Improved energy system optimization control method and system based on incremental learning

Abstract

The invention discloses an incremental learning-based optimization control method and system for a modified energy system, and the method comprises the steps: obtaining the field collection data of an original energy system, and carrying out the processing through a multiple data processing method, and obtaining an original energy system data set; training a pre-selected data model by using the normalized original energy system data set, and selecting a system energy consumption and temperature model by comparing fitting effects of different types of data models; for the transformed new energy system, adopting an incremental learning algorithm to form a new energy system energy consumption and temperature model which can be simultaneously suitable for old knowledge and new knowledge; the optimal operation mode, the operation state and the optimal system control parameters under different working conditions are solved by utilizing the energy consumption and temperature model of the new energy system according to different optimization objectives and matching with corresponding optimization algorithms, so that energy conservation is realized. According to the method, the new energy system model is rapidly formed by forward migration of old energy system knowledge, and the problem of disastrous forgetting is avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of energy system energy saving, and in particular to a reconstruction energy system optimization control method and system based on incremental learning. BACKGROUND

[0002] With the continuous development of economy and the continuous growth of population, energy problems have become one of the most urgent challenges faced by modern society. In order to improve the energy efficiency of energy system and achieve the goal of energy saving, some old energy systems of enterprises face the urgent need of upgrading and reconstruction. However, the reconstruction of new energy system requires a large amount of investment, which makes many enterprises prefer to choose to upgrade the old energy system to a new high-efficiency energy system. At the same time, energy enterprises may also need to reconstruct the energy system in order to meet market demand, expand business scale and other purposes, so as to expand the capacity of the system. However, in the process of energy system reconstruction, a series of technical problems have appeared, mainly including:

[0003] The data model of the old energy system is constructed based on its specific structure and operation characteristics. With the upgrading of the energy system to a new high-efficiency energy system, its structure, operation mechanism and thermodynamic law have changed significantly, resulting in that the data model of the old energy system is no longer applicable to the new high-efficiency energy system. If the model is retrained for the new high-efficiency energy system, not only the training cycle is long, a large amount of time, manpower and computing resources are needed, but also catastrophic forgetting problem will occur in the training process, that is, in the process of learning the knowledge of the new high-efficiency energy system, the new model will gradually lose the thermodynamic knowledge contained in the old energy system, and these knowledge still has important significance for comprehensively and accurately understanding and controlling the energy system.

[0004] Therefore, how to accurately fit the thermodynamic law of the new high-efficiency energy system and effectively avoid catastrophic forgetting and retain the relevant knowledge of the old energy system has become a key technical problem to be solved in the field of energy system. SUMMARY

[0005] The present application aims at the problems in the prior art, and provides a reconstruction energy system optimization control method and system based on incremental learning, which utilizes the positive transfer of old energy system knowledge and quickly forms a new energy system model, solves the problem of time-consuming re-modeling, retains the knowledge of old energy system and avoids catastrophic forgetting.

[0006] In order to achieve the above purpose, the present application has the following technical solutions:

[0007] In a first aspect, a reconstruction energy system optimization control method based on incremental learning is provided, comprising:

[0008] The original energy system field collection data is acquired, and the original energy system field collection data is processed by using multiple data processing methods to obtain an original energy system data set;

[0009] The original energy system data set is normalized, and the original energy system data set after the normalization processing is used to train a pre-selected data model. By comparing the fitting effects of different types of data models, the best fitting effect is selected as the system energy consumption and temperature model.

[0010] For the modified new energy system, an incremental learning algorithm is used based on the selected system energy consumption and temperature model to form a new energy system energy consumption and temperature model that can be applied to both old and new knowledge.

[0011] Using the new energy system energy consumption and temperature model that can be applied to both old and new knowledge, according to different optimization objectives, the corresponding optimization algorithm is matched to solve the best operation mode, operation state and best system control parameters under different working conditions, so as to realize energy saving.

[0012] As a preferred scheme, it also includes adjusting the hyperparameters of the incremental learning algorithm, and selecting the best incremental learning algorithm from different types of incremental learning algorithms through effect comparison to form a new energy system energy consumption and temperature model that can be applied to both old and new knowledge data.

[0013] As a preferred scheme, the incremental learning algorithm includes any one or a combination of multiple of the following methods: a regularization-based incremental learning algorithm, a dynamic adjustment strategy-based incremental learning algorithm, a memory replay and complementary learning system method.

[0014] As a preferred scheme, the regularization-based incremental learning algorithm strengthens the constraint when updating the model weight through the regularization method, so as to realize the learning of new tasks on the premise of maintaining existing knowledge; the regularization-based incremental learning algorithm includes any one or a combination of multiple of the following methods: a parameter regularization method, a distribution regularization method, and a regularization method from the Bayesian perspective.

[0015] As a preferred scheme, the dynamic adjustment strategy-based incremental learning algorithm adjusts the network structure dynamically to adapt to the changing environment. The dynamic adjustment strategy-based incremental learning algorithm selectively trains the network and expands the network when necessary to adapt to the learning of new tasks.

[0016] As a preferred scheme, the memory replay and complementary learning system method includes two parts of hippocampus and neocortical system. The hippocampus shows short-term adaptability and allows learning of new knowledge. The learned new knowledge is put back into the neocortical system over time to maintain long-term memory.

[0017] As a preferred scheme, the energy system includes any one or a combination of the following: a refrigeration and freezing system, a thermal power generation system, and a data center refrigeration system.

[0018] The raw energy system field collection data is obtained by a sensor, and the raw energy system field collection data includes any one or a combination of the following thermodynamic parameters: temperature, pressure, flow force, and power.

[0019] The data processing method includes data denoising, data filtering, data repair, and data reduction. The data denoising method includes any one or a combination of the following: a moving average method, a smoothing filter method, and an outlier processing method. The data filtering method includes a classifier-based or keyword-based data filtering method. The data reduction method includes any one or a combination of the following: principal component analysis, factor analysis, and singular value decomposition.

[0020] As a preferred scheme, the training of the data model uses any one or a combination of the following: an artificial neural network (ANN), an extreme gradient boosting (XGBoost), a support vector regression (SVR), a gated recurrent unit (GRU), a long short-term memory network (LSTM), and a light gradient boosting machine (LightGBM).

[0021] When training the pre-selected data model using the normalized raw energy system dataset, the root mean square error (RMSE) and the mean absolute percentage error (MAPE) are used to continuously adjust the hyperparameters in the data model, and the fitting effects of different types of data models are compared to select the system energy consumption and temperature model.

[0022] The hyperparameters in the data model include the learning rate, the number of weak learners, and the maximum tree depth of the XGBoost. The initial learning rate, the batch size, and the number of iterations of the ANN. The sample sampling ratio and the column sampling ratio in the LightGBM.

[0023] As a preferred scheme, in the step of solving the optimal operating mode, operating state, and optimal system control parameters under different operating conditions according to different optimization objectives and combining corresponding optimization algorithms to achieve energy saving, the optimization algorithm includes any one or a combination of the following: Newton's method, gradient descent method, branch and bound algorithm, background segmentation algorithm, dynamic programming algorithm, genetic algorithm, particle swarm optimization algorithm, and simulated annealing algorithm.

[0024] In a second aspect, an energy system optimization control system based on incremental learning is provided, which includes:

[0025] A dataset construction module is configured to obtain raw energy system field collection data and process the raw energy system field collection data using multiple data processing methods to obtain a raw energy system dataset.

[0026] The data model selection module is configured to normalize the original energy system dataset, train a pre-selected data model using the normalized original energy system dataset, and select the data model with the best fitting effect as the system energy consumption and temperature model by comparing the fitting effects of different types of data models.

[0027] The incremental learning module is configured to use an incremental learning algorithm to form a new energy system energy consumption and temperature model that can be applied to both old and new knowledge based on the selected system energy consumption and temperature model for the reformed new energy system.

[0028] The model matching output module is configured to use the new energy system energy consumption and temperature model that can be applied to both old and new knowledge to solve the optimal operation mode, operation state and optimal system control parameters under different working conditions according to different optimization objectives, so as to achieve energy saving.

[0029] Compared with the prior art, the present application has at least the following beneficial effects:

[0030] The incremental learning algorithm used in the present application can learn new knowledge while preserving old knowledge, avoiding the problem of catastrophic forgetting. The method of the present application simulates the idea of lifelong learning of human beings, selects the best machine learning model to establish the initial system energy consumption and temperature model for the energy system reformed for various reasons, and uses the incremental learning algorithm to update the model after the energy system is reformed, so as to learn new knowledge (the characteristics of the reformed system) while preserving old knowledge (the characteristics of the original system). Using the new energy system energy consumption and temperature model that can be applied to both old and new knowledge, the optimal operation mode, operation state and optimal system control parameters under different working conditions are solved according to different optimization objectives, so as to achieve efficient operation of the energy system. The method of the present application establishes an incremental model that can continuously learn, preserves the accuracy of the knowledge of the old energy system while ensuring the accuracy of the thermodynamic fitting of the reformed system, and avoids the problem of catastrophic forgetting. After obtaining the incremental learning model that can accurately predict the thermodynamic knowledge of the new system and the knowledge of the old system, the genetic algorithm and other optimization algorithms can be used to quickly realize intelligent control and energy saving optimization of the energy system. The incremental learning-based efficient modeling method for the reformed energy system of the present application can quickly model the reformed energy system, thereby quickly realizing optimization and energy saving control, shortening the training period and greatly saving system energy consumption. BRIEF DESCRIPTION OF DRAWINGS

[0031] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and other related drawings can be obtained by those skilled in the art without creative labor.

[0032] Figure 1 A cold plate type liquid cooling data center refrigeration system structure schematic diagram before the embodiment of the present application is modified;

[0033] Figure 2 A cold plate type liquid cooling data center refrigeration system structure schematic diagram after the embodiment of the present application is modified;

[0034] Figure 3 A flow chart of the energy system optimization control method based on incremental learning of the embodiment of the present application is shown in the figure.

[0035] Figure 4 An EWC regularization incremental learning algorithm schematic diagram of the embodiment of the present application is shown in the figure.

[0036] In the figure, 1 is a cooling tower, 2 is a primary side filter, 3 is a plate heat exchanger, 4 is a No. 1 three-way valve, 5 is a primary side water pump, 6 is a primary side bypass valve, 7 is a secondary side filter, 8 is a server, 9 is a secondary side water pump, 10 is a water chilling unit, 11 is a No. 2 three-way valve, 12 is an LCU backboard door, 13 is a wind cooling side water pump, 14 is a wind cooling side filter, and 15 is a wind cooling side bypass valve. DETAILED DESCRIPTION

[0037] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, other embodiments can also be obtained by those skilled in the art without creative labor.

[0038] The embodiment of the present application proposes an energy system optimization control method based on incremental learning, which comprises the following steps:

[0039] S1. Collecting data on site of the original energy system and processing the collected data on site of the original energy system by using multiple data processing methods to obtain an original energy system data set;

[0040] S2. Normalizing the original energy system data set and training a pre-selected data model by using the normalized original energy system data set, and selecting the best fitting effect as the system energy consumption and temperature model by comparing the fitting effects of different types of data models;

[0041] S3. For the new energy system after modification, using an incremental learning algorithm based on the selected system energy consumption and temperature model to form a new energy system energy consumption and temperature model that can be applied to both old knowledge and new knowledge;

[0042] S4. Using the new energy system energy consumption and temperature model which can be applied to both old knowledge and new knowledge, according to different optimization objectives, the corresponding optimization algorithm is matched to solve the optimal operation mode, operation state and optimal system control parameters under different working conditions, so as to realize energy saving.

[0043] In a possible implementation, after step S3, further comprising adjusting the hyperparameters of the incremental learning algorithm with accuracy, forgetting rate and the like indicators, and selecting the best incremental learning algorithm from different types of incremental learning algorithms through effect comparison, to form a new energy system energy consumption and temperature model which can be applied to both old knowledge data and new knowledge data.

[0044] Further, the incremental learning algorithm includes any one or a combination of more than one of a regularization-based incremental learning algorithm, a dynamic adjustment strategy-based incremental learning algorithm, a memory replay and complementary learning system method.

[0045] The regularization-based incremental learning algorithm strengthens the constraint when updating the model weight through the regularization method, so as to realize the learning of new tasks on the premise of maintaining existing knowledge; the regularization-based incremental learning algorithm includes SI, R-EWC, Rwalk and the like parameter regularization methods; LwF, SLNID, GL-GAN and the like distribution regularization methods; and PVI and the like methods for regularization from the Bayesian perspective.

[0046] The dynamic adjustment strategy-based incremental learning algorithm adjusts the network structure dynamically to adapt to the changing environment, selectively trains the network, and expands the network when necessary to adapt to the learning of new tasks. The dynamic adjustment strategy-based incremental learning algorithm includes CWR, DEN, PN, ACLF and the like methods.

[0047] The memory replay and complementary learning system method includes hippocampus and neocortex systems, the hippocampus shows short-term adaptability and allows rapid learning of new knowledge, and the new knowledge is put back into the neocortex system over time to maintain long-term memory. The memory replay and complementary learning system algorithm includes EMR, LGM, MER and the like episodic memory methods, and BIIL, GDM and the like dual memory learning methods.

[0048] The incremental learning algorithm of the embodiment of the application can also be called a continuous learning algorithm or a lifelong learning algorithm.

[0049] In a possible implementation, the energy system in step S1 includes any one or a combination of more than one of a refrigeration and refrigeration system, a thermal power generation system and a data center refrigeration system.

[0050] The original energy system field collection data is acquired by a sensor, and the acquired original energy system field collection data includes any one or a combination of multiple of thermodynamic parameters such as temperature, pressure, flow force and power;

[0051] The data processing method includes data denoising, data filtering, data repairing and data reduction, the data denoising method includes any one or a combination of multiple of sliding average method, smoothing filter method and processing outlier method, the data filtering method includes a classifier-based or keyword-based data filtering method, and the data reduction method includes any one or a combination of multiple of principal component analysis, factor analysis and singular value decomposition method.

[0052] In a possible implementation, the training of the data model in step S2 adopts any one or a combination of multiple of artificial neural network (ANN), extreme gradient boosting (XGBoost), support vector regression (SVR), gated recurrent unit (GRU), long short-term memory network (LSTM) and light gradient boosting machine (LightGBM).

[0053] When the pre-selected data model is trained by using the normalized original energy system data set, the hyperparameters in the data model are constantly adjusted by using root mean square error (RMSE) and mean absolute percentage error (MAPE) indexes, and the fitting effects of different types of data models are compared to select the system energy consumption and temperature model.

[0054] The hyperparameters in the data model include learning rate, weak learner number and tree maximum depth of XGBoost, initial learning rate, batch size and iteration number of ANN, and sample sampling ratio and column sampling ratio in LightGBM.

[0055] In a possible implementation, the reconstruction form of the original old energy system includes but is not limited to adding cooling towers, water chillers and water pumps and other energy equipment due to factory expansion, and adding valves to realize the interaction between multiple subsystems in the old energy system and to realize the energy efficiency improvement of the old energy system due to policy requirements.

[0056] In a possible implementation, in the step of solving the optimal operation mode, operation state and optimal system control parameter under different working conditions according to different optimization objectives to realize energy saving, the optimization algorithm includes but is not limited to mathematical methods such as Newton method and gradient descent method, accurate algorithms such as branch and bound algorithm, background segmentation algorithm and dynamic programming, and meta-heuristic algorithms such as genetic algorithm, particle swarm algorithm and simulated annealing algorithm.

[0057] The following content takes the reconstruction of an actual cold plate liquid cooling data center refrigeration system as an example to describe the application of the reconstruction energy system optimization control method based on incremental learning of the embodiment of the application.

[0058] As Figure 1 With Figure 2 As shown in the figure, the cold plate type liquid cooling data center refrigeration system of the embodiment of the application includes an external refrigeration device cooling tower 1, which utilizes the direct / indirect contact of the cold carrier with air to perform cold and heat exchange to generate steam, and the steam volatilization takes away heat to achieve the principles of evaporative heat dissipation, convective heat transfer and radiative heat transfer to dissipate the waste heat generated in industry or in refrigeration and air conditioning; the primary side filter 2, the secondary side filter 7 and the air cooling side filter 14 play the role of filtering impurities in the pipeline; the plate heat exchanger 3 performs heat exchange on the primary side fluid and the secondary side fluid of the liquid cooling part, cools the secondary side hot fluid by the cold fluid generated by the primary side refrigeration device, and further cools the server; the primary side water pump 5, the secondary side water pump 9 and the air cooling side water pump 13 provide energy for fluid flow; the primary side bypass valve 6 and the air cooling side bypass valve 15 play the role of adjusting cold; the server 8 stores a variety of chips, including high-power density chips cooled by liquid through direct contact with the cold plate and low-power density chips cooled by air through the fan in the air-liquid heat exchanger 12; the external refrigeration device water chiller 10 cools the air cooling part of the cold carrier through the vapor compression refrigeration cycle; the LCU backplane door (air-liquid heat exchanger) 12 stores a fan, which brings the cold fluid in the pipeline into the computer room to cool the server through the fan; the first three-way valve 4 and the second three-way valve 11 are components added to the modified refrigeration system, the old refrigeration system only cools the air cooling part through the water chiller 10, and the cooling tower 1 only cools the liquid cooling part, the two three-way valves connect the air cooling part and the liquid cooling part to form three modes of low temperature mode, normal mode and high temperature mode; the normal mode is shown in the figure Figure 1 As shown in the figure, the cold plate type liquid cooling data center refrigeration system of the embodiment of the application includes an external refrigeration device cooling tower 1, which utilizes the direct / indirect contact of the cold carrier with air to perform cold and heat exchange to generate steam, and the steam volatilization takes away heat to achieve the principles of evaporative heat dissipation, convective heat transfer and radiative heat transfer to dissipate the waste heat generated in industry or in refrigeration and air conditioning; the primary side filter 2, the secondary side filter 7 and the air cooling side filter 14 play the role of filtering impurities in the pipeline; the plate heat exchanger 3 performs heat exchange on the primary side fluid and the secondary side fluid of the liquid cooling part, cools the secondary side hot fluid by the cold fluid generated by the primary side refrigeration device, and further cools the server; the primary side water pump 5, the secondary side water pump 9 and the air cooling side water pump 13 provide energy for fluid flow; the primary side bypass valve 6 and the air cooling side bypass valve 15 play the role of adjusting cold; the server 8 stores a variety of chips, including high-power density chips cooled by liquid through direct contact with the cold plate and low-power density chips cooled by air through the fan in the air-liquid heat exchanger 12; the external refrigeration device water chiller 10 cools the air cooling part of the cold carrier through the vapor compression refrigeration cycle; the LCU backplane door (air-liquid heat exchanger) 12 stores a fan, which brings the cold fluid in the pipeline into the computer room to cool the server through the fan; the first three-way valve 4 and the second three-way valve 11 are components added to the modified refrigeration system, the old refrigeration system only cools the air cooling part through the water chiller 10, and the cooling tower 1 only cools the liquid cooling part, the two three-way valves connect the air cooling part and the liquid cooling part to form three modes of low temperature mode, normal mode and high temperature mode; the normal mode is shown in the figure

[0059] As Figure 3 The energy system optimization control method based on incremental learning of the embodiment of the application includes the following steps:

[0060] 1) Figure 1 With Figure 2 The typical cold plate type liquid cooling data center refrigeration system in the prior art is provided with a variety of sensors, which acquire Figure 1The original energy system before the transformation collects data on site, and the data collected by the sensors include thermodynamic parameters such as temperature, pressure, flow force, power, frequency, etc. After obtaining the data of the refrigeration system before the transformation, multiple data processing methods are used, including data denoising methods such as moving average method, smoothing filter method, and outlier processing method, data filtering methods based on classifier and keyword, data reduction methods such as principal component analysis, factor analysis, and singular value decomposition, to form the original system data set.

[0061] 2) The obtained original data set is normalized by the following method:

[0062]

[0063] wherein x0 is the normalized data; x is the original data; x min is the minimum value in the original data; x max is the maximum value of the original data.

[0064] The hyperparameters in the data models such as BPNN, LSTM, and GRU are continuously adjusted according to the RMSE accuracy index, and the system energy consumption and temperature model with the best fitting effect is selected by comparing different types of data models. The input parameters are environmental temperature and humidity, air cooling part refrigerating capacity, liquid cooling part refrigerating capacity, fan and pump speed, and chiller load, and the output parameters are system power, PUE, and chip temperature. The calculation method of RMSE is as follows:

[0065]

[0066] wherein RMSE is the root mean square error, which can evaluate the model accuracy, N is the data quantity of the test set, Yi is the actual value of the system energy consumption and temperature, is the predicted value of the neural network model.

[0067] 3) To improve the PUE of the data center refrigeration system, the refrigeration system is transformed from the original old cold plate liquid cooling data center refrigeration system shown in Figure 1 to the new cold plate liquid cooling data center refrigeration system shown in Figure 2 , forming three modes of low temperature mode, normal mode, and high temperature mode. At this time, the system energy consumption and temperature data model established in step 2) is no longer applicable to the low temperature mode and the high temperature mode. Re-collecting data for the low temperature mode and the high temperature mode and re-training the models for these two modes is time-consuming and laborious, and three models are not conducive to subsequent optimization control, which will multiply the optimization time. The old energy system belongs to a kind of operation mode called normal mode in the new energy system, so the data of the old energy system still has value, which is old knowledge in incremental learning. The data of the low temperature mode and the high temperature mode of the new energy system after transformation are new knowledge in incremental learning.

[0068] The model suitable for the original old cold plate liquid cooling data center refrigeration system formed in step 2) is combined with an incremental learning algorithm, and EWC based on regularization is taken as an example to form a model that can accurately predict the thermodynamic behavior of the original old system (regular mode) and the thermodynamic behavior of the low temperature mode and the regular mode in the new system. In the objective function of the EWC model, a penalty term for the model parameters between new and old tasks is included, thereby effectively alleviating the forgetting of the relevant knowledge in the previously learned knowledge. The effect diagram of EWC is shown in Figure 4 . The circles represent that the hyperparameters here have good prediction accuracy for the corresponding tasks, and the intersection of the two circles indicates that the hyperparameters here have good performance for all tasks. The data model parameters of the regular mode learned in step 2) are θ A . At this time, if the low temperature mode and high temperature mode data collected are independently modeled, the regular mode knowledge will be forgotten (shown by the dense dashed line in the figure). If the EWC method is used to restrict the model hyperparameters, strong constraints are applied to the hyperparameters important for regular mode prediction, and weak constraints are applied to the hyperparameters not important for regular mode prediction, the hyperparameters that can accurately predict the regular mode and the new high temperature mode / low temperature mode can be found (shown by the sparse dashed line in the figure). If the same strength is used to constrain the hyperparameters, the effect of the two tasks (regular mode and high temperature mode / low temperature mode) may not reach the best (shown by the solid line in the figure). The objective function of EWC is as follows:

[0069]

[0070] , wherein L(θ) is the EWC loss function, θ represents the hyperparameter, L B (θ) is the loss function of the high temperature mode / low temperature mode task, λ represents the trade-off index of new and old tasks, i represents the index of the parameter, and F represents the Fisher matrix, which can determine the importance of the hyperparameter to the old task (regular mode).

[0071] 4) Adjust the hyperparameters in the EWC incremental learning algorithm and select the best parameter value to retain the old knowledge of the original energy system (regular mode) and accurately fit the data of the new energy system (low temperature mode / high temperature mode) to learn new knowledge, and form a new system energy consumption and temperature model based on the incremental learning algorithm that can accurately fit both old knowledge data and new knowledge data;

[0072] 5) Use the new system energy consumption and temperature model based on the incremental learning algorithm to calculate the best operating mode / operating state and the best system control parameters under different operating conditions according to the optimization goal of minimizing the energy consumption of the new energy system, so as to achieve maximum energy saving.

[0073] Another embodiment of the present application also provides a reconstruction energy system optimization control system based on incremental learning, comprising:

[0074] A data set construction module is configured to acquire original energy system field collection data and process the original energy system field collection data by using multiple data processing methods to obtain original energy system data sets.

[0075] A data model selection module is configured to normalize the original energy system data sets and train the pre-selected data models by using the normalized original energy system data sets, and select the best fitting effect as the system energy consumption and temperature model by comparing the fitting effects of different types of data models.

[0076] An incremental learning module is configured to form a new energy system energy consumption and temperature model that can be applied to both old knowledge and new knowledge by using an incremental learning algorithm based on the selected system energy consumption and temperature model for the new energy system after reconstruction.

[0077] A model matching output module is configured to use the new energy system energy consumption and temperature model that can be applied to both old knowledge and new knowledge to solve the optimal operation mode, operation state and optimal system control parameters under different working conditions according to different optimization objectives, so as to achieve energy saving.

[0078] Another embodiment of the present application provides an electronic device, comprising:

[0079] A memory is configured to store at least one instruction, and a processor is configured to execute the instruction stored in the memory to implement the reconstruction energy system optimization control method based on incremental learning.

[0080] Another embodiment of the present application provides a computer readable storage medium, wherein the computer readable storage medium stores at least one instruction, and the at least one instruction is executed by a processor in an electronic device to implement the reconstruction energy system optimization control method based on incremental learning.

[0081] For example, the instruction stored in the memory can be divided into one or more modules / units, which are stored in the computer readable storage medium and executed by the processor to complete the reconstruction energy system optimization control method based on incremental learning. The one or more modules / units can be a series of computer readable instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the server.

[0082] The electronic device can be a smart phone, a notebook, a palm computer, a cloud server, and the like. The electronic device can include, but is not limited to, a processor, a memory. Those skilled in the art can understand that the electronic device can further include more or less components, or combine certain components, or different components, for example, the electronic device can further include an input / output device, a network access device, a bus, and the like.

[0083] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and the like. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0084] The memory can be an internal storage unit of the server, such as a hard disk or a memory of the server. The memory can also be an external storage device of the server, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, and the like. Further, the memory can include both the internal storage unit and the external storage device of the server. The memory is used to store the computer readable instructions and other programs and data required by the server. The memory can also be used to temporarily store data that has been output or will be output.

[0085] It should be noted that the information interaction, execution process, and the like between the above module units, since based on the same concept as the method embodiments, the specific functions and the technical effects brought about can be referred to the method embodiment part, and will not be described here.

[0086] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of each functional unit and module is exemplified, and in actual application, the above-mentioned functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or software. In addition, the specific names of each functional unit and module are only for easy distinction, and do not limit the protection scope of the application. The specific working process of the unit and module in the system can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0087] The integrated unit, if realized in the form of a software functional unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing related hardware, and the computer program can be stored in a computer-readable storage medium. When the processor executes the computer program, the steps of each method embodiment described above can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or some intermediate forms. The computer-readable medium at least includes any entity or device that can carry a computer program code to a photographing device / terminal equipment, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. For example, a U disk, a mobile hard disk, a magnetic disk or an optical disk, etc.

[0088] In the above embodiments, the description of each embodiment has its own emphasis, and the parts not described or recorded in detail in a certain embodiment can be referred to the related description of other embodiments.

[0089] The above-described embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A method for optimizing the control of an energy system based on incremental learning, characterized in that, include: Data collected on-site from the original energy system was obtained, and the data was processed using multiple data processing methods to obtain the original energy system dataset. The original energy system dataset is normalized, and a pre-selected data model is trained using the normalized original energy system dataset. By comparing the fitting effects of different types of data models, the model with the best fitting effect is selected as the system energy consumption and temperature model. For the modified new energy system, an incremental learning algorithm is used based on the selected system energy consumption and temperature model to form a new energy system energy consumption and temperature model that can be applied to both old and new knowledge. By utilizing a new energy system energy consumption and temperature model that is applicable to both old and new knowledge, and by matching appropriate optimization algorithms with different optimization objectives, the optimal operating mode, operating state, and optimal system control parameters under different working conditions can be solved to achieve energy conservation.

2. The method for optimizing and controlling an energy system based on incremental learning according to claim 1, characterized in that, It also includes adjusting the hyperparameters of the incremental learning algorithm and selecting the best incremental learning algorithm from different types of incremental learning algorithms through effect comparison, so as to form a new energy system energy consumption and temperature model that can be applied to both old and new knowledge data.

3. The method for optimizing and controlling a modified energy system based on incremental learning according to claim 2, characterized in that, Incremental learning algorithms include any one or more combinations of methods based on regularized incremental learning algorithms, incremental learning algorithms based on dynamic adjustment strategies, memory playback, and complementary learning systems.

4. The energy system optimization control method based on incremental learning according to claim 3, characterized in that, Regularized incremental learning algorithms strengthen constraints when updating model weights through regularization methods, thereby enabling the learning of new tasks while maintaining existing knowledge. The regularized incremental learning algorithm includes any one or more combinations of parametric regularization methods, distribution regularization methods, and regularization methods from a Bayesian perspective.

5. The energy system optimization control method based on incremental learning according to claim 3, characterized in that, Incremental learning algorithms based on dynamic adjustment strategies adapt to constantly changing environments by dynamically adjusting the network structure. These algorithms selectively train the network and expand it as needed to adapt to learning new tasks.

6. The method for optimizing control of an energy system based on incremental learning according to claim 3, characterized in that, The memory replay and complementary learning system consists of two parts: the hippocampus and the neocortex. The hippocampus exhibits short-term adaptability and allows the learning of new knowledge. The learned knowledge is then put back into the neocortex over time to maintain long-term memory.

7. The method for optimizing control of an energy system based on incremental learning according to claim 1, characterized in that, Energy systems include any one or more combinations of refrigeration and cooling systems, thermal power generation systems, and data center cooling systems; Data collected on-site from the primary energy system is acquired through sensors. The acquired data includes one or more thermodynamic parameters such as temperature, pressure, fluidity, and power. Data processing methods include data denoising, data filtering, data repair, and data reduction; data denoising methods include any or a combination of moving average, smoothing filter, and outlier removal methods; data filtering methods include classifier-based or keyword-based data filtering methods; data reduction methods include any or a combination of principal component analysis, factor analysis, and singular value decomposition methods.

8. The method for optimizing control of an energy system based on incremental learning according to claim 1, characterized in that, The training of the data model employs any one or a combination of artificial neural networks (ANN), extreme gradient boosting (XGBoost), support vector regression (SVR), gated recurrent units (GRU), long short-term memory networks (LSTM), and lightweight gradient boosting machines (LightGBM). When training a pre-selected data model using the normalized raw energy system dataset, the hyperparameters within the data model are continuously adjusted using the root mean square error (RMSE) and mean absolute percentage error (MAPE) indices. The system energy consumption and temperature models are selected by comparing the fitting effects of different types of data models. The hyperparameters in the data model include the learning rate, number of weak learners, and maximum tree depth of XGBoost; the initial learning rate, batch size, and number of iterations of ANN; and the sample sampling ratio and column sampling ratio in LightGBM.

9. The method for optimizing control of an energy system based on incremental learning according to claim 1, characterized in that, In the step of solving the optimal operating mode, operating state, and optimal system control parameters under different working conditions according to different optimization objectives and corresponding optimization algorithms to achieve energy saving, the optimization algorithms include any one or more combinations of Newton's method, gradient descent method, branch and bound algorithm, background segmentation algorithm, dynamic programming algorithm, genetic algorithm, particle swarm algorithm, and simulated annealing algorithm.

10. A modified energy system optimization control system based on incremental learning, characterized in that, include: The dataset construction module is used to acquire field-collected data from the original energy system and process the field-collected data from the original energy system using multiple data processing methods to obtain the original energy system dataset. The data model selection module is used to normalize the original energy system dataset and train a pre-selected data model using the normalized original energy system dataset. By comparing the fitting effects of different types of data models, the model with the best fitting effect is selected as the system energy consumption and temperature model. The incremental learning module is used to generate a new energy system energy consumption and temperature model that can be applied to both old and new knowledge by using an incremental learning algorithm based on the selected system energy consumption and temperature model for the modified new energy system. The model matching output module is used to utilize energy consumption and temperature models of new energy systems that are applicable to both old and new knowledge. Based on different optimization objectives, it combines corresponding optimization algorithms to solve for the optimal operating mode, operating state, and optimal system control parameters under different operating conditions, so as to achieve energy saving.

Images