of the sample, and output indoor temperature prediction information and energy consumption information.

[0018] The training module is used to train the target model according to indoor temperature prediction information, energy consumption information, historical control sequence information, historical environmental factor information and disturbance parameter information to obtain the trained target model.

[0019] A fifth aspect of the present disclosure provides an electronic device, comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.

[0020] According to the multi-objective optimization control method, training method and device of the building radiant energy supply system provided by the present invention, by taking the lowest energy supply cost of the building radiant energy supply system and the highest indoor temperature control accuracy as the objective function, based on the target constraint conditions, the acquired target time period and the historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiant energy supply system corresponding to the target time period for the target building are processed, so that the target control sequence information of the building radiant energy supply system can be generated, and then the operating parameters of the building radiant energy supply system can be controlled based on the target control sequence information so that the indoor temperature of the target building reaches the target threshold while minimizing the energy supply cost, thereby improving the control accuracy and control efficiency of the indoor temperature of the target building, thereby improving the energy utilization efficiency, reducing resource waste, and improving the indoor environmental quality of the building, while comprehensively considering factors such as changes in electricity prices and user behavior, thereby improving energy efficiency, reducing energy supply costs, and improving user comfort. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The above contents and other purposes, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0022] Figure 1 A flowchart schematically shows a multi-objective optimization control method for a building radiation energy supply system according to an embodiment of the present disclosure;

[0023] Figure 2 A schematic diagram of a long short-term memory network layer according to an embodiment of the present disclosure is schematically shown;

[0024] Figure 3 A system diagram of a target model according to an embodiment of the present disclosure is schematically shown;

[0025] Figure 4 A schematic diagram showing a comparison of the effects of a control method in the related art and a control method based on MPC;

[0026] Figure 5 A flowchart schematically shows a method for training a target model of a building radiant energy supply system according to an embodiment of the present disclosure;

[0027] Figure 6 A structural block diagram schematically shows a multi-objective optimization control device for a building radiation energy supply system according to an embodiment of the present disclosure;

[0028] Figure 7 A structural block diagram schematically shows a training device for a target model of a building radiation energy supply system according to an embodiment of the present disclosure; and

[0029] Figure 8A block diagram of an electronic device suitable for implementing a multi-objective optimization control method for a building radiation energy supply system according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0030] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0031] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0032] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0033] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0034] In the technical solution of the present disclosure, the user information (including but not limited to user personal information, user image information, user device information, such as location information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with relevant laws, regulations and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0035] The energy supply terminal of the building based on the embedded heat-carrying fluid pipeline has a large energy storage potential, and the heat storage / release of the enclosure structure is adjustable, which is conducive to the implementation of flexible energy use in the building. The maximum energy supply potential of the energy supply terminal is determined by its thermal parameters, while its actual heating / cooling capacity is determined by the control method. Due to the large randomness of the outdoor temperature of the building and the volatility caused by occupancy, the actual heating / cooling capacity often deviates from the target during the operation of the building.

[0036] In the process of realizing the present disclosure, it is found that the related technologies mainly use the rule-based control RBC (Rule-Based Control) method to control the indoor temperature of the building. The rule-based control relies on preset rules and schedules to operate the building system, including timing control, temperature setting and manual adjustment, etc. Its advantage is that it is simple and easy to implement, but it often cannot adapt to rapidly changing environments or complex system requirements. And the rule-based control methods, such as on / off and PID control methods, although simple to implement, are not ideal for the regulation effect of embedded tubular building systems with large energy storage. Advanced control methods for large energy flexible terminals and high dynamic energy supply systems need to be studied in depth.

[0037] In view of this, an embodiment of the present disclosure provides a multi-objective optimization control method for a building radiant energy supply system, comprising: in response to receiving an energy supply demand for a target time period, obtaining historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiant energy supply system for a target building corresponding to the target time period; taking the lowest energy supply cost of the building radiant energy supply system and the highest indoor temperature control accuracy as the objective function, based on objective constraints, processing the historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiant energy supply system to generate target control sequence information of the building radiant energy supply system; based on the target control sequence information, controlling the operating parameters of the building radiant energy supply system so that the indoor temperature of the target building reaches a target threshold while minimizing the energy supply cost; wherein the target threshold is determined according to user needs, and the user needs include energy supply needs and temperature needs.

[0038] Figure 1 The flowchart of the multi-objective optimization control method of the building radiant energy supply system according to the embodiment of the present disclosure is schematically shown.

[0039] like Figure 1 As shown, the multi-objective optimization control method 100 of the building radiant energy supply system of this embodiment includes operations S110 to S130.

[0040] In operation S110 , in response to receiving an energy supply demand for a target period, historical control sequence information, historical environmental factor information, and disturbance parameter information of a building radiant energy supply system for a target building corresponding to the target period are acquired.

[0041] In operation S120, taking the lowest energy supply cost of the building radiant energy supply system and the highest indoor temperature control accuracy as the objective function, based on the target constraints, the historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiant energy supply system are processed to generate the target control sequence information of the building radiant energy supply system.

[0042] In operation S130, based on the target control sequence information, the operating parameters of the building radiant energy supply system are controlled so that the target building indoor temperature reaches the target threshold while minimizing the energy supply cost.

[0043] According to an embodiment of the present disclosure, user demand may include energy demand and temperature demand, and the temperature demand may characterize the temperature that the target building is expected to reach. Energy demand may include cooling demand and heating demand. The building radiation energy supply system of the target building may characterize a system in which water pipes with chilled water / hot water are embedded inside the enclosure structure of the target building. The target building may be heated / cooled based on the building radiation energy supply system. The target period may characterize a specific period of time. For example, the target period may be a quarter. For another example, the target period may be a month.

[0044] According to an embodiment of the present disclosure, the historical control sequence information may characterize the control sequence of the building radiation energy supply system in the target period. The historical environmental factor information may characterize the environmental factors of the target building in the target period, and the environmental factors may include the historical indoor temperature, historical outdoor temperature and solar radiation of the target building. The disturbance parameter information may characterize the thermal disturbance caused by the occupancy of people in the target building.

[0045] According to an embodiment of the present disclosure, historical control sequence information, historical environmental factor information, and disturbance parameter information of the building radiation energy supply system for the target building corresponding to the target period can be obtained according to the energy supply demand of the target period. For example, when the target period is the second month of the fourth quarter, historical control sequence information, historical environmental factor information, and disturbance parameter information of the building radiation energy supply system of the target building in the same period of the last three years can be obtained.

[0046] According to the embodiments of the present disclosure, the energy supply of the building radiation energy supply system in the target period can be taken as heating as an example, and the energy supply cost of the building radiation energy supply system is the lowest and the indoor temperature control accuracy is the highest as the objective function. As shown in formula (1).

[0047] (1)

[0048] Wherein, N represents the number of moments included in the target period; Indicates the maximum thermal power of the heat pump; express The coefficient of performance of the heat pump at the moment; express The target control signal input at all times; express Electricity price at any time; express The measured indoor temperature at the moment; express Target temperature at the moment; represents the first weight; Represents the second weight.

[0049] According to an embodiment of the present disclosure, the performance coefficient of the heat pump may be affected by the environment. The target control signal may represent the target control signal of multiple time steps in the target control sequence. When the time step is infinitely small, one time step may be considered as one moment. The target control signal may be used to control the valve opening of the electric valve.

[0050] According to an embodiment of the present disclosure, the target constraint condition may include a state vector constraint condition, an output constraint condition, a target control signal constraint condition, a target control signal change rate constraint condition, a temperature constraint condition, and a temperature change rate constraint condition. The state vector constraint condition is as follows: formula (2), the output constraint condition is as follows: formula (3), the target control signal constraint condition is as follows: formula (4), the target control signal change rate constraint condition is as follows: formula (5), the temperature constraint condition is as follows: formula (6), and the temperature change rate constraint condition is as follows: formula (7).

[0051] (2)

[0052] (3)

[0053] (4)

[0054] (5)

[0055] (6)

[0056] (7)

[0057] in, represents the state vector, represents the output vector, represents the perturbation vector, represents the first functional relationship, represents the second functional relationship, Indicates the lower limit of the input target control signal, Indicates the upper limit of the input target control signal, Indicates the lower limit of the input target control signal change rate, Indicates the upper limit of the input target control signal change rate, Indicates the lower limit of indoor temperature. Indicates the upper limit of indoor temperature. Indicates the lower limit of the rate of change of indoor temperature. Indicates the upper limit of the indoor temperature change rate.

[0058] According to the embodiments of the present disclosure, the target control sequence information can characterize the control signals of the energy supply equipment in the building radiation energy supply system at each moment in the target period. Taking the lowest energy supply cost of the building radiation energy supply system and the highest indoor temperature control accuracy as the objective function, based on the target constraint conditions, by processing the historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiation energy supply system, the target control sequence information of the building radiation energy supply system can be generated.

[0059] According to an embodiment of the present disclosure, the target threshold can be determined according to user needs. The operating parameters of the building radiation energy supply system can be controlled based on the target control sequence information, that is, the energy supply equipment can be controlled by the control signal of the energy supply equipment in the building radiation energy supply system at each time step of the target period, so as to control the operating parameters of the building radiation energy supply system so that the indoor temperature of the target building reaches the target threshold while minimizing the energy supply cost.

[0060] According to an embodiment of the present disclosure, by taking the lowest energy supply cost of the building radiant energy supply system and the highest indoor temperature control accuracy as the objective function, based on the target constraint conditions, the acquired target time period and the historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiant energy supply system corresponding to the target time period for the target building are processed, and the target control sequence information of the building radiant energy supply system can be generated. Then, based on the target control sequence information, the operating parameters of the building radiant energy supply system can be controlled so that the indoor temperature of the target building reaches the target threshold while minimizing the energy supply cost, thereby improving the control accuracy and control efficiency of the indoor temperature of the target building, thereby improving the energy utilization efficiency, reducing resource waste, and improving the indoor environmental quality of the building. At the same time, factors such as electricity price changes and user behavior are comprehensively considered, thereby improving energy efficiency, reducing energy supply costs, and improving user comfort.

[0061] According to an embodiment of the present disclosure, the lowest energy supply cost of the building radiant energy supply system and the highest indoor temperature control accuracy are taken as the objective function, and based on the target constraint conditions, the historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiant energy supply system are processed to generate the target control sequence information of the building radiant energy supply system, including: inputting the historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiant energy supply system into the target model, and outputting indoor temperature prediction information and energy consumption information; taking the lowest energy supply cost of the building radiant energy supply system and the highest indoor temperature control accuracy as the objective function, and based on the target constraint conditions, processing the indoor temperature prediction information to generate the target control sequence information of the building radiant energy supply system.

[0062] According to an embodiment of the present disclosure, the indoor temperature prediction information may characterize the indoor temperature of at least one time step in the target period. For example, when the target period is 24 hours and one time step is 2 hours, the indoor temperature prediction information may include the indoor temperature of 12 time steps.

[0063] According to an embodiment of the present disclosure, the energy consumption information may represent the energy required to be consumed when the target building reaches the indoor temperature prediction information. In this embodiment, the consumed energy may be electrical energy.

[0064] According to an embodiment of the present disclosure, the target model can be obtained by training the historical control sequence information, the historical environmental factor information and the disturbance parameter information with the historical control sequence information as a label. By inputting the historical control sequence information, the historical environmental factor information and the disturbance parameter information of the building radiation energy supply system into the target model, the target model can output the indoor temperature prediction information and the energy consumption information after processing the historical control sequence information, the historical environmental factor information and the disturbance parameter information. The indoor temperature prediction information can include the indoor temperature at multiple time steps in the future. The energy consumption information can also include multiple energy consumptions corresponding to the indoor temperature at multiple time steps.

[0065] According to an embodiment of the present disclosure, by inputting historical control sequence information, historical environmental factor information and disturbance parameter information into a target model and processing them, the target model can output indoor temperature prediction information and energy consumption information based on a period of time in the future, and by processing the indoor temperature prediction information and energy consumption information based on the objective function and objective constraints, a target control sequence for the building radiation function system can be generated, thereby improving the accuracy of the indoor temperature prediction information.

[0066] According to an embodiment of the present disclosure, historical control sequence information, historical environmental factor information and disturbance parameter information of a building radiation energy supply system are input into a target model, and indoor temperature prediction information and energy consumption information are output, including: using a first layer of long short-term memory network layer to process the historical control sequence information, historical environmental factor information and disturbance parameter information to obtain first time series characteristic parameter information; using a first layer of random discard layer to process the first time series characteristic parameter information to obtain first target time series characteristic parameter information; using a second layer of long short-term memory network layer to process the first target time series characteristic parameter information to obtain second time series characteristic parameter information; using a second layer of random discard layer to process the second time series characteristic parameter information to obtain second target time series characteristic parameter information; using a third layer of long short-term memory network layer to process the second target time series characteristic parameter information to obtain third time series characteristic parameter information; using a fully connected layer to process the third time series characteristic parameter information to obtain indoor temperature prediction information and energy consumption information.

[0067] According to an embodiment of the present disclosure, the target model may include three layers of long short-term memory network layers, two layers of random drop layers, and one fully connected layer. The three layers of long short-term memory network layers and the two layers of random drop layers exist in an alternating manner, and the fully connected layer is the last layer, that is, the structure of the target model is the first layer of long short-term memory network layer-the first layer of random drop layer-the second layer of long short-term memory network layer-the second layer of random drop layer-the third layer of long short-term memory network layer-the fully connected layer. The long short-term memory network layer is mainly composed of three gates, namely the forgetting gate, the input gate, and the output gate. The three gates jointly determine the transmission of information in neurons. The long short-term memory network layer can effectively select memory, update and output information through the gating mechanism of the three gates.

[0068] Figure 2 A schematic diagram of a long short-term memory network layer according to an embodiment of the present disclosure is schematically shown.

[0069] like Figure 2 The working principle of the long short-term memory network layer shown in Figure 8 is as follows.

[0070] (8)

[0071] in, represents the output of the forget gate, represents the Sigmoid activation function, represents the output of the previous time step, represents the output of the current time step, represents the input of the current time step, represents the weight of the forget gate, represents the weight of the input gate, represents the weight of the candidate cell state, The weight representing the weight of the output gate, represents the bias of the forget gate, represents the bias of the input gate, represents the bias of the candidate cell state, represents the bias of the output gate, represents the output of the input gate, represents the candidate cell state, represents the cell state at the previous time step, represents the cell state at the current time step, represents the output of the output gate, Represents the activation function.

[0072] According to an embodiment of the present disclosure, the output range of the Sigmoid activation function is . The output range of the activation function is .

[0073] According to the embodiments of the present disclosure, based on the Sigmoid activation function, the information flow can be smoothly controlled, and the output of the forget gate Determines the proportion of information from the previous time step that needs to be retained, the output of the forget gate The range is [0,1], 0 means completely discarded, 1 means completely retained. Similarly, the output of the input gate The range of is also [0,1], 0 means completely discarded, 1 means completely retained. The output of the output gate The range of is also [0,1], 0 means completely discarded, and 1 means completely retained. The long short-term memory network layer can effectively retain long-term historical dependency information at each time step through such a mechanism, solving the gradient vanishing problem of traditional recurrent neural networks (RNNs).

[0074] According to an embodiment of the present disclosure, the output of the current time step Using the Rectified Linear Unit (ReLU) as the activation function can make the output no longer limited to the range of [0,1]. The output of the current time step is Contains historical information and new information generated at the current time step, and the cell state at the current time step The difference is that the cell state at the current time step is It stores the long-term information accumulated by successive generations of cells, while the output of the previous time step Pay more attention to the short-term output of the current time step.

[0075] According to the embodiments of the present disclosure, the first layer of long short-term memory network layer can be used to process the historical control sequence information, the historical environmental factor information and the disturbance parameter information, so as to obtain the first time series characteristic parameter information. The first layer of random drop layer can be used to process the first time series characteristic parameter information output by the first layer of long short-term memory network layer to obtain the first target time series characteristic parameter information; the second layer of long short-term memory network layer can be used to process the first target time series characteristic parameter information output by the first layer of random drop layer to obtain the second time series characteristic parameter information. The second layer of random drop layer can be used to process the second time series characteristic parameter information output by the second layer of long short-term memory network layer to obtain the second target time series characteristic parameter information; the third layer of long short-term memory network layer can be used to process the second target time series characteristic parameter information output by the second layer of random drop layer to obtain the third time series characteristic parameter information. The third time series characteristic parameter information output by the third layer of long short-term memory network layer can be processed by the fully connected layer to obtain the indoor temperature prediction information and energy consumption information.

[0076] According to the embodiments of the present disclosure, the three-layer long short-term memory network layer can more efficiently capture deeper temporal features in time series data. The two-layer random dropout layer can randomly drop some neurons in each training batch to reduce the overfitting of the target model. Finally, the fully connected layer integrates the features extracted in the previous layers, thereby improving the accuracy of the final output indoor temperature prediction information and energy consumption information.

[0077] According to the embodiments of the present disclosure, a discrete genetic algorithm can be used to solve the objective function, thereby obtaining the target control sequence information. In the discrete genetic algorithm, the roulette method can be used for the selection of individuals. The following will describe in detail the process of using the discrete genetic algorithm to solve the objective function and obtain the target control sequence information.

[0078] According to an embodiment of the present disclosure, the lowest energy supply cost of the building radiant energy supply system and the highest indoor temperature control accuracy are taken as objective functions, and based on the objective constraint conditions, the indoor temperature prediction information is processed to generate the target control sequence information of the building radiant energy supply system, including: determining the objective function value of each individual in the k-th generation I individuals based on the indoor temperature prediction information, energy consumption information and the objective function; for each individual, determining the fitness of each individual according to the difference between the objective function value of each individual and the objective function value of I individuals; determining the probability of each individual in the k-th generation I individuals being selected according to the fitness of each individual and the sum of the fitness of I individuals; determining the cumulative probability according to the probability of the first i individuals being selected; determining the selected individual according to the cumulative probability; crossovering the selected individuals to generate the k+1-th generation I individuals; determining the target control sequence information of the building radiant energy supply system according to the k+1-th generation I individuals and the predetermined mutation probability.

[0079] According to an embodiment of the present disclosure, the objective function value of each individual in the kth generation I individuals can be determined based on the indoor temperature prediction information, the energy consumption information and the objective function, so that the maximum objective function value in the I individuals and the minimum objective function value in the I individuals can be determined. Wherein, I and k are both integers greater than or equal to 1.

[0080] According to an embodiment of the present disclosure, the objective function value difference of each individual represents the difference between the maximum objective function value of I individuals and the objective function value of each individual. The objective function value difference of I individuals represents the difference between the maximum objective function value and the minimum objective function value of I individuals. For each individual, the fitness of each individual can be determined according to the objective function value difference of each individual and the objective function value difference of I individuals, as shown in the following formula (9).

[0081] (9)

[0082] in, Indicates Daidi The fitness of an individual, Indicates Daidi Individuals, Indicates Daidi The objective function value of each individual, For the The maximum objective function value among the individuals, , I represents the total number of individuals; Indicates The minimum objective function value among I individuals, .

[0083] According to an embodiment of the present disclosure, the probability of each individual being selected among I individuals of the kth generation can be determined according to the fitness of each individual and the sum of the fitness of I individuals, as shown in the following formula (10).

[0084] (10)

[0085] in, Indicates Daidi The probability of an individual being selected.

[0086] According to an embodiment of the present disclosure, the cumulative probability can be determined according to the probability of the first i individuals being selected, as shown in the following formula (11).

[0087] (11)

[0088] in, represents the cumulative probability, Indicates The probability that the jth individual is selected in this generation.

[0089] According to the embodiments of the present disclosure, the selected individuals can be determined according to the cumulative probability. The individuals with the larger cumulative probability are determined as the selected individuals, and there can be multiple individuals selected. For example, two individuals with the larger cumulative probability can be determined as the selected individuals.

[0090] According to the embodiment of the present disclosure, the selected individuals may be crossovered to generate I individuals of the k+1th generation. The individuals may be crossovered to perform multi-point crossover, and the number of points of the multi-point crossover is as follows: Formula (12).

[0091] (12)

[0092] Among them, m represents the number of multi-point crossover points, α represents the first coefficient, and N represents the number of individual gene fragments.

[0093] According to an embodiment of the present disclosure, the first coefficient may be set based on experience. Generally, the value of the first coefficient may be 0.1 to 0.2.

[0094] According to an embodiment of the present disclosure, the selected individuals are crossed to generate I individuals of the k+1th generation. For example, two selected individuals are crossed to generate two individuals of the k+1th generation, as shown in the following formulas (13) and (14).

[0095] (13)

[0096] (14)

[0097] in, represents the first Generation (offspring) individual 1, represents the first Generation (offspring) individual 2, Indicates the first Generation (parent) individual 1, Indicates the first Generation (parent) individual 2, Indicates Daidi The individual gene fragments, .

[0098] According to an embodiment of the present disclosure, target control sequence information of a building radiation energy supply system is determined based on I individuals of the k+1th generation and a predetermined mutation probability, including: determining S gene fragments of I individuals of the k+1th generation and a predetermined mutation probability, (15).

[0099] (15)

[0100] in, is a random number uniformly distributed between 0 and 1. Represents the predetermined mutation probability.

[0101] According to an embodiment of the present disclosure, the target control sequence information of the building radiation energy supply system may include multiple target control signals. Daidi The individual The gene fragment can be a target control signal in the target control sequence information of the building radiation energy supply system. Therefore, when the target control sequence information of the building radiation energy supply system is determined based on the S gene fragments of the I individual of the k+1th generation, the target control sequence information of the building radiation energy supply system can be determined, that is, The S gene fragments of one individual of a generation can constitute the target control sequence information of the building radiation energy supply system.

[0102] According to an embodiment of the present disclosure, The population of the next generation (offspring) is as follows:

[0103] (16)

[0104] in, Indicates The population of a generation (offspring) includes all offspring individuals.

[0105] According to the embodiments of the present disclosure, The next iteration is carried out with the population of the next generation (offspring), so that the generation.

[0106] According to the embodiments of the present disclosure, a discrete genetic algorithm is used to solve the objective function, and through each population iteration, the accuracy of generating the target control sequence information of the building radiation energy supply system is improved, thereby improving the control accuracy of the indoor temperature.

[0107] Figure 3 A system diagram of a target model according to an embodiment of the present disclosure is schematically shown.

[0108] like Figure 3As shown, the MPC controller 310 and the target building 320. The MPC controller 310 is a device using a data-driven model predictive control (MPC) method. The historical control sequence information, historical environmental factor information, and disturbance parameter information of the target building 320 can be input into the target model for processing, so that the indoor temperature prediction information and energy consumption information can be obtained, and the indoor temperature prediction information and energy consumption information are input into the MPC controller 310. Based on the dynamic electricity price, constraint conditions, indoor temperature prediction information and energy consumption information, the objective function is used for optimization and solution, and the final target control sequence information can be obtained. The target control sequence information is used to control the operating parameters of the target building 320 so that the indoor temperature of the target building 320 reaches the target threshold while minimizing the energy consumption cost.

[0109] According to an embodiment of the present disclosure, the temperature control effect index can be characterized by the degree of violation of the indoor temperature and the absolute value deviation between the indoor temperature and the target threshold. The degree of violation of the indoor temperature can be calculated using the following formula (17). The absolute value deviation between the indoor temperature and the target threshold can be calculated using the following formula (18).

[0110] (17)

[0111] (18)

[0112] in, Indicates the degree of violation of indoor temperature, represents the actual measured value of the indoor temperature at time k, represents the target threshold of indoor temperature at time k, represents the upper limit of the indoor temperature at time k, represents the lower limit of the indoor temperature at time k, Indicates the absolute value deviation between the indoor temperature at time k and the target threshold.

[0113] According to an embodiment of the present disclosure, the economic performance index can be represented by energy cost. The energy cost can be calculated using the following formula (19).

[0114] (19)

[0115] in, represents energy cost, represents the power consumption at time k, represents the electricity price at time k.

[0116] According to an embodiment of the present disclosure, the energy flexible utilization index can be characterized by a flexible factor. The flexible factor can be calculated using the following formula (20).

[0117] (20)

[0118] in, represents the flexibility factor, Indicates the low price time, Indicates high electricity price times.

[0119] According to the embodiment of the present disclosure, the room temperature control effect index and the economic index are normalized based on the results of the RBC control method. The energy flexibility index is within the range of 0-1, and no normalization is performed. Direct comparison is sufficient. The comparison results are as follows: Figure 4 shown.

[0120] Figure 4 The effect comparison diagram of the control method in the related art and the control method based on MPC is schematically shown.

[0121] like Figure 4 As shown in the figure, the MPC-based control method has a significant performance improvement when applied to embedded tube buildings compared to the traditional rule-based control (RBC) method. In terms of the degree of violation of the indoor temperature constraint limit, the MPC-based control method has a 79% performance improvement; in terms of the absolute value deviation between the room temperature and the set value temperature, the MPC-based control method has a 5.5% performance improvement; in terms of economic indicators, the MPC-based control method saves 47% of energy costs; in terms of energy flexibility indicators, the MPC-based control method improves energy flexibility by 28%.

[0122] Figure 5 The flowchart of the training method of the target model of the building radiant energy supply system according to the embodiment of the present disclosure is schematically shown.

[0123] like Figure 5 As shown, the training method 500 of the target model of the building radiant energy supply system of this embodiment includes operations S510 to S580.

[0124] In operation S510 , a target-feature-target pair data set is acquired.

[0125] In operation S520, the target feature target pair data set is processed using a first layer long short-term memory network to output a sample first time series feature target pair sequence.

[0126] In operation S530, the sample first time series feature target pair sequence is processed using a first random discard layer to output the sample first target time series feature target pair sequence.

[0127] In operation S540, the first target temporal feature parameter information of the sample is processed using the second long short-term memory network layer to output the second temporal feature parameter information of the sample.

[0128] In operation S550, the second temporal feature parameter information of the sample is processed by using the second random discarding layer, and the second target temporal feature parameter information of the sample is output.

[0129] In operation S560, the second target temporal feature parameter information of the sample is processed using the third long short-term memory network layer, and the third temporal feature parameter information of the sample is output.

[0130] In operation S570, the third time series feature parameter information of the sample is processed using a fully connected layer to output indoor temperature prediction information and energy consumption information.

[0131] In operation S580, a target model is trained according to the indoor temperature prediction information, the energy consumption information, the historical control sequence information, the historical environmental factor information, and the disturbance parameter information to obtain a trained target model.

[0132] According to an embodiment of the present disclosure, the target model includes three layers of long short-term memory network layers, two layers of random drop layers and one layer of fully connected layer, and the machine structure of the target model is a first layer of long short-term memory network layer, a first layer of random drop layer, a second layer of long short-term memory network layer, a second layer of random drop layer, a third layer of long short-term memory network layer and a fully connected layer.

[0133] According to the embodiments of the present disclosure, the first layer of long short-term memory network can be used to process the acquired target feature target pair data set, and output the sample first time series feature target pair sequence. The first layer of random discard layer can be used to process the first time series feature parameter information output by the first layer of long short-term memory network layer, and output the first target time series feature parameter information; the second layer of long short-term memory network layer can be used to process the first target time series feature parameter information output by the first layer of random discard layer, and output the second time series feature parameter information. The second layer of random discard layer is used to process the second time series feature parameter information output by the second layer of long short-term memory network layer, and output the second target time series feature parameter information; the third layer of long short-term memory network layer is used to process the second target time series feature parameter information output by the second layer of random discard layer, and output the third time series feature parameter information. The third time series feature parameter information output by the third layer of long short-term memory network layer is processed by the fully connected layer, and the indoor temperature prediction information and energy consumption information are output.

[0134] According to an embodiment of the present disclosure, the target model can use the ReLU function as the output activation function, which helps to speed up the nonlinear learning. The training of the target model selects the mean square error (MSE) as the loss function to guide the target model to continuously adjust the weights and bias values ​​to reduce the prediction error. The number of training iterations (epochs for short) of the target model is set to 100, and the training batch size is 50. Use the trained target model to predict the new data set, and denormalize the indoor temperature prediction information to restore it to the original temperature scale to make the result interpretable. The root mean square error (RMSE) of the target model is used as the evaluation index to quantify the difference between the indoor temperature prediction information result and the actual indoor temperature information.

[0135] According to the embodiments of the present disclosure, the three-layer long short-term memory network layer can more efficiently capture deeper temporal features in time series data. The two-layer random dropout layer can randomly drop some neurons in each training batch to reduce the overfitting of the target model. Finally, the fully connected layer integrates the features extracted in the previous layers, thereby improving the accuracy of the final output indoor temperature prediction information and energy consumption information.

[0136] According to an embodiment of the present disclosure, a method for constructing a target feature target pair data set includes: the target feature target pair data set is generated by standardizing a feature target pair data set in a target format; the feature target pair data set in a target format is generated by format conversion of the feature target pair data set; the feature target pair data set is determined based on a custom function and a feature set, with the indoor temperature of the current time step as the target value; the feature set is determined for each time step by including historical control sequence information of the t time steps before the current time step, disturbance parameter information of the t time steps before the current time step, historical outdoor environmental factors of the t time steps before the current time step, and historical indoor environmental factors of the t-1 time steps before the current time step; the custom function is constructed based on the historical control sequence information, historical environmental factor information, and disturbance parameter information of the building radiation energy supply system.

[0137] According to an embodiment of the present disclosure, by preprocessing the historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiation power supply system for the target building corresponding to the target time period, other abnormal data such as extremely high or extremely low temperatures can be cleaned, thereby improving the effect of model training and the accuracy of the output indoor temperature prediction information.

[0138] According to an embodiment of the present disclosure, a custom function may be written based on historical control sequence information, historical environmental factor information, and disturbance parameter information of a building radiant energy supply system.

[0139] According to an embodiment of the present disclosure, a rolling window method may be used to create a target-feature-target pair dataset. The creation process of the target-feature-target pair dataset will be described below.

[0140] According to an embodiment of the present disclosure, for each time step, a feature set may be determined including historical control sequence information of the t time steps before the current time step, disturbance parameter information of the t time steps before the current time step, historical outdoor environmental factors of the t time steps before the current time step, and historical indoor environmental factors of the t-1 time steps before the current time step. The feature set may be combined into an input vector containing multi-step time series features.

[0141] According to an embodiment of the present disclosure, a feature-target pair data set may be determined based on a custom function and a feature set, with the indoor temperature at the current time step as the target value.

[0142] According to an embodiment of the present disclosure, a feature target pair dataset in a target format may be generated by format conversion of the feature target pair dataset. The feature target pair dataset in the target format may be in a sequence format required by the target model, that is, the feature set and the target temperature threshold of each time step are organized into a continuous time step sequence.

[0143] According to an embodiment of the present disclosure, a target feature target pair data set can be generated by standardizing a feature target pair data set in a target format. Standardization can improve the ability of the target model to process data. All feature target pair data sets in target formats are standardized to a distribution with zero mean and unit variance to ensure that the data is compared on the same scale, which helps optimize the convergence speed and reduce the training time.

[0144] According to an embodiment of the present disclosure, the target feature target pair data set can be divided into a training set and a test set, using 80% of the target feature target pair data set as training data and 20% of the target feature target pair data set as test data, the training data is used for learning and fitting the target model, and the test data is used for performance evaluation of the target model. Through this division, the target model can be verified on unseen test data to ensure that the target model has generalization ability.

[0145] According to an embodiment of the present disclosure, by using a rolling window method to create a target-feature-target-pair dataset, the efficiency of constructing the target-feature-target-pair dataset can be improved and the diversity of the target-feature-target-pair dataset can be improved.

[0146] Based on the above-mentioned multi-objective optimization control method based on the building radiation energy supply system, the present disclosure also provides a multi-objective optimization control device for the building radiation energy supply system. Figure 6 The device is described in detail.

[0147] Figure 6 The structural block diagram of the multi-objective optimization control device of the building radiation energy supply system according to the embodiment of the present disclosure is schematically shown.

[0148] like Figure 6 As shown, the multi-objective optimization control device 600 of the building radiant energy supply system of this embodiment includes a first acquisition module 610 , a first processing module 620 and a control module 630 .

[0149] The first acquisition module 610 is used to, in response to receiving the energy supply demand for the target period, acquire the historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiation energy supply system for the target building corresponding to the target period. In one embodiment, the first acquisition module 610 can be used to perform the operation S110 described above, which will not be repeated here.

[0150] The first processing module 620 is used to process the historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiant energy supply system based on the target constraint conditions, taking the lowest energy supply cost of the building radiant energy supply system and the highest indoor temperature control accuracy as the objective function, and generate the target control sequence information of the building radiant energy supply system. In one embodiment, the first processing module 620 can be used to perform the operation S120 described above, which will not be repeated here.

[0151] The control module 630 is used to control the operating parameters of the building radiation energy supply system based on the target control sequence information so that the indoor temperature of the target building reaches the target threshold while minimizing the energy supply cost; wherein the target threshold is determined according to user needs, and the user needs include energy supply needs and temperature needs. In one embodiment, the control module 630 can be used to perform the operation S130 described above, which will not be repeated here.

[0152] According to an embodiment of the present disclosure, the first processing module 620 includes: a first processing sub-module and a second processing sub-module.

[0153] The first processing submodule is used to input the historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiation energy supply system into the target model, and output indoor temperature prediction information and energy consumption information, wherein the target model is obtained by training the historical control sequence information, historical environmental factor information and disturbance parameter information with the historical control sequence information as a label.

[0154] The second processing submodule is used to process indoor temperature prediction information and energy consumption information based on target constraints, taking the lowest energy supply cost of the building radiation energy supply system and the highest indoor temperature control accuracy as the objective function, and generate target control sequence information of the building radiation energy supply system.

[0155] According to an embodiment of the present disclosure, the target model includes three long short-term memory network layers, two random drop layers and one fully connected layer; the first processing submodule includes: a first processing unit, a second processing unit, a third processing unit, a fourth processing unit, a fifth processing unit and a sixth processing unit.

[0156] The first processing unit is used to process the historical control sequence information, the historical environmental factor information and the disturbance parameter information by using the first long short-term memory network layer to obtain the first time series feature parameter information.

[0157] The second processing unit is used to process the first time series feature parameter information by using the first random discard layer to obtain the first target time series feature parameter information.

[0158] The third processing unit is used to process the first target time series feature parameter information using the second long short-term memory network layer to obtain second time series feature parameter information.

[0159] The fourth processing unit is used to process the second time series characteristic parameter information by using the second random discarding layer to obtain the second target time series characteristic parameter information.

[0160] The fifth processing unit is used to process the second target time series feature parameter information using the third long short-term memory network layer to obtain third time series feature parameter information.

[0161] The sixth processing unit is used to process the third time series feature parameter information using a fully connected layer to obtain indoor temperature prediction information and energy consumption information.

[0162] According to an embodiment of the present disclosure, the second processing submodule includes: a seventh processing unit, an eighth processing unit, a ninth processing unit, a tenth processing unit, an eleventh processing unit, a twelfth processing unit and a thirteenth processing unit.

[0163] The seventh processing unit is used to determine the objective function value of each individual in the kth generation I individuals based on the indoor temperature prediction information, the energy consumption information and the objective function; wherein I and k are both integers greater than or equal to 1.

[0164] The eighth processing unit is used to determine the fitness of each individual according to the difference in objective function values ​​of each individual and the difference in objective function values ​​of I individuals; wherein the difference in objective function values ​​of each individual represents the difference between the maximum objective function value among I individuals and the objective function value of each individual; the difference in objective function values ​​of I individuals represents the difference between the maximum objective function value and the minimum objective function value among I individuals.

[0165] The ninth processing unit is used to determine the probability of each individual being selected among the I individuals of the kth generation according to the fitness of each individual and the sum of the fitness of I individuals.

[0166] The tenth processing unit is used to determine the cumulative probability according to the probability of the first i individuals being selected.

[0167] The eleventh processing unit is used to determine the selected individuals according to the cumulative probability.

[0168] The twelfth processing unit is used to crossover the selected individuals to generate I individuals of the k+1th generation.

[0169] The thirteenth processing unit is used to determine the target control sequence information of the building radiation energy supply system according to the I individuals of the k+1th generation and the predetermined mutation probability.

[0170] According to an embodiment of the present disclosure, a thirteenth processing unit includes: a first processing sub-unit and a second processing sub-unit.

[0171] The first processing subunit is used to determine S gene segments of I individuals in the k+1th generation according to I individuals in the k+1th generation and a predetermined mutation probability.

[0172] The second processing subunit is used to determine the target control sequence information of the building radiation energy supply system according to the S gene fragments of the I individual of the k+1th generation.

[0173] Based on the training method of the target model of the building radiation energy supply system, the present disclosure also provides a training device for the target model of the building radiation energy supply system. Figure 7 The device is described in detail.

[0174] Figure 7 The structural block diagram of the training device of the target model of the building radiation energy supply system according to the embodiment of the present disclosure is schematically shown.

[0175] like Figure 7 As shown, the training device 700 of the target model of the building radiation energy supply system of this embodiment includes a second acquisition module 710, a second processing module 720, a third processing module 730, a fourth processing module 740, a fifth processing module 750, a sixth processing module 760, a seventh processing module 770 and a training module 780.

[0176] According to an embodiment of the present disclosure, the target model includes three long short-term memory network layers, two random dropout layers and one fully connected layer.

[0177] The second acquisition module 710 is used to acquire a target feature target pair data set. In one embodiment, the second acquisition module 710 can be used to perform the operation S510 described above, which will not be described in detail here.

[0178] The second processing module 720 is used to process the target feature target pair data set using the first layer of long short-term memory network, and output a sample first time series feature target pair sequence. In one embodiment, the second processing module 720 can be used to perform the operation S520 described above, which will not be repeated here.

[0179] The third processing module 730 is used to process the sample first time series feature target pair sequence using the first random discard layer, and output the sample first target time series feature target pair sequence. In one embodiment, the third processing module 730 can be used to perform the operation S530 described above, which will not be repeated here.

[0180] The fourth processing module 740 is used to process the sample first target temporal characteristic parameter information using the second long short-term memory network layer, and output the sample second temporal characteristic parameter information. In one embodiment, the fourth processing module 740 can be used to perform the operation S540 described above, which will not be repeated here.

[0181] The fifth processing module 750 is used to process the sample second time series characteristic parameter information using the second random discarding layer, and output the sample second target time series characteristic parameter information. In one embodiment, the fifth processing module 750 can be used to perform the operation S550 described above, which will not be repeated here.

[0182] The sixth processing module 760 is used to process the sample second target temporal characteristic parameter information using the third long short-term memory network layer, and output the sample third temporal characteristic parameter information. In one embodiment, the sixth processing module 760 can be used to perform the operation S560 described above, which will not be repeated here.

[0183] The seventh processing module 770 is used to process the sample third target time series characteristic parameter information using a fully connected layer, and output indoor temperature prediction information and energy consumption information. In one embodiment, the seventh processing module 770 can be used to perform the operation S570 described above, which will not be repeated here.

[0184] The training module 780 is used to train the target model according to the indoor temperature prediction information, energy consumption information, historical control sequence information, historical environmental factor information and disturbance parameter information to obtain the trained target model. In one embodiment, the training module 780 can be used to perform the operation S580 described above, which will not be repeated here.

[0185] According to an embodiment of the present disclosure, the second acquisition module 710 includes: a first construction submodule, a second construction submodule, a third construction submodule, a fourth construction submodule and a fifth construction submodule.

[0186] The first construction submodule is used to generate a target feature target pair data set by standardizing the feature target pair data set according to the target format.

[0187] The second construction submodule is used for generating a feature target pair data set in a target format by converting the format of the feature target pair data set.

[0188] The third construction submodule is used for the feature target pair data set which is determined according to the custom function and the feature set, taking the indoor temperature at the current time step as the target value.

[0189] The fourth construction submodule is used to determine the feature set for each time step, including the historical control sequence information of the t time steps before the current time step, the disturbance parameter information of the t time steps before the current time step, the historical outdoor environmental factors of the t time steps before the current time step, and the historical indoor environmental factors of the t-1 time steps before the current time step.

[0190] The fifth construction submodule is used to construct the custom function based on the historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiation energy supply system.

[0191] According to an embodiment of the present disclosure, any multiple modules among the modules, submodules, units, and subunits can be combined into one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the modules, submodules, units, and subunits can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware, and firmware, or in any appropriate combination of any of them. Alternatively, at least one of the modules, submodules, units, and subunits can be at least partially implemented as a computer program module, and when the computer program module is run, the corresponding function can be executed.

[0192] Figure 8 A block diagram of an electronic device suitable for implementing a multi-objective optimization control method for a building radiation energy supply system according to an embodiment of the present disclosure is schematically shown.

[0193] like Figure 8 As shown, the electronic device 800 according to an embodiment of the present disclosure includes a processor 801, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 802 or a program loaded from a storage part 808 into a RAM (Random Access Memory). The processor 801 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a dedicated microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 801 may also include an onboard memory for caching purposes. The processor 801 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0194] In RAM 803, various programs and data required for the operation of electronic device 800 are stored. Processor 801, ROM 802 and RAM 803 are connected to each other via bus 804. Processor 801 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 802 and / or RAM 803. It should be noted that the program can also be stored in one or more memories other than ROM 802 and RAM 803. Processor 801 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in the one or more memories.

[0195] According to an embodiment of the present disclosure, the electronic device 800 may further include an input / output (I / O) interface 805, which is also connected to the bus 804. The electronic device 800 may further include one or more of the following components connected to the input / output (I / O) interface 805: an input portion 806 including a keyboard, a mouse, etc.; an output portion 807 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 808 including a hard disk, etc.; and a communication portion 809 including a network interface card such as a LAN card, a modem, etc. The communication portion 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to the input / output (I / O) interface 805 as needed. A removable medium 811, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 810 as needed, so that the computer program read therefrom is installed into the storage portion 808 as needed.

[0196] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0197] It will be appreciated by those skilled in the art that the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways, even if such combinations or combinations are not explicitly described in the present disclosure. In particular, without departing from the spirit and teachings of the present disclosure, the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways. All of these combinations and / or combinations fall within the scope of the present disclosure.

[0198] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. A multi-objective optimization control method for a building radiation energy supply system, characterized in that: The method comprises: In response to receiving an energy supply demand for a target period, acquiring historical control sequence information, historical environmental factor information, and disturbance parameter information of a building radiation energy supply system for a target building corresponding to the target period; Taking the lowest energy cost of the building radiation energy supply system and the highest indoor temperature control accuracy as the objective function, based on the objective constraint conditions, the historical control sequence information, the historical environmental factor information and the disturbance parameter information of the building radiation energy supply system are processed to generate the target control sequence information of the building radiation energy supply system; Based on the target control sequence information, the operating parameters of the building radiation energy supply system are controlled so that the indoor temperature of the target building reaches a target threshold while minimizing the energy supply cost; wherein the target threshold is determined according to user needs, and the user needs include the energy supply demand and temperature demand.

2. The method according to claim 1, characterized in that The objective function is to minimize the energy cost of the building radiation energy supply system and maximize the indoor temperature control accuracy. Based on the target constraint conditions, the historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiation energy supply system are processed to generate the target control sequence information of the building radiation energy supply system, including: Inputting the historical control sequence information, the historical environmental factor information and the disturbance parameter information of the building radiation energy supply system into a target model, and outputting indoor temperature prediction information and energy consumption information, wherein the target model is obtained by training the historical control sequence information, the historical environmental factor information and the disturbance parameter information with the historical control sequence information as a label; Taking the lowest energy supply cost and the highest temperature control accuracy of the building radiation energy supply system as the objective function, based on the target constraint conditions, the indoor temperature prediction information and the energy consumption information are processed to generate the target control sequence information of the building radiation energy supply system.

3. The method according to claim 2, characterized in that The target model includes three long short-term memory network layers, two random dropout layers and one fully connected layer; The step of inputting the historical control sequence information, the historical environmental factor information and the disturbance parameter information of the building radiation energy supply system into a target model and outputting indoor temperature prediction information and energy consumption information comprises: Using the first long short-term memory network layer to process the historical control sequence information, the historical environmental factor information and the disturbance parameter information, to obtain first time series feature parameter information; Processing the first time series characteristic parameter information by using the first random discard layer to obtain first target time series characteristic parameter information; Processing the first target time series feature parameter information using the second long short-term memory network layer to obtain second time series feature parameter information; Processing the second time series characteristic parameter information by using the second random discard layer to obtain second target time series characteristic parameter information; Processing the second target time series feature parameter information using the third long short-term memory network layer to obtain third time series feature parameter information; The third time series characteristic parameter information is processed by using the fully connected layer to obtain the indoor temperature prediction information and the energy consumption information.

4. The method according to claim 3, characterized in that The objective function is to minimize the energy supply cost of the building radiation energy supply system and maximize the temperature control accuracy, and based on the target constraint conditions, process the indoor temperature prediction information and the energy consumption information to generate the target control sequence information of the building radiation energy supply system, including: Based on the indoor temperature prediction information, the energy consumption information and the objective function, determine the objective function value of each individual in the kth generation I individuals; wherein I and k are both integers greater than or equal to 1; For each individual, the fitness of each individual is determined according to the objective function value difference of each individual and the objective function value difference of the I individuals; wherein the objective function value difference of each individual represents the difference between the maximum objective function value of the I individuals and the objective function value of each individual; the objective function value difference of the I individual represents the difference between the maximum objective function value and the minimum objective function value of the I individual; Determine the probability of each individual in the k-th generation I individuals being selected according to the fitness of each individual and the sum of the fitness of the I individuals; Determine the cumulative probability based on the probability of the first i individuals being selected; Determining the selected individual according to the cumulative probability; Crossover the selected individuals to generate I individuals of the k+1th generation; According to the k+1th generation I individuals and the predetermined mutation probability, the target control sequence information of the building radiation energy supply system is determined.

5. The method according to claim 4, characterized in that The step of determining target control sequence information of the building radiation energy supply system according to the k+1th generation I individuals and the predetermined mutation probability includes: Determining S gene segments of the I individual of the k+1th generation according to the I individual of the k+1th generation and the predetermined mutation probability; The target control sequence information of the building radiation energy supply system is determined based on the S gene fragments of the I individuals of the k+1th generation.

6. A method for training a target model of a building radiation energy supply system, characterized in that: The target model includes three long short-term memory network layers, two random dropout layers and one fully connected layer, and the method includes: Obtain a target feature target pair data set; Using the first layer of the long short-term memory network to process the target feature target pair data set, and output a sample first time series feature target pair sequence; Processing the sample first time series feature target pair sequence by using the first random discard layer, and outputting the sample first target time series feature target pair sequence; Processing the first target temporal characteristic parameter information of the sample by using the second long short-term memory network layer, and outputting the second temporal characteristic parameter information of the sample; Processing the second temporal characteristic parameter information of the sample by using the random discarding layer of the second layer, and outputting the second target temporal characteristic parameter information of the sample; Processing the second target temporal characteristic parameter information of the sample by using the third long short-term memory network layer, and outputting the third temporal characteristic parameter information of the sample; Processing the third time series characteristic parameter information of the sample by using the fully connected layer, and outputting indoor temperature prediction information and energy consumption information; A target model is trained according to the indoor temperature prediction information, the energy consumption information, the historical control sequence information, the historical environmental factor information and the disturbance parameter information to obtain a trained target model.

7. The method according to claim 6, characterized in that The method for constructing the target feature target pair data set includes: The target feature target pair data set is generated by standardizing the feature target pair data set according to the target format; The feature target pair data set in the target format is generated by converting the format of the feature target pair data set; The feature target pair data set is based on a custom function and is determined according to the feature set with the indoor temperature at the current time step as the target value; The feature set is determined for each time step by including historical control sequence information of t time steps before the current time step, disturbance parameter information of t time steps before the current time step, historical outdoor environmental factors of t time steps before the current time step, and historical indoor environmental factors of t-1 time steps before the current time step; The user-defined function is constructed based on historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiation energy supply system.

8. A multi-objective optimization control device for a building radiation energy supply system, characterized in that: The device comprises: A first acquisition module is used to, in response to receiving an energy supply demand for a target period, acquire historical control sequence information, historical environmental factor information and disturbance parameter information of a building radiation energy supply system for a target building corresponding to the target period; The first processing module is used to take the lowest energy supply cost and the highest temperature control accuracy of the building radiant energy supply system as the objective function, and based on the target constraint conditions, process the historical control sequence information, historical environmental factor information and disturbance parameter information of the building radiant energy supply system to generate the target control sequence information of the building radiant energy supply system; A control module is used to control the operating parameters of the building radiation energy supply system based on the target control sequence information so that the indoor temperature of the target building reaches a target threshold while minimizing the energy supply cost; wherein the target threshold is determined according to user needs, and the user needs include temperature requirements and the energy supply requirements.

9. A training device for a target model of a building radiation energy supply system, characterized in that: The target model includes three long short-term memory network layers, two random dropout layers and one fully connected layer, and the device includes: The second acquisition module is used to acquire a target feature target pair data set; A second processing module is used to process the target feature target pair data set using the first layer of the long short-term memory network, and output a sample first time series feature target pair sequence; A third processing module is used to process the sample first time series feature target pair sequence by using the first layer of the random discard layer, and output the sample first target time series feature target pair sequence; A fourth processing module is used to process the first target temporal characteristic parameter information of the sample by using the second long short-term memory network layer, and output the second temporal characteristic parameter information of the sample; A fifth processing module, used to process the second temporal characteristic parameter information of the sample by using the second random discarding layer, and output the second target temporal characteristic parameter information of the sample; A sixth processing module, configured to process the second target temporal characteristic parameter information of the sample by using the third long short-term memory network layer, and output the third temporal characteristic parameter information of the sample; a seventh processing module, configured to process the third target time series characteristic parameter information of the sample by using the fully connected layer, and output indoor temperature prediction information and energy consumption information; The training module is used to train the target model according to the indoor temperature prediction information, the energy consumption information, the historical control sequence information, the historical environmental factor information and the disturbance parameter information to obtain the trained target model.

10. An electronic device, comprising: one or more processors; a memory for storing one or more computer programs, It is characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 5 or 6 to 7.

Images