Refrigeration system design method based on digital twinning

Abstract

This invention belongs to the field of refrigeration system design technology and discloses a refrigeration system design method based on digital twins. This method discretizes a multi-loop parallel evaporator into a distributed parameter node model along the refrigerant flow direction and establishes the heat transfer equations for each node. A Kalman filter algorithm is used to estimate the outlet dryness of each loop. A weighted performance coefficient for all operating conditions is constructed using representative operating conditions obtained from hourly cooling load clustering throughout the year and their time percentage weights. A differential evolution algorithm is used to jointly optimize the number of loops, pipe diameter, orifice diameter of the distributor, and the maximum flow area of ​​the electronic expansion valve. The optimal parameters are written into the parameterized twin, and after being put into operation, online identification and real-time status updates are achieved through sensor signals. This invention solves the problem of significant decrease in system performance coefficients caused by uneven refrigerant distribution under partial load conditions in existing design methods.

Description

Technical Field

[0001] This invention relates to the field of refrigeration system design technology, and more particularly to a refrigeration system design method based on digital twins. Background Technology

[0002] Multi-loop parallel evaporators are a common structure used in refrigeration systems to increase heat exchange area and are widely used in commercial refrigeration systems. Current technology for designing multi-loop parallel evaporators typically uses the single-point outlet superheat under rated load conditions as the sole design constraint, employing fixed empirical parameters to determine the number of loops, pipe diameter, and distributor orifice diameter. This design process is disconnected from the actual operating conditions of the system, making it impossible to quantitatively assess system performance under partial load conditions during the design phase.

[0003] The aforementioned design method results in significant non-uniformity of refrigerant mass flow rate at the inlet of each parallel loop during partial operation at 40% to 80% of rated load due to the inherent distribution deviation of the distributor structure. This leads to an expansion of the superheated area in some loops, wet vapor carryover in others, a decrease in overall heat exchange efficiency, and a significantly lower weighted performance coefficient (GSP) than the design expectation throughout the year. Current technology uses only single-point outlet superheat as the expansion valve control feedback signal, failing to identify the cumulative impact of loop-level distribution deviations on overall performance. This results in the refrigeration system operating inefficiently for extended periods under partial load conditions. Therefore, a refrigeration system design method capable of predicting and optimizing overall performance during the design phase is urgently needed. Summary of the Invention

[0004] The purpose of this invention is to provide a digital twin-based refrigeration system design method to solve the problems existing in the prior art.

[0005] This invention provides a design method for a refrigeration system based on digital twins, comprising the following steps: Step 1: Discretize each loop of the multi-loop parallel evaporator into M computational nodes along the refrigerant flow direction, with a node length of... ,in For the total length of a single loop, Establish the refrigerant-side energy conservation equation at each node. Solve for the refrigerant dryness at each node by simultaneously solving the nodal equations of all N parallel loops. Construct a digital twin model of a distributed parameter evaporator; Step 2: Define the outlet dryness vector of each loop as a state variable. The temperature, pressure, and pressure drop signals at the evaporator's main outlet pipe manifold are defined as observables. , with equation State estimation of the outlet dryness of each loop is performed, and the Kalman gain matrix is ​​calculated. Observation matrix and noise covariance matrix Identification via Monte Carlo method offline; Step 3: Discretize the annual operating conditions into Q representative load levels, each load level The corresponding weight is ,satisfy The performance coefficient is weighted by all operating conditions. Maximizing the number of evaporator loops N and the single-loop pipe diameter is the optimization objective. Orifice diameter of the separator head plate Maximum flow area of ​​electronic expansion valve To design variables, a differential evolution algorithm is used to jointly optimize the above parameters in the digital twin model, with adjacent generations... The iteration terminates when the improvement is less than 0.005. Step four: Write the optimal design parameters into the digital twin model to generate a parameterized twin. Integrate the evaporator outlet temperature, suction pressure, discharge pressure, and compressor current signals into the Kalman filter state estimation module from step two. Complete the process noise covariance matrix within 72 hours of system operation. Online adaptive identification.

[0006] By integrating the distributed parameter evaporator digital twin model, Kalman filter loop dryness state estimation, full-condition weighted performance coefficient objective function, and differential evolution algorithm in the design phase, joint optimization of the number of evaporator loops, single-loop pipe diameter, distributor orifice plate diameter, and maximum flow area of ​​electronic expansion valve is achieved. This significantly improves the refrigerant distribution uniformity of each parallel loop under partial load conditions and substantially improves the annual weighted performance coefficient. It solves the technical problem of long-term inefficient operation of the system under partial load conditions caused by the reliance on fixed empirical parameters in existing design methods.

[0007] Furthermore, in step one, in the two-phase region, the first... The overall heat transfer coefficient of each node The Gungor-Winterton modified formula was used for calculation, and the Dittus-Boelter single-association formula was used for calculation in the superheated region.

[0008] Furthermore, in step three, the cooling capacity... The compressor input power is obtained by summing the heat transfer integrals of all nodes in all loops under this load level using a distributed parameter digital twin model. Calculated by interpolation of compressor characteristic maps from the digital twin model.

[0009] Furthermore, in step three, the population size of the differential evolution algorithm is 60, the maximum number of iterations is 200, and the search range of the design variables is... .

[0010] Furthermore, in step three, the number Q of representative load levels is extracted by clustering the hourly cooling load data for 8760 hours throughout the year, and the weights corresponding to each load level are determined. It is determined by the proportion of time that each cluster load level accounts for in the total annual operating time.

[0011] Furthermore, in step three, the orifice plate of the distributor uses different orifice diameters for the outer loop and the inner loop of the evaporator. The values ​​of the orifice diameters of the outer loop and the inner loop are independently determined in the optimization process of the differential evolution algorithm, forming a non-uniform orifice diameter differentiated distributor design.

[0012] Furthermore, in step two, the Kalman filter state estimation module simulates different refrigerant distribution deviation scenarios by injecting virtual sensor signals into the digital twin model during the design phase, outputting the dryness distribution estimation error of each loop and the observation matrix. The finite difference method is used for offline computation in the digital twin model.

[0013] Furthermore, in step four, the evaporator outlet temperature sensor is a Pt100 type with a noise standard deviation of 0.15K, and the pressure transmitter noise standard deviation is 1.5kPa. These calibration values ​​are used as the sensor noise covariance matrix. The initial input value.

[0014] The beneficial effects of this invention are: Firstly, this invention employs a distributed parameter evaporator model that discretizes each loop into M nodes during the design phase. Combined with the Gungor-Winterton correction formula and the Dittus-Boelter correlation partitioning calculation of the comprehensive heat transfer coefficient, it can accurately predict the refrigerant dryness distribution along each loop. This allows for the identification and elimination of refrigerant distribution deviations between loops caused by the consistent design of the distributor orifice diameter during the design phase, thus avoiding the coexistence of overheating and wet vapor carryover in some loops.

[0015] Secondly, this invention constructs a weighted performance coefficient for all operating conditions using Q representative operating conditions obtained by clustering the hourly cooling load throughout the year and their time proportion weights as the optimization target. It uses a differential evolution algorithm in a digital twin environment to jointly optimize the number of loops, pipe diameter, orifice diameter of the liquid distributor, and maximum flow area of ​​the electronic expansion valve. This breaks through the limitation of existing design methods that only target the single-point performance under rated operating conditions, so that the performance of the design scheme under partial load conditions can be taken into account. The weighted performance coefficient for the whole year is significantly improved compared with the design scheme based on empirical parameters.

[0016] Third, this invention connects the parameterized twin generated during the design phase with the temperature, pressure, and compressor current sensor signals of the physical refrigeration system. Through the Kalman filter state estimation module, the process noise covariance matrix is ​​identified online within 72 hours after the system is put into operation. This realizes the digital twin closed-loop handover from the design phase to the operation phase, enabling the twin to continuously provide real-time estimates of the refrigerant dryness distribution in each loop during the operation phase, providing continuous data support for active refrigerant allocation and control. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the distributed parameter evaporator digital twin model of the present invention; Figure 2 This is a schematic diagram of the Kalman filter loop dryness state estimation module of the present invention; Figure 3 This is a schematic diagram of the full-condition weighted performance coefficient optimization process of the present invention; Figure 4 This is a schematic diagram illustrating the closed-loop handover between the parametric twin design and operation phases of the present invention. Detailed Implementation

[0019] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. It should also be noted that, to make the embodiments more comprehensive, the following embodiments are the best and preferred embodiments, and those skilled in the art can use other alternative methods to implement some well-known technologies; moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0020] It should be noted that the use of terms such as "an embodiment," "an embodiment," "an exemplary embodiment," and "some embodiments" in the specification indicates that the described embodiment may include a specific feature, structure, or characteristic, but not every embodiment necessarily includes that specific feature, structure, or characteristic. Furthermore, when a specific feature, structure, or characteristic is described in connection with an embodiment, implementing such a feature, structure, or characteristic in conjunction with other embodiments (whether explicitly described or not) should be within the knowledge of those skilled in the art.

[0021] Generally, terms can be understood at least partly from their use in context. For example, depending at least partly on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in a singular sense, or a combination of features, structures, or characteristics in a plural sense. Additionally, the term "based on" can be understood not necessarily to convey an exclusive set of factors, but rather, alternatively, depending at least partly on the context, to allow for the presence of other factors that are not necessarily explicitly described.

[0022] See Figures 1 to 4 As shown This invention addresses the problem in existing technologies where uneven refrigerant distribution in multi-loop evaporators under partial load conditions leads to a significant decrease in the coefficient of performance (COP). It proposes a digital twin-based refrigeration system design method, the core of which lies in combining a distributed parameter evaporator model with a Kalman filter state estimation algorithm to collaboratively optimize the evaporator loop geometry parameters and the orifice size of the electronic expansion valve during the design phase, rather than relying on fixed empirical parameters for static design.

[0023] In existing commercial refrigeration systems, multi-loop parallel evaporators are a common structure for increasing heat exchange area. However, during certain operating periods at 40% to 80% of rated load, the refrigerant mass flow rate at the inlet of each parallel loop exhibits a non-uniformity of 5% to 18% due to the inherent distribution deviation of the distributor structure. This results in an expansion of the superheated area in some loops and the occurrence of wet vapor carryover in others, leading to an overall decrease in heat exchange efficiency of 20% to 35%. Existing design methods only use the single-point outlet superheat as the expansion valve control feedback signal, which cannot identify the cumulative impact of loop-level distribution deviations on overall performance during the design phase.

[0024] The specific embodiments of the present invention include the following steps.

[0025] The first step is to construct a digital twin model of the distributed parameter evaporator. Each evaporator loop is discretized into M computational nodes along the refrigerant flow direction, with each node having a length of... ,in This is the total length of a single loop. For the first... Establish the refrigerant-side energy conservation equation at each node, with the specific expression as follows:

[0026] in For the first Refrigerant mass flow rate at each node, in units of ; The difference in specific enthalpy between the inlet and outlet of the refrigerant at this node, in units of ; The overall heat transfer coefficient of this node, in units of The Gungor-Winterton modified formula is used for calculation in the two-phase region, and the Dittus-Boelter single-connection formula is used for calculation in the superheated region. This refers to the outer circumference of the heat exchange tube, in units of... ; The air inlet temperature of the evaporator, in units of ; For the corresponding node pressure The refrigerant saturation temperature is given by the following unit: For all Solve the equations of all nodes in the parallel loop simultaneously to obtain the refrigerant dryness distribution along the path. The spatial field constitutes the state space of the digital twin model.

[0027] The second step is to establish a loop-level refrigerant dryness state estimation module based on Kalman filtering. In the digital twin model, the outlet dryness vector of each loop is defined as a state variable. The available system measurements are defined as observables. This includes the temperature, pressure, and pressure drop signals at the evaporator's main outlet manifold. The Kalman filter state update equation is:

[0028] in For the first The observed-corrected posterior state estimate vector at time step , with dimension . ; The prior state vector is predicted by the distributed parameter model; Here is the Kalman gain matrix, where Let be the prior error covariance matrix. To map state variables to the observation matrix in the observation space, The sensor noise covariance matrix is ​​shown above; all matrices were identified offline using the Monte Carlo method during the design phase. This module simulates different refrigerant distribution deviation scenarios using virtual sensor injection during the design phase, outputting the estimated error of dryness distribution in each loop to evaluate the observability of the design scheme under partial load conditions.

[0029] The third step is to construct a weighted performance coefficient optimization objective function for all operating conditions. Based on the hourly cooling load data of the target building, the annual operating conditions of the refrigeration system are discretized into... Representative load levels, each load level The corresponding occurrence probability weight is ,satisfy The weighted performance coefficient under all operating conditions is defined as follows:

[0030] in The performance coefficient is a dimensionless, weighted average for the entire year. For the first The actual cooling capacity of the evaporator under a given load level, in units of... It is obtained by integrating and summing the distributed parameter model over all loop nodes; The compressor input power, in units of Calculated by interpolation of compressor characteristic maps in the digital twin model; The input power of the evaporator fan, in units of .by Maximizing the number of evaporator loops is the optimization objective. Single-loop pipe diameter Orifice diameter of the separator head plate Maximum flow area of ​​electronic expansion valve To design the variables, a differential evolution algorithm is used to jointly optimize the above parameters in a digital twin environment. The iteration termination condition is the number of adjacent generations. Improvement amount less than .

[0031] The fourth step is to output the design scheme and generate a parametric twin. After optimization, the optimal design parameter set is written into the digital twin model to generate a parametric twin that can be updated synchronously with the physical manufacturing data. After the refrigeration system is put into operation, the twin continuously receives evaporator outlet temperature, suction pressure, discharge pressure and compressor current signals, and updates the state estimate in real time through the Kalman filter module in the second step, providing a basis for active refrigerant distribution and control during operation.

[0032] Example 1 The rated refrigeration capacity of a certain cold storage project is The refrigerant is R404A, and the design point for the evaporation temperature is... The condensation temperature is Annual load analysis shows that the system operates at 50% to 70% of its rated load for 62% of the year. The specific process for evaporator design using the method of this invention is as follows: Collect hourly cooling load data for the cold storage facility for 8760 hours throughout the year, and extract the data after clustering. A representative load level, corresponding to a weight vector The load level is linearly distributed from 35% of the rated value to 100%.

[0033] The initial design scheme sets the number of evaporator loops. Pipe diameter The separator uses a uniform orifice plate structure, and the orifice diameter of the electronic expansion valve is based on empirical values. Discretize each loop into Establish a distributed parameter digital twin model for each node.

[0034] Transient simulations were run on 12 representative operating conditions in the digital twin model. The results show that under 50% load, the outlet dryness vector of the six parallel circuits is... Among them, loops 2 and 4 experienced overheating, while loops 3 and 5 remained in a wet steam state, with a dryness difference of 0.16 between the loops. The initial design scheme under 50% load... The initial value is 3.21. .

[0035] The differential evolution optimization algorithm is initiated, with a population size of 60, a maximum number of iterations of 200, and the search range of the design variables set as follows: , , , After 47 iterations, the algorithm converged, and the optimal design parameters were found to be... Single loop, pipe diameter The diameter of the four loop orifices on the outer side of the separator orifice plate is taken as follows: The diameter of the four loop holes on the inner side is taken as follows: Differentiated non-uniform orifice design, maximum flow area of ​​electronic expansion valve .

[0036] The distributed parameter model is reconstructed using optimal parameters, and the outlet dryness vector of each loop is narrowed under 50% load conditions. The dryness difference between circuits decreased from 0.16 to 0.03 under 50% load conditions. The performance coefficient was increased from 3.21 to 3.89, representing the weighted performance coefficient under all operating conditions. It increased from 3.47 to 4.03, an increase of 16.1%.

[0037] The optimal design parameters are written into a parameterized digital twin. This twin, along with signals from four Pt100 temperature sensors, two pressure transmitters, and one compressor current transformer of the physical refrigeration system, is then connected to a Kalman filter state estimation module. The process noise covariance matrix is ​​completed within the first 72 hours after the system begins actual operation. Online adaptive identification, observation matrix The identification was performed offline in the digital twin model using the finite difference method. The standard deviation of the noise of each sensor was taken as the laboratory calibration value, and the standard deviation of the manifold outlet temperature sensor was [value missing]. The standard deviation of the pressure transmitter is This completes the digital twin closed-loop handover from the design phase to the operation phase.

[0038] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0039] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A digital-twin-based refrigeration system design method, characterized in that, Includes the following steps: Step one, each loop of the multi-loop parallel evaporator is discretized into M calculation nodes along the direction of refrigerant flow, and the node length is wherein is the total length of the single loop, and for nodes, the refrigerant side energy conservation equation is established Solve the refrigerant dryness of each node by simultaneously solving all node equations of all N parallel loops Build a digital twin model of the distributed parameter evaporator; Step two, define the outlet dryness vector of each circuit as a state variable Define the temperature, pressure and pressure drop signals at the total outlet header of the evaporator as observation variables Equation State estimation of the outlet dryness of each circuit, Kalman gain matrix Observation matrix Noise covariance matrix Offline identification by Monte Carlo method; Step 3: Discretize the annual operating conditions into Q representative load levels, each load level The corresponding weight is ,satisfy The performance coefficient is weighted by the full operating conditions. Maximizing the number of evaporator loops N and the diameter of a single loop pipe is the optimization objective. Orifice diameter of the separator head plate Maximum flow area of ​​electronic expansion valve To design variables, a differential evolution algorithm is used to jointly optimize the above parameters in the digital twin model, with adjacent generations... The iteration terminates when the improvement is less than 0.

005. Step four: Write the optimal design parameters into the digital twin model to generate a parameterized twin. Integrate the evaporator outlet temperature, suction pressure, discharge pressure, and compressor current signals into the Kalman filter state estimation module from step two. Complete the process noise covariance matrix within 72 hours of system operation. Online adaptive identification.

2. The refrigeration system design method based on digital twins according to claim 1, characterized in that, In step one, in the two-phase region, the first The overall heat transfer coefficient of each node The Gungor-Winterton modified formula was used for calculation, and the Dittus-Boelter single-association formula was used for calculation in the superheated region.

3. The refrigeration system design method based on digital twins according to claim 1, characterized in that, In step three, the cooling capacity The compressor input power is obtained by summing the heat transfer integrals of all nodes in all loops under this load level using a distributed parameter digital twin model. Calculated by interpolation of compressor characteristic maps from the digital twin model.

4. The refrigeration system design method based on digital twins according to claim 1, characterized in that, In step three, the population size of the differential evolution algorithm is 60, the maximum number of iterations is 200, and the search range of the design variables is... .

5. The refrigeration system design method based on digital twins according to claim 1, characterized in that, In step three, the number Q of representative load levels is extracted by clustering the hourly cooling load data for 8760 hours throughout the year, and the weights corresponding to each load level are determined. It is determined by the proportion of time that each cluster load level accounts for in the total annual operating time.

6. The refrigeration system design method based on digital twins according to claim 1, characterized in that, In step three, the orifice plate of the distributor uses different orifice diameters for the outer loop and the inner loop of the evaporator. The values ​​of the orifice diameters of the outer loop and the inner loop are independently determined in the optimization process of the differential evolution algorithm, forming a non-uniform orifice diameter differentiated distributor design.

7. The refrigeration system design method based on digital twins according to claim 1, characterized in that, In step two, the Kalman filter state estimation module simulates different refrigerant distribution deviation scenarios by injecting virtual sensor signals into the digital twin model during the design phase, outputting the dryness distribution estimation error of each loop and the observation matrix. The finite difference method is used for offline computation in the digital twin model.

8. The refrigeration system design method based on digital twins according to claim 1, characterized in that, In step four, the evaporator outlet temperature sensor is a Pt100 type with a noise standard deviation of 0.15K, and the pressure transmitter noise standard deviation is 1.5kPa. These calibration values ​​are used as the sensor noise covariance matrix. The initial input value.

Images