Refrigeration equipment energy saving optimization method and system based on deep learning algorithm

Abstract

The application provides a refrigeration equipment energy-saving optimization method and system based on a deep learning algorithm, and relates to the field of refrigeration energy saving.The method comprises the following steps: obtaining original data information, and performing abnormality processing and normalization processing to obtain target data information; according to a time series prediction method of a long short-term memory network (LSTM), a prediction model is constructed by introducing an attention mechanism to output prediction information; deep learning training is performed based on the target data information to obtain a decision model, and the decision model is used to make decisions with the maximum unit energy consumption refrigeration capacity as the target according to refrigeration demand and pre-acquired meteorological prediction data, and decision information is output to optimize and control the refrigeration equipment.The application improves the timeliness of equipment control, and through early estimation of refrigeration demand and early setting of equipment operation, the control of the refrigeration equipment can be forward-looking, the timeliness of control is improved, and energy consumption loss caused by lagging control is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of refrigeration energy saving, in particular to a refrigeration equipment energy saving optimization method and system based on a deep learning algorithm. BACKGROUND

[0002] With the continuous growth of global energy consumption, energy saving and emission reduction has become the focus of social attention, and refrigeration equipment, as the main part of building energy consumption, accounts for about 40%-60% of the total building consumption throughout the year, and in hot summer, in some areas, it can even be as high as 70% or more. Therefore, energy saving optimization of refrigeration equipment is of great significance to reduce energy consumption.

[0003] In related technologies, the technical method for energy saving optimization of refrigeration equipment has variable frequency technology by adjusting the speed of the compressor to match the actual refrigeration demand, such as variable frequency air conditioner, which can automatically adjust the speed of the compressor according to the change of indoor temperature to ensure that it runs at a lower speed during operation, thereby avoiding the energy loss caused by the frequent start-stop of traditional fixed frequency air conditioners running at a higher speed; there is also intelligent control technology that uses more sensitive integrated sensors and automation technology to realize real-time monitoring and rapid adjustment of air conditioning system operation state. Related technologies mainly focus on two aspects: one is to reduce the energy loss caused by frequent start-stop of the equipment by controlling the low-frequency continuous operation of the compressor; the other is to avoid the occurrence of overheating or overcooling by sensitive sensing of the external environment and automatic adjustment to reduce unnecessary waste.

[0004] However, the related technology has the following defects: (1) the existing "variable frequency technology" mainly focuses on reducing the operating power of the compressor and ensuring low-frequency continuous operation to achieve energy saving, mainly taking the operating power of the compressor as the control variable to achieve the energy saving goal of reducing energy consumption, but it ignores the influence of the change of evaporation temperature and condensation temperature during equipment operation on the refrigeration rate. According to the operation principle of refrigeration equipment, the refrigeration effect of refrigerant has a certain nonlinear relationship with the high and low of evaporation temperature and condensation temperature, and the cooling capacity of unit refrigerant is also different, and the refrigeration capacity under unit energy consumption will also change. (2) With the wide application of new technologies, integrated sensors and automation technology have also been added to the refrigeration equipment control system to realize real-time monitoring and automatic adjustment, but the adjustment mechanism mostly relies on fixed settings, lacks flexibility, and due to the lack of demand prediction for future refrigeration capacity, the control is usually based on the current state, resulting in a certain hysteresis of the adjustment of the control system, which is prone to overcooling, resulting in waste of cooling capacity and ultimately causing energy loss. SUMMARY

[0005] In view of the deficiencies of the prior art, the application provides a refrigeration equipment energy-saving optimization method and system based on a deep learning algorithm, and solves the problem of unsatisfactory refrigeration equipment regulation and control effect caused by inaccurate refrigeration capacity demand prediction.

[0006] To achieve the above object, the application is implemented by the following technical solutions:

[0007] In a first aspect, the application provides a refrigeration equipment energy-saving optimization method based on a deep learning algorithm, which comprises: obtaining historical and time-series-arranged original data information, wherein the original data information comprises refrigeration output of the refrigeration equipment, environmental meteorological data, and daily operation data of the refrigeration equipment; performing abnormality processing and normalization processing on the original data information to obtain target data information; based on the target data information, introducing an attention mechanism to construct a prediction model according to a time series prediction method of a long short-term memory network (LSTM) to output prediction information to predict refrigeration demand in a future time period; performing deep learning training based on the target data information to obtain a decision model, and performing decision-making with the maximum refrigeration output per unit energy consumption as the target according to the refrigeration demand and pre-acquired meteorological prediction data to output decision information; and performing regulation and optimization on the refrigeration equipment according to the decision information, wherein the decision information is used to represent various values of equipment operation that maximize the refrigeration output per unit energy consumption.

[0008] In a second aspect, the application provides a refrigeration equipment energy-saving optimization system based on a deep learning algorithm, which comprises: a data acquisition module, a data processing module, a refrigeration demand prediction module, a refrigeration equipment operation decision-making module, and a refrigeration equipment regulation and optimization module.

[0009] Specifically, the data acquisition module is configured to acquire historical and time-series-arranged original data information, wherein the original data information comprises refrigeration output of the refrigeration equipment, environmental meteorological data, and daily operation data of the refrigeration equipment; the data processing module is configured to perform abnormality processing and normalization processing on the original data information to obtain target data information; the refrigeration demand prediction module is configured to introduce an attention mechanism to construct a prediction model according to a time series prediction method of a long short-term memory network (LSTM) based on the target data information to output prediction information to predict refrigeration demand in a future time period; the refrigeration equipment operation decision-making module is configured to perform deep learning training based on the target data information to obtain a decision model, and perform decision-making with the maximum refrigeration output per unit energy consumption as the target according to the refrigeration demand and pre-acquired meteorological prediction data to output decision information; and the refrigeration equipment regulation and optimization module is configured to perform regulation and optimization on the refrigeration equipment according to the decision information, wherein the decision information is used to represent various values of equipment operation that maximize the refrigeration output per unit energy consumption.

[0010] In a third aspect, an electronic device is provided, which comprises a processor, a memory, and a program stored in the memory and executable on the processor, and the program, when executed by the processor, implements the method for optimizing energy saving of a refrigeration device based on a deep learning algorithm in the first aspect.

[0011] In a fourth aspect, a computer readable storage medium is provided, which stores a program or instructions, and the program or instructions, when executed by a processor, implements the method for optimizing energy saving of a refrigeration device based on a deep learning algorithm in the first aspect.

[0012] The present application provides a method and system for optimizing energy saving of a refrigeration device based on a deep learning algorithm. Compared with the prior art, the present application has the following beneficial effects:

[0013] The present application optimizes energy saving of a refrigeration device in two stages. In the first stage, the present application predicts the refrigeration demand in a future time period based on a time series prediction method of a long short-term memory network (LSTM), collects original data information in time series, and performs abnormality processing and normalization processing to obtain target data information. According to the combination of the time series prediction method and the attention mechanism, a prediction model is constructed to predict the refrigeration demand in the future time period through model analysis. The predicted refrigeration demand is used as the basis for subsequent regulation and control decisions, so that the regulation and control are forward-looking, more timely, and achieve on-demand refrigeration. In the second stage, the present application performs deep learning training based on the target data information to obtain a decision model. The target data information is input into the model as input data for autonomous learning and repeated training. Finally, the decision information under different environmental conditions and refrigeration demands is inferred to represent the values of the device operation that maximize the refrigeration output per unit energy consumption, and subsequent regulation and control decisions of the refrigeration device are made. BRIEF DESCRIPTION OF DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative effort.

[0015] Figure 1 is a flowchart of a method for optimizing energy saving of a refrigeration device based on a deep learning algorithm provided by the present application;

[0016] Figure 2 is Figure 1 is an exemplary flowchart of S130 in

[0017] Figure 3is a structural schematic diagram of a refrigeration equipment energy-saving optimization system based on a deep learning algorithm provided by an embodiment of the present application.

[0018] Figure 4 is a structural schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0019] In order to make the objectives, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application are described clearly and completely. Obviously, the described embodiments are some of the embodiments of the present application, but not 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 protection scope of the present application.

[0020] It should be noted that, in this document, relational terms such as first and second and the like can only be used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that these entities or operations exist in any actual relationship or order. Moreover, the terms "comprising", "including", or any other variant thereof are intended to cover a non-exclusive inclusion, so that a process, method, article or apparatus that includes a list of elements does not include only those elements, but can also include other elements not expressly listed or inherent to such process, method, article or apparatus. Without more limitations, an element defined by the statement "including a..." does not exclude the existence of additional identical elements in the process, method, article or apparatus that includes the element.

[0021] The embodiments of the present application provide a refrigeration equipment energy-saving optimization method and system based on a deep learning algorithm, and solve the problem of unsatisfactory refrigeration equipment regulation effect caused by inaccurate refrigeration capacity demand prediction.

[0022] The technical solutions in the embodiments of the present application are as follows to solve the above technical problems:

[0023] With the continuous growth of global energy consumption, energy saving and emission reduction has become the focus of social attention, and refrigeration equipment is the main part of building energy consumption. The annual consumption of refrigeration equipment accounts for about 40%-60% of the total building consumption, and in hot summer, the consumption in some areas can even be as high as 70% or more. Therefore, energy-saving optimization of refrigeration equipment is of great significance to reduce energy consumption.

[0024] In the related art, the technical method for energy saving optimization of refrigeration equipment has a variable frequency technology for matching the actual refrigeration demand by adjusting the rotating speed of the compressor, such as a variable frequency air conditioner. This technology can automatically adjust the rotating speed of the compressor according to the change of indoor temperature to ensure that the compressor continuously operates at a lower rotating speed during operation, thereby avoiding the energy loss caused by the frequent start-stop of the traditional fixed frequency air conditioner. There is also an intelligent control technology for realizing real-time monitoring and rapid adjustment of the operating state of the air conditioning system by using more sensitive integrated sensors and automation technology. The related technology mainly focuses on two aspects: one is to reduce the energy loss caused by the frequent start-stop of the equipment by controlling the low-frequency continuous operation of the compressor; the other is to avoid the occurrence of overheating or overcooling by sensitive sensing of the external environment and automatic adjustment, thereby reducing unnecessary waste.

[0025] However, the related art has the following defects: (1) The existing "variable frequency technology" mainly focuses on reducing the operating power of the compressor and ensuring low-frequency continuous operation to achieve energy saving. The operating power of the compressor is mainly used as a control variable to achieve the energy saving goal of reducing energy consumption, but the influence of the change of evaporation temperature and condensation temperature on the refrigeration rate during equipment operation is ignored. According to the operation principle of the refrigeration equipment, the refrigeration effect of the refrigerant has a certain nonlinear relationship with the evaporation temperature and the condensation temperature. The cooling capacity of unit refrigerant varies with the temperature, and the refrigeration capacity per unit energy consumption also changes. (2) With the wide application of new technologies, integrated sensors and automation technology are added to the control system of the refrigeration equipment to realize real-time monitoring and automatic adjustment. However, the adjustment mechanism mostly relies on fixed settings and lacks flexibility. Moreover, due to the lack of demand prediction for future refrigeration capacity, the control is usually based on the current state, resulting in a certain lag in the adjustment of the control system and the occurrence of overcooling, which leads to waste of cooling capacity and ultimately causes energy loss.

[0026] Deep learning algorithm is a way to simulate the way the human brain processes information by building multiple layers of neural networks, so as to realize feature learning and pattern recognition of data and achieve efficient task solving capability. It can help the model learn relevant features by training a large amount of historical data, and through fine-tuning, the model can better adapt to the specific application environment. Applying this model to refrigeration equipment can provide decision basis for the automatic adjustment of the refrigeration equipment, so as to achieve more accurate regulation and control, thereby saving energy.

[0027] In order to better understand the above technical solution, the above technical solution will be described in detail in combination with the drawings in the specification and the specific embodiments.

[0028] First, a refrigeration equipment energy saving optimization method based on deep learning algorithm provided by the embodiments of the present application will be introduced.

[0029] A flowchart of a refrigeration equipment energy-saving optimization method based on a deep learning algorithm is provided in the embodiments of the present application, as shown in Figure 1 The refrigeration equipment energy-saving optimization method based on the deep learning algorithm can include the following steps S110-S150.

[0030] S110, historical and time-series arranged original data information is acquired, the original data information including refrigeration output of the refrigeration equipment, environmental meteorological data, and daily operation data of the refrigeration equipment.

[0031] S120, the original data information is subjected to abnormality processing and normalization processing to obtain target data information.

[0032] S130, based on the target data information, a prediction model is constructed according to a time series prediction method of a long short-term memory network (LSTM) with an attention mechanism to output prediction information to predict refrigeration demand in a future time period.

[0033] S140, a deep learning training is performed based on the target data information to obtain a decision model, and a decision is made with a maximum refrigeration output per unit energy consumption as a target according to the refrigeration demand and pre-acquired meteorological prediction data to output decision information.

[0034] S150, the refrigeration equipment is regulated and optimized according to the decision information, wherein the decision information is used to represent various values of equipment operation that maximize the refrigeration output per unit energy consumption.

[0035] The above is a specific implementation of the refrigeration equipment energy-saving optimization method based on the deep learning algorithm provided in the embodiments of the present application. It can be understood that the present application includes two stages for energy-saving optimization of the refrigeration equipment. In the first stage, the present application predicts refrigeration demand in a future time period based on a time series prediction method of a long short-term memory network (LSTM), collects original data information in a time series, and performs abnormality processing and normalization processing to obtain target data information. A prediction model is constructed by combining the time series prediction method with an attention mechanism to predict the refrigeration demand in the future time period through model analysis, and the predicted refrigeration demand is used as a basis for subsequent regulation and decision making, so that the regulation is forward-looking, more timely, and achieves on-demand refrigeration.

[0036] Further, in the second stage, a deep learning training is performed based on the target data information to obtain a decision model. The target data information is input into the model as input data for autonomous learning and repeated training, and finally the decision information under different environmental conditions and refrigeration demands is inferred to represent various values of equipment operation that maximize the refrigeration output per unit energy consumption, and subsequent regulation and decision making of the refrigeration equipment are performed.

[0037] In one example, the environmental meteorological data includes environmental temperature, environmental humidity and light intensity, and the daily operation data includes energy consumption value, compressor operation power, evaporation temperature, condensation temperature and expansion valve operation value. It can be understood that the present application considers the different effects of evaporation temperature and condensation temperature on refrigeration effect, and on the basis of considering energy saving by reducing compressor power, the regulation of evaporation temperature and condensation temperature during equipment operation is increased to increase the refrigeration output per unit energy consumption, so as to improve the energy efficiency ratio to achieve energy saving.

[0038] In some embodiments, the foregoing original data information is subjected to abnormality processing and normalization processing to obtain target data information, that is, the foregoing S120 can specifically include the following steps:

[0039] S210, the missing values in the original data information are marked and filled in a linear filling manner;

[0040] S220, the abnormal values in the data are identified by the box plot method, after determining the abnormal values, the abnormal values are set as null values, and linear supplement is performed to ensure the integrity and continuity of the data;

[0041] S230, the data is subjected to normalization processing to ensure the consistency of the data, and target data information is obtained; wherein the target data information includes target refrigeration information, target environmental information and target operation information corresponding to refrigeration output, environmental meteorological data and daily operation data respectively.

[0042] In the embodiments of the present application, it can be understood that the original data information obtained is first subjected to data preprocessing to complete cleaning processing; specifically, the missing values can be marked and filled in a linear filling manner, the abnormal values are identified by the box plot method, the abnormal points are detected by using the quartile range of the box plot, after finding the abnormal values, the abnormal values are first set as null values, and then linearly supplemented. After ensuring the integrity and continuity of the data through the above operations, all the data is subjected to normalization processing to ensure the consistency of the data, so as to facilitate subsequent analysis and prediction.

[0043] In some embodiments, please refer to Figure 2 , the foregoing target data information is used to construct a prediction model based on a long short-term memory network (LSTM) time series prediction method with an attention mechanism to output prediction information to predict the refrigeration demand in a future time period, that is, the foregoing S130 can specifically include the following steps:

[0044] S310, time and target environmental information are selected as feature variables, and the feature variables and target refrigeration information are determined as input information;

[0045] S320, select an activation function, a loss function and an optimization algorithm, and determine the number of LSTM units, the number of layers and the neural network layer for output mapping to obtain an LSTM model structure;

[0046] S330, introduce an attention mechanism to the LSTM model structure by calculating an attention score or a weight distribution to obtain a target neural network structure;

[0047] S340, training the target neural network structure based on the input information to obtain a prediction model;

[0048] S350, outputting prediction information based on the prediction model to represent the refrigeration demand in a future time period.

[0049] In the embodiments of the present application, it can be understood that the target environment information is a factor that may affect the refrigeration amount, and the present application selects the characteristic variables and the target refrigeration information as the input information, and can also calculate the Pearson correlation coefficient for screening; the present application first determines the preliminary LSTM model structure, and then introduces the attention mechanism, so that the model can automatically learn and pay attention to important features in the input information, thereby highlighting the role of important features. The present application constructs a model by combining the above-mentioned time series prediction method based on long short-term memory network LSTM and the attention mechanism, and uses historical data for training, which can predict the refrigeration demand in a future time period, and output it as part of the input data in the next stage.

[0050] In one example, the foregoing training of the target neural network structure based on the input information to obtain the prediction model, that is, the foregoing S340 can specifically include the following steps:

[0051] S410, inputting the input information into the target LSTM model structure;

[0052] S420, learning the long-term dependence of the time series corresponding to the input information through iteration of time steps and connection between LSTM units, and extracting key features;

[0053] S430, in the model training process, adjusting the network parameters to minimize the prediction error to obtain the prediction model.

[0054] In some embodiments, the foregoing deep learning training based on the target data information to obtain a decision model, and according to the refrigeration demand and the pre-acquired meteorological prediction data, the maximum refrigeration amount per unit energy consumption is taken as the target to make a decision, and the decision information is output, that is, the foregoing S140 can specifically include the following steps:

[0055] S510, dividing the target data information into a test set and a validation set according to a preset proportion;

[0056] S520, input the test set into the preset deep learning framework for training to obtain an initial model;

[0057] S530, verifying the initial model based on the verification set, and continuously adjusting internal parameters of the initial model according to a verification result obtained to obtain a decision model; wherein the decision model is used to determine refrigeration output of the refrigeration equipment under different environments and different operating states per unit energy consumption;

[0058] S540, inputting the refrigeration demand and the obtained meteorological prediction data into the decision model to make a decision with the maximum refrigeration output per unit energy consumption as a target, and outputting decision information.

[0059] In the embodiments of the present application, it can be understood that the present application constructs a decision model based on a deep learning algorithm to find out the values of the refrigeration equipment operation that maximize the refrigeration output per unit energy consumption, and then makes a decision in combination with the prediction result of the previous stage to determine the values of the refrigeration equipment operation that maximize the refrigeration output per unit energy consumption.

[0060] It should be noted that, since the target data information includes target refrigeration information, target environment information and target operation information corresponding to the refrigeration output, the environmental meteorological data and the daily operation data of the refrigeration equipment respectively, in the process of inputting the test set into the deep learning framework for training, the model can learn the relationship among the refrigeration output, the environmental meteorological data and the daily operation data of the refrigeration equipment, so that after determining the decision model, the model can make a decision based on the refrigeration demand at a future time and the obtained meteorological prediction data to output the decision information.

[0061] In one example, the foregoing regulating and optimizing the refrigeration equipment according to the decision information, that is, the foregoing S150 includes: determining a subsequent operation plan of the refrigeration equipment according to the decision information to regulate and optimize; wherein the subsequent operation plan includes a device start time, an initial operating power of a compressor, an initial opening degree of an expansion valve, a cooling water circulation flow rate and a speed of an air conditioning fan.

[0062] In some embodiments, the present application provides a refrigeration equipment energy-saving optimization system 600 based on a deep learning algorithm, as shown in Figure 3 The refrigeration equipment energy-saving optimization system 600 can include the following modules:

[0063] A data acquisition module 610 is configured to acquire historical and time-series-arranged original data information, the original data information including refrigeration output of the refrigeration equipment, environmental meteorological data, and daily operation data of the refrigeration equipment;

[0064] A data processing module 620 is configured to perform abnormality processing and normalization processing on the original data information to obtain target data information;

[0065] The refrigeration demand amount prediction module 630 is configured to, based on the target data information, construct a prediction model according to a time series prediction method of a long short-term memory network (LSTM) and by introducing an attention mechanism, to output prediction information to predict a refrigeration demand amount in a future time period.

[0066] The refrigeration equipment operation decision module 640 is configured to, based on the target data information, train a decision model through deep learning, and according to the refrigeration demand amount and pre-acquired meteorological prediction data, make a decision with a maximum refrigeration output per unit energy consumption as a target, and output decision information.

[0067] The refrigeration equipment regulation and optimization module 650 is configured to regulate and optimize the refrigeration equipment according to the decision information, where the decision information is used to represent various numerical values of equipment operation that maximize the refrigeration output per unit energy consumption.

[0068] According to embodiments of the present application, any multiple modules of the data acquisition module 610, the data processing module 620, the refrigeration demand amount prediction module 630, the refrigeration equipment operation decision module 640 and the refrigeration equipment regulation and optimization module 650 can be combined in one module, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of the modules can be combined with at least part of the functions of other modules, and implemented in one module.

[0069] In some embodiments, the data processing module 620 can be specifically configured to:

[0070] Label and fill in missing values in the original data information in a linear filling manner;

[0071] Identify abnormal values in the data through a box plot method, set the abnormal values as null values after determining the abnormal values, and perform linear supplementation to ensure the integrity and continuity of the data;

[0072] Perform normalization processing on the data to ensure the consistency of the data, to obtain target data information; where the target data information includes target refrigeration information, target environment information and target operation information corresponding to refrigeration output, environmental meteorological data and daily operation data respectively.

[0073] In some embodiments, the refrigeration demand amount prediction module 630 can specifically include:

[0074] The first selection determination unit 631 is configured to select time and target environment information as feature variables, and determine the feature variables and the target refrigeration information as input information;

[0075] The second selection determination unit 632 is configured to select an activation function, a loss function and an optimization algorithm, and determine a number of LSTM units, a number of layers and a neural network layer for output mapping, so as to obtain an LSTM model structure.

[0076] The attention introduction unit 633 is configured to introduce an attention mechanism into the LSTM model structure by calculating an attention score or a weight distribution, so as to obtain a target neural network structure.

[0077] The training unit 634 is configured to train the target neural network structure based on input information, so as to obtain a prediction model.

[0078] The output unit 635 is configured to output prediction information based on the prediction model, so as to represent a refrigeration demand in a future time period.

[0079] In some embodiments, the training unit 634 can be specifically configured to:

[0080] input the input information into the target LSTM model structure;

[0081] learn long-term dependencies of a time sequence corresponding to the input information through iteration of time steps and connection between LSTM units, and extract key features;

[0082] In the model training process, the prediction model is obtained by adjusting network parameters to minimize a prediction error.

[0083] In some embodiments, the refrigeration equipment operation decision module 640 can be specifically configured to:

[0084] divide the target data information into a test set and a validation set according to a preset proportion;

[0085] input the test set into a preset deep learning framework for training, so as to obtain an initial model;

[0086] verify the initial model based on the validation set, and constantly adjust internal parameters of the initial model according to a verification result, so as to obtain a decision model; wherein the decision model is used to determine refrigeration output of a unit energy consumption of the refrigeration equipment under different environments and different operation states;

[0087] input the refrigeration demand and obtained meteorological prediction data into the decision model, so as to make a decision with a maximum unit energy consumption refrigeration amount as a target, and output decision information.

[0088] Figure 3 Each module in the system has a function of implementing each step in the foregoing refrigeration equipment energy-saving optimization method based on a deep learning algorithm, and can achieve a corresponding technical effect. For brevity, the foregoing will not be described again.

[0089] In some embodiments, the present application provides an electronic device, a structural schematic diagram of which is shown in Figure 4

[0090] The electronic device can include a processor 710 and a memory 720 storing computer program instructions.

[0091] In particular, the processor 710 described above can include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or can be configured to implement one or more integrated circuits that embody the embodiments of the present application.

[0092] The memory 720 can include a mass storage for data or instructions. By way of example and not limitation, the memory 720 can include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. Where appropriate, the memory 720 can include removable or non-removable (or fixed) media. Where appropriate, the memory 720 can be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, the memory 720 is non-volatile solid-state memory.

[0093] The memory 720 can include read-only memory (ROM), random access memory (RAM), a disk storage medium device, an optical storage medium device, a flash memory device, an electrical, optical, or other physical / tangible memory storage device. Thus, generally, the memory 720 includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions that, when executed (e.g., by one or more processors), cause performance of any of the operations described above in the embodiments of the method for energy saving optimization of a refrigeration device based on a deep learning algorithm.

[0094] The processor 710 implements the method for energy saving optimization of a refrigeration device based on a deep learning algorithm in any of the embodiments described above by reading and executing the computer program instructions stored in the memory 720.

[0095] In one example, the electronic device can further include a communication interface 730 and a bus 700. As shown in Figure 4 The processor 710, the memory 720, and the communication interface 730 are connected through the bus 700 and complete communication therebetween, as shown in

[0096] ​The communication interface 730 is mainly configured to implement the communication between the modules, devices, units and / or equipment in the embodiments of the present application.

[0097] Bus 700 includes hardware, software, or both, to couple and / or interface the components of the online data traffic billing device to each other and / or to other peripheral devices. By way of example, and not limitation, bus can be an Accelerated Graphics Port (AGP) or other graphics bus, Enhanced Industry Standard Architecture (EISA) bus, Front Side Bus (FSB), HyperTransport, Industry Standard Architecture (ISA) bus, InfiniBand, Low Pin Count (LPC) bus, Memory Bus, Micro Channel Architecture (MCA) bus, Peripheral Component Interconnect (PCI) bus, PCI Express (PCI-X) bus, Serial Advanced Technology Attachment (SATA) bus, Video Electronics Standards Association local (VLB) bus, or another suitable bus or a combination of two or more of these. Where appropriate, bus 700 can include one or more buses. Although the present application is described and illustrated with a particular bus, it is not intended to be limited to this arrangement.

[0098] In addition, in combination with the above-mentioned embodiment of the refrigeration equipment energy-saving optimization method based on the deep learning algorithm, the present embodiment can provide a computer storage medium to realize. The computer storage medium has computer program instructions stored thereon; the computer program instructions are executed by the processor to realize any one of the above-mentioned embodiments of the refrigeration equipment energy-saving optimization method based on the deep learning algorithm.

[0099] It should be understood that the present application is not limited to the particular configurations and processes described above and illustrated in the drawings. For the sake of brevity, detailed descriptions of known methods are omitted. In the above-mentioned embodiments, several specific steps are described and illustrated as examples. However, the method process of the present application is not limited to the specific steps described and illustrated, and those skilled in the art can make various changes, modifications and additions, or change the order between steps, after understanding the spirit of the present application.

[0100] The functional blocks shown in the structural block diagrams above can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application specific integrated circuits (ASICs), appropriate firmware, plug-ins, functional cards, and the like. When implemented in software, the elements of the present application are program or code segments that are used to perform the required tasks. The program or code segments can be stored in a machine-readable medium or transmitted through a data signal carried in a carrier wave over a transmission medium or communication link. The "machine-readable medium" can include any medium that can store or transfer information. Examples of the machine-readable medium include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, and the like. The code segments can be downloaded via a computer network such as the Internet, an intranet, and the like.

[0101] It is also necessary to note that the exemplary embodiments mentioned in the present application describe some methods or systems based on a series of steps or devices. However, the present application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiments, or in an order different from the embodiments, or several steps can be performed simultaneously.

[0102] The above describes the aspects of the present application with reference to the flowcharts and / or block diagrams of the methods, apparatuses (systems) and computer program products according to the embodiments of the present disclosure. It should be understood that each block in the flowcharts and / or block diagrams, and the combination of the blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus, to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus enable the implementation of the functions / actions specified in one or more blocks of the flowcharts and / or block diagrams. It should also be understood that each block in the block diagrams and / or flowcharts, and the combination of the blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware that performs the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0103] In summary, compared with the prior art, the present application has the following beneficial effects:

[0104] 1. The present application combines the time series prediction method based on long short-term memory network with the attention mechanism to construct a prediction model to predict the refrigeration demand in a future time period through model analysis, thereby improving the accuracy of cold load prediction, providing more accurate data support for the regulation and control of equipment, and improving the accuracy of regulation and control. The regulation and control is forward-looking, more timely, and achieves on-demand refrigeration.

[0105] 2、The application carries out deep learning training based on target data information, obtains a decision model, inputs the target data information into the model as input data for autonomous learning and repeated training, finally infers decision information under different environmental conditions and refrigeration demands, finds the optimal equipment operation value, and then adjusts the equipment operation setting to improve the unit energy consumption refrigeration output, thereby improving the energy efficiency ratio of the refrigeration equipment to achieve energy saving.

[0106] 3、The application improves the timeliness of equipment regulation. Through early estimation of refrigeration demand and early setting of equipment operation, the regulation of the refrigeration equipment can be forward-looking, and the timeliness of the regulation can be improved, thereby reducing the energy consumption loss caused by regulation lag.

[0107] The above embodiments are only used to illustrate the technical solutions of the application, rather than limit them; although the application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features can be replaced by equivalents; 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 application.

Claims

1. A method for energy-saving optimization of refrigeration equipment based on deep learning algorithms, characterized in that, include: Acquire historical raw data information arranged in time series, including the refrigeration output of refrigeration equipment, environmental meteorological data, and daily operation data of refrigeration equipment; The original data information is subjected to anomaly processing and normalization to obtain the target data information; Based on the target data information, an attention mechanism is introduced to construct a prediction model according to the time series prediction method of Long Short-Term Memory Network (LSTM) to output prediction information to predict the cooling demand in the future time period. Deep learning training is performed based on the target data information to obtain a decision model. Based on the cooling demand and the pre-acquired meteorological forecast data, a decision is made with the goal of maximizing the cooling capacity per unit of energy consumption, and the decision information is output. Based on the decision information, the refrigeration equipment is adjusted and optimized, wherein the decision information is used to characterize various values ​​of the equipment operation that maximize the output of refrigeration capacity per unit of energy consumption; The process of performing anomaly processing and normalization on the original data information to obtain the target data information includes: Missing values ​​in the original data are marked and filled using a linear fill method; Outliers in the data are identified using box plots. After identifying outliers, they are set to null values ​​and linearly padded to ensure the integrity and continuity of the data. The data is normalized to ensure consistency and obtain the target data information; The target data information includes target refrigeration information, target environmental information, and target operational information corresponding to the refrigeration output, the environmental meteorological data, and the daily operation data, respectively. Based on the target data information, and using the Long Short-Term Memory (LSTM) network time series prediction method, an attention mechanism is introduced to construct a prediction model to output prediction information to predict the cooling demand in future time periods, including: Select time and the target environmental information as feature variables, and determine the feature variables and the target cooling information as input information; Select the activation function, loss function, and optimization algorithm, and determine the number of LSTM units, the number of layers, and the neural network layers used for output mapping to obtain the LSTM model structure; By calculating attention scores or weight distributions, an attention mechanism is introduced into the LSTM model structure to obtain the target neural network structure. The target neural network structure is trained based on the input information to obtain a prediction model; Based on the prediction model, predictive information is output to characterize the cooling demand in the future time period; The step of training the target neural network structure based on the input information to obtain a prediction model includes: The input information is input into the LSTM model structure; The long-term dependencies of the time series corresponding to the input information are learned by iterating through time steps and connecting the LSTM units, and key features are extracted. During model training, the prediction model is obtained by adjusting the network parameters to minimize the prediction error. The decision model is trained using deep learning based on the target data information. Based on the cooling demand and pre-acquired meteorological forecast data, a decision is made with the objective of maximizing cooling capacity per unit of energy consumption, and the decision information is output, including: The target data information is divided into a test set and a verification set according to a preset ratio; The test set is input into a preset deep learning framework for training to obtain an initial model; The initial model is tested based on the validation set, and the internal parameters of the initial model are continuously adjusted according to the test results to obtain a decision model; wherein, the decision model is used to determine the refrigeration output of the refrigeration equipment under different environments and different operating conditions based on the unit energy consumption. The cooling demand and the acquired meteorological forecast data are input into the decision model, and a decision is made with the goal of maximizing the cooling capacity per unit of energy consumption, and the decision information is output.

2. The energy-saving optimization method for refrigeration equipment based on deep learning algorithms as described in claim 1, characterized in that, The environmental meteorological data includes ambient temperature, ambient humidity, and light intensity, while the daily operation data includes energy consumption, compressor operating power, evaporation temperature, condensation temperature, and expansion valve operating values.

3. The energy-saving optimization method for refrigeration equipment based on deep learning algorithms as described in claim 1, characterized in that, The step of adjusting and optimizing the refrigeration equipment based on the decision information includes: Based on the decision information, the subsequent operation plan of the refrigeration equipment is determined for regulation and optimization; wherein, the subsequent operation plan includes equipment start-up time, compressor initial operating power, expansion valve initial opening, cooling water circulation flow rate and air conditioning fan speed.

4. A deep learning-based energy-saving optimization system for refrigeration equipment, based on the deep learning-based energy-saving optimization method for refrigeration equipment according to any one of claims 1-3, characterized in that, include: The data acquisition module is used to acquire historical raw data information arranged in time series. The raw data information includes the refrigeration output of the refrigeration equipment, environmental meteorological data, and daily operation data of the refrigeration equipment. The data processing module is used to perform anomaly processing and normalization on the raw data information to obtain the target data information; The cooling demand prediction module is used to construct a prediction model based on the target data information, according to the time series prediction method of Long Short-Term Memory Network (LSTM), by introducing an attention mechanism, so as to output prediction information to predict the cooling demand in the future time period. The refrigeration equipment operation decision module is used to perform deep learning training based on the target data information to obtain a decision model, and to make a decision based on the refrigeration demand and the pre-acquired meteorological forecast data, with the goal of maximizing the refrigeration capacity per unit of energy consumption, and output decision information. The refrigeration equipment control and optimization module is used to control and optimize the refrigeration equipment based on the decision information, wherein the decision information is used to characterize various values ​​of equipment operation that maximize the output of refrigeration capacity per unit of energy consumption.

5. An electronic device, characterized in that, include: A processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the energy-saving optimization method for refrigeration equipment based on a deep learning algorithm as described in any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program or instructions that, when executed by a processor, implement the energy-saving optimization method for refrigeration equipment based on a deep learning algorithm as described in any one of claims 1 to 3.

Images