Refrigerating system energy-saving control method and system based on digital twinning

Abstract

The invention relates to the technical field of refrigerating system intelligent control and energy conservation, in particular to a refrigerating system energy-saving control method and system based on digital twinning, and the method comprises the steps: multi-dimensional perception and working condition recognition: collecting running state data and external environment data of a refrigerating system; cleaning the operation state data and the external environment data; whether the environment temperature change rate or the tail end heat load change rate exceeds a conventional fluctuation threshold value or not is judged; if yes, generating a working condition abrupt change trigger signal; active detection type control decision: outputting characteristic parameters of perturbation excitation signals; real-time identification of degradation characteristics: calculating degradation parameters representing the physical health state of the equipment; performing dynamic digital twinborn evolution, calculating an operation parameter combination with the highest energy efficiency ratio, and feeding back the operation parameter combination to the active detection type control decision step; according to the method, online dynamic extraction of the health state of the equipment is realized, and the problem that a traditional static model is not accurate due to the fact that physical degradation of the equipment cannot be sensed is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent control and energy saving of refrigeration systems, in particular to a refrigeration system energy saving control method and system based on digital twinning. BACKGROUND

[0002] The refrigeration system is the core facility of modern building environment regulation and industrial cooling, and its energy efficiency level directly determines the overall operation cost and energy utilization efficiency. The system is mainly composed of a water chiller unit, a compressor, a heat exchanger and an end execution mechanism, and realizes heat transfer and dissipation through the thermodynamic cycle of refrigerant. The existing control technology mainly depends on the static ideal model or fixed empirical rule of the equipment at the time of factory shipment, and does not fully consider the physical degradation characteristics of the equipment in the whole life cycle. With the passage of running time, the increase of thermal resistance caused by the fouling of the heat exchanger surface and the decrease of volumetric efficiency caused by the mechanical wear of the compressor make the actual physical characteristics of the system deviate seriously from the preset static model, causing model misalignment. At the same time, the traditional passive control mode is difficult to identify the real health status of the equipment in real time without interrupting the business operation, and cannot perceive the shrinkage of the safety boundary caused by aging. In the face of drastic changes in environmental temperature or sudden changes in terminal load, if the invalid theoretical model is still blindly pursued for high energy efficiency, it is easy to cause the running parameters to break through the physical safety limit and cause faults such as high-voltage trip, which is difficult to balance the safety and economy of the system. Therefore, there is an urgent need for a solution to solve the problems in the prior art. SUMMARY

[0003] The purpose of the present application is to provide a refrigeration system energy saving control method and system based on digital twinning, which overcomes the defects of the prior art that relies on static models and cannot adapt to the physical degradation of equipment, causing control misalignment and safety hazards, and can real-time correct model deviation and dynamically reconstruct safety boundary through active detection without interrupting operation, so as to realize energy efficiency optimization in the whole life cycle while ensuring the safety of system operation. Specifically, the technical scheme of the present application is as follows:

[0004] A refrigeration system energy saving control method based on digital twinning, comprising:

[0005] Multi-dimensional perception and working condition identification, collecting running state data and external environment data of the refrigeration system; cleaning the running state data and external environment data; judging whether the environmental temperature change rate or the terminal heat load change rate exceeds the normal fluctuation threshold; if yes, generating a working condition mutation trigger signal;

[0006] The active detection type control decision receives the benchmark control strategy generated by the dynamic digital twin evolution step and the operation constraint condition generated by the safety boundary dynamic reconstruction step; in response to the working condition mutation trigger signal, a perturbation excitation signal is superimposed on the benchmark control strategy; a composite control instruction is generated and issued to the physical execution mechanism; and the characteristic parameters of the perturbation excitation signal are output;

[0007] The degradation feature real-time identification receives the characteristic parameters of the perturbation excitation signal; actual response data of the physical execution mechanism to the composite control instruction are collected; the actual response data are compared with the pre-stored theoretical response benchmark; and the degradation parameters representing the physical health state of the equipment are calculated;

[0008] The dynamic digital twin evolution receives the degradation parameters; the boundary conditions and coefficient values of the thermodynamic equation group of the virtual mathematical model are corrected by using the degradation parameters; global optimization is performed based on the corrected virtual mathematical model; the operation parameter combination with the highest energy efficiency ratio is calculated, which is defined as the benchmark control strategy and fed back to the active detection type control decision step;

[0009] The safety boundary dynamic reconstruction performs limit simulation based on the corrected virtual mathematical model; the operation safety red line is dynamically adjusted according to the simulation result; the operation constraint condition is generated and sent to the active detection type control decision step to limit the generation range of the composite control instruction.

[0010] Preferably, the multi-dimensional perception and working condition identification in step 1 comprises:

[0011] The steady state determination logic and the transient trigger logic are configured;

[0012] When the monitored environmental temperature change rate or the end heat load change rate exceeds the conventional fluctuation threshold, it is determined that the working condition has mutated;

[0013] The high-frequency response mode is activated and the working condition mutation trigger signal is output to the active detection type control decision step;

[0014] The operation state data include temperature, pressure and power.

[0015] Preferably, the active detection type control decision in step 2 comprises:

[0016] The active detection mechanism is constructed, and the perturbation excitation signal is a sine signal or a pseudo-random signal with a specific frequency;

[0017] The amplitude range of the perturbation excitation signal is set to be within a predetermined percentage of the total frequency, so as to ensure that no user-perceptible refrigeration temperature fluctuation is caused;

[0018] The thermodynamic dynamic response inside the physical system is excited by using the perturbation excitation signal, so as to expose the current physical characteristics of the system;

[0019] The composite control instruction takes into account both the regulation target and the detection target.

[0020] Preferably, the real-time degradation feature recognition in step 3 comprises:

[0021] Comparing the actual response data with the theoretical response benchmark;

[0022] Calculating the phase lag and amplitude attenuation between the actual response data and the theoretical response benchmark;

[0023] Infer the degradation parameters based on the phase lag and amplitude attenuation;

[0024] The degradation parameters include a fouling thermal resistance coefficient reflecting the degree of fouling of the heat exchanger, and a mechanical wear coefficient reflecting the decline in compressor volumetric efficiency;

[0025] If it is determined that the actual response data lags behind the theoretical response benchmark, it is determined that the fouling thermal resistance coefficient increases; if it is determined that the actual response data has amplitude attenuation relative to the theoretical response benchmark, it is determined that the mechanical wear coefficient increases.

[0026] Preferably, the dynamic digital twin evolution in step 4 comprises:

[0027] Model correction: using the fouling thermal resistance coefficient to correct the heat transfer coefficient of the heat exchanger, and using the mechanical wear coefficient to correct the compressor volumetric efficiency, real-time correction is performed on the compressor polytropic efficiency equation and the heat exchanger heat transfer equation, so that the virtual mathematical model reproduces the current fouling and worn equipment physical properties;

[0028] Global optimization: taking the minimum total energy consumption of the system as the objective function, and taking meeting the current end cold load demand as the constraint condition, a global optimization algorithm is used to perform iterative calculation within the feasible operating domain of the equipment;

[0029] The benchmark control strategy includes compressor frequency, condenser fan speed, and electronic expansion valve opening degree.

[0030] Preferably, the safety boundary dynamic reconstruction in step 5 comprises:

[0031] Using the corrected virtual mathematical model to simulate the exhaust gas pressure and temperature under full load operation state;

[0032] If the simulation result shows that the exhaust gas pressure exceeds the safety threshold, the maximum compressor frequency or the minimum condenser air volume allowed is actively contracted;

[0033] Form a safety envelope that changes dynamically with the health state of the equipment, and use it as a constraint condition for operation.

[0034] A digital twin-based energy-saving control system for a refrigeration system, comprising:

[0035] A multi-dimensional perception and data preprocessing unit is configured to collect data and identify a working condition mutation, and output a working condition mutation trigger signal;

[0036] An active detection type control decision unit is connected with the multi-dimensional perception and data preprocessing unit, configured to receive a reference control strategy, an operation constraint condition and the working condition mutation trigger signal, superimpose a perturbation excitation signal to generate a composite control instruction;

[0037] A degradation feature real-time identification unit is connected with the active detection type control decision unit, configured to calculate a degradation parameter according to a characteristic parameter of the perturbation excitation signal and actual response data;

[0038] A dynamic digital twin evolution unit is connected with the degradation feature real-time identification unit, configured to correct a model and calculate the reference control strategy according to the degradation parameter;

[0039] A safety boundary dynamic reconstruction unit is connected with the dynamic digital twin evolution unit, configured to generate the operation constraint condition according to the corrected virtual mathematical model and feed back to the active detection type control decision unit.

[0040] Compared with the prior art, the present application has the following beneficial effects:

[0041] 1. The present application excites the thermodynamic dynamic response of the system by superimposing the perturbation excitation signal, and uses the phase lag and amplitude attenuation characteristics between the actual response and the theoretical reference to deduce the fouling thermal resistance and the degree of mechanical wear; this method converts the physical aging phenomenon which is difficult to measure directly into an electrical signal feature which can be accurately measured, realizes online dynamic extraction of the equipment health status, and solves the problem that the traditional static model cannot sense the physical degradation of the equipment and is inaccurate;

[0042] 2. The present application uses the identified degradation parameter to correct the virtual mathematical model in real time, so that the model is changed from an idealized formula to an accurate mapping of the physical entity, and global optimization is performed based on the high-fidelity model; this mechanism ensures that each optimization calculation is based on the current real physical capability of the equipment, so as to excavate the real energy efficiency limit that can be achieved under the condition of equipment aging, establish a new control base point, and avoid false optimization of the control strategy caused by model deviation;

[0043] 3. The present application uses the corrected model to simulate the evolution of physical parameters under extreme working conditions, and when it is predicted that the exhaust pressure and other indicators may exceed the safety limit, the upper limit of the compressor frequency or the lower limit of the air volume is actively contracted, and a safety envelope line that changes dynamically with the health status of the equipment is generated; this realizes the dynamic balance between safety and energy efficiency while ensuring the continuous and stable operation of the system, and effectively avoids high-pressure trip accidents caused by blindly pursuing theoretical efficiency;

[0044] 4. The present application adopts perturbation excitation signal superposition technology, uses building thermal inertia as a low-pass filter, realizes parameter identification under the condition that the user is not aware, without shutdown or large changes in working conditions; Cooperate with the dual-mode logic of steady-state judgment and transient triggering, only activate active detection when the working condition changes, which not only ensures the rapid response ability to extreme working conditions, but also avoids unnecessary disturbance and waste of computing resources caused by frequent detection. BRIEF DESCRIPTION OF DRAWINGS

[0045] The present application will be further explained in conjunction with the accompanying drawings and examples:

[0046] Figure 1 is a flow chart of the method of the present application;

[0047] Figure 2 is a structural diagram of the system of the present application. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be further described in detail below in conjunction with specific examples.

[0049] Example 1:

[0050] Please refer to Figure 1 A digital twin-based energy-saving control method for a refrigeration system, comprising:

[0051] Multi-dimensional perception and working condition identification, collecting running state data and external environment data of the refrigeration system; cleaning the running state data and external environment data; determining whether the environmental temperature change rate or the terminal heat load change rate exceeds the normal fluctuation threshold; if so, generating a working condition mutation trigger signal;

[0052] Active detection type control decision, receiving the reference control strategy generated by the dynamic digital twin evolution step, and the operation constraint condition generated by the safety boundary dynamic reconstruction step; superimposing a perturbation excitation signal on the reference control strategy in response to the working condition mutation trigger signal; generating a composite control instruction and issuing it to the physical execution mechanism; outputting the characteristic parameters of the perturbation excitation signal;

[0053] Real-time identification of degradation characteristics, receiving the characteristic parameters of the perturbation excitation signal; collecting actual response data of the physical execution mechanism to the composite control instruction; comparing the actual response data with the pre-stored theoretical response reference; calculating the degradation parameters representing the physical health state of the equipment;

[0054] Dynamic digital twin evolution, receiving the degradation parameters; using the degradation parameters to modify the boundary conditions and coefficient values of the thermodynamic equation set of the virtual mathematical model; performing global optimization based on the modified virtual mathematical model; calculating the operating parameter combination with the highest energy efficiency ratio, defining it as the reference control strategy and feeding it back to the active detection type control decision step;

[0055] The safety boundary is dynamically reconstructed, limit simulation is performed based on the corrected virtual mathematical model, the operation safety red line is dynamically adjusted according to the simulation result, and operation constraint conditions are generated and sent to the active detection type control decision step to limit the generation range of the composite control instruction.

[0056] In the operation logic of the embodiment, a closed-loop coupled control architecture is constructed, and the core design concept is to solve the industry common problem that the static model is inaccurate due to physical degradation in the whole life cycle operation of the refrigeration system; when the multi-dimensional perception and working condition identification step is performed, the system establishes a connection interface between the physical world and the digital control logic, and identifies the key time nodes that need high-precision model calibration through real-time data stream analysis; in the data acquisition and processing stage, the logic core continuously receives the sensor data array of the refrigeration unit, including the inlet and outlet water temperature of the water chiller, the suction and discharge pressure of the compressor, and the environmental meteorological station data; after data cleaning based on sliding average filtering or Kalman filtering, the system focuses on monitoring the time series change slope of the environmental temperature or the terminal load; when the change rate value calculated by the algorithm exceeds the preset normal fluctuation threshold, for example, in the scene of sudden pressure change before summer rain or sudden increase of heat load due to sudden increase of computing power in the data center, it is determined that the current working condition enters the mutation state; the confidence of the original static model decreases in this state, and the system immediately generates and outputs a working condition mutation trigger signal to activate the subsequent active calibration process;

[0057] The logic flow goes to the active detection type control decision step, which is the core of decision processing, breaks the limitation of traditional control which only passively executes preset instructions, and introduces an active excitation mechanism; this unit receives the reference control strategy calculated by the upper optimization layer, that is, the theoretically optimal operating point, and the operation constraint conditions from the safety module, that is, the physical safety boundary; once the working condition mutation trigger signal is captured, the step starts the signal modulation logic, and actively superimposes a small perturbation excitation signal on the basis waveform of the reference control strategy; the superposition operation of the signal generates the composite control instruction finally issued to the physical execution mechanism; in this process, the system synchronously outputs the frequency, amplitude characteristic parameters of the artificially applied excitation signal to provide accurate reference for the subsequent identification link;

[0058] After the physical device executes the instructions, the degradation feature real-time recognition step uses the input and output feedback data stream of the system to quantify the health of the device; this step receives the just-applied excitation signal feature as an input reference and synchronously collects the temperature or pressure fluctuation curve actually exhibited by the physical system as an output response; by deeply comparing and analyzing the actual response data and the theoretically stored response reference in time and frequency domains, the algorithm can identify the difference between the two; based on this difference feature, the current degradation parameter is calculated by reverse deduction; the degradation parameter mentioned in this embodiment specifically refers to a set of indexes for quantifying the performance degradation of the physical components of the refrigeration device, and the value thereof is derived by analyzing the response lag and amplitude attenuation characteristics of the system to a specific excitation signal, and is used to accurately reflect physical facts such as the increase of the heat resistance of the heat exchanger surface due to dirt or the decrease of the volumetric efficiency due to internal leakage of the compressor;

[0059] After obtaining the real device state data, the dynamic digital twin evolution step is responsible for performing the operation of eliminating model drift; this step receives the latest degradation parameter and substitutes it into the thermodynamic equation set of the virtual mathematical model to perform online dynamic correction on the key coefficients; the corrected virtual model is transformed from an idealized mathematical formula into a high-fidelity mapping of the physical entity in the digital space; based on this mapping model, the system takes the energy efficiency ratio maximization as the objective function, performs global optimization iteration in the feasible region, calculates the optimal operating parameter combination that can actually be realized under the current device aging state, and feeds it back to the control decision link as the updated reference control strategy, establishing a new control base point;

[0060] At the end of the closed loop, the safety boundary dynamic reconstruction step guarantees the safety of the system operation; this step uses the corrected high-fidelity model to perform forward-looking prediction simulation; it simulates the extreme physical parameter evolution that may occur if the device is running at full capacity under the current aging state; once the simulation result shows that a certain parameter will break through the physical safety limit, the step will trigger the boundary contraction algorithm, reduce the allowed operating range of the system, and generate more stringent operating constraint conditions to send to the control decision step, thereby forcibly limiting the generation range of the control instruction on the physical level;

[0061] A closed-loop mechanism of control as detection and model as constraint is constructed, eliminating the contradiction between model static and device dynamic degradation in traditional digital twin control; by actively applying perturbation excitation, the system can real-time perceive the degree of fouling and wear of itself without interrupting the business operation, and calibrate the control model and safety boundary in real time accordingly; this mechanism ensures that the system always operates at the real energy efficiency optimal point throughout its life cycle, and can automatically identify the decrease of safety margin caused by device aging, actively contract the operating boundary, and effectively avoid high-pressure trip accidents caused by blindly pursuing theoretical high efficiency.

[0062] Embodiment 2:

[0063] The multi-dimensional perception and working condition recognition in Step 1 includes:

[0064] The steady state determination logic and transient state triggering logic are configured;

[0065] When the monitored environmental temperature change rate or end heat load change rate exceeds the conventional fluctuation threshold, it is determined that the working condition has changed abruptly;

[0066] The high-frequency response mode is activated and a working condition change triggering signal is output to the active probing control decision step;

[0067] The operating state data includes temperature, pressure, and power.

[0068] In this embodiment, the multi-dimensional perception and working condition recognition step uses a dual-mode logic judgment mechanism to improve the environmental adaptability of the system; the steady state determination logic and transient state triggering logic are pre-configured in the system; in the conventional operation scenario, the system executes the steady state logic and maintains a low-frequency sampling strategy to optimize the allocation of computing resources; as soon as a significant fluctuation in the key indicators is detected, the transient state logic takes over the execution; the conventional fluctuation threshold defined in this embodiment is a time series change rate limit value that distinguishes between normal adjustment fluctuations of the system and external environmental changes, which is obtained based on historical operation big data statistical analysis; the setting logic of this threshold is usually 1 to 2 times the standard deviation of the parameter change rate in the normal working condition, and its technical effect is to serve as a judgment basis for working condition recognition, aiming to filter out sensor white noise and only respond to working condition changes with clear physical meaning;

[0069] When the monitored environmental temperature change rate or end heat load change rate exceeds the conventional fluctuation threshold, it is determined that the working condition has changed abruptly; in response to this determination, the system activates the high-frequency response mode, increases the data acquisition frequency to several times that of the conventional mode, and generates a working condition change triggering signal to send to the subsequent step; the system simultaneously starts a reset timer to continuously monitor the environmental temperature change rate and the end heat load change rate; if the above two change rates are continuously below the conventional fluctuation threshold within a preset stable determination time window, for example, 30 minutes, the system determines that the working condition has returned to normal, and automatically stops outputting the working condition change triggering signal, closes the superposition of the perturbation excitation signal, and restores the data acquisition frequency to the low-frequency steady state mode;

[0070] ​​Meanwhile, if the system is in a low-frequency steady-state mode for a long time, for example, more than 24 hours, the system will forcibly activate a primary active detection process to prevent slow and gradual degradation in the long-term steady state from being missed; in addition, the operating state data collected in the embodiment covers temperature data reflecting the thermodynamic state, including the evaporation temperature and the condensation temperature; pressure data, including the suction pressure and the exhaust pressure; and electric power data reflecting the energy consumption characteristics, ensuring the completeness of the data source in the thermodynamic and kinetic dimensions;

[0071] By distinguishing between steady state and transient state logic, a dynamic balance of computing resources and response speed is achieved; the system reduces the computing load when running smoothly, and quickly improves the sensing sensitivity at the critical moment of working condition mutation; this threshold-triggered mechanism ensures that the active detection action is only performed at the most necessary moment, which not only guarantees the rapid response ability to extreme working conditions, but also avoids unnecessary disturbance to the system caused by frequent detection.

[0072] Embodiment 3:

[0073] The active detection type control decision in step 2 includes:

[0074] An active detection mechanism is constructed, and the perturbation excitation signal is a sine signal or a pseudo-random signal of a specific frequency;

[0075] The amplitude range of the perturbation excitation signal is set within a predetermined percentage of the total frequency to ensure that it does not cause user-perceptible refrigeration temperature fluctuations;

[0076] The perturbation excitation signal is used to excite the thermodynamic dynamic response inside the physical system, thereby exposing the current physical characteristics of the system;

[0077] The composite control instruction takes into account both the adjustment target and the detection target.

[0078] In the embodiment, the technical core of the active detection type control decision step lies in constructing a non-intrusive parameter identification excitation mechanism; when generating the control instruction, the signal superposition synthesis technology path is adopted; the system selects a sine wave or a pseudo-random sequence of a specific frequency as the perturbation excitation signal; in order to strictly balance the contradiction between detection accuracy and user comfort, the system sets clear quantitative constraints on the strength of the signal; the perturbation excitation signal defined in the embodiment is a small-amplitude fluctuation signal artificially constructed with specific spectral characteristics, and its amplitude range is strictly limited to to ; and the lower limit of the amplitude of the perturbation excitation signal needs to meet , wherein The minimum action resolution step of the physical actuator, such as a frequency converter or an electronic expansion valve, for example, the frequency converter is 0.5 Hz, and the valve is 10 steps, to prevent the excitation signal from being submerged by quantization noise due to being lower than the hardware response dead zone, and to ensure the effectiveness of the detection; the physical basis of the amplitude setting is that the amplitude is sufficient to drive the actuator to produce physical action that can be captured by the sensor; combined with the high-frequency characteristics, the signal will be smoothed and filtered by the huge thermal inertia of the building envelope, thereby ensuring that the end environment temperature does not fluctuate perceptibly to the user;

[0079] During execution, the perturbation excitation signal is linearly superimposed on the reference control strategy, and the generated composite control instruction has dual functional properties: its direct current component or low frequency component continues to drive the refrigeration system to meet the cooling demand and achieve the regulation target; and its high-frequency perturbation component acts as a detection wave to excite the dynamic response of the compressor and the internal fluid of the heat exchanger, thereby achieving the detection target; this design enables the system physical properties hidden behind the steady-state data, such as the thermal inertia time constant and the heat transfer coefficient, to be excited and exposed;

[0080] Innovatively, the control and detection functions are synchronously coupled; compared to the defect that the traditional method must be stopped or greatly changed in working condition to measure the equipment characteristics, the embodiment, through the fine design of the perturbation excitation signal, uses the building thermal inertia as a natural low-pass filter, and realizes online and dynamic extraction of the physical characteristics of the refrigeration system on the premise that the user is completely unaware, thereby providing key data support for subsequent precise modeling.

[0081] Embodiment 4:

[0082] The real-time degradation feature recognition in step 3 includes:

[0083] Comparing the actual response data with the theoretical response reference;

[0084] Calculating the phase lag amount and the amplitude attenuation amount between the actual response data and the theoretical response reference;

[0085] Based on the phase lag amount and the amplitude attenuation amount, the degradation parameters are deduced reversely;

[0086] The degradation parameters include a dirt thermal resistance coefficient reflecting the degree of fouling of the heat exchanger and a mechanical wear coefficient reflecting the decrease in the volumetric efficiency of the compressor;

[0087] If it is determined that the actual response data lags relative to the theoretical response reference, it is determined that the dirt thermal resistance coefficient increases; if it is determined that the actual response data attenuates in amplitude relative to the theoretical response reference, it is determined that the mechanical wear coefficient increases.

[0088] In this embodiment, the degradation feature real-time identification step performs a logic process similar to dynamic system identification; when the physical system is excited by a perturbation excitation signal, the actual response data recorded by the sensor, such as the fluctuation curve of the condensing pressure, is introduced into the analysis module; the module performs a reference dynamic generation step: calling the initial non-degradation mathematical model stored in the system device factory, the external environment data collected in step 1 is input as the boundary condition into the initial model; superimpose the same perturbation excitation signal on the initial model as the physical system for online simulation, and calculate the dynamic theoretical response reference that the device should exhibit under the current working condition when the device is in an ideal healthy state;

[0089] The processing logic mainly focuses on the comparison of the dynamic characteristics of the signals; the system performs cross-correlation analysis or fast Fourier transform analysis to accurately calculate the phase lag and amplitude attenuation of the actual waveform relative to the theoretical waveform; these two indicators have a clear mapping relationship with physical degradation: the system constructs a multi-parameter coupled identification matrix. Since the increase of the dirt thermal resistance and the aggravation of the mechanical wear will have a nonlinear coupling effect on the phase and amplitude of the response signal, this step does not use single feature mapping, but uses least squares method or maximum likelihood estimation method; specifically, the system constructs a differential algebraic equation DAE based on the conservation of mass and energy to describe the dynamic characteristics of the refrigerant as an observation model:

[0090]

[0091] wherein, is the density of the refrigerant, is the partial derivative of the refrigerant density with respect to pressure, is the effective volume of the condenser, is the dynamic response of the condensing pressure, is the compressor discharge mass flow rate, is the refrigerant flow rate out of the condenser;

[0092] The introduction of the to-be-identified parameters constructs the correlation equation:

[0093] The mapping relationship between the discharge mass flow rate and the mechanical wear coefficient is:

[0094]

[0095] The mapping relationship between the condenser heat transfer process and the dirt thermal resistance coefficient is reflected through the influence on the condensing pressure change rate, and its dynamic thermal balance equation is:

[0096]

[0097] Based on the above differential equation, the observation vector With parameter vector Jacobian matrix The specific form is defined as follows:

[0098]

[0099] The partial derivative terms By applying parameter perturbations to the reference model The numerical solution of the above differential equation system is obtained by solving it. With the goal of minimizing the combined residuals of the actual response and the theoretical response in terms of phase and amplitude, the current fouling thermal resistance coefficient and mechanical wear coefficient are obtained by solving them together.

[0100] In this embodiment, the fouling thermal resistance coefficient is defined as a physical index that quantifies the degree of fouling on the surface of the condenser or evaporator and its hindering effect on heat transfer efficiency; the mechanical wear coefficient is defined as a physical index that quantifies the degree of loss in actual gas delivery relative to theoretical gas delivery caused by increased internal clearance or valve wear of the compressor; for example, when the system detects a significant phase lag in the response time of the condensing pressure to changes in the compressor frequency, the logic determines that the thermal resistance of the heat exchange tube wall has increased, thereby calculating the increased fouling thermal resistance coefficient; similarly, if the pressure response amplitude is lower than the theoretical prediction, the increase in the mechanical wear coefficient is derived.

[0101] The physical aging phenomenon, which is difficult to measure directly, is transformed into an electrical signal characteristic that can be measured precisely. This identification method based on dynamic response analysis has higher sensitivity than analysis that relies solely on steady-state data. It can accurately distinguish whether the performance fluctuation is caused by environmental changes or by performance degradation caused by the hardware degradation of the device itself, providing an accurate and quantitative basis for model correction.

[0102] Example 5:

[0103] Step 4, the dynamic digital twin evolution, includes:

[0104] Model correction: The heat transfer coefficient of the heat exchanger is corrected by using the fouling thermal resistance coefficient, and the volumetric efficiency of the compressor is corrected by using the mechanical wear coefficient. The compressor multivariable efficiency equation and the heat transfer equation of the heat exchanger are corrected in real time, so that the virtual mathematical model can reproduce the physical characteristics of the equipment that is currently fouled and worn.

[0105] Global optimization: Taking the minimum total system energy consumption as the objective function and meeting the current terminal cooling load demand as the constraint, the global optimization algorithm is used to perform iterative calculations within the feasible operating domain of the equipment;

[0106] The baseline control strategy includes compressor frequency, condenser fan speed, and electronic expansion valve opening.

[0107] In this embodiment, the dynamic digital twin adaptive correction step includes two core sub-processes: parameter mapping update and global optimization. During the model correction stage, the system uses the fouling thermal resistance coefficient and mechanical wear coefficient identified in the previous steps to update the parameters of the core physical equations of the virtual model. Specifically, the operation logic is as follows: the fouling thermal resistance coefficient is substituted into the heat transfer coefficient term of the heat exchanger heat transfer equation, and the mechanical wear coefficient is substituted as a correction term into the compressor multivariable efficiency equation. Specifically, for the heat exchanger heat transfer equation... ,in The system updates the overall heat transfer coefficient based on the logarithmic mean temperature difference. The calculation formula is: ,in The thermal resistance of the pipe wall is usually ignored or treated as a constant in calculations because its value is extremely small. ,in The thermal resistance coefficient of the currently identified fouling. and These are the heat transfer coefficients of the inner and outer surfaces of the pipe, calculated in real time based on the current refrigerant flow rate and fan air volume. Specifically, for heat exchange on the outer surface of the pipe, the Colburn coefficient is used. Calculations are performed using factor correlations; for heat transfer inside the tube, if it is a single-phase flow region, the Dittus-Boelter correlation is used:

[0108]

[0109] For two-phase flow regions, in condensation or evaporation processes, the Shah correlation is used:

[0110]

[0111] in The heat transfer coefficient at the liquid phase surface is... For dryness, To compare pressures, the system uses the Reynolds number calculated in real time. With dryness Automatically switch the above correlations to ensure and The calculation accuracy is improved; for the compressor's variable efficiency equation, the system uses volumetric efficiency. ,in The product of the factory baseline coefficient and (1 - the currently identified mechanical wear coefficient) is used to reflect the volumetric efficiency reduction caused by increased wear; this operation transforms the virtual model from describing the factory standard equipment to accurately describing the currently operating equipment with defects, eliminating the deviation between the model and the physical entity;

[0112] During the global optimization phase, the system performs optimization tasks based on this modified high-fidelity model. The optimization logic sets the total system energy consumption, i.e., minimizing the sum of the energy consumption of the compressor, fan, and water pump, as the objective function, and takes meeting the terminal cooling load demand as a hard constraint that must be followed. Based on this, a method such as sequential quadratic programming is adopted. Or Particle Swarm Optimization Global optimization algorithms are used to perform multi-dimensional search and iteration within the parameter space allowed by the equipment. Finally, a set of parameters that are optimal in terms of energy efficiency under the current health condition of the equipment is calculated, including the compressor operating frequency, the condenser fan speed and the opening of the electronic expansion valve. This set of parameters constitutes a new benchmark control strategy.

[0113] Adaptive control optimization is achieved. Traditional optimization control is often based on a fixed model, which can easily give theoretically optimal but actually suboptimal or even infeasible instructions after the equipment ages. This embodiment ensures that every optimization calculation is based on the actual physical capabilities of the equipment at this moment, thereby uncovering the true energy efficiency limit that can be achieved under the conditions of equipment aging and avoiding false optimization of the control strategy.

[0114] Example 6:

[0115] Step 5, dynamic reconstruction of the security boundary, includes:

[0116] The modified virtual mathematical model was used to simulate the exhaust pressure and temperature under full-load operation.

[0117] If the simulation results show that the exhaust pressure exceeds the safety threshold, then actively reduce the maximum allowable compressor frequency or the minimum condenser air volume;

[0118] A safety envelope that dynamically changes with the health status of the equipment is formed and used as an operational constraint.

[0119] In this embodiment, the dynamic reconstruction step of the safety boundary focuses on the pre-management of risks. This step uses a modified virtual mathematical model that includes information on dirt accumulation and wear to perform simulations under extreme working conditions. The system sets the virtual equipment to operate at full load and, in conjunction with the current or predicted ambient temperature, simulates and calculates key safety indicators, especially the compressor's exhaust pressure and exhaust temperature.

[0120] Logical judgment mechanism intervention: If the simulation results show that, under the current level of fouling, full-load operation will cause the exhaust pressure to exceed the physical safety threshold, such as the high-pressure switch action value, the system will determine that the original operating boundary is no longer safe; the system will automatically perform boundary contraction action, reducing the upper limit of the maximum allowable compressor frequency or increasing the lower limit of the minimum allowable condensing air volume;

[0121] The contraction action is executed using an iterative approximation algorithm: the system sets a frequency adjustment step size, such as 2Hz, and an airflow adjustment step size, such as 5%; when the simulation result exceeds the limit, a frequency adjustment step size is subtracted from the current maximum frequency upper limit, or an airflow adjustment step size is added to the current minimum condensing airflow lower limit; after adjustment, the virtual mathematical model is called again for simulation verification; if the result still exceeds the limit, the above deduction or addition operation is repeated until the exhaust pressure calculated by simulation is lower than the physical safety threshold; this process forms a dynamically changing safety envelope; the safety envelope mentioned in this embodiment specifically refers to a closed region in a multi-dimensional parameter space, the boundary of which defines the parameter limit set for which the equipment can operate stably for a long time without triggering fault protection under the current healthy state; this envelope is transformed into specific operating constraints and forcibly fed back to the control decision unit;

[0122] This solution addresses the safety issues associated with operating with faulty components. In cases of heat exchanger blockage or other problems, blindly pursuing higher cooling capacity can easily lead to high-pressure tripping and business interruption. This embodiment utilizes the predictive capabilities of digital twins to anticipate the shrinking safety boundaries caused by equipment aging and proactively limits the aggressiveness of control strategies. Thus, in extreme high-temperature weather, it sacrifices some unnecessary adjustment speeds or maximum cooling capacity in exchange for continuous and uninterrupted stable system operation, achieving a dynamic balance between safety and energy efficiency.

[0123] Example 7:

[0124] Please see Figure 2 A digital twin-based energy-saving control system for refrigeration systems includes:

[0125] The multi-dimensional sensing and data preprocessing unit is used to collect data and identify sudden changes in operating conditions, and output a signal triggering a sudden change in operating conditions.

[0126] An active detection-type control decision unit, connected to a multi-dimensional sensing and data preprocessing unit, is used to receive baseline control strategies, operating constraints, and trigger signals for sudden changes in operating conditions, and superimpose perturbation excitation signals to generate composite control commands.

[0127] The degradation feature real-time identification unit is connected to the active detection control decision unit and is used to calculate degradation parameters based on the characteristic parameters of the perturbation excitation signal and the actual response data.

[0128] A dynamic digital twin evolutionary unit, connected to a real-time degradation feature identification unit, is used to correct the model based on degradation parameters and calculate the baseline control strategy.

[0129] The safety boundary dynamic reconstruction unit, connected to the dynamic digital twin evolution unit, is used to generate operating constraints based on the modified virtual mathematical model and feed them back to the active probing control decision unit.

[0130] This embodiment discloses a hardware system architecture for executing the above method; the system consists of five tightly coupled functional units:

[0131] The multi-dimensional sensing and data preprocessing unit serves as the system's sensing interface. It integrates sensor interfaces and data cleaning algorithms, and has built-in recognition logic for environmental changes, enabling it to send a trigger signal for changes in operating conditions to the system.

[0132] The active detection control decision unit acts as the processing and decision-making core of the system. It is not only responsible for executing conventional control, but also has the ability to actively explore. After receiving a sudden change signal, it injects the perturbation excitation signal for detection into the control loop to generate composite commands to drive the physical device.

[0133] The real-time degradation feature identification unit acts as the system's diagnostic analysis module. It is connected to the control decision unit, acquires the characteristics of the detection signal, and compares it with the actual feedback of the physical system to quantify the degradation parameters of the equipment.

[0134] The dynamic digital twin evolutionary unit acts as the system's digital modeling and optimization module. It receives degradation parameter reports, corrects its internal mathematical model, and calculates the optimal baseline control strategy accordingly.

[0135] The safety boundary dynamic reconfiguration unit acts as the system's risk assessment module. It performs risk simulation based on the revised model, defines the latest operating constraints, and feeds these constraints back to the control decision unit to form closed-loop control.

[0136] By designing modular hardware or software units, the complex control logic of the digital twin is decoupled into five independent functional blocks: perception, decision-making, identification, evolution, and security. At the same time, through clear signal flow directions, namely trigger signals, degradation parameters, and constraints, organic collaboration between the units is achieved. This architecture not only facilitates the engineering deployment and maintenance of the system, but also ensures the efficient closed-loop flow of data between the physical entity and the digital twin, supporting energy efficiency optimization functions throughout the entire life cycle.

[0137] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A digital-twin-based energy-saving control method for a refrigeration system, characterized by, Comprise: Multi-dimensional perception and working condition recognition, collect the running state data and external environment data of the refrigeration system; Clean the running state data and external environment data; judge whether the environmental temperature change rate or the end heat load change rate exceeds the conventional fluctuation threshold; if so, generate a working condition mutation trigger signal; Active detection type control decision, receive the reference control strategy generated by the dynamic digital twin evolution step, and the operation constraint condition generated by the safety boundary dynamic reconstruction step; in response to the working condition mutation trigger signal, superimpose a perturbation excitation signal on the reference control strategy; generate a composite control instruction and issue it to the physical execution mechanism; output the characteristic parameters of the perturbation excitation signal; Degradation feature real-time identification, receive the characteristic parameters of the perturbation excitation signal; Collect the actual response data of the physical execution mechanism to the composite control instruction; Compare the actual response data with the pre-stored theoretical response reference; Calculate the degradation parameter representing the physical health status of the equipment; Dynamic digital twin evolution, receive the degradation parameter; Use the degradation parameter to modify the boundary conditions and coefficient values of the thermodynamic equation set of the virtual mathematical model; perform global optimization based on the modified virtual mathematical model; Calculate the operation parameter combination with the highest energy efficiency ratio, define it as the reference control strategy and feed back to the active detection type control decision step; Safety boundary dynamic reconstruction, perform limit simulation based on the modified virtual mathematical model; dynamically adjust the operation safety red line according to the simulation results; generate operation constraint conditions and send them to the active detection type control decision step to limit the generation range of the composite control instruction.

2. The digital-twin-based energy-saving control method for a refrigeration system according to claim 1, characterized in that, The multi-dimensional perception and working condition recognition in step 1 comprises: Configure steady-state judgment logic and transient-state trigger logic; When the environmental temperature change rate or the end heat load change rate is monitored to exceed the conventional fluctuation threshold, it is determined that the working condition has mutated; Activate the high-frequency response mode and output the working condition mutation trigger signal to the active detection type control decision step; The running state data includes temperature, pressure and power.

3. The digital-twin-based energy-saving control method for a refrigeration system according to claim 1, characterized in that, The active detection type control decision in step 2 comprises: Build an active detection mechanism, the perturbation excitation signal is a sine signal or a pseudo-random signal of a specific frequency; Set the amplitude range of the perturbation excitation signal within a predetermined percentage of the total frequency to ensure that it does not cause user-perceptible refrigeration temperature fluctuations; Use the perturbation excitation signal to excite the thermodynamic dynamic response inside the physical system, thereby exposing the current physical characteristics of the system; The composite control instruction takes into account both the adjustment target and the detection target.

4. The digital-twin-based energy-saving control method for a refrigeration system according to claim 1, characterized in that, The degradation feature real-time identification in step 3 comprises: Compare the actual response data with the theoretical response reference; Calculate the phase lag and amplitude attenuation between the actual response data and the theoretical response reference; Based on the phase lag and amplitude attenuation, deduce the degradation parameter in reverse; The degradation parameter includes the dirt heat resistance coefficient reflecting the degree of fouling of the heat exchanger, and the mechanical wear coefficient reflecting the decline in compressor volumetric efficiency; If it is determined that the actual response data lags behind the theoretical response reference, it is determined that the dirt heat resistance coefficient increases; if it is determined that the actual response data has amplitude attenuation relative to the theoretical response reference, it is determined that the mechanical wear coefficient increases.

5. The digital-twin-based energy-saving control method of a refrigeration system according to claim 1, characterized in that, The dynamic digital twin evolution in step 4 comprises: Model correction: the heat transfer coefficient of the heat exchanger is corrected by the fouling thermal resistance coefficient, and the compressor volumetric efficiency is corrected by the mechanical wear coefficient. The real-time correction of the compressor polytropic efficiency equation and the heat exchanger heat transfer equation makes the virtual mathematical model reproduce the physical characteristics of the current fouling and worn equipment; Global optimization: taking the minimum total energy consumption of the system as the objective function and meeting the current terminal cooling load demand as the constraint condition, the global optimization algorithm is used for iterative calculation in the feasible operation domain of the equipment; The reference control strategy includes the compressor frequency, the condenser fan speed and the electronic expansion valve opening degree.

6. The digital-twin-based energy-saving control method of a refrigeration system according to claim 1, wherein, The dynamic reconstruction of the safety boundary in step 5 includes: Using the corrected virtual mathematical model to simulate the exhaust pressure and temperature under full load operation state; If the simulation result shows that the exhaust pressure exceeds the safety threshold, the maximum compressor frequency or the minimum condenser air volume allowed is actively contracted; Forming a safety envelope line that changes dynamically with the health state of the equipment, and taking it as the operation constraint condition.

7. A digital-twin-based energy-saving control system for a refrigeration system, applied to the digital-twin-based energy-saving control method for a refrigeration system according to any one of claims 1 to 6, characterized in that, It includes: A multi-dimensional perception and data preprocessing unit is used to collect data and identify working condition mutations, and output working condition mutation trigger signals; An active detection type control decision unit is connected with the multi-dimensional perception and data preprocessing unit, used to receive the reference control strategy, the operation constraint condition and the working condition mutation trigger signal, and to superimpose a perturbation excitation signal to generate a composite control instruction; A real-time degradation feature recognition unit is connected with the active detection type control decision unit, used to calculate the degradation parameters according to the characteristic parameters of the perturbation excitation signal and the actual response data; A dynamic digital twin evolution unit is connected with the real-time degradation feature recognition unit, used to correct the model according to the degradation parameters and calculate the reference control strategy; A safety boundary dynamic reconstruction unit is connected with the dynamic digital twin evolution unit, used to generate the operation constraint condition according to the corrected virtual mathematical model and feed back to the active detection type control decision unit.

Images