Building energy control method and system

Abstract

The invention relates to the technical field of energy control, in particular to a building energy control method and system, reference noise parameters are introduced in an intermediate control parameter generation process, the reference noise parameters are dynamically updated based on an optimization function value, and under the condition that target load information is met, the building energy control efficiency is improved. According to the method, more reference control parameters are searched as much as possible, so that the target control parameters with high efficiency and stability are screened out from the multiple reference control parameters, control parameter selection errors caused by single data or experience deviation can be avoided, the target control parameters can better adapt to the long-term operation characteristics of the building energy system, and the stability of the building energy system is improved. The service life of equipment is prolonged, the operation and maintenance cost is reduced, and a powerful guarantee is provided for long-term efficient operation of a building energy system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of energy control, in particular to a building energy control method and system. BACKGROUND

[0002] With the increasing global energy crisis, the building field as a key field of energy consumption, the improvement of energy utilization efficiency has become one of the core issues of industry development. At present, the building energy system usually contains air conditioning, lighting, elevator, heating and other energy-consuming equipment. The running load of these devices is affected by many factors such as seasonal change, personnel flow, external environment temperature, equipment aging degree, etc., showing significant nonlinearity, volatility and uncertainty characteristics, which brings great challenges to the precise control of building energy.

[0003] In the optimization and screening of energy control parameters, the existing method usually only constructs an optimization function according to a single optimization target, without considering multi-dimensional indicators such as energy utilization efficiency and equipment operation stability, resulting in poor usability of the optimized control parameters. At the same time, in the screening of the final energy control parameters, the existing technology relies on real-time data or experience judgment, which is difficult to screen out control parameters with stability and high efficiency in long-term operation, resulting in problems such as high energy consumption, insufficient control accuracy and shortened equipment life of building energy system, which cannot meet the demand of modern building for energy fine management.

[0004] Therefore, in the building energy control scene, how to improve the precision and stability of building energy control has become a problem to be solved. SUMMARY

[0005] In order to solve the above technical problems, the technical scheme adopted by the present application is a building energy control method, which comprises the following steps: S1, determining the target load information corresponding to the target building according to the historical load information of each energy-consuming device in the target building; S2, obtaining the intermediate control parameter corresponding to the target load information according to the target load information, the reference noise parameter and the trained control parameter generation model, wherein the reference noise parameter is initially set as an initial noise parameter; S3, determining the optimization function value according to the intermediate control parameter, the trained discriminant model and the preset optimization function; S4, optimizing the intermediate control parameter according to the optimization function value to obtain a reference control parameter; S5, updating the reference noise parameter according to the optimization function value, returning to step S2 until a preset condition is met, and obtaining a plurality of reference control parameters; S6, screening a target control parameter for the target building from the reference control parameters according to historical control parameters.

[0006] The application further provides a building energy control system, which comprises: a load obtaining module, configured to determine target load information corresponding to the target building according to historical load information of each energy-consuming device in the target building; a parameter generating module, configured to generate an intermediate control parameter corresponding to the target load information according to the target load information, a reference noise parameter and a trained control parameter generation model, wherein the reference noise parameter is initially set as an initial noise parameter; a function value obtaining module, configured to determine an optimization function value according to the intermediate control parameter, a trained discriminant model and a preset optimization function; a parameter optimizing module, configured to optimize the intermediate control parameter according to the optimization function value to obtain a reference control parameter; an iterative processing module, configured to update the reference noise parameter according to the optimization function value, and return to step S2 until a preset condition is met to obtain a plurality of reference control parameters; a parameter determining module, configured to screen a target control parameter for the target building from the reference control parameters according to historical control parameters.

[0007] The application has at least the following beneficial effects: the reference noise parameter is introduced in the intermediate control parameter generation process, and the reference noise parameter is dynamically updated based on the optimization function value, so that more reference control parameters can be searched as much as possible under the condition of meeting the target load information, the target control parameter with high efficiency and stability is screened from the plurality of reference control parameters, the selection error of the control parameter caused by single data or experience deviation can be avoided, the target control parameter can better adapt to the long-term operation characteristics of the building energy system, the service life of the equipment is prolonged, the operation and maintenance cost is reduced, and strong guarantee is provided for the long-term and efficient operation of the building energy system. BRIEF DESCRIPTION OF DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0009] Figure 1 a flowchart of a building energy control method provided for the first embodiment of the application; Figure 2A schematic diagram of a building energy control system provided for embodiment two of the present application. DETAILED DESCRIPTION

[0010] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.

[0011] It should be noted that the terms "first", "second", and the like in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or sequence. It can be understood that the above-described terms for distinguishing similar objects can be interchanged under appropriate circumstances, so that the present application can also be implemented in other embodiments in addition to the above-described illustrated embodiments or described embodiments. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or server including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to the process, method, product, or device.

[0012] Embodiment one The embodiment one provides a building energy control method, which comprises the following steps, as shown in the figure: Figure 1 S1, determining target load information corresponding to a target building according to historical load information of each energy-consuming device in the target building.

[0013] The target building refers to a building that needs to be managed and controlled in terms of energy, and the target building includes a plurality of energy-consuming devices. The types of the energy-consuming devices include heating, ventilation, and air conditioning main machines, water pumps, fans, lighting circuits, etc. The historical load information of the energy-consuming devices includes historical load values corresponding to the energy-consuming devices at the 1st preset time point to the Mth preset time point, respectively. The time interval between adjacent preset time points is fixed as T, and the specific value of T can be set by the implementer according to the actual situation.

[0014] The target load information is a predicted load value corresponding to the target building at the M+1th preset time point.

[0015] In a specific embodiment, S1 comprises the following steps: S11, for any energy-consuming device in the target building, obtaining environmental parameter information associated with the energy-consuming device through a plurality of preset sensors associated with the energy-consuming device. ​

[0016] S12, obtain the device state information corresponding to the energy-consuming equipment through the digital twin model corresponding to the target building.

[0017] S13, predict the reference load information corresponding to the energy-consuming equipment according to the historical load information, the environmental parameter information and the device state information corresponding to the energy-consuming equipment.

[0018] S14, add the reference load information corresponding to each energy-consuming equipment respectively, and take the addition result as the target load information corresponding to the target building.

[0019] The types of the preset sensors include temperature sensors, humidity sensors, light intensity sensors, carbon dioxide concentration sensors, etc. The environmental parameter information is obtained by splicing the collection values of each preset sensor associated with the energy-consuming equipment at the Mth preset time point. It should be noted that the preset sensors associated with different energy-consuming equipment may be different. Therefore, in order to unify the size of the environmental parameter information for different energy-consuming equipment, the size of the environmental parameter information is set to 1xN, N is the number of types of preset sensors. By default, for any type of sensor, a single energy-consuming equipment is associated with at most one preset sensor of this type. When the energy-consuming equipment is not associated with a preset sensor of a certain type, the collection value corresponding to this type in the environmental parameter information is 0.

[0020] The digital twin model constructs a virtual mapping of the energy system of the target building through high-fidelity three-dimensional visualization technology combined with real-time data flowing in, so that the device state information corresponding to the energy-consuming equipment can be obtained through the digital twin model corresponding to the target building. The device state information can include start-stop state, running power, frequency, duration, etc.

[0021] When predicting the reference load information corresponding to the energy-consuming equipment according to the historical load information, the environmental parameter information and the device state information corresponding to the energy-consuming equipment, a time series prediction model based on the Transformer architecture can be used. The time series prediction model can be a long short-term memory network model, a time domain convolution network model, etc. The architecture and training process of the time series prediction model based on the Transformer architecture are prior art and will not be described here.

[0022] S2, obtain the intermediate control parameter corresponding to the target load information according to the target load information, the reference noise parameter and the trained control parameter generation model, wherein the reference noise parameter is initially set as the initial noise parameter.

[0023] The reference noise coefficient is used to introduce random information to enhance variability in subsequent optimization of the intermediate control parameter, so as to obtain a more comprehensive reference control parameter and improve completeness of the reference control parameter.

[0024] The control parameter generation model can include a full connection layer and an up-sampling layer. The full connection layer is used to map the input data into a multi-channel feature tensor, and the up-sampling layer is used to increase the dimension of the feature tensor while reducing the number of channels to obtain the intermediate control parameter.

[0025] The dimension of the intermediate control parameter can represent information such as a cold start-stop state, a water pump frequency, a valve opening, and a water temperature set point. Since the number of dimensions of the intermediate control parameter is large, the dimensions of the intermediate control parameter can form a large solution space, that is, there can be multiple control parameters that can meet the demand of the target load information.

[0026] Specifically, the reference noise parameter can be a variance parameter of a Gaussian function, and a mean parameter of the Gaussian function is 0 by default. The reference noise parameter can control the probability of generating a non-zero noise value. The larger the reference noise parameter, the greater the probability of generating a non-zero noise value. The result after the target load information is processed by the reference noise parameter is usually more different from the target load information.

[0027] The implementer can set the characteristic value of Poisson noise or the change probability of salt and pepper noise as the reference noise parameter according to actual conditions, that is, the noise introduced by the target load information can be various forms of noise.

[0028] In this embodiment, the initial noise parameter is set to 0, that is, no noise is added by default in the first iteration.

[0029] In a specific embodiment, S2 includes the following steps: S21, determining reference noise according to the reference noise parameter.

[0030] S22, sampling the reference noise to obtain a noise vector.

[0031] S23, inputting the target load information and the noise vector after splicing to the trained control parameter generation model to obtain the intermediate control parameter.

[0032] When the reference noise parameter is a variance parameter of a Gaussian function, the reference noise is Gaussian noise. Sampling the reference noise to obtain a noise vector can mean sampling the reference noise L times to obtain a noise vector of 1xL size. After the target load information and the noise vector are spliced, the control parameter generation model is used for generation processing to obtain the intermediate control parameter. In this embodiment, L is set to 63, and the specific value of L can be set by the implementer according to actual conditions.

[0033] S3, determining an optimization function value according to the intermediate control parameter, the trained discriminant model, and a preset optimization function.

[0034] The discriminant model adopts a regression prediction model in this embodiment, and can output a corresponding load regression prediction result according to an input control parameter. The architecture of the discriminant model includes an encoder and a full connection layer, and the discriminant model can be trained by using a mean square error loss function.

[0035] The preset optimization function can calculate an optimization function value according to the intermediate control parameter, the output of the trained discriminant model, and the like. The optimization function value can indicate an optimization process of the intermediate control parameter.

[0036] In a specific embodiment, the preset optimization function includes a first function item and a second function item, and S3 includes the following steps: S31, inputting the intermediate control parameter into the trained discriminant model to output intermediate load information.

[0037] S32, calculating a first optimization value according to the intermediate load information, the target load information, and the first function item.

[0038] S33, calculating a second optimization value according to the intermediate control parameter, the historical control parameter corresponding to the target building, and the second function item.

[0039] S34, determining the optimization function value according to the first optimization value and the second optimization value.

[0040] The intermediate load information corresponds to the intermediate control parameter input into the trained discriminant model. The first function item can be used to calculate the mean square error of the intermediate load information and the target load information. It can be known that the first optimization value can be used to supervise the difference between the intermediate load information and the target load information to be as small as possible, so that the load that can be supported by the energy supply side through the intermediate control parameter and the target load information of the energy consumption equipment of the target building are close enough to avoid energy waste caused by the energy supply side providing additional energy supply, or the energy supply side providing energy that cannot meet the load demand of the target building, affecting the normal operation of the target building.

[0041] The second function item can be used to calculate the mean square error of the intermediate control parameter and the historical control parameter. It can be known that the second optimization value can be used to supervise the difference between the intermediate control parameter and the historical control parameter to be as small as possible, so that the control parameter of the energy supply side changes in a small range, reducing the adjustment loss of the energy supply side. The historical control parameter can be the target control parameter corresponding to the Mth preset time point.

[0042] It can be seen that the optimization function value is determined by the first optimization value and the second optimization value, which not only enables the energy supply side to provide energy that can support the target building target load information, but also enables the reference control parameter to better adapt to the long-term operation characteristics of the building energy system, prolongs the service life of the equipment, reduces the operation and maintenance cost, and provides strong guarantee for the long-term efficient operation of the building energy system.

[0043] In an embodiment, S34 comprises the following steps: S341, multiplying the first optimization value and a preset first weight to obtain a first weighted result.

[0044] S342, multiplying the second optimization value and a preset second weight to obtain a second weighted result, wherein the preset first weight is much larger than the preset second weight.

[0045] S343, adding the first weighted result and the second weighted result to obtain the optimization function value.

[0046] Wherein, the first optimization value and the second optimization value are obtained by weighted addition to obtain the optimization function value, and the preset first weight is much larger than the preset second weight, so that when the intermediate control parameter is optimized, the energy provided by the energy supply side to support the target load information of the target building is prioritized to avoid affecting the normal operation of the target building.

[0047] S4, optimizing the intermediate control parameter according to the optimization function value to obtain the reference control parameter.

[0048] Wherein, the reference control parameter can be a control parameter that meets the energy provided by the energy supply side to support the target load information.

[0049] Specifically, the process of optimizing the intermediate control parameter according to the optimization function value can use a heuristic optimization search algorithm, such as particle swarm optimization algorithm, genetic algorithm, etc., or use simulated annealing algorithm, gradient descent method, etc. It is known to those skilled in the art that any optimization algorithm in the prior art for optimizing the intermediate control parameter falls within the scope of the present application, and will not be described here.

[0050] Specifically, when the continuous K iterations do not obtain new reference control parameters, it can be considered that all control parameters that can meet the demand of the target load information have been obtained, and the iteration can be stopped. K is set to 5 in this embodiment, and the implementer can set the value of K according to the actual situation.

[0051] S5, updating the reference noise parameter according to the optimization function value, returning to execute step S2 until a preset condition is met to obtain a plurality of reference control parameters.

[0052] In an embodiment, S5 comprises the following steps: S51, mapping an intermediate noise parameter according to the second optimization value.

[0053] S52, updating the reference noise parameter with the intermediate noise parameter, and returning to perform S2 until a preset condition is met, to obtain a plurality of reference control parameters, wherein the preset condition is that K consecutive iterations do not obtain a new reference control parameter, and K is a positive integer.

[0054] In an embodiment, the mapping function corresponding to the mapping of the intermediate noise parameter according to the second optimization value can be a logarithmic function, specifically ln(x+1), where x is the second optimization value. It can be seen that the larger the second optimization value, the worse the convergence effect of the second function term, and at this time, the influence of noise should be expanded in order to obtain more reference control parameters and guarantee the completeness of the control parameters. Therefore, the larger the second optimization value, the larger the intermediate noise parameter. The smaller the second optimization value, the better the convergence effect of the second function term, and at this time, it is considered that the reference control parameter is good enough, and the influence of noise can be reduced to end the iteration process as soon as possible. Therefore, the smaller the second optimization value, the smaller the intermediate noise parameter.

[0055] S6, selecting a target control parameter for the target building from the plurality of reference control parameters according to historical control parameters.

[0056] In an embodiment, the historical control parameter can be an actual control parameter corresponding to the target building at each historical time point, and there can be a plurality of historical control parameters.

[0057] In an embodiment, S6 comprises the following steps: S61, clustering a plurality of historical control parameters to obtain a plurality of parameter cluster sets.

[0058] S62, for any reference control parameter, determining the number of historical control parameters contained in the parameter cluster set to which the reference control parameter belongs as a reference number corresponding to the reference control parameter.

[0059] S63, taking the reference control parameter with the largest reference number as the target control parameter.

[0060] In an embodiment, the clustering can be based on the Euclidean distance between the historical control parameters. In this embodiment, the clustering can use a density-based spatial clustering algorithm with noise (Density-Based Spatial Clustering of Applications with Noise, DBSCAN).

[0061] Specifically, the more historical control parameters a parameter cluster set contains, the more frequently the control parameter range represented by the parameter cluster set is used in long-term operation. Therefore, the control parameter range represented by the parameter cluster set can be considered more reliable in long-term operation. Thus, the reference control parameter with the largest number of references is used as the target control parameter. Then, the target control parameter that combines high efficiency and stability is selected from multiple reference control parameters. This not only avoids the selection error of control parameter due to single data or experience bias, but also makes the target control parameter better adapt to the long-term operating characteristics of the building energy system, extends the service life of equipment, and reduces operation and maintenance costs.

[0062] This embodiment introduces a reference noise parameter during the intermediate control parameter generation process and dynamically updates the reference noise parameter based on the optimized function value. This allows for the searching of as many reference control parameters as possible while meeting the target load information, enabling the selection of the target control parameter that combines high efficiency and stability from multiple reference control parameters. This not only avoids control parameter selection errors caused by single data or experience biases, but also allows the target control parameter to better adapt to the long-term operating characteristics of the building energy system, extending equipment lifespan, reducing operation and maintenance costs, and providing strong support for the long-term efficient operation of the building energy system.

[0063] Example 2 This second embodiment provides a building energy control system, such as Figure 2 As shown, the building energy control system includes: The load acquisition module 21 is used to determine the target load information corresponding to the target building based on the historical load information of each energy-consuming device in the target building.

[0064] The parameter generation module 22 is used to generate a model based on the target load information, reference noise parameters and trained control parameters to obtain intermediate control parameters corresponding to the target load information, wherein the reference noise parameters are initially set as initial noise parameters.

[0065] The function value acquisition module 23 is used to determine the optimization function value based on the intermediate control parameters, the trained discrimination model, and the preset optimization function.

[0066] The parameter optimization module 24 is used to optimize the intermediate control parameters according to the optimization function value to obtain reference control parameters.

[0067] The iterative processing module 25 is used to update the reference noise parameters according to the optimized function value, return to the execution step S2, until the preset conditions are met, and obtain several reference control parameters.

[0068] The parameter determination module 26 is configured to select a target control parameter for the target building from the reference control parameters according to historical control parameters.

[0069] In an embodiment, the load obtaining module 21 comprises: The environmental parameter obtaining submodule is configured to obtain environmental parameter information associated with the energy-consuming device through a plurality of preset sensors associated with the energy-consuming device in the target building.

[0070] The device state obtaining submodule is configured to obtain device state information corresponding to the energy-consuming device through the digital twin model corresponding to the target building.

[0071] The load information prediction submodule is configured to predict reference load information corresponding to the energy-consuming device according to historical load information, environmental parameter information, and device state information corresponding to the energy-consuming device.

[0072] The load information synthesizing submodule is configured to add the reference load information corresponding to each energy-consuming device to obtain the target load information corresponding to the target building.

[0073] In an embodiment, the parameter generation module 22 comprises: The noise determination submodule is configured to determine reference noise according to the reference noise parameter.

[0074] The noise sampling submodule is configured to sample the reference noise to obtain a noise vector.

[0075] The parameter generation submodule is configured to input the target load information and the noise vector into the trained control parameter generation model to obtain the intermediate control parameter.

[0076] In an embodiment, the preset optimization function comprises a first function term and a second function term, and the function value obtaining module 23 comprises: The parameter discrimination submodule is configured to input the intermediate control parameter into the trained discrimination model to output intermediate load information.

[0077] The first optimization value calculation submodule is configured to calculate a first optimization value according to the intermediate load information, the target load information, and the first function term.

[0078] The second optimization value calculation submodule is configured to calculate a second optimization value according to the intermediate control parameter, historical control parameters corresponding to the target building, and the second function term.

[0079] The optimization function value determining submodule is configured to determine the optimization function value according to the first optimization value and the second optimization value.

[0080] In an embodiment, the optimization function value determining submodule comprises: The first weighting unit is configured to multiply the first optimization value by a preset first weight to obtain a first weighted result.

[0081] The second weighting unit is configured to multiply the second optimization value by a preset second weight to obtain a second weighted result, wherein the preset first weight is much greater than the preset second weight.

[0082] The weighted sum unit is configured to add the first weighted result and the second weighted result to obtain the optimization function value.

[0083] In an embodiment, the iteration processing module 25 comprises: The parameter mapping submodule is configured to map an intermediate noise parameter according to the second optimization value.

[0084] The parameter updating submodule is configured to update the reference noise parameter with the intermediate noise parameter, and return to step S2 until the preset condition is met to obtain a plurality of reference control parameters, wherein the preset condition is that no new reference control parameter is obtained in K consecutive iterations, and K is a positive integer.

[0085] In an embodiment, the parameter determining module 26 comprises: The clustering processing submodule is configured to perform clustering processing on the plurality of historical control parameters to obtain a plurality of parameter clustering sets.

[0086] The reference number determining submodule is configured to determine, for any reference control parameter, a number of historical control parameters contained in the parameter clustering set to which the reference control parameter belongs as a reference number corresponding to the reference control parameter.

[0087] The target control parameter determining submodule is configured to take the reference control parameter with the largest reference number as the target control parameter.

[0088] It should be noted that the information interaction, execution process, and the like between the above modules, submodules, and units are based on the same concept as the method embodiments, and the specific functions and technical effects brought by them can be referred to the method embodiments part, which will not be described here.

[0089] The above merely describes the preferred embodiments of the present application, and is not intended to limit the present application in any form. Although the present application has been described above with the preferred embodiments, it is not intended to limit the present application, and any person skilled in the art can make some changes or modifications to the above disclosed technical contents without departing from the technical solution of the present application, and the equivalent embodiments with equivalent changes and modifications are still within the scope of the technical solution of the present application.

Claims

1. A building energy control method, characterized in that, The building energy control method includes the following steps: S1, determine the target load information corresponding to the target building based on the historical load information of each energy-consuming device in the target building; S2, Based on the target load information, reference noise parameters and trained control parameters, a model is generated to obtain intermediate control parameters corresponding to the target load information, wherein the reference noise parameters are initially set as initial noise parameters; S3, determine the optimization function value based on the intermediate control parameters, the trained discrimination model, and the preset optimization function; S4, optimize the intermediate control parameters based on the optimization function value to obtain reference control parameters; S5, update the reference noise parameter according to the optimized function value, return to step S2, until the preset condition is met, and obtain several reference control parameters; S6. Based on historical control parameters, select target control parameters for the target building from various reference control parameters.

2. The building energy control method according to claim 1, characterized in that, S1 includes the following steps: S11, for any energy-consuming device in the target building, environmental parameter information associated with the energy-consuming device is obtained through a number of preset sensors associated with the energy-consuming device; S12, Obtain the equipment status information corresponding to the energy-consuming equipment through the digital twin model corresponding to the target building; S13, Based on the historical load information, environmental parameter information and equipment status information corresponding to the energy-consuming equipment, the reference load information corresponding to the energy-consuming equipment is predicted; S14, add up the reference load information corresponding to each energy-consuming device, and use the sum as the target load information corresponding to the target building.

3. The building energy control method according to claim 1, characterized in that, S2 includes the following steps: S21, Determine the reference noise based on the reference noise parameters; S22, Sample the reference noise to obtain a noise vector; S23, the target load information and the noise vector are concatenated and input into the trained control parameter generation model to obtain the intermediate control parameters.

4. The building energy control method according to claim 1, characterized in that, The preset optimization function includes a first function term and a second function term. S3 includes the following steps: S31, input the intermediate control parameters into the trained discrimination model and output the intermediate load information; S32, calculate the first optimized value based on the intermediate load information, the target load information, and the first function term; S33, calculate the second optimized value based on the intermediate control parameters, the historical control parameters corresponding to the target building, and the second function term; S34, determine the optimization function value based on the first optimization value and the second optimization value.

5. The building energy control method according to claim 4, characterized in that, S34 includes the following steps: S341, Multiply the first optimized value and the preset first weight to obtain the first weighted result; S342, multiply the second optimized value by the preset second weight to obtain the second weighted result, wherein the preset first weight is much larger than the preset second weight; S343, add the first weighted result and the second weighted result to obtain the optimized function value.

6. The building energy control method according to claim 5, characterized in that, S5 includes the following steps: S51, based on the second optimized value, the intermediate noise parameters are mapped to obtain the intermediate noise parameters; S52, update the reference noise parameter with the intermediate noise parameter, return to step S2, until the preset condition is met, and obtain several reference control parameters, wherein the preset condition is that no new reference control parameter is obtained in K consecutive iterations, and K is a positive integer.

7. The building energy control method according to claim 1, characterized in that, S6 includes the following steps: S61, cluster several historical control parameters to obtain several parameter cluster sets; S62, for any reference control parameter, determine the number of historical control parameters contained in the parameter cluster set to which the reference control parameter belongs as the reference number corresponding to the reference control parameter; S63, the reference control parameter with the largest number of references is used as the target control parameter.

8. A building energy control system, characterized in that, The building energy control system includes: The load acquisition module is used to determine the target load information corresponding to the target building based on the historical load information of each energy-consuming device in the target building; The parameter generation module is used to generate a model based on the target load information, reference noise parameters and trained control parameters to obtain intermediate control parameters corresponding to the target load information, wherein the reference noise parameters are initially set as initial noise parameters; The function value acquisition module is used to determine the optimization function value based on the intermediate control parameters, the trained discrimination model, and the preset optimization function. The parameter optimization module is used to optimize the intermediate control parameters based on the optimization function value to obtain reference control parameters; The iterative processing module is used to update the reference noise parameters according to the optimization function value, return to the execution step S2, until the preset conditions are met, and obtain several reference control parameters; The parameter determination module is used to select target control parameters for the target building from various reference control parameters based on historical control parameters.

9. The building energy control system according to claim 8, characterized in that, The load acquisition module includes: The environmental parameter acquisition submodule is used to acquire environmental parameter information associated with any energy-consuming device in the target building through several preset sensors associated with the energy-consuming device. The equipment status acquisition submodule is used to acquire the equipment status information corresponding to the energy-consuming equipment through the digital twin model corresponding to the target building; The load information prediction submodule is used to predict the reference load information corresponding to the energy-consuming equipment based on the historical load information, environmental parameter information and equipment status information corresponding to the energy-consuming equipment. The load information integration submodule is used to add up the reference load information corresponding to each energy-consuming device, and use the sum as the target load information corresponding to the target building.

10. The building energy control system according to claim 8, characterized in that, The parameter generation module includes: The noise determination submodule is used to determine the reference noise based on the reference noise parameters; The noise sampling submodule is used to sample the reference noise to obtain a noise vector; The parameter generation submodule is used to concatenate the target load information and the noise vector and input them into the trained control parameter generation model to obtain the intermediate control parameters.

Images