Working medium proportion optimization method of heat pump unit

Abstract

The invention relates to a working medium proportion optimization method of a heat pump unit. According to the method, multiple sample mixed working media are obtained, the matching proportions of multiple working media in different sample mixed working media are different, and based on each sample mixed working medium, operation parameters, corresponding to each sample mixed working medium, of the heat pump unit are obtained, and the multiple sample mixed working media and the operation parameters corresponding to the sample mixed working media serve as parameter samples, the initial model is trained based on the parameter samples, a process model of the heat pump unit is obtained, and the optimal proportion of the mixed working media of the heat pump unit is obtained based on the process model. According to the method, the parameter sample for training the initial model is obtained by obtaining the actually measured operation data corresponding to each sample mixed working medium, and the process model suitable for the heat pump unit under different use working conditions is trained to obtain the optimal ratio of the mixed working medium, so that the efficiency and the precision of screening the mass ratio of the non-azeotropic mixed working medium by the heat pump unit are improved.

Description

Technical Field

[0001] The present disclosure relates to the fields of thermodynamics and energy technology, and in particular to a method for optimizing the working fluid ratio of a heat pump unit. Background Art

[0002] With the continuous development of industrial heat pump technology, its application scenarios are also constantly expanding. A wider temperature range is a major problem currently faced. Conventional single working fluids cannot meet the requirements of a wider temperature range due to the limitations of their physical and chemical properties.

[0003] Non-azeotropic mixtures can meet wider temperature range requirements by adjusting the mass ratio of two or even more different working fluids. However, the current screening methods for the mass ratio of non-azeotropic mixtures still have problems with insufficient accuracy and efficiency. Summary of the Invention

[0004] In order to overcome the problems existing in the related art, the present disclosure provides a method for optimizing the working fluid ratio of a heat pump unit.

[0005] According to a first aspect of an embodiment of the present disclosure, a method for optimizing a working fluid ratio of a heat pump unit is provided, comprising:

[0006] Acquire multiple sample mixed working fluids, wherein different sample mixed working fluids contain different ratios of multiple working fluids;

[0007] Based on each of the sample mixed working fluids, obtaining the operating parameters of the heat pump unit corresponding to each of the sample mixed working fluids, wherein the plurality of sample mixed working fluids and the operating parameters corresponding to each of the sample mixed working fluids are used as parameter samples;

[0008] Training the initial model based on the parameter sample to obtain a process model of the heat pump unit;

[0009] Based on the process model, an optimal ratio of the mixed working fluid of the heat pump unit is obtained.

[0010] In some embodiments, the obtaining of multiple sample mixed working fluids, wherein different sample mixed working fluids have different ratios of multiple working fluids, includes:

[0011] Mixing at least two working fluids according to different ratios to obtain a plurality of initial sample mixed working fluids;

[0012] According to the working condition parameters required by the heat pump unit and the physical property parameters of the initial sample mixed working fluid, the initial sample mixed working fluid is initially screened to obtain a plurality of the sample mixed working fluids.

[0013] In some embodiments, based on each of the sample mixed working fluids, obtaining the operating parameters of the heat pump unit corresponding to each of the sample mixed working fluids, wherein the plurality of the sample mixed working fluids and the operating parameters corresponding to each of the sample mixed working fluids are used as parameter samples, includes:

[0014] Inputting a plurality of sample mixed working fluids into the heat pump unit in sequence for operation test;

[0015] Obtaining operating parameters corresponding to each of the sample mixed working fluids, wherein the operating parameters include the physical property parameters of the sample mixed working fluid, the fluctuation parameters of the heat pump unit, and the operating condition parameters of the heat pump unit;

[0016] Each sample mixed working medium and the corresponding operating parameter are used as the parameter sample.

[0017] In some embodiments, the initial model includes a transformer model with a six-layer stacked architecture.

[0018] In some embodiments, the training of the initial model based on the parameter sample to obtain the process model of the heat pump unit includes:

[0019] Preprocessing the operating parameters corresponding to each sample mixed working medium in the parameter sample to obtain a preprocessed parameter sample;

[0020] Using a multi-head attention mechanism to process the preprocessed parameter samples and configure associated weights;

[0021] Setting weight parameters for the parameters in the preprocessed parameter sample;

[0022] Constructing a feedforward neural network to enhance the nonlinear expression capability of the preprocessed parameter samples and the corresponding weight parameters;

[0023] Iterative optimization is performed using the coefficient of performance, volumetric heating capacity, global warming potential, and flammability index of the heat pump unit as optimization targets;

[0024] Determining whether the sample mixed working fluid in the iterative optimization result meets the fluctuation requirement of the fluctuation parameter of the heat pump unit during operation;

[0025] If the sample mixed working fluid in the iterative optimization result meets the volatility requirement of the fluctuation parameter of the heat pump unit during operation, the mixed working fluid ratio is output as the optimal ratio of the mixed working fluid of the heat pump unit;

[0026] If the sample mixed working fluid in the iterative optimization result does not meet the volatility requirement of the fluctuation parameter of the heat pump unit during operation, the proportion of the sample mixed working fluid is deleted and training of the next group of sample mixed working fluids is carried out;

[0027] The initial model after training is subjected to knowledge distillation processing to obtain the process model of the heat pump unit.

[0028] In some embodiments, preprocessing the operating parameters corresponding to each sample mixed working medium in the parameter samples to obtain the preprocessed parameter samples includes:

[0029] Performing time-series coding processing on the operating parameters corresponding to each sample mixed working medium;

[0030] Sampling the operating parameters after the time series encoding processing at preset intervals to obtain sampling data;

[0031] Using the Clausius-Clapeyron equation to screen qualified data from the sampled data;

[0032] The qualified data is corrected for outliers to obtain the preprocessed parameter samples.

[0033] In some embodiments, sampling the operating parameters after the time series encoding processing at a preset interval to obtain sampled data includes:

[0034] Using a sliding window technique, time-series sampling is performed on the operating parameter after the time-series coding process at the preset interval to obtain first sampling data;

[0035] Sampling the fluctuation of the fluctuation parameter of the heat pump unit in the first sampling data to obtain second sampling data;

[0036] The first sampling data and the second sampling data are used as the sampling data.

[0037] In some embodiments, the multi-head attention mechanism is used to process the preprocessed parameter samples and configure the associated weights, including: using the multi-head attention mechanism to determine the weight relationship between the critical temperature and the performance coefficient, the ambient temperature and the performance coefficient, and the relationship between temperature slip and pressure.

[0038] In some embodiments, weight parameters are set for the parameters in the preprocessed parameter sample, including: setting the weight parameters of the coefficient of performance, volumetric heating capacity, temperature glide, evaporation temperature / condensation temperature, bubble point temperature / dew point temperature, global warming potential value and flammability index in the preprocessed parameter sample to decrease in sequence.

[0039] In some embodiments, the physical property parameters include bubble point temperature, dew point temperature, global warming potential value, and flammability index; the fluctuation parameters of the heat pump unit include performance coefficient, volumetric heating capacity and temperature glide; and the operating condition parameters include evaporation temperature and condensing temperature.

[0040] The technical solutions provided by the embodiments of the present disclosure may have the following beneficial effects:

[0041] The present disclosure provides a method for optimizing the working fluid ratio of a heat pump unit, by obtaining a plurality of sample mixed working fluids, wherein the ratio of the plurality of working fluids in different sample mixed working fluids is different, and based on each sample mixed working fluid, obtaining the operating parameters of the heat pump unit corresponding to each sample mixed working fluid, wherein the plurality of sample mixed working fluids and the operating parameters corresponding to each sample mixed working fluid are used as parameter samples, and an initial model is trained based on the parameter samples to obtain a process model of the heat pump unit, and based on the process model, an optimal ratio of the mixed working fluids of the heat pump unit is obtained. The method for optimizing the working fluid ratio of a heat pump unit provided by the present disclosure obtains the measured operating data corresponding to each sample mixed working fluid, obtains the parameter samples for training the initial model, and then trains a process model of the heat pump unit suitable for different operating conditions, and through the process model of the heat pump unit, the optimal ratio of the mixed working fluids can be obtained, thereby improving the efficiency and accuracy of the heat pump unit in screening the mass ratio of non-azeotropic mixed working fluids.

[0042] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0044] Figure 1 The present invention is a flow chart showing a method for optimizing the working fluid ratio of a heat pump unit according to an exemplary embodiment.

[0045] Figure 2 The present invention is a flow chart showing a method for optimizing the working fluid ratio of a heat pump unit according to an exemplary embodiment.

[0046] Figure 3 The present invention is a flow chart showing a method for optimizing the working fluid ratio of a heat pump unit according to an exemplary embodiment.

[0047] Figure 4 The present invention is a flow chart showing a method for optimizing the working fluid ratio of a heat pump unit according to an exemplary embodiment.

[0048] Figure 5The present invention is a flow chart showing a method for optimizing the working fluid ratio of a heat pump unit according to an exemplary embodiment.

[0049] Figure 6 The present invention is a flow chart showing a method for optimizing the working fluid ratio of a heat pump unit according to an exemplary embodiment.

[0050] Figure 7 The present invention is a flow chart showing a method for optimizing the working fluid ratio of a heat pump unit according to an exemplary embodiment.

[0051] Figures 8a-8b The present invention is a flow chart showing a method for optimizing the working fluid ratio of a heat pump unit according to an exemplary embodiment. DETAILED DESCRIPTION

[0052] Exemplary embodiments will be described in detail herein, examples of which are illustrated in the accompanying drawings. In the following description, when referring to the drawings, like numbers in different figures represent like or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all possible embodiments consistent with the present invention. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present invention, as detailed in the appended claims.

[0053] With the continuous development of industrial heat pump technology, its application scenarios are also expanding. A wider temperature range is currently a major challenge. Conventional single refrigerants cannot meet the requirements of such a wide temperature range due to the limitations of their physical and chemical properties. Non-azeotropic mixtures can meet the requirements of a wider temperature range by adjusting the mass ratio of two or even more different refrigerants. However, the current screening methods for the mass ratio of non-azeotropic mixtures still have problems with accuracy and efficiency.

[0054] In order to solve the above problems, the present disclosure provides a method for optimizing the working fluid ratio of a heat pump unit, by obtaining a plurality of sample mixed working fluids, wherein the ratio of the plurality of working fluids in different sample mixed working fluids is different, and based on each sample mixed working fluid, the operating parameters of the heat pump unit corresponding to each sample mixed working fluid are obtained, wherein the plurality of sample mixed working fluids and the operating parameters corresponding to each sample mixed working fluid are used as parameter samples, and an initial model is trained based on the parameter samples to obtain a process model of the heat pump unit, and based on the process model, the optimal ratio of the mixed working fluid of the heat pump unit is obtained. The method for optimizing the working fluid ratio of a heat pump unit provided by the present disclosure obtains the measured operating data corresponding to each sample mixed working fluid, obtains the parameter samples for training the initial model, and then trains a process model of the heat pump unit suitable for different operating conditions. Through the process model of the heat pump unit, the optimal ratio of the mixed working fluid can be obtained, thereby improving the efficiency and accuracy of the heat pump unit in screening the mass ratio of non-azeotropic mixed working fluids.

[0055] The exemplary embodiment of the present disclosure provides a method for optimizing the working fluid ratio of a heat pump unit, such as Figure 1 As shown, including:

[0056] S100: Acquire multiple sample mixed working fluids, wherein different sample mixed working fluids have different ratios of multiple working fluids.

[0057] S200 . Based on each sample mixed working medium, obtain operating parameters of the heat pump unit corresponding to each sample mixed working medium, wherein a plurality of sample mixed working media and the operating parameters corresponding to each sample mixed working medium are used as parameter samples.

[0058] S300: Train the initial model based on the parameter sample to obtain a process model of the heat pump unit.

[0059] S400: Based on the process model, an optimal ratio of the mixed working fluid of the heat pump unit is obtained.

[0060] In step S100, the sample mixed working fluid is a non-azeotropic mixed working fluid. A non-azeotropic mixed working fluid is a mixture of two or more pure working fluids with different boiling points. The gas and liquid phase compositions of the mixtures are always different during the phase transition process, and there is a certain degree of glide in the boiling point temperature. The sample mixed working fluids in this embodiment all have different ratios of working fluids.

[0061] like Figure 2 As shown, in step S100, a plurality of sample mixed working fluids are obtained, wherein different sample mixed working fluids have different ratios of the plurality of working fluids, including:

[0062] S210: Mix at least two working fluids according to different ratios to obtain a plurality of initial sample mixed working fluids.

[0063] S220 , performing an initial screening of the initial sample mixed working fluid according to the operating parameters required by the heat pump unit and the physical property parameters of the initial sample mixed working fluid to obtain a plurality of sample mixed working fluids.

[0064] In step S210, at least two working fluids are mixed according to different ratios. For example, working fluid 1 and working fluid 2 are mixed according to mass ratios of 1:99, 2:98, ..., 98:2, and 99:1 to generate 99 initial sample mixed working fluids, namely, initial sample mixed working fluid 1 to initial sample mixed working fluid 99.

[0065] In step S220, the physical properties of each initial sample mixed refrigerant are initially screened based on the design operating conditions of the heat pump unit, such as the evaporation temperature and condensation temperature required by the heat pump unit. If the physical properties of the initial sample mixed refrigerant are insufficient to enable the heat pump unit to achieve the required evaporation temperature and condensation temperature, the initial sample mixed refrigerant is deleted and the next initial sample mixed refrigerant is screened. The physical properties of the initial sample mixed refrigerant include bubble point temperature, dew point temperature, global warming potential (GWP), and flammability penalty.

[0066] In this embodiment, the initial sample mixed working fluids are screened based on the actual design operating conditions of the heat pump unit. The screened sample mixed working fluids are suitable for the operation of the heat pump unit under these actual operating conditions. Therefore, the initial sample mixed working fluids can be screened based on the different design operating conditions of the heat pump unit to obtain sample mixed working fluids suitable for different operating conditions. Furthermore, by screening the initial sample mixed working fluids, the data quality of the sample mixed working fluids can be improved, training efficiency can be increased, and the generalization ability of the model can be enhanced.

[0067] In step S200, each sample mixed working medium corresponds to a set of operating parameters of the heat pump unit.

[0068] like Figure 3 As shown, in step S200, based on each sample mixed working medium, the operating parameters of the heat pump unit corresponding to each sample mixed working medium are obtained, wherein the multiple sample mixed working media and the operating parameters corresponding to each sample mixed working medium are used as parameter samples, including:

[0069] S310, inputting a plurality of sample mixed working fluids into the heat pump unit in sequence for operation test.

[0070] S320: Obtain operating parameters corresponding to each sample mixed working fluid, wherein the operating parameters include physical parameters of the sample mixed working fluid, fluctuation parameters of the heat pump unit, and operating parameters of the heat pump unit.

[0071] S330: Take each sample mixed working fluid and its corresponding operating parameters as parameter samples.

[0072] In steps S310 and S320, a sample mixed working fluid is input into the heat pump unit for actual operation testing. Only one sample mixed working fluid is input each time, and the operating parameters of the heat pump unit corresponding to the sample mixed working fluid can be obtained. The physical properties of the sample mixed working fluid in the operating parameters include: bubble point temperature, dew point temperature, global warming potential value, and flammability index. The fluctuating parameters of the heat pump unit in the operating parameters include: coefficient of performance (COP), volumetric heating capacity (VHC), and temperature glide. The operating parameters of the heat pump unit in the operating parameters include: evaporation temperature and condensing temperature.

[0073] In step S330, each sample mixed working medium and its operating parameters actually obtained from operating in the heat pump unit are used as parameter samples.

[0074] In this embodiment, actual operating parameters corresponding to each sample mixed working fluid are obtained through actual heat pump unit testing, without the need for manual data labeling. This can improve the authenticity of the data in the parameter samples and support continuous optimization of the model.

[0075] In step S300, the initial model is trained based on the parameter sample, and the initial model includes a transformer model with a six-layer stacked architecture.

[0076] In this embodiment, a transformer model with a six-layer stacked architecture is used as the initial model, and a multi-head attention mechanism can be used to improve the feature extraction capability of the model and improve training efficiency.

[0077] like Figure 4 As shown, in step S300, the initial model is trained based on the parameter sample to obtain the process model of the heat pump unit, including:

[0078] S410 , preprocessing the operating parameters corresponding to each sample mixed working medium in the parameter sample to obtain a preprocessed parameter sample.

[0079] S420: Use a multi-head attention mechanism to process the preprocessed parameter samples and configure the associated weights.

[0080] S430: Setting weight parameters for the parameters in the preprocessed parameter sample.

[0081] S440, constructing a feedforward neural network to enhance the nonlinear expression capability of the preprocessed parameter samples and the corresponding weight parameters.

[0082] S450: The coefficient of performance, volumetric heating capacity, global warming potential, and flammability index of the heat pump unit are used as optimization targets for iterative optimization.

[0083] S460: Determine whether the sample mixed working fluid in the iterative optimization result meets the fluctuation requirement of the heat pump unit's fluctuation parameters during operation. If the sample mixed working fluid in the iterative optimization result meets the fluctuation requirement of the heat pump unit's fluctuation parameters during operation, execute step S470; if the sample mixed working fluid in the iterative optimization result does not meet the fluctuation requirement of the heat pump unit's fluctuation parameters during operation, execute step S480.

[0084] S470: Output the mixed working fluid ratio as the optimal ratio of the mixed working fluid of the heat pump unit.

[0085] S480: Delete the sample mixed working fluid ratio, perform training on the next set of sample mixed working fluids, and execute step S420.

[0086] S490: Perform knowledge distillation on the trained initial model to obtain a process model of the heat pump unit.

[0087] In step S410, Figure 5 As shown, the operating parameters corresponding to each sample mixed working medium in the parameter sample are preprocessed to obtain the preprocessed parameter sample, including:

[0088] S510: Perform time-series coding processing on the operating parameters corresponding to each sample mixed working medium.

[0089] S520: Sampling the operating parameters after the time series coding processing at preset intervals to obtain sampling data.

[0090] S530. Use the Clausius-Clapeyron equation to screen qualified data from the sampled data.

[0091] S540: Correct outliers on the qualified data to obtain preprocessed parameter samples.

[0092] In step S510, the operating parameters corresponding to each sample mixed working fluid, including the physical properties of the sample mixed working fluid (bubble point temperature, dew point temperature, global warming potential value, flammability index), the fluctuation parameters of the heat pump unit (performance coefficient, volumetric heating capacity and temperature glide), and the operating parameters of the heat pump unit (evaporation temperature and condensing temperature) are encoded in a time series manner and scaled into a multi-dimensional time series matrix.

[0093] In step S520, the preset interval can be obtained based on empirical values, for example, 2 to 3 times of the operating cycle of the heat pump unit can be selected, and this embodiment does not impose any specific limitation.

[0094] like Figure 6As shown, in step S520, the operating parameters after the time series coding process are sampled at a preset interval to obtain sampled data, including:

[0095] S610 , using a sliding window technology to perform time series sampling on the operating parameters after time series coding processing at preset intervals to obtain first sampling data.

[0096] S620: Sampling the fluctuation of the fluctuation parameter of the heat pump unit in the first sampling data to obtain second sampling data.

[0097] S630: Use the first sampling data and the second sampling data as sampling data.

[0098] For example, the multi-dimensional time series matrix can be sampled using a sliding window technique at preset intervals of 2 to 3 times the operating cycle of the heat pump unit to obtain first sampled data. Simultaneously, the fluctuation of the heat pump unit's fluctuation parameters (coefficient of performance, volumetric heating capacity, and temperature glide) in the first sampled data is sampled to obtain second sampled data.

[0099] The volatility of the fluctuating parameters of the heat pump unit, that is, the volatility characteristics of parameters such as the coefficient of performance, volumetric heating capacity, and temperature glide. Among them, the coefficient of performance is the core indicator for measuring the energy efficiency of the heat pump unit system. It is defined as the ratio of the effective energy (cooling capacity or heating capacity) provided by the heat pump unit system to the input energy (such as power consumption). The smaller its volatility, the better the match between the working fluid characteristics and the working conditions. The volumetric heating capacity is a key parameter for measuring the amount of heating that the heat pump unit system can provide per unit time and per unit compressor displacement. It directly reflects the energy carrying efficiency of the working fluid in the system. The smaller its volatility, the better the uniformity of the working fluid. Temperature glide is the core characteristic of non-azeotropic mixtures during the phase change (evaporation / condensation) process. It refers to the temperature change range of the mixed working fluid from the initial phase change (bubble point) to the complete phase change (dew point). The smaller its volatility, the higher the stability of the working fluid components and the more stable the system operation.

[0100] In this embodiment, by performing time series sampling on the operating parameters after time series coding processing, the time series dynamics of the operating parameters can be captured and the time dependency of the operating parameters can be retained.

[0101] In step S530, the Clausius-Clapeyron equation is used to screen qualified data from the sampled data, thereby ensuring the authenticity of the data.

[0102] In step S540, the trend term and the residual term are separated by STL decomposition, and the qualified data are corrected for outliers to ensure data validity, and the preprocessed parameter samples are obtained.

[0103] In this embodiment, by performing time series encoding processing on the operating parameters corresponding to the sample mixed working fluid, performing time series sampling on the obtained multi-dimensional time series matrix, screening the sampled data and correcting the outliers, the time series characteristics of the parameter sample data can be accurately captured, the correctness and validity of the data can be ensured, and a reliable training sample data foundation can be provided for the initial model.

[0104] In step S420, a multi-head attention mechanism is used to process the preprocessed parameter samples and configure the associated weights, including: using the multi-head attention mechanism to determine the weight relationship between the critical temperature and the performance coefficient, the ambient temperature and the performance coefficient, and the relationship between the temperature slip and the pressure.

[0105] For example, an eight-head attention mechanism can be used, which is not specifically limited in this embodiment. A multi-head attention mechanism is used to determine the weighted associations between critical temperature and performance coefficient, and between ambient temperature and performance coefficient. The greater the impact on performance coefficient volatility, the greater the association with the performance coefficient. The relationship between temperature glide and pressure is also determined.

[0106] In step S430, weight parameters are set for the parameters in the preprocessed parameter sample, including: setting the weight parameters of the coefficient of performance, volumetric heating capacity, temperature glide, evaporation temperature / condensation temperature, bubble point temperature / dew point temperature, global warming potential value and flammability index in the preprocessed parameter sample to decrease in sequence.

[0107] In step S440 , a feedforward neural network (FNN) is constructed to scale the preprocessed parameter samples and their corresponding weight parameters from 512 to 2048 to 512 dimensions to enhance the nonlinear expression capability of the local features in the parameters.

[0108] In step S450, if Figure 7 As shown in the figure, the coefficient of performance, volumetric heating capacity, global warming potential, and flammability index of the heat pump unit are taken as optimization targets, and the iterative optimization steps include:

[0109] S710. Design a loss function using the coefficient of performance, volumetric heating capacity, global warming potential, and flammability index of the heat pump unit as optimization targets.

[0110] S720 uses the Newton-Raphson optimization algorithm and the Hessian matrix to adjust the learning rate to solve the gradient oscillation problem caused by the scale difference of physical parameters.

[0111] S730. Perform a time series analysis on the narrow point parameters of the sample mixed working fluid during operation to determine its fluctuation characteristics.

[0112] S740: Re-divide the weights of the parameters in the preprocessed parameter samples to achieve full parameter target optimization, thereby completing the iterative optimization of the initial model.

[0113] In step S710, the coefficient of performance, volumetric heating capacity, global warming potential, and flammability index of the heat pump unit are used as optimization targets. The requirements for the coefficient of performance, volumetric heating capacity, global warming potential, flammability index and other parameters of the heat pump unit are different under different operating conditions of the heat pump unit. Specific requirements for the coefficient of performance, volumetric heating capacity, global warming potential, flammability index and other parameters of the heat pump unit can be designed in combination with specific actual operating conditions, and a specific loss function can be designed.

[0114] In this embodiment, multi-objective collaborative optimization can improve the comprehensiveness of the working fluid design, and the Newton-Raphson algorithm is used to solve the gradient problem, so that the initial model can achieve stable and efficient convergence. The timing analysis of the narrow point parameters can enhance the dynamic adaptability of the initial model, and the weights can be redivided to achieve full parameter optimization, ultimately achieving efficient and stable system optimization.

[0115] In steps S460-S490, a determination is made as to whether the sample mixed refrigerant in the iterative optimization results meets the fluctuation requirements of the heat pump unit's fluctuation parameters during operation. Specifically, a determination is made as to whether the coefficient of performance, volumetric heating capacity, and temperature glide parameters of the sample mixed refrigerant output by the initial model meet the required fluctuation characteristics during operation of the initial model. In this embodiment, the required fluctuation characteristics include requiring that the fluctuation characteristics of the coefficient of performance, volumetric heating capacity, and temperature glide parameters tend to be stable.

[0116] If the sample mixed working fluid in the iterative optimization result meets the fluctuation requirement of the fluctuation parameter of the heat pump unit during operation, the mixed working fluid ratio is output as the optimal ratio of the mixed working fluid of the heat pump unit.

[0117] If the sample mixed working fluid in the iterative optimization result does not meet the fluctuation requirement of the fluctuation parameter of the heat pump unit during operation, the proportion of the sample mixed working fluid is deleted, and the next group of sample mixed working fluids is trained, and step S420 is continued.

[0118] Finally, the trained initial model undergoes knowledge distillation to produce a simplified process model for the heat pump unit. This process model simulates the actual operation of the heat pump unit and identifies the optimal ratio of the mixed working fluid by determining the fluctuation characteristics of the heat pump unit's parameters during operation.

[0119] The present disclosure provides a method for optimizing the working fluid ratio of a heat pump unit, by obtaining a plurality of sample mixed working fluids, wherein the ratio of the plurality of working fluids in different sample mixed working fluids is different, and based on each sample mixed working fluid, obtaining the operating parameters of the heat pump unit corresponding to each sample mixed working fluid, wherein the plurality of sample mixed working fluids and the operating parameters corresponding to each sample mixed working fluid are used as parameter samples, and an initial model is trained based on the parameter samples to obtain a process model of the heat pump unit, and based on the process model, an optimal ratio of the mixed working fluids of the heat pump unit is obtained. The method for optimizing the working fluid ratio of a heat pump unit provided by the present disclosure obtains the measured operating data corresponding to each sample mixed working fluid, obtains the parameter samples for training the initial model, and then trains a process model of the heat pump unit suitable for different operating conditions, and through the process model of the heat pump unit, the optimal ratio of the mixed working fluids can be obtained, thereby improving the efficiency and accuracy of the heat pump unit in screening the mass ratio of non-azeotropic mixed working fluids.

[0120] For ease of understanding, a specific embodiment is given below to describe the working fluid ratio optimization method of the heat pump unit of the present application. Figure 8a-Figure 8b As shown, Figure 8a The flowchart of steps S801-S811 is shown as an example. Figure 8b The flowchart of steps S812-S823 is exemplarily illustrated.

[0121] S801. Mix at least two working fluids according to different ratios to obtain a plurality of initial sample mixed working fluids.

[0122] S802. Performing an initial screening of the initial sample mixed working fluid according to the operating parameters required by the heat pump unit and the physical property parameters of the initial sample mixed working fluid to obtain a plurality of sample mixed working fluids.

[0123] S803. Input various sample mixed working fluids into the heat pump unit in sequence for operation test.

[0124] S804: Obtain operating parameters corresponding to each sample mixed working fluid, wherein the operating parameters include physical parameters of the sample mixed working fluid, fluctuation parameters of the heat pump unit, and operating parameters of the heat pump unit.

[0125] S805: Take each sample mixed working fluid and its corresponding operating parameters as parameter samples.

[0126] S806: Perform time-series coding processing on the operating parameters corresponding to each sample mixed working medium.

[0127] S807: Using a sliding window technique, perform time series sampling on the operating parameters after time series coding at preset intervals to obtain first sampled data;

[0128] S808, sampling the fluctuation of the fluctuation parameter of the heat pump unit in the first sampling data to obtain second sampling data;

[0129] S809: Use the first sampling data and the second sampling data as sampling data.

[0130] S810. Use the Clausius-Clapeyron equation to screen qualified data from the sampled data.

[0131] S811. Correct outliers on qualified data to obtain preprocessed parameter samples.

[0132] S812. Use the multi-head attention mechanism to process the preprocessed parameter samples and configure the associated weights.

[0133] S813. Set weight parameters for the parameters in the preprocessed parameter sample.

[0134] S814. Construct a feedforward neural network to enhance the nonlinear expression capability of the preprocessed parameter samples and the corresponding weight parameters.

[0135] S815. Design a loss function using the coefficient of performance, volumetric heating capacity, global warming potential, and flammability index of the heat pump unit as optimization targets.

[0136] S816, using the Newton-Raphson optimization algorithm, using the Hessian matrix to adjust the learning rate, solves the gradient oscillation problem caused by the scale difference of physical parameters.

[0137] S817. Perform a time series analysis on the narrow point parameters of the sample mixed working fluid during operation to determine its fluctuation characteristics.

[0138] S818. Re-divide the weights of the parameters in the preprocessed parameter samples to achieve full parameter target optimization, thereby completing the iterative optimization of the initial model.

[0139] S819: Determine whether the sample mixed working fluid in the iterative optimization result meets the fluctuation requirements of the heat pump unit's fluctuating parameters during operation. If the sample mixed working fluid in the iterative optimization result meets the fluctuation requirements of the heat pump unit's fluctuating parameters during operation, execute step S820; if the sample mixed working fluid in the iterative optimization result does not meet the fluctuation requirements of the heat pump unit's fluctuating parameters during operation, execute step S821.

[0140] S820: Output the mixed working fluid ratio as the optimal ratio of the mixed working fluid of the heat pump unit.

[0141] S821. Delete the ratio of the sample mixed working fluid, perform training on the next set of sample mixed working fluids, and execute step S420.

[0142] S822: Perform knowledge distillation on the trained initial model to obtain a process model of the heat pump unit.

[0143] S823. Based on the process model, the optimal ratio of the mixed working fluid of the heat pump unit is obtained.

[0144] In the above steps, although Figure 8a and Figure 8b In the embodiment, step S811 and step S812 are two separate steps, but during the execution process, step S811 and step S812 are two steps that are executed continuously, that is, step S812 is executed after step S811 is executed.

[0145] The disclosed method for optimizing the working fluid ratio of a heat pump unit uses machine learning and other methods to further optimize the mass ratio of the mixed working fluid based on the fluctuations of the heat pump unit's fluctuating parameters under different operating conditions, thereby maintaining high operating efficiency of the mixed working fluid under complex operating conditions. This method can dynamically determine the mass ratio of the heat pump unit's non-azeotropic mixed working fluid based on different heat pump unit models and operating conditions, thereby improving the efficiency and accuracy of heat pump working fluid optimization.

[0146] The disclosed method for optimizing the working fluid ratio of a heat pump unit leverages the multi-threaded and sequential nature of the Transformer model to perform a time-series analysis of the fluctuations in various parameters of a sample working fluid mixture during actual operation. This allows for the selection of sample working fluid mixtures while ensuring the operating efficiency of the heat pump unit. Furthermore, distillation and simplification of the Transformer model improves the model's operational efficiency and screening accuracy.

[0147] Other embodiments of the present invention will readily occur to those skilled in the art after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the invention being indicated by the following claims.

[0148] It should be understood that the present invention is not limited to the exact construction described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

Claims

1. A method for optimizing the working fluid ratio of a heat pump unit, characterized in that: include: Acquire multiple sample mixed working fluids, wherein different sample mixed working fluids contain different ratios of multiple working fluids; Based on each of the sample mixed working fluids, obtaining the operating parameters of the heat pump unit corresponding to each of the sample mixed working fluids, wherein the plurality of sample mixed working fluids and the operating parameters corresponding to each of the sample mixed working fluids are used as parameter samples; Training the initial model based on the parameter sample to obtain a process model of the heat pump unit; Based on the process model, an optimal ratio of the mixed working fluid of the heat pump unit is obtained.

2. The method for optimizing the working fluid ratio of a heat pump unit according to claim 1, characterized in that: The obtaining of multiple sample mixed working fluids, wherein different sample mixed working fluids have different ratios of multiple working fluids, includes: Mixing at least two working fluids according to different ratios to obtain a plurality of initial sample mixed working fluids; According to the operating parameters required by the heat pump unit and the physical property parameters of the initial sample mixed working fluid, the initial sample mixed working fluid is initially screened to obtain a plurality of the sample mixed working fluids.

3. The method for optimizing the working fluid ratio of a heat pump unit according to claim 2, characterized in that: The step of obtaining, based on each of the sample mixed working fluids, an operating parameter of the heat pump unit corresponding to each of the sample mixed working fluids, wherein the plurality of sample mixed working fluids and the operating parameters corresponding to each of the sample mixed working fluids are used as parameter samples, includes: Inputting a plurality of sample mixed working fluids into the heat pump unit in sequence for operation test; Obtaining operating parameters corresponding to each of the sample mixed working fluids, wherein the operating parameters include the physical property parameters of the sample mixed working fluid, the fluctuation parameters of the heat pump unit, and the operating condition parameters of the heat pump unit; Each sample mixed working medium and the corresponding operating parameter are used as the parameter sample.

4. The method for optimizing the working fluid ratio of a heat pump unit according to claim 1, characterized in that: The initial model includes a transformer model with a six-layer stacked architecture.

5. The method for optimizing the working fluid ratio of a heat pump unit according to claim 4, characterized in that: The training of the initial model based on the parameter sample to obtain the process model of the heat pump unit includes: Preprocessing the operating parameters corresponding to each sample mixed working medium in the parameter sample to obtain a preprocessed parameter sample; Using a multi-head attention mechanism to process the preprocessed parameter samples and configure associated weights; Setting weight parameters for the parameters in the preprocessed parameter sample; Constructing a feedforward neural network to enhance the nonlinear expression capability of the preprocessed parameter samples and the corresponding weight parameters; Iterative optimization is performed using the coefficient of performance, volumetric heating capacity, global warming potential, and flammability index of the heat pump unit as optimization targets; Determining whether the sample mixed working fluid in the iterative optimization result meets the fluctuation requirement of the fluctuation parameter of the heat pump unit during operation; If the sample mixed working fluid in the iterative optimization result meets the volatility requirement of the fluctuation parameter of the heat pump unit during operation, the mixed working fluid ratio is output as the optimal ratio of the mixed working fluid of the heat pump unit; If the sample mixed working fluid in the iterative optimization result does not meet the volatility requirement of the fluctuation parameter of the heat pump unit during operation, the proportion of the sample mixed working fluid is deleted and training of the next group of sample mixed working fluids is carried out; The initial model after training is subjected to knowledge distillation processing to obtain the process model of the heat pump unit.

6. The method for optimizing the working fluid ratio of a heat pump unit according to claim 5, characterized in that: The preprocessing of the operating parameters corresponding to each sample mixed working medium in the parameter samples to obtain the preprocessed parameter samples includes: Performing time-series coding processing on the operating parameters corresponding to each sample mixed working medium; Sampling the operating parameters after the time series encoding processing at preset intervals to obtain sampling data; Using the Clausius-Clapeyron equation to screen qualified data from the sampled data; The qualified data is corrected for outliers to obtain the preprocessed parameter samples.

7. The method for optimizing the working fluid ratio of a heat pump unit according to claim 6, characterized in that: The sampling of the operating parameters after the time series encoding processing at a preset interval to obtain sampled data includes: Using a sliding window technique, time-series sampling is performed on the operating parameter after the time-series coding process at the preset interval to obtain first sampling data; Sampling the fluctuation of the fluctuation parameter of the heat pump unit in the first sampling data to obtain second sampling data; The first sampling data and the second sampling data are used as the sampling data.

8. The method for optimizing the working fluid ratio of a heat pump unit according to claim 5, characterized in that: The multi-head attention mechanism is used to process the preprocessed parameter samples and configure the associated weights, including: using the multi-head attention mechanism to determine the weight relationship between the critical temperature and the performance coefficient, the ambient temperature and the performance coefficient, and the relationship between temperature slip and pressure.

9. The method for optimizing the working fluid ratio of a heat pump unit according to claim 5, characterized in that: Setting weight parameters for the parameters in the preprocessed parameter sample includes setting the weight parameters of the coefficient of performance, volumetric heating capacity, temperature glide, evaporation temperature / condensation temperature, bubble point temperature / dew point temperature, global warming potential value and flammability index in the preprocessed parameter sample to decrease in sequence.

10. The method for optimizing the working fluid ratio of a heat pump unit according to claim 3, characterized in that: The physical property parameters include bubble point temperature, dew point temperature, global warming potential value, and flammability index; The fluctuating parameters of the heat pump unit include coefficient of performance, volumetric heating capacity and temperature glide; The operating parameters include evaporation temperature and condensation temperature.

Images