Water chilling unit fault diagnosis method based on interaction of large language model and simulation system

Abstract

The invention discloses a water chilling unit fault diagnosis method based on interaction of a large language model and a simulation system, and relates to the technical field of comprehensive energy, and the water chilling unit fault diagnosis method based on interaction of the large language model and the simulation system comprises the steps: obtaining target operation data of a water chilling unit; inputting the target operation data into a fault diagnosis model to obtain a fault diagnosis result output by the fault diagnosis model; wherein the fault diagnosis model is a model obtained by using simulated fault data as training data and training by using a domain adaptation migration technology, and the simulated fault data is obtained based on interaction of a large language model and a simulation system. By applying the scheme provided by the embodiment of the invention, the accuracy of fault diagnosis of the water chilling unit can be improved.

Description

Technical Field

[0001] The present invention relates to the field of integrated energy technology, and in particular to a chiller fault diagnosis method based on interaction between a large language model and a simulation system. Background Art

[0002] As the core component of the air conditioning refrigeration system, the chiller plays an important role. In the air conditioning system, the energy consumption of the chiller accounts for about 50% of the total energy consumption of the air conditioning system. Due to the high efficiency and large cooling capacity of the chiller, engineers will give priority to it in the high-efficiency machine room.

[0003] Since chillers are expensive, their maintenance has become an important part of energy-saving control. During the maintenance process, the essence is to diagnose the operation of the chiller, to facilitate timely warning of faults and to assist maintenance workers in troubleshooting.

[0004] Currently, there are many methods for fault diagnosis of chillers, such as applying AI models to model the historical operation data of chillers and using them for real-time fault diagnosis tasks. However, this method has the problem of relying on historical fault samples in actual engineering applications. When there are fewer faults in the historical operation of the chiller, the accuracy of the trained model in fault diagnosis is limited. Summary of the invention

[0005] The present invention aims to solve one of the technical problems in the related art at least to a certain extent. To this end, one object of the present invention is to propose a chiller fault diagnosis method based on the interaction of a large language model and a simulation system to improve the accuracy of chiller fault diagnosis.

[0006] According to a first aspect of an embodiment of the present invention, a chiller fault diagnosis method based on interaction between a large language model and a simulation system is provided, the method comprising:

[0007] Obtain target operating data of the chiller;

[0008] Inputting the target operation data into a fault diagnosis model to obtain a fault diagnosis result output by the fault diagnosis model;

[0009] The fault diagnosis model is a model trained by using simulated fault data as training data and utilizing domain adaptation migration technology, wherein the simulated fault data is obtained based on the interaction between a large language model and a simulation system.

[0010] According to a second aspect of an embodiment of the present invention, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned chiller fault diagnosis method based on the interaction between a large language model and a simulation system is implemented.

[0011] According to a third aspect of an embodiment of the present invention, there is provided an electronic device, comprising: a memory and a processor; a computer program is stored on the memory, and when the computer program is executed by the processor, the above-mentioned chiller fault diagnosis method based on the interaction of a large language model and a simulation system is implemented.

[0012] In the solution provided by the embodiment of the present invention, fault parameters can be obtained and simulated fault data can be generated when the user inputs prompt text. Faults occurring in the actual environment can be simulated according to the prompt text, and the model can be trained using domain adaptation migration technology, so that a large number of fault sample libraries can be obtained to make up for the lack of fault samples in the historical operation process, so that the model can learn the ability to identify faults from simulated fault data, so that when the target data is input, fault diagnosis can be implemented according to the parameters in the target operation data, thereby improving the accuracy of chiller fault diagnosis using a neural network model and reducing the degree of dependence on historical fault data of the chiller.

[0013] The simulated fault data is obtained based on the interaction between the large language model and the simulation system. While generating more data through the large model, the simulated fault data can more accurately simulate the conditions of a real chiller when a fault occurs under the processing of the simulation system, thereby providing the accuracy of the model in identifying faults when training the model.

[0014] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flow chart of a chiller fault diagnosis method based on interaction between a large language model and a simulation system provided by an embodiment of the present invention;

[0016] Figure 2 is a schematic diagram of a first user interface for interacting with a large language model provided by an embodiment of the present invention;

[0017] Figure 3 is a schematic diagram of a second user interface for interacting with a large language model provided by an embodiment of the present invention;

[0018] Figure 4 is a schematic diagram of a third user interface for interacting with a large language model provided by an embodiment of the present invention;

[0019] Figure 5 It is a flowchart of a model training method provided by an embodiment of the present invention;

[0020] Figure 6It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0021] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and should not be construed as limiting the present invention.

[0022] The following describes a chiller fault diagnosis method, storage medium, and electronic device based on interaction between a large language model and a simulation system according to an embodiment of the present invention with reference to the accompanying drawings.

[0023] In one embodiment of the present invention, see Figure 1 , a chiller fault diagnosis method based on the interaction of a large language model and a simulation system is provided, and the method includes the following steps S101-S102.

[0024] S101: Obtain target operation data of the chiller;

[0025] S102: inputting the target operation data into the fault diagnosis model to obtain the fault diagnosis result output by the fault diagnosis model;

[0026] Among them, the fault diagnosis model is: a model trained by using simulated fault data as training data and using domain adaptation migration technology, and the simulated fault data is obtained based on the interaction between a large language model and a simulation system.

[0027] The target operation data may be operation data detected in real time, including real-time status parameters of the chiller, such as real-time chilled water flow, cooling water flow, and the like.

[0028] The fault diagnosis result may include a description of whether a fault exists, or a specific fault type.

[0029] Specifically, the interactive process is: the large language model obtains fault parameters according to the prompting project, the simulation system processes the fault parameters, and the simulated fault data is obtained according to the processing result of the simulation system. See the following embodiment for details.

[0030] In the solution provided by the embodiment of the present invention, fault parameters can be obtained and simulated fault data can be generated when the user inputs prompt text. Faults occurring in the actual environment can be simulated according to the prompt text, and the model can be trained using domain adaptation migration technology, so that a large number of fault sample libraries can be obtained to make up for the lack of fault samples in the historical operation process, so that the model can learn the ability to identify faults from simulated fault data, so that when the target data is input, fault diagnosis can be implemented according to the parameters in the target operation data, thereby improving the accuracy of chiller fault diagnosis using a neural network model and reducing the degree of dependence on historical fault data of the chiller.

[0031] The simulated fault data is obtained based on the interaction between the large language model and the simulation system. While generating more data through the large model, under the processing of the simulation system, the simulated fault data can more accurately simulate the conditions when a real operating chiller fails, thereby providing the accuracy of model fault identification when training the model.

[0032] Specifically, the fault diagnosis model domain is trained using adaptive transfer technology, including:

[0033] The preset neural network model is trained using simulated fault data as source domain data and historical operation data of the chiller as target domain data.

[0034] Specifically, the features extracted from the source domain data and the target domain data can be mapped to the same feature space to form an adversarial relationship, and a label predictor can be used to obtain the classification results of the mapping features after the source domain data and the target domain data are mapped; and a domain discriminator can be used to discriminate the domain to which each mapping feature belongs, that is, to determine whether a certain mapping feature is source domain data or target domain data before mapping.

[0035] When calculating the loss, the loss of the label predictor and the loss of the domain discriminator are calculated comprehensively, and the model parameters of the network model are adjusted based on the comprehensively calculated loss. The loss of the label predictor is the difference between the fault type of the classification result predicted by the label predictor and the fault type of the labeled source domain data. The labeled fault type can be generated by a large language model, see the subsequent embodiments; the loss of the domain discriminator is whether the domain judged by the domain discriminator is the domain to which each mapping feature actually belongs.

[0036] In one embodiment of the present invention, the fault diagnosis model is trained in the following manner.

[0037] Obtain historical operating data of the chiller;

[0038] Get the prompt text entered by the user;

[0039] Inputting the prompt text into the large language model, obtaining the fault parameters output by the large language model and conforming to the fault indicated by the prompt text;

[0040] Input the fault parameters into the simulation system to obtain the performance coefficient output by the simulation system;

[0041] obtaining simulated failure data based on performance coefficients;

[0042] The preset neural network model is trained using simulated fault data as source domain data and historical operation data of the chiller as target domain data.

[0043] The historical operation data refers to the state parameters representing the operation state of the chiller during the historical operation of the chiller, which may include state parameters representing a fault or no fault. The fault parameter in step S103 is a parameter of the same type as the simulated fault parameter.

[0044] The working parameters may include: chilled water flow, cooling water flow, chilled water return temperature, cooling water return temperature, host temperature, etc., see the following embodiments. The host refers to the air conditioner host.

[0045] The user end may be a user interface provided to the user, so that the user enters a prompt text in a designated text box on the user interface.

[0046] Large language model, the full name is Large Language Model (LLM).

[0047] LLM refers to a large-scale language model. A language model is a machine learning model used to understand and generate natural language text. LLM is usually used to describe deep learning models with billions to hundreds of billions of parameters, which can learn the statistical laws of large-scale language data to generate natural and fluent text, or perform various natural language processing tasks. Wenxinyiyan, ChatGPT, etc. are applications developed based on LLM. They can generate fluent, logical, and creative text content, and can even have natural conversations with humans. Specifically, a large language model can be a Generative Pre-trained Transformer (GPT) model based on Transformer, or an Enhanced Representation through Knowledge Integration (ERNIE) model based on knowledge integration.

[0048] Based on prompt engineering, the large language model can respond to prompt text described in natural language, including online prompts and offline prompts.

[0049] For example, you can tell the large language model in the prompt text: "Generate a "device" object for me", specify the type of state parameters that describe "device", and set some initial values ​​of the state parameters. The state parameters of the output of the large language model can be obtained through Figure 2 The interface shown is displayed.

[0050] Figure 2 In the example, the output of the large language model includes the prompt word "device object" and different state parameters that are set to initial values, including:

[0051] Refrigeration season flow, parameter value is 100; chilled water flow, parameter value is 50; cooling water flow, parameter value is 30; chilled water return temperature, parameter value is 12; cooling water return temperature, parameter value is 25; host temperature setting value, parameter value is 9; evaporator heat exchange area, parameter value is 10; condenser heat exchange area, parameter value is 5.

[0052] In this case, the large language model can be instructed in the prompt text to output the state parameters of the device object with the fault as the fault parameter, and the fault parameter has a parameter value different from the initial one.

[0053] The acquisition of historical operation data can be achieved by pre-setting measuring equipment. For example, pre-setting temperature sensors can be used to obtain historical state parameters such as the host temperature, cooling water return temperature, and chilled water return temperature of the chiller during its historical operation, thereby forming historical operation data based on the historical state parameters.

[0054] In step S105, the neural network model may be trained in the following manner.

[0055] In labeled training, historical operating data and simulated fault data can be annotated to generate labels, such as marking the presence or absence of a fault and the type of fault in the label. Specifically, manual labeling or large language model output labeling can be used.

[0056] The simulated fault data and historical operation data can be respectively input into the preset neural network model to obtain the fault type predicted by the neural network model, or to express the prediction result of whether there is a fault or not. The loss is calculated by the difference between the output and the annotation of the neural network model, and the model parameters of the neural network model are adjusted according to the calculated loss until the model converges, thereby realizing the training of the neural network model.

[0057] In one embodiment, the neural network model is a model set for a device type.

[0058] When diagnosing chillers of different equipment types, the fault diagnosis models corresponding to the equipment types are pre-trained. When diagnosing, the appropriate fault diagnosis model is selected from the trained models according to the equipment type. In this way, different chillers can be effectively distinguished when the working states and corresponding fault parameters are different when they fail.

[0059] In unlabeled training, source domain data from non-chillers, that is, simulated fault data with fault labels, can be used, combined with target domain unlabeled data of target devices, that is, historical operation data, combined with domain adaptation migration technology to achieve the migration of fault knowledge. Among them, unlabeled data can be used as fault-free sample data.

[0060] Specifically, the neural network model may be a domain adversarial neural network (DANN) model. The model includes a feature extraction layer and a fault classification layer. The feature extraction layer is used to extract features according to input data, and the fault classification layer is used to determine the fault type according to the extracted features.

[0061] In addition, during training, the simulated fault data and the historical operation data may be standardized to obtain sample fault data corresponding to the simulated fault data and the historical operation data; and the preset neural network model may be trained based on the sample fault data.

[0062] The standardization process can include the use of data conversion tools and middleware to convert simulated fault data and historical operation data of different formats and structures into a unified format. The data standardization process can include data cleaning, data format conversion and normalization to ensure data consistency and availability.

[0063] The specific implementation method of obtaining simulated fault data is described below.

[0064] In one embodiment, the performance coefficient output by the simulation system may be obtained in the following manner.

[0065] Inputting fault parameters and equipment type of the chiller into the simulation system, and determining a target simulation model corresponding to the equipment type among simulation models constructed for various chillers included in the simulation system;

[0066] The performance coefficient corresponding to the fault parameters output by the target simulation model is obtained.

[0067] The equipment type of the chiller can be identified by the host model, specifications, brand and other identification information.

[0068] Determine the target simulation model corresponding to the equipment type: When constructing simulation models for various chillers, the equipment type of the chiller corresponding to the established simulation model is recorded in advance, so that the target simulation model can be obtained by matching the equipment type.

[0069] Specifically, the target simulation model includes:

[0070] COP=A1+A2T L,in +A3T S,in +A4V L +A5V S

[0071] Among them, COP represents the coefficient of performance; the fault parameters include: the inlet temperature T of the main evaporator side L,in , Main engine condenser side inlet temperature T S,in , Circulating water flow rate V on the main engine evaporator side L , Circulating water flow rate V on the main engine condenser side S ; A1, A2, A3, A4, A5 are preset weight coefficients.

[0072] Among them, the inlet temperature on the main engine evaporator side is the chilled water return temperature; the inlet temperature on the main engine condenser side is the cooling water return temperature; the circulating water flow on the main engine evaporator side is the chilled water flow, and the circulating water flow on the main engine condenser side is the cooling water flow.

[0073] The coefficient of performance (COP), also known as the refrigeration performance coefficient, refers to the amount of cooling that can be obtained per unit power consumption.

[0074] The simulation models corresponding to the chillers of different equipment types can all calculate the coefficient of performance in a manner similar to the above formula. The difference is that the specific values ​​of A1, A2, A3, A4, and A5 are different in different simulation models.

[0075] Specifically, the weight coefficient is obtained according to the parameter fitting result of the sample performance data of the corresponding device type.

[0076] The sample performance data may include: for a certain chiller, fault parameters and corresponding performance coefficients recorded at multiple moments in a faulty working state.

[0077] That is, for chillers of different equipment types, different data sets of different chillers are obtained through manufacturer samples or actual project data, so that the corresponding relationship between the performance coefficient and each fault parameter can be obtained from each data set and substituted into the above formula for parameter fitting.

[0078] In this way, the environmental parameters input by the user end can be obtained; and simulated fault data including fault parameters, environmental parameters and performance coefficients can be generated.

[0079] Among them, the environmental parameters are parameters that simulate the actual environment, which may specifically be the room temperature, the heat transfer coefficient inside and outside the chiller, etc.

[0080] There is a difference in performance coefficient between a chiller with a fault and a chiller without a fault. Thus, by calculating the performance coefficient in simulation data and training the neural network model, the neural network model can learn the ability to detect faults based on performance changes of the chiller.

[0081] The simulation system can simulate the operation of the chiller according to the above simulation model, and obtain the performance generated during the simulated operation, such as power and energy consumption, under the operating conditions represented by the fault parameters and environmental parameters, and add it to the simulated fault data.

[0082] Different simulation systems can provide different communication interfaces for inputting fault parameters and environmental parameters, such as the Trnsystor interface of the TRNSYS system and the eppy interface of the EnergyPlus system.

[0083] The following describes a specific implementation method of inputting the prompt text into the large language model and obtaining the fault parameters.

[0084] In one implementation, a prompt text describing the fault type is input into a large language model, and a fault parameter corresponding to the fault type output by the large language model is obtained;

[0085] In this manner, the prompt text may include the fault type specified by the user, so that the large language model outputs fault parameters that meet the characteristics of this type of fault.

[0086] For example, using the "equipment" object in the aforementioned embodiment, when it is necessary to simulate the fault type of "chilled water flow rate is too low", the corresponding prompt text may be "Simulate the fault of chilled water flow rate is too low, change the parameters of this object".

[0087] The output of the corresponding large language model is as follows Figure 3 Relative to the initial values ​​of the parameters of the "device" object, the large language model adjusts the output parameter values ​​to adapt to the fault type newly input by the user.

[0088] In the output, "Simulated fault: Chilled water flow is too low" is noted after the equipment object to explain the fault type extracted by the large language model from the prompt text.

[0089] The corresponding fault parameters include: cooling season flow, parameter value is 100; chilled water flow, parameter value is 20, and the parameter value is annotated in the output as a fault value, the normal value should be 50; cooling water flow, parameter value is 30; chilled water return temperature, parameter value is 12; cooling water return temperature, parameter value is 25; host temperature setting value, parameter value is 9; evaporator heat exchange area, parameter value is 10; condenser heat exchange area, parameter value is 5.

[0090] In addition, the large language model can output how the fault parameters are adjusted and generated, such as Figure 3 "Note" text in the text: In the above equipment object, I have changed the value of "Chilled Water Flow" to 20 to simulate the fault condition of too low chilled water flow. In actual situations, this value should be adjusted according to the specific fault conditions.

[0091] In another implementation, a prompt text that does not describe the fault type is input into the large language model to obtain a fault parameter corresponding to the random fault type output by the large language model.

[0092] For example, using the "device" object in the aforementioned embodiment, the prompt text may be "Think of a failure of a chiller and express it by adjusting the parameters of this object."

[0093] The output of the corresponding large language model is as follows Figure 4 As shown in the figure, the simulated fault of the equipment object is: the cooling water return temperature is too high. This is a fault generated by the large language model, and the fault type is not specified by the user.

[0094] The corresponding fault parameters include: cooling season flow, parameter value is 100; chilled water flow, parameter value is 50; cooling water flow, parameter value is 30; chilled water return temperature, parameter value is 12; cooling water return temperature, parameter value is 35, and it is noted that the parameter value is the fault value, the normal value should be 25; host temperature setting value, parameter value is 9; evaporator heat exchange area, parameter value is 10; condenser heat exchange area, parameter value is 5.

[0095] The corresponding "Note" text is: In this simulated fault, I adjusted the value of "Cooling Water Return Temperature" from 25 to 35 to indicate the fault condition of too high cooling water return temperature. This may be caused by reduced cooling tower efficiency, insufficient cooling water flow, or other cooling system-related problems. In actual applications, such a fault may cause reduced cooling efficiency or equipment damage, which requires timely investigation and repair.

[0096] By generating random fault types, data other than user input and user experience can be generated to enrich training samples.

[0097] In one embodiment of the present invention, when inputting a prompt text, the prompt text and the search information corresponding to the prompt text may be input into a large language model to obtain fault parameters in the search enhanced reasoning result output by the large language model.

[0098] When retrieval enhancement is achieved through retrieval information, a knowledge base for storing the retrieval information can be preset, and the retrieval information is the characteristics of different faults of the chiller and related parameter value information, so that the large language model can learn how to set reasonable parameter values ​​that represent the faults through the retrieval information.

[0099] The specific retrieval process may include: dividing the prompt text into text blocks, using an encoder to embed the text blocks into feature vectors, generating an index based on the vector to obtain context information in the knowledge base as input to the large language model, so that the large language model outputs retrieval enhanced reasoning results, and generates answers to the prompt text in the results based on the context.

[0100] Retrieval enhancement technology is used to strengthen the large language model's understanding of specific domain knowledge and enhance the reliability of fault object parameters, thereby being able to handle the problem of training sample coverage when some fault information has a deep industry verticality.

[0101] The following Figure 5 The flowchart diagram shown illustrates the overall process of the model training method provided by an embodiment of the present invention.

[0102] In the data preparation stage, auxiliary technologies including prompt engineering and retrieval enhancement are used to enable the large language model to output fault parameters according to the user's prompt text, and interact with the simulation system to input the fault parameters into the simulation model through the communication interface, that is, the simulation model of the chiller constructed by the simulation system, to generate the performance coefficient, that is, the simulated fault data obtained based on the simulation system, including fault label, chilled water flow, cooling water flow, chilled water inlet temperature, chilled water outlet temperature, cooling water inlet temperature, chilled water outlet temperature, etc., and the generated performance coefficient, so as to obtain the source domain data; the target domain data is the historical operation data of the chiller.

[0103] In the data application stage, the source domain data and the target domain data are input into the DANN deep neural network model for model training to obtain an instantiated model, i.e., a fault diagnosis model. Real-time data is input into the instantiated model for model reasoning to obtain a judgment result, which may include the fault type detected by the model.

[0104] In one embodiment of the present invention, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, any of the above-mentioned chiller fault diagnosis methods based on the interaction of a large language model and a simulation system is implemented.

[0105] In one embodiment of the present invention, an electronic device is provided, comprising: a memory and a processor; a computer program is stored on the memory, and when the computer program is executed by the processor, any of the above-mentioned chiller fault diagnosis methods based on the interaction of a large language model and a simulation system is implemented.

[0106] Figure 6 It is a structural block diagram of an electronic device according to an embodiment of the present invention.

[0107] like Figure 6 As shown, the electronic device 600 includes: a processor 601 and a memory 603. The processor 601 and the memory 603 are connected, such as through a bus 602. Optionally, the electronic device 600 may further include a transceiver 604. It should be noted that in actual applications, the transceiver 604 is not limited to one, and the structure of the electronic device 600 does not constitute a limitation on the embodiments of the present invention.

[0108] Processor 601 may be a CPU (Central Processing Unit), a general purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of the present invention. Processor 601 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0109] The bus 602 may include a path to transmit information between the above components. The bus 602 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 602 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 6 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0110] The memory 603 is used to store a computer program corresponding to the chiller fault diagnosis method based on the interaction between a large language model and a simulation system according to the above embodiment of the present invention, and the computer program is controlled and executed by the processor 601. The processor 601 is used to execute the computer program stored in the memory 603 to implement the contents shown in the above method embodiment.

[0111] Among them, the electronic device 600 includes but is not limited to: mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 6 The electronic device 600 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0112] It should be noted that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection portion with one or more wirings (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0113] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0114] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0115] In the description of the present invention, it is to be understood that the terms “center”, “longitudinal”, “lateral”, “length”, “width”, “thickness”, “up”, “down”, “front”, “back”, “left”, “right”, “vertical”, “horizontal”, “top”, “bottom”, “inside”, “outside”, “clockwise”, “counterclockwise”, “axial”, “radial”, “circumferential”, etc., indicating orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0116] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0117] In the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0118] In the present invention, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may mean that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. Moreover, a first feature being "above", "above" or "above" a second feature may mean that the first feature is directly above or obliquely above the second feature, or simply means that the first feature is higher in level than the second feature. A first feature being "below", "below" or "below" a second feature may mean that the first feature is directly below or obliquely below the second feature, or simply means that the first feature is lower in level than the second feature.

[0119] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A chiller fault diagnosis method based on interaction between a large language model and a simulation system, characterized in that: The method comprises: Obtain target operating data of the chiller; Inputting the target operation data into a fault diagnosis model to obtain a fault diagnosis result output by the fault diagnosis model; The fault diagnosis model is a model trained by using simulated fault data as training data and utilizing domain adaptation migration technology, wherein the simulated fault data is obtained based on the interaction between a large language model and a simulation system.

2. The method according to claim 1, characterized in that: The fault diagnosis model domain is trained using adaptive migration technology, including: The preset neural network model is trained using the simulated fault data as source domain data and the historical operation data of the chiller as target domain data.

3. The method according to claim 2, characterized in that The fault diagnosis model is trained in the following manner: Obtaining historical operation data of the chiller; Get the prompt text entered by the user; Inputting the prompt text into a large language model to obtain fault parameters output by the large language model that are consistent with the fault represented by the prompt text; Inputting the fault parameters into a simulation system to obtain a performance coefficient output by the simulation system; obtaining simulated fault data based on the performance coefficient; The preset neural network model is trained using the simulated fault data as source domain data and the historical operation data of the chiller as target domain data.

4. The method according to claim 3, characterized in that The step of inputting the fault parameter into a simulation system to obtain a performance coefficient output by the simulation system includes: Inputting the fault parameter and the equipment type of the chiller into a simulation system, and determining a target simulation model corresponding to the equipment type among simulation models constructed for a variety of chillers included in the simulation system; A performance coefficient corresponding to the fault parameter output by the target simulation model is obtained.

5. The method according to claim 4, characterized in that The target simulation model comprises: COP=A1+A2T L,in +A3T S,in +A4V L +A5V S Wherein, COP represents the coefficient of performance; the fault parameters include: the inlet temperature T of the main engine evaporator side L,in , Main engine condenser side inlet temperature T S,in , Circulating water flow rate V on the main engine evaporator side L , Circulating water flow rate V on the main engine condenser side S ; A1, A2, A3, A4, A5 are preset weight coefficients; the weight coefficients are obtained based on the parameter fitting results of the sample performance data corresponding to the device type.

6. The method according to claim 3, characterized in that The obtaining of simulated fault data based on the performance coefficient comprises: Acquire environmental parameters input by the user terminal; generating simulated fault data including the fault parameter, the environmental parameter and the performance coefficient; The step of inputting the prompt text into a large language model to obtain a fault parameter output by the large language model that is consistent with the fault indicated by the prompt text includes: Inputting the prompt text describing the fault type into a large language model to obtain a fault parameter corresponding to the fault type output by the large language model; and / or, Inputting the prompt text that does not describe the fault type into a large language model to obtain a fault parameter corresponding to the random fault type output by the large language model; and / or, The prompt text and the search information corresponding to the prompt text are input into a large language model to obtain the fault parameters in the search enhanced reasoning result output by the large language model.

7. The method according to claim 2, characterized in that The method of using the simulated fault data as source domain data and the historical operation data of the chiller as target domain data to train a preset neural network model includes: Performing standardization processing on the simulated fault data and the historical operation data to obtain sample fault data corresponding to the simulated fault data and the historical operation data; A preset neural network model is trained based on the sample fault data.

8. The method according to claim 4, characterized in that The neural network model is a domain adversarial neural network (DANN) model, and the neural network model is a model set for the device type.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the chiller fault diagnosis method based on the interaction between a large language model and a simulation system as described in any one of claims 1 to 8 is implemented.

10. An electronic device, characterized in that: include: Memory, processor; The memory stores a computer program, and when the computer program is executed by the processor, the chiller fault diagnosis method based on the interaction of a large language model and a simulation system as described in any one of claims 1 to 8 is implemented.

Images