Model optimization method and device for heating ventilation air conditioner, electronic equipment and storage medium
By introducing business constraints and multi-task learning into the HVAC control model, and combining self-supervised and supervised training, the model parameters were optimized, solving the problem of temperature and humidity mismatch in the HVAC control model, and achieving higher control accuracy and business adaptability.
Patent Information
- Application Number
- CN202511041597.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-28
- Publication Date
- 2025-11-07
AI Technical Summary
Existing HVAC control models do not match the required or standard temperature and humidity within the facility, resulting in insufficient control accuracy.
By introducing business constraints as regularization terms into the first loss function of the base model and employing a multi-task learning strategy to optimize model parameters, efficient fine-tuning of parameters is achieved through a combination of self-supervised and supervised training.
It improves the accuracy of HVAC control and ensures that the model meets business needs and industry standards in practical applications.
Smart Images
Figure CN120911280A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of heating ventilation air conditioning control, and particularly relates to a model optimization method and device for heating ventilation air conditioning, an electronic device and a storage medium. BACKGROUND
[0002] Commercial centers, shopping plazas and other facilities are equipped with heating ventilation equipment. The heating ventilation equipment is used to provide heating, ventilation and air conditioning for the facilities.
[0003] At present, a control model of heating ventilation air conditioning is used for control. However, when the heating ventilation control is performed by using the control model, the temperature and humidity in the facility do not meet the requirements or standards. How to improve the accuracy of the control of the heating ventilation air conditioning becomes a problem to be solved. SUMMARY
[0004] The present application provides a model optimization method and device for heating ventilation air conditioning, an electronic device and a storage medium, to solve the problem of insufficient control accuracy of the control model of the heating ventilation air conditioning.
[0005] According to an aspect of the present application, a model optimization method for heating ventilation air conditioning is provided, comprising:
[0006] obtaining a basic model for heating ventilation air conditioning control;
[0007] determining a business constraint condition according to historical statistical data;
[0008] determining a second loss function according to the business constraint condition and a first loss function of the basic model;
[0009] performing parameter efficient fine-tuning according to the second loss function and the basic model to obtain a parameter efficient fine-tuning model.
[0010] According to another aspect of the present application, a model optimization device for heating ventilation air conditioning is provided, comprising:
[0011] a basic model obtaining module configured to obtain a basic model for heating ventilation air conditioning control;
[0012] a business constraint determining module configured to determine a business constraint condition according to historical statistical data;
[0013] a loss function determining module configured to determine a second loss function according to the business constraint condition and a first loss function of the basic model;
[0014] a parameter efficient fine-tuning module configured to perform parameter efficient fine-tuning according to the second loss function and the basic model to obtain a parameter efficient fine-tuning model.
[0015] According to another aspect of the present application, an electronic device is provided, comprising:
[0016] at least one processor; and
[0017] a memory communicatively connected with the at least one processor; wherein
[0018] the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to perform the model optimization method of the HVAC according to any one of the embodiments of the present application.
[0019] According to another aspect of the present application, a computer readable storage medium is provided, which stores computer instructions for enabling a processor to implement the model optimization method of the HVAC according to any one of the embodiments of the present application when executed by the processor.
[0020] The technical solution of the embodiments of the present application acquires a basic model for HVAC control, determines a business constraint condition according to historical statistical data, determines a second loss function according to the business constraint condition and a first loss function of the basic model, and performs parameter efficient fine-tuning according to the second loss function and the basic model to obtain a parameter efficient fine-tuning model. Compared with the current situation of insufficient accuracy of HVAC control, the technical solution provided by the embodiments of the present application can modify the first loss function of the basic model according to the business constraint condition to obtain the second loss function. Based on the second loss function, the parameter efficient fine-tuning of the basic model can be performed to fine-tune the basic model in combination with the business constraint condition, and the parameter efficient fine-tuning model obtained can meet the requirement of the business constraint condition and improve the control accuracy of the HVAC.
[0021] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS
[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced as follows. 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 labor.
[0023] Figure 1 is a flowchart of a model optimization method of an HVAC provided by an embodiment of the present application;
[0024] Figure 2 is a flowchart of another model optimization method of an HVAC provided by an embodiment of the present application;
[0025] Figure 3 Fig. 1 is a structural schematic diagram of a model optimization device for a heating, ventilation and air conditioning according to an embodiment of the present application;
[0026] Figure 4 Fig. 2 is a structural schematic diagram of an electronic device for implementing a model optimization method for a heating, ventilation and air conditioning according to an embodiment of the present application. DETAILED DESCRIPTION
[0027] In order to enable persons skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by persons skilled in the art without creative labor should fall within the scope of protection of the present application.
[0028] It should be noted that the terms "first", "second", and the like in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or a chronological sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but can include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0029] Commercial centers, shopping plazas and other facilities are equipped with heating, ventilation and air conditioning equipment. The heating, ventilation and air conditioning equipment is used to provide heating, ventilation and air conditioning for the facilities.
[0030] Currently, a control model of a heating, ventilation and air conditioning is used for control. However, when the heating, ventilation and air conditioning is controlled by the control model, there is a situation that the temperature and humidity in the facility do not meet the requirements or standards.
[0031] The inventor found that the Fine-tuning method only focuses on the optimization of model parameters, but does not fully consider the optimization strategy of business constraints and multi-task fine-tuning, resulting in that the heating, ventilation and air conditioning control according to the control parameters output by the control model does not meet the business requirements or standards. How to improve the accuracy of the control of the heating, ventilation and air conditioning becomes a problem to be solved.
[0032] The technical scheme provided by the embodiment of the present application introduces a service constraint as a regularization term into a first loss function used by a basic model, and adopts a multi-task learning strategy to simultaneously optimize multiple related tasks, thereby improving the feasibility and precision of Fine-tuning.
[0033] Figure 1 is a flowchart of a model optimization method for a heating ventilation air conditioner provided by the embodiment of the present application. The embodiment can be applicable to the case of fine-tuning training of a basic model for heating ventilation air conditioner control. The method can be executed by a model optimization device for a heating ventilation air conditioner. The model optimization device for a heating ventilation air conditioner can be realized in the form of hardware and / or software. The model optimization device for a heating ventilation air conditioner can be configured in an electronic device such as a personal computer or a server. Figure 1 As shown in the figure, it comprises:
[0034] S110, a basic model for heating ventilation air conditioner control is acquired.
[0035] The initial model can be trained using general air conditioner control data to obtain a basic model for heating ventilation air conditioner control. The general air conditioner control data is not limited to a specific scenario. Any data that can be used to train the initial model for heating ventilation air conditioner control can be used.
[0036] S120, a service constraint condition is determined according to historical statistical data.
[0037] Optionally, the service constraint condition is determined according to historical statistical data. This can be implemented as:
[0038] The service constraint condition is determined according to the service requirement, static parameters and historical statistical data of the target scenario.
[0039] The collection of data includes static parameters and historical statistical data. Optionally, all static parameters that can affect the performance of the model are collected from the relevant business field. For example, in a heating ventilation air conditioner (HVAC) system, the maximum and minimum temperature limits of the equipment, the humidity control range, etc. are acquired as static parameters.
[0040] In addition to the static parameters, a preliminary statistical analysis can be performed on a small amount of historical data to identify potential dynamic service constraints. This can be achieved by performing a simple statistical analysis on the operation logs, sensor readings, etc. data in the past period of time.
[0041] Based on the above collected information, rules or constraint conditions reflecting the characteristics of the service are defined. For example, for a heating ventilation air conditioner system, the temperature fluctuation range should be within ±2 degrees; for a fluid transmission system, the maximum allowable flow rate is set to a specific value.
[0042] Optionally, an algorithm tool such as clustering analysis in machine learning, anomaly detection, etc. can be used to automatically extract additional business constraints from the data, and then discover implicit rules that are not obvious but are crucial to the system operation.
[0043] The above embodiment can accurately determine the business constraint condition according to the demand and historical statistics data, and then improve the accuracy of the model obtained by fine-tuning based on the business constraint condition.
[0044] S130, determining a second loss function according to the business constraint condition and a first loss function of the base model.
[0045] Optionally, determining a second loss function according to the business constraint condition and a first loss function of the base model can be implemented as:
[0046] determining a constraint regular expression according to the business constraint condition;
[0047] determining a second loss function according to the constraint regular expression and the first loss function, the first loss function being a loss function of the base model.
[0048] All identified business constraints are respectively converted into mathematical expressions suitable for being integrated into the loss function. For example, the temperature fluctuation range can be expressed as an interval constraint, and the flow rate upper limit can be directly used as a maximum value limit. According to different business constraint properties, a suitable regularization method is selected. The regularization method can be: L2 regularization for preventing overfitting, KL divergence for measuring the difference between two probability distributions, etc. Ensure that the selected regularization method can effectively reflect and enforce the corresponding business constraint.
[0049] The above embodiment can represent the business constraint through the constraint regular expression, so that the second loss function can accurately train the model based on the business constraint condition, and improve the training accuracy.
[0050] Optionally, determining a second loss function according to the constraint regular expression and the first loss function includes:
[0051] adding a constraint expression to the constraint regular expression in the first loss function;
[0052] configuring a parameter weight in the first loss function and a weight of the constraint expression.
[0053] The converted business constraint, i.e., the constraint regular expression, is added to the first loss function as a regular term. Not only can the model consider specific business requirements during training, but it can also help the model better adapt to actual application scenarios. To balance the relationship between the original task target and the newly introduced business constraints, the weight coefficients of each regular term need to be set reasonably. Usually, multiple experiments are needed to find the best configuration.
[0054] The above implementation can configure weights for the parameter and the constraint expression, and then determine the weight of each parameter in the second loss function according to the importance of the business constraint. Through definable weights, flexible training is achieved, and the applicability of training to different business constraint requirements is improved.
[0055] S140, according to the second loss function and the base model, performing parameter efficient fine-tuning to obtain a parameter efficient fine-tuning model.
[0056] A portion of the validation set data can be used to preliminarily test the modified model, check whether the expected business constraints are correctly implemented, and evaluate their impact on the performance of the parameter efficient fine-tuning model. According to the test results, adjust the definition of the business constraint, the conversion method, and the weight in the loss function until the satisfactory model performance and business compliance are achieved.
[0057] Parameter efficient fine-tuning is used to achieve training effects with a small amount of computing resources. To improve the efficiency of Fine-tuning and reduce the required computing resources, parameter efficient fine-tuning (PEFT) techniques such as Lora technology or Adapter technology are used for parameter efficient fine-tuning. Among them, the LoRA technology is to update the weights by adding a low-rank matrix to the model (only a small number of parameters are added, similar to a "fine-tuning shortcut"), which allows the model to quickly adapt to new tasks. The Adapter technology is to insert a small trainable module (similar to a "plug-in") inside the model, and only adjust these parts rather than the entire model.
[0058] Further, before determining the second loss function according to the first loss function of the base model and the business constraint, it further includes:
[0059] The base model is subjected to self-supervised training to obtain a self-supervised model.
[0060] Self-supervised training is used to enable the base model to learn the basic rules based on time series. Unlabeled time series data can be used to enhance the learning ability of the model (let the model learn like "guessing a riddle"). Even without label information (such as temperature, humidity, etc.), the model can learn an effective feature representation, preparing for subsequent tasks. However, by using MaskedTime-series Modeling method, some segments in the time series are randomly masked, and the model is trained to predict these masked parts. This helps the model learn the dynamic relationships and patterns within the time series. For example, HVAC in seasonal changes, trend correlations, etc.
[0061] The above embodiment can realize fine-tuning training of the base model through self-supervised learning, and improve the accuracy of the model.
[0062] After self-supervised training of the base model, further comprising:
[0063] According to the control data of the HVAC in the target scene, the self-supervised model is supervised trained to obtain a supervised fine-tuning model.
[0064] After completing the preliminary self-supervised fine-tuning, the HVAC specific business target can be turned to. The actual business data set with clear labels (such as temperature and humidity at fixed time) is used to further optimize the model. The goal of this step is to improve the performance of the model in specific business scenarios, like "professional training", which enables the model to switch from general analysis to specific tasks (such as predicting tomorrow's temperature).
[0065] Correspondingly, the second loss function is determined according to the business constraint condition and the first loss function of the base model, comprising:
[0066] The second loss function is determined according to the business constraint condition and the first loss function of the supervised fine-tuning model.
[0067] The above embodiment can further fine-tune the model using supervised training method, thereby improving the accuracy of the parameter efficient fine-tuning model and making it more suitable for HVAC control scenarios.
[0068] Further, after obtaining the parameter efficient fine-tuning model, further comprising:
[0069] According to the output result of the parameter efficient fine-tuning model when performing multiple temperature control tasks, the business demand is verified; according to the verification result, the weight in the second loss function is adjusted until the output result of the parameter efficient fine-tuning model is verified successfully.
[0070] The verification methods can include but are not limited to: multi-dimensional evaluation system (not only looking at accuracy, but also looking at running delay, memory occupation, etc.), automatic AB testing (simultaneously running new and old models to compare effects) or online learning mechanism (continuously collecting feedback data to optimize the model after deployment).
[0071] If the evaluation result shows that the performance of the model does not meet the expectation or needs to be improved in some aspects, we need to go back to the Fine-tuning stage and make corresponding adjustments according to the feedback. For example, it may be necessary to increase or modify certain business constraints or adjust the weights of model parameters. The above process may be repeated several times until the model can stably meet the business requirements. The ultimate goal is to ensure that the model can reliably operate under complex conditions in the real world.
[0072] The above implementation realizes model training in different business weight scenarios through adjustment of the parameter weight in the second loss function, and improves ease of use.
[0073] The model optimization method of the heating ventilation air conditioner provided in the embodiment of the present application comprises the following steps: obtaining a basic model for heating ventilation air conditioner control; determining a business constraint condition according to historical statistical data; determining a second loss function according to a first loss function of the basic model and the business constraint condition; and performing parameter efficient fine-tuning according to the second loss function and the basic model to obtain a parameter efficient fine-tuning model. Compared with the current situation of insufficient accuracy of heating ventilation air conditioner control, the model optimization method of the heating ventilation air conditioner provided in the embodiment of the present application can modify the first loss function of the basic model according to the business constraint condition to obtain the second loss function. The parameter efficient fine-tuning of the basic model based on the second loss function can fine-tune the basic model in combination with the business constraint condition, and the obtained parameter efficient fine-tuning model can meet the requirements of the business constraint condition, thereby improving the control accuracy of the heating ventilation air conditioner.
[0074] The model optimization method of the heating ventilation air conditioner provided in the embodiment of the present application customizes the training scheme of the model according to actual business requirements (such as the accuracy requirement of predicted temperature and humidity, temperature control response speed, hardware limitation, etc.). For example, the temperature fluctuation range during equipment operation, the upper limit of fluid transmission rate and the like are important business constraints. In order to ensure that the model works under these constraints, these rules are converted into mathematical expressions and added as regularization terms to the loss function. This can help the model learn how to make accurate predictions while complying with business rules. Based on the business constraints, the pre-trained basic model is fine-tuned (Fine-tuning) to make the model more suitable for specific business environment. Not only can the accuracy of the model be improved, but also the compliance with industry standards and regulatory requirements in actual application can be ensured.
[0075] Figure 2Another flowchart of a model optimization method for a heating, ventilation and air conditioning system is provided in the embodiments of the present application, which comprises:
[0076] In S210, a basic model for heating, ventilation and air conditioning control is obtained.
[0077] In S220, a business constraint condition is determined according to a business requirement, static parameters and historical statistical data of a target scene.
[0078] In S230, a constraint regular expression is determined according to the business constraint condition.
[0079] In S240, a constraint expression is added to the constraint regular expression in the first loss function, and a parameter weight in the first loss function and a weight of the constraint expression are configured.
[0080] In S250, a self-supervised training is performed on the basic model to obtain a self-supervised model.
[0081] In S260, a supervised fine-tuning model is obtained by performing a supervised training on the self-supervised model according to control data of the heating, ventilation and air conditioning system in the target scene.
[0082] In S270, a second loss function is determined according to the business constraint condition and a first loss function of the supervised fine-tuning model.
[0083] In S280, a parameter efficient fine-tuning model is obtained by performing a parameter efficient fine-tuning according to the second loss function and the supervised fine-tuning model.
[0084] In S290, a business requirement verification is performed according to an output result of the parameter efficient fine-tuning model when performing a plurality of temperature control tasks, and a weight in the second loss function is adjusted according to a verification result until the output result of the parameter efficient fine-tuning model is verified successfully.
[0085] Figure 3 A structure diagram of a model optimization device for a heating, ventilation and air conditioning system is provided in the embodiments of the present application, which is applicable to the case of fine-tuning training of a basic model for heating, ventilation and air conditioning control, for example, Figure 3 As shown in the figure, the device comprises a basic model obtaining module 31, a business constraint determining module 32, a loss function determining module 33 and a parameter efficient fine-tuning module 34.
[0086] The basic model obtaining module 31 is used to obtain a basic model for heating, ventilation and air conditioning control.
[0087] The business constraint determining module 32 is used to determine a business constraint condition according to historical statistical data.
[0088] The loss function determination module 33 is configured to determine a second loss function according to the business constraint condition and a first loss function of the base model.
[0089] The parameter efficient fine-tuning module 34 is configured to perform parameter efficient fine-tuning according to the second loss function and the base model to obtain a parameter efficient fine-tuning model.
[0090] In the above embodiment, optionally, the loss function determination module 33 is configured to:
[0091] determine a constraint regular expression according to the business constraint condition;
[0092] determine the second loss function according to the constraint regular expression and the first loss function, the first loss function being a loss function of the base model.
[0093] In the above embodiment, optionally, the loss function determination module is configured to determine the second loss function according to the constraint regular expression and the first loss function, including:
[0094] adding a constraint expression in the first loss function.
[0095] configuring a parameter weight in the first loss function and a weight of the constraint expression.
[0096] In the above embodiment, optionally, further comprising a self-supervised learning module configured to perform self-supervised training on the base model to obtain a self-supervised model before determining the second loss function according to the business constraint condition and the first loss function of the base model.
[0097] In the above embodiment, optionally, further comprising a supervised learning module configured to perform supervised training on the self-supervised model according to control data of a heating, ventilation and air conditioning in a target scene to obtain a supervised fine-tuning model after performing self-supervised training on the base model.
[0098] Correspondingly, the loss function determination module 33 is configured to:
[0099] determine the second loss function according to the business constraint condition and a first loss function of the supervised fine-tuning model.
[0100] In the above embodiment, optionally, further comprising an iterative optimization module configured to perform business demand verification according to results of outputs of the parameter efficient fine-tuning model when performing a plurality of temperature control tasks after obtaining the parameter efficient fine-tuning model.
[0101] adjust weights in the second loss function according to the verification results until the results of the outputs of the parameter efficient fine-tuning model are verified successfully.
[0102] On the basis of the above-mentioned embodiments, optionally, the service constraint determination module 32 is used for:
[0103] According to the service demand, static parameters and historical statistical data of the target scene, the service constraint condition is determined.
[0104] The model optimization device for the heating, ventilation and air conditioning provided in the embodiments of the present application comprises a basic model acquisition module 31, a service constraint determination module 32 and a loss function determination module 33. The basic model acquisition module 31 is used for acquiring a basic model for the control of the heating, ventilation and air conditioning. The service constraint determination module 32 is used for determining a service constraint condition according to historical statistical data. The loss function determination module 33 is used for determining a second loss function according to the service constraint condition and a first loss function of the basic model. The parameter efficient fine-tuning module 34 is used for performing parameter efficient fine-tuning according to the second loss function and the basic model, so as to obtain a parameter efficient fine-tuning model. Compared with the current situation of insufficient control accuracy of the heating, ventilation and air conditioning, the model optimization device for the heating, ventilation and air conditioning provided in the embodiments of the present application can modify the first loss function of the basic model according to the service constraint condition, so as to obtain the second loss function. The parameter efficient fine-tuning of the basic model based on the second loss function can fine-tune the basic model in combination with the service constraint condition, so that the parameter efficient fine-tuning model obtained can meet the requirement of the service constraint condition, and the control accuracy of the heating, ventilation and air conditioning is improved.
[0105] The model optimization device for the heating, ventilation and air conditioning provided in the embodiments of the present application can execute the model optimization method for the heating, ventilation and air conditioning provided in any of the embodiments of the present application, and has the function modules and beneficial effects corresponding to the execution method.
[0106] Figure 4 is a structural schematic diagram of an electronic device provided by the embodiments of the present application. The electronic device 10 is intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular telephones, smart phones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not intended to limit the implementations of the present application described and / or claimed in this document.
[0107] As Figure 4As shown, the electronic device 10 includes at least one processor 11, and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., communicatively connected to the at least one processor 11, where the memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer programs stored in the read-only memory (ROM) 12 or loaded into the random access memory (RAM) 13 from the storage unit 18. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.
[0108] Various components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc., an output unit 17, such as various types of displays, a speaker, etc., a storage unit 18, such as a magnetic disk, an optical disk, etc., and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunication networks.
[0109] The processor 11 can be various general and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 performs various methods and processes described above, such as the model optimization method for heating, ventilation, and air conditioning.
[0110] In some embodiments, the model optimization method for heating, ventilation, and air conditioning can be implemented as a computer program tangibly embodied in a computer readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the model optimization method for heating, ventilation, and air conditioning described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to perform the model optimization method for heating, ventilation, and air conditioning by any other appropriate means, such as by means of firmware.
[0111] Various implementations of the systems and techniques described above can be realized in digital electronic circuitry, integrated circuitry, specially designed application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.
[0112] Computer programs used to implement the model optimization method of heating, ventilation, and air conditioning of the present application can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the computer program running on the processor implements the functions / operations specified in the flow diagrams and / or block diagrams. The computer program can execute entirely on a machine, partly on a machine, partly on a machine as a stand-alone software package, partly on a machine and partly on a remote machine or entirely on a remote machine or server.
[0113] The embodiments of the application also provide a computer readable storage medium, which stores computer instructions for causing a processor to execute a model optimization method of heating, ventilation, and air conditioning, comprising:
[0114] Obtaining a basic model for heating, ventilation, and air conditioning control;
[0115] Determining a business constraint condition according to historical statistical data;
[0116] Determining a second loss function according to the business constraint condition and a first loss function of the basic model;
[0117] Performing parameter efficient fine-tuning according to the second loss function and the basic model to obtain a parameter efficient fine-tuning model.
[0118] On the basis of the above-mentioned embodiments, optionally, determining a second loss function according to the business constraint condition and a first loss function of the basic model comprises:
[0119] Determining a constraint regular expression according to the business constraint condition;
[0120] Determine a second loss function according to the constraint regular expression and a first loss function, the first loss function being a loss function of the base model.
[0121] On the basis of the above-mentioned embodiment, optionally, determining a second loss function according to the constraint regular expression and a first loss function, comprising:
[0122] Adding a constraint expression in the constraint regular expression in the first loss function;
[0123] Configuring a parameter weight in the first loss function and a weight of the constraint expression.
[0124] On the basis of the above-mentioned embodiment, optionally, before determining a second loss function according to the business constraint condition and a first loss function of the base model, further comprising:
[0125] Performing self-supervised training on the base model to obtain a self-supervised model.
[0126] On the basis of the above-mentioned embodiment, optionally, after performing self-supervised training on the base model, further comprising:
[0127] Performing supervised training on the self-supervised model according to control data of heating, ventilation and air conditioning in a target scene to obtain a supervised fine-tuning model;
[0128] Correspondingly, determining a second loss function according to the business constraint condition and a first loss function of the base model, comprising:
[0129] Determining a second loss function according to the business constraint condition and a first loss function of the supervised fine-tuning model.
[0130] On the basis of the above-mentioned embodiment, optionally, after obtaining the parameter efficient fine-tuning model, further comprising:
[0131] Verifying a business requirement according to an output result of the parameter efficient fine-tuning model when performing a plurality of temperature control tasks;
[0132] Adjusting a weight in the second loss function according to a verification result until the output result of the parameter efficient fine-tuning model is verified successfully.
[0133] On the basis of the above-mentioned embodiment, optionally, determining a business constraint condition according to historical statistical data, comprising:
[0134] Determining a business constraint condition according to a business requirement, a static parameter and historical statistical data of a target scene.
[0135] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. A computer-readable storage medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of a machine-readable storage medium will include one or more lines of a program of instructions in a transitory signal, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
[0136] To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.
[0137] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.
[0138] The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system, to solve the defects of large management difficulty and weak business scalability in traditional physical host and VPS service.
[0139] It should be understood that the various forms of flow shown above can be reordered, added to, or have steps deleted. For example, the steps described in the present application can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions of the present application can be achieved, which are not limited herein.
[0140] The above detailed description does not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present application shall be included in the protection scope of the present application.
Claims
1. A model optimization method for heating, ventilation, and air conditioning, characterized by, The method comprises the following steps: obtaining a base model for HVAC control; determining a business constraint condition according to historical statistical data; determining a second loss function according to a first loss function of the base model and the business constraint condition; performing parameter efficient fine-tuning according to the second loss function and the base model to obtain a parameter efficient fine-tuning model.
2. The method of claim 1, wherein, The method further comprises the following steps before determining the second loss function according to the first loss function of the base model and the business constraint condition: determining a constraint regular expression according to the business constraint condition; determining the second loss function according to the constraint regular expression and the first loss function, wherein the first loss function is a loss function of the base model.
3. The method of claim 2, wherein, The method further comprises the following steps of determining the second loss function according to the constraint regular expression and the first loss function: adding a constraint expression to the constraint regular expression in the first loss function; configuring a parameter weight in the first loss function and a weight of the constraint expression.
4. The method of claim 3, wherein, The method further comprises the following step before determining the second loss function according to the first loss function of the base model and the business constraint condition: performing self-supervised training on the base model to obtain a self-supervised model.
5. The method of claim 4, wherein, The method further comprises the following step after performing self-supervised training on the base model: performing supervised training on the self-supervised model according to control data of the HVAC in a target scene to obtain a supervised fine-tuning model; correspondingly, the method further comprises the following step of determining the second loss function according to the first loss function of the base model and the business constraint condition: determining the second loss function according to the first loss function of the base model and the business constraint condition.
6. The method of claim 5, wherein, The method further comprises the following step after obtaining the parameter efficient fine-tuning model: verifying a business requirement according to an output result of the parameter efficient fine-tuning model when performing multiple temperature control tasks; adjusting a weight in the second loss function according to a verification result until the output result of the parameter efficient fine-tuning model is verified successfully.
7. The method of claim 1, wherein, The method further comprises the following step of determining the business constraint condition according to historical statistical data: determining the business constraint condition according to a business requirement, a static parameter and historical statistical data of a target scene.
8. A model optimization device for HVAC systems, characterized in that, The method comprises the following steps: obtaining a base model for HVAC control; determining a business constraint condition according to historical statistical data; determining a second loss function according to a first loss function of the base model and the business constraint condition; performing parameter efficient fine-tuning according to the second loss function and the base model to obtain a parameter efficient fine-tuning model.
9. An electronic device, comprising: The electronic device comprises: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to perform the HVAC model optimization method in any one of claims 1-7.
10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions for enabling the processor to perform the HVAC model optimization method in any one of claims 1-7 when executed.