Energy-saving control method and system for water chilling unit

Abstract

The invention discloses an energy-saving control method and system for a water chilling unit, and relates to the technical field of energy-saving control. During operation of the system, a data set is synchronously collected from a water chilling unit system through a multi-channel asynchronous sampling mechanism, time sequence compression and redundancy removal are carried out, an embedded perturbation calculation mechanism is used for calculating and generating a non-dominant control core coefficient, and the non-dominant control core coefficient is used for controlling the energy-saving control of the water chilling unit; a three-dimensional coefficient space is converted and output through an entropy state change structure and is used for constructing a state balance atlas, comprehensively calculating a control state index SEEI, evaluating the current energy state offset degree of a system, determining whether intervention is carried out or not, carrying out rule search and nonlinear modeling based on the control state index SEEI value, generating an adjustment matrix, and carrying out state balance analysis. And starting a data reconstruction micro-strategy of a short-time historical window, receiving an adjustment matrix, converting the adjustment matrix into a device-level instruction, executing an action through an edge controller, feeding back a response error epsilon (t) in real time, and predicting a potential performance degradation trend based on long-time system operation data.

Description

Technical Field

[0001] The present invention relates to the technical field of energy-saving control, and in particular to an energy-saving control method and system for a chiller. Background Art

[0002] As the core equipment of air conditioning and industrial cooling systems, the operating efficiency of chillers directly determines the energy consumption and economic efficiency of the entire system. Traditional chillers rely primarily on fixed parameter settings and simplified PID control logic to maintain the refrigeration cycle. Their control methods often use extensive adjustment based on return water temperature and compressor load. However, with the improvement of building energy efficiency standards and the strengthening of industrial energy conservation policies, improving the energy responsiveness of chillers under variable loads and complex boundary conditions has become a key research topic in energy-saving control.

[0003] Although some current chiller control systems have introduced variable frequency drives, PID self-tuning, and cloud-based remote monitoring capabilities, they still have significant deficiencies in multi-source data integration capabilities, control decision granularity, and execution feedback loop accuracy. Existing systems generally use single-dimensional signals such as apparent temperature, current, and voltage to derive adjustment parameters, ignoring the nonlinear interactive coupling relationship between heat channels, flow control disturbances, and load matching. This leads to delayed strategy response, insensitive disturbance identification, large load forecast deviations, and an inability to respond flexibly under complex boundary conditions. In addition, traditional systems lack high-dimensional indicator evaluation mechanisms, making it difficult to accurately determine whether the system is currently in a skewed or high-consumption area. There is also a lack of an evolutionary judgment mechanism for potential performance degradation. When cooling efficiency decreases, timely intervention and maintenance are often not possible, resulting in increased energy consumption and increased equipment aging.

[0004] The fundamental reason for response hysteresis and strategy mismatch in existing chiller control methods lies in structural flaws at the information processing level: low representation dimensionality, isolated response units, and a broken feedback mechanism. When heat flux slope fluctuates dramatically, water flow disturbances suddenly increase, or load distribution changes dramatically, traditional systems are unable to identify precursors to perturbations from multiple sources. They lack the ability to globally assess energy state shifts and often intervene only after frequent compressor starts and stops, energy consumption increases, and heat transfer efficiency plummets. Furthermore, the lack of time-series modeling capabilities for operating state trends makes it difficult to provide early warning and intervention when the equipment remains in a subhealthy state for extended periods. This can easily lead to abnormalities such as frequent compressor starts and stops, ineffective pump operation, unstable condensing pressure, and large fluctuations in energy consumption, severely impacting system energy efficiency and equipment lifespan. Therefore, there is an urgent need to develop an energy-saving chiller control system with multi-parameter entropy state perception, exponential state assessment, adaptive control, and long-term evolutionary maintenance capabilities to address the current system's shallow perception, extensive control, and lack of evolutionary control. Summary of the Invention

[0005] In view of the deficiencies in the prior art, the present invention provides an energy-saving control method and system for a chiller, which solves the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: an energy-saving control system for a chiller, comprising a multi-source parameter acquisition module, an entropy conversion module, a system energy state balance module, an adaptive strategy generation module, an edge execution response module, and a dynamic evolution maintenance module;

[0007] The multi-source parameter acquisition module is used to synchronously collect dynamic heat flux imbalance data, flow control disturbance data, and power consumption matching data from the chiller system using a multi-channel, asynchronous sampling mechanism, and perform time series compression and redundancy removal.

[0008] The entropy conversion module is used to input the three sets of collected data into the model, use the embedded perturbation calculation mechanism to calculate and generate the non-explicit control core coefficients, and output the three-dimensional coefficient space through the entropy state change structure transformation to construct the state equilibrium map;

[0009] The system energy state balance module is used to comprehensively calculate the control state index SEEI, evaluate the current energy state deviation of the system, and decide whether to intervene;

[0010] The adaptive strategy generation module is used to perform rule search and nonlinear modeling based on the control state index SEEI value, generate a regulation matrix, and enable micro-strategy reconstruction based on data from a short-term historical window;

[0011] The edge execution response module is used to receive the adjustment matrix and convert it into device-level instructions, execute actions through the edge controller and provide real-time feedback of the response error ε(t);

[0012] The dynamic evolution maintenance module is used to predict potential performance degradation trends based on long-term system operation data and implement early maintenance intervention plan formulation.

[0013] Preferably, the multi-source parameter acquisition module includes a heat flux sensing unit, a flow disturbance acquisition unit and a work load sensing unit;

[0014] The heat flux sensing unit is used to place a bidirectional thermocouple probe pair at the inlet and outlet of the return pipe to collect the temperature difference between the outlet and return water: the outlet and return water temperature difference t g , use a micro-pitch thermocouple array at both ends of the internal channel of the refrigerant evaporator or condenser to collect the slope of the change in the heat transfer rate of the main heat exchanger: heat transfer slope s ht , collect the refrigerant channel temperature gradient change through the heat flux diaphragm sensor: refrigerant temperature gradient rate ΔT, and collect the heat flux density per unit time through the RT temperature probe: unit heat flux density forming a dynamic heat flux imbalance data set;

[0015] The flow disturbance acquisition unit is used to collect the flow velocity v of the main channel of the water flow by deploying an ultrasonic flow meter in the main pipeline section of the water system. f and velocity variance γ v , by setting pressure ports in the outlet and return sections and placing differential pressure sensors to collect the rate of change of the inlet and outlet pressure difference δ pr The dynamic throttling resistance coefficient λ is collected by the electronic valve encoder and torque detector installed in the bypass valve control section. t , forming a flow control disturbance data group;

[0016] The load estimation unit is used to collect input power, current, and voltage, and predict load demand in combination with the system operation log. The real-time input power e is obtained through the energy meters deployed at the compressor and fan ports. in , the compressor efficiency η is obtained by calculating the power meter and the operation log model c , use the temperature and humidity integrated sensor to obtain the wet bulb temperature h w , the current cooling load ψ is obtained by load calculation l , forming a power consumption matching data group;

[0017] The data is compressed for redundancy and de-drifted to form a standardized input vector.

[0018] Preferably, the entropy conversion module includes a tolerance elimination unit, an index calculation engine unit and a multi-scale entropy value conversion unit;

[0019] The tolerance elimination unit is used to introduce dynamic sliding window filtering and median reconstruction strategies to identify input data with unreasonable fluctuations, including instrument noise and data jumps, and perform filtering and tolerance elimination.

[0020] The indicator calculation engine unit is used to mathematically reconstruct the dynamic heat flux imbalance data group, the flow control disturbance data group, and the power consumption matching data group, and extract the key control coefficients: heat flux imbalance factor TDI, flow control disturbance coefficient FDPC, and power consumption partial load matching coefficient PCLM;

[0021] The multi-scale entropy conversion unit is used to convert the original physical unit value into a dimensionless entropy value structure;

[0022] The heat flux imbalance factor TDI is calculated using the following formula:

[0023]

[0024] Where TDI represents the heat flux imbalance factor, which is used to describe the degree of heat transfer imbalance and thermal response deviation. g Indicates the return water temperature difference, s ht represents the heat transfer slope, ΔT represents the refrigerant temperature gradient rate, Represents unit heat flux density.

[0025] Preferably, the flow control disturbance coefficient FDPC is calculated by the following formula:

[0026]

[0027] Where FDPC represents the flow control disturbance coefficient, which is used to describe the difficulty of disturbance and response of water system, v f represents the flow velocity of the main channel of water flow, γ v represents the flow velocity variance, δ pr Indicates the rate of change of the inlet and outlet water pressure difference, λ t Indicates the dynamic throttling resistance coefficient.

[0028] Preferably, the power consumption partial load matching coefficient PCLM is calculated by the following formula:

[0029]

[0030] Where PCLM represents the power consumption partial load matching coefficient, which is used to measure the matching between power consumption and actual load, e in Indicates the real-time input power, η c Indicates the compressor efficiency, h w represents the wet bulb temperature, ψ l Indicates the current cooling load.

[0031] Preferably, the system energy state balance module includes a comprehensive modeling logic unit and a state level division unit;

[0032] The comprehensive modeling logic unit is used to construct the three coefficients into exponential form, and the following formula processor is integrated to calculate and obtain: control state index SEEI;

[0033]

[0034] Where SEEI represents the control state index, TDI represents the heat flux imbalance factor, FDPC represents the flow control disturbance coefficient, and PCLM represents the power consumption load matching coefficient.

[0035] The state level classification unit is used to obtain an evaluation strategy scheme by comparing the control state index SEEI with the preset standard state threshold Z and the preset standard state threshold X:

[0036] SEEI≤X indicates that the system is in the steady-state operation zone. The three indices of heat flux, flow state, and power consumption are all within the normal perturbation range, with no drastic fluctuations or over-regulation, and no intervention adjustment is required.

[0037] X < SEEI ≤ Z indicates that the system is in the offset warning area. The system enters a skewed state due to short-term heat surge, water flow disturbance, or load jump. The system shows an amplified response to small disturbances, or enters a skewed state due to short-term heat surge, water flow disturbance, or load jump. The speed of the frequency converter is reduced by 3 - 5 Hz to slow down the refrigeration rhythm, avoid high-amplitude excitation of the convection temperature difference in a short time, slightly reduce the flow rate of the main circuit, improve the hydraulic stability of the branch, and reduce the load jitter of the main heat exchanger.

[0038] SEEI > Z indicates that the system is in the energy consumption abnormal area, with phenomena such as serious mismatch of the cooling load, frequent start and stop of the compressor, and ineffective high-frequency operation of the water pump. The system is in a high-consumption state or there is a risk of equipment operation. It is recommended to immediately enter the "strong intervention mode". The frequency of the main compressor's frequency conversion is forced to drop by > 10 Hz, the air volume of the cooling tower fan drops suddenly, and the condensation temperature is restored to 2℃ below the set target value within 60 seconds. The water pump opens the branch bypass channel to relieve the peak value of the main circulation pressure difference.

[0039] Preferably, the adaptive strategy generation module includes a fast disturbance adaptation and strategy switching unit.

[0040] The fast disturbance adaptation and strategy switching unit is used to generate an optimal control strategy matrix CFM* based on the current SEEI value, equipment working conditions, and environmental disturbances, drive the underlying execution, and monitor the disturbance trigger factors in real time, including sudden changes in temperature and humidity, power consumption volatility, and large jumps in the cooling load. If a detected sudden disturbance is activated, it immediately enters the "fast sliding window strategy switching mechanism", enables a micro-strategy for data reconstruction of the short-term historical window, generates a temporary alternative control matrix within 200 milliseconds and overrides the current main strategy, and sends an interference mark to the edge execution response module in parallel to enter the temporary observation state.

[0041] Strategy matrix generation: ν r represents the speed of the main compressor frequency converter, α e represents the opening degree of the electronic expansion valve, φ l represents the cooling load distribution factor, ω f represents the speed of the cooling tower fan.

[0042] Preferably, the edge execution response module includes a response feedback closed-loop unit.

[0043] The response feedback closed-loop unit is used to receive the adjustment matrix output by the adaptive strategy generation module, analyze each control instruction and send it to the compressor, electronic expansion valve, water pump, and fan, collect and compare the real-time response status after execution, construct the equipment response error ε(t). If the continuous response deviation exceeds the preset threshold, it automatically feeds back to the adaptive strategy generation module for strategy correction, and supports the quick replacement of the current strategy, parameter rollback, or equipment fault marking, realizing the closed-loop control link of the execution of the control instruction.

[0044] Preferably, the dynamic evolution maintenance module includes a performance trend modeling unit, a residual performance evaluation unit and a maintenance plan output unit;

[0045] The performance trend modeling unit is used to construct a system performance degradation model and identify performance degradation trends using historical time series data of the heat flux imbalance factor TDI, the flow control disturbance coefficient FDPC, the power consumption load matching coefficient PCLM, and the control state index SEEI as input;

[0046] The residual performance evaluation unit is used to calculate the remaining energy efficiency cycle of the current system based on the trend model and output the residual performance index R(t), which reflects the degree of deviation of the current operating state from the optimal operating condition;

[0047] The maintenance plan output unit is used to predict the latest time window DL of equipment failure based on the R(t) decline rate and deviation threshold, output the recommended maintenance time node and the required maintenance parts list, and form an advance maintenance intervention strategy.

[0048] A method for energy-saving control of a chiller comprises the following steps:

[0049] Step 1: Synchronously collect dynamic heat flux imbalance data, flow control disturbance data, and power consumption matching data from the chiller system using a multi-channel, asynchronous sampling mechanism, and perform time series compression and redundancy removal.

[0050] Step 2: Input the three sets of collected data into the model, use the embedded perturbation calculation mechanism to calculate and generate the non-explicit control core coefficients, and output the three-dimensional coefficient space through entropy state change structure transformation to construct the state equilibrium map;

[0051] Step 3: Comprehensively calculate the control state index SEEI to evaluate the current energy state deviation of the system and decide whether to intervene;

[0052] Step 4: Based on the control state index SEEI value, rule search and nonlinear modeling are performed to generate a regulation matrix, and the micro-strategy is reconstructed using data from a short historical window.

[0053] Step 5: Receive the adjustment matrix and convert it into device-level instructions, execute the action through the edge controller and provide real-time feedback of the response error ε(t);

[0054] Step 6: Based on long-term system operation data, predict potential performance degradation trends and formulate proactive maintenance intervention plans.

[0055] The present invention provides an energy-saving control method and system for a chiller, which has the following beneficial effects:

[0056] (1) When the system is running, a multi-channel, asynchronous sampling mechanism is used to synchronously collect data groups from the chiller system, and time series compression and redundancy removal are performed. The embedded perturbation calculation mechanism is used to calculate and generate non-explicit control core coefficients. The three-dimensional coefficient space is output through entropy state change structure transformation to construct a state balance map. The control state index SEEI is comprehensively calculated to evaluate the current energy state deviation of the system and decide whether to intervene. Based on the control state index SEEI value, rule search and nonlinear modeling are performed to generate a regulation matrix. The data of the short-term historical window is enabled to reconstruct the micro-strategy. The regulation matrix is ​​received and converted into device-level instructions. The edge controller executes the action and feedbacks the response error ε(t) in real time. Based on the long-term system operation data, the potential performance degradation trend is predicted.

[0057] (2) The present invention realizes high-frequency dynamic monitoring and data modeling of the three dimensions of heat flux response, fluid disturbance and power consumption matching during the operation of the chiller through the high degree of coordination of the multi-source parameter acquisition module, entropy conversion module, system energy state balance module, adaptive strategy generation module, edge execution response module and dynamic evolution maintenance module, and constructs a complete 1 identification-strategy generation-execution feedback-maintenance evolution closed-loop control link. The system can accurately extract the non-explicit control core coefficients (TDI, FDPC, PCLM) under different operating conditions, and then comprehensively calculate the control state index SEEI, automatically determine whether the system is in a steady state, skewed state or abnormal state, and realize adaptive generation and rapid switching of control strategies. Finally, the system completes the energy-saving control task of the whole process from "parameter perception-state evaluation-strategy intervention-evolution maintenance", and improves the stability, response sensitivity and energy efficiency level of the overall chiller operation.

[0058] (3) Compared with the existing energy-saving control method based on static temperature difference and electric power control, the present invention breaks the traditional control mode of "single parameter drive + fixed strategy execution" and introduces multiple sets of heterogeneous physical parameter group modeling, multi-scale entropy value conversion mechanism and non-weighted index construction path for the first time, which can effectively capture the complex multi-source coupling effect within the system. By constructing three key control factors, TDI, FDPC and PCLM, a three-dimensional energy state characterization from energy flux-fluid disturbance-load response is achieved, which makes up for the technical shortcomings of traditional systems that cannot identify transient disturbances and cannot accurately intervene in load offsets. At the same time, the introduction of the state index SEEI can realize the hierarchical evaluation of the operating health status, avoid the control blind spots of "over-regulation" or "error delay", and make the strategy scheduling closer to the dynamic changes of actual operating load and environmental disturbances.

[0059] (4) Through the six-module closed-loop control mechanism of the present invention, not only the energy-saving control accuracy and strategy execution response speed of the chiller are significantly improved, but also the energy consumption fluctuation amplitude and equipment operation fatigue are effectively reduced. Specifically, the average response time is shortened by more than 30%, the energy efficiency ratio of the unit cooling load is improved by about 10-15%, the frequency of compressor start and stop is significantly reduced, the risk of condenser overload is reduced, and the vibration fluctuation of the heat exchanger is significantly suppressed. The system supports rapid switching of micro-window strategies for short-term sudden load changes and trend prediction maintenance for long-term operating status, which greatly improves the operating stability and life cycle utilization of the chiller. In summary, the system not only realizes the transformation and upgrading from traditional energy-saving control to "intelligent identification + dynamic adjustment + evolutionary optimization", but also provides a universal and adaptive energy-saving control technology path for the operation of chilled water systems in complex scenarios such as large buildings, data centers, and industrial cooling. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 This is a block diagram of an energy-saving control system for a chiller according to the present invention;

[0061] Figure 2 This is a schematic diagram of the steps of an energy-saving control method for a chiller according to the present invention;

[0062] Figure 3 This is a time-varying trend diagram of the control state index SEEI of an energy-saving control system for a chiller according to the present invention. DETAILED DESCRIPTION

[0063] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0064] Example 1

[0065] The present invention provides an energy-saving control system for a chiller. Figure 1 , including multi-source parameter acquisition module, entropy conversion module, system energy state balance module, adaptive strategy generation module, edge execution response module and dynamic evolution maintenance module;

[0066] The multi-source parameter acquisition module is used to synchronously collect dynamic heat flux imbalance data, flow control disturbance data, and power consumption matching data from the chiller system using a multi-channel, asynchronous sampling mechanism, and perform time series compression and redundancy removal.

[0067] The entropy conversion module is used to input the three sets of collected data into the model, use the embedded perturbation calculation mechanism to calculate and generate the non-explicit control core coefficients, and output the three-dimensional coefficient space through the entropy state change structure transformation to construct the state equilibrium map;

[0068] The system energy state balance module is used to comprehensively calculate the control state index SEEI, evaluate the current energy state deviation of the system, and decide whether to intervene;

[0069] The adaptive strategy generation module is used to perform rule search and nonlinear modeling based on the control state index SEEI value, generate a regulation matrix, and enable micro-strategy reconstruction based on data from a short-term historical window;

[0070] The edge execution response module is used to receive the adjustment matrix and convert it into device-level instructions, execute actions through the edge controller and provide real-time feedback of the response error ε(t);

[0071] The dynamic evolution maintenance module is used to predict potential performance degradation trends based on long-term system operation data and implement early maintenance intervention plan formulation.

[0072] In this embodiment, a dynamic heat flux imbalance data set, a flow control disturbance data set, and a power consumption matching data set are synchronously collected from the chiller system using a multi-channel, asynchronous sampling mechanism, and time series compression and redundancy removal are performed. The three sets of collected data are input into the model, and a non-explicit control core coefficient is calculated using an embedded perturbation calculation mechanism. The three-dimensional coefficient space is output through entropy state change structure transformation for constructing a state balance map. The control state index SEEI is comprehensively calculated, the current energy state deviation of the system is evaluated, and a decision is made whether to intervene. Rule search and nonlinear modeling are performed based on the control state index SEEI value to generate a regulation matrix. The data reconstruction micro-strategy in the short-term historical window is enabled. The regulation matrix is ​​received and converted into device-level instructions. The action is executed by the edge controller and the response error ε(t) is fed back in real time. Based on long-term system operation data, potential performance degradation trends are predicted, and advance maintenance intervention plans are formulated.

[0073] Example 2

[0074] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the multi-source parameter acquisition module includes a heat flux sensing unit, a ,flow disturbance acquisition unit and a work load sensing unit;

[0075] The heat flux sensing unit is used to place a bidirectional thermocouple probe pair at the inlet and outlet of the return pipe to collect the temperature difference between the outlet and return water: the outlet and return water temperature difference t g , use a micro-pitch thermocouple array at both ends of the internal channel of the refrigerant evaporator or condenser to collect the slope of the change in the heat transfer rate of the main heat exchanger: heat transfer slope s ht, collect the refrigerant channel temperature gradient change through the heat flux diaphragm sensor: refrigerant temperature gradient rate ΔT, and collect the heat flux density per unit time through the RT temperature probe: unit heat flux density forming a dynamic heat flux imbalance data set;

[0076] The flow disturbance acquisition unit is used to collect the flow velocity v of the main channel of the water flow by deploying an ultrasonic flow meter in the main pipeline section of the water system. f and velocity variance γ v , by setting pressure ports in the outlet and return sections and placing differential pressure sensors to collect the rate of change of the inlet and outlet pressure difference δ pr , the dynamic throttling resistance coefficient λ is collected through the electronic valve encoder and torque detector installed in the bypass valve control section t , forming a flow control disturbance data group;

[0077] The load estimation unit is used to collect input power, current, and voltage, and predict load demand in combination with the system operation log. The real-time input power e is obtained through the energy meters deployed at the compressor and fan ports. in , the compressor efficiency η is obtained by calculating the power meter and the operation log model c , use the temperature and humidity integrated sensor to obtain the wet bulb temperature h w , the current cooling load ψ is obtained by load calculation l , forming a power consumption matching data group;

[0078] The data is compressed for redundancy and de-drifted to form a standardized input vector.

[0079] The entropy conversion module includes a tolerance elimination unit, an indicator calculation engine unit and a multi-scale entropy value conversion unit;

[0080] The tolerance elimination unit is used to introduce dynamic sliding window filtering and median reconstruction strategies to identify input data with unreasonable fluctuations, including instrument noise and data jumps, and perform filtering and tolerance elimination.

[0081] The indicator calculation engine unit is used to mathematically reconstruct the dynamic heat flux imbalance data group, the flow control disturbance data group, and the power consumption matching data group, and extract the key control coefficients: heat flux imbalance factor TDI, flow control disturbance coefficient FDPC, and power consumption partial load matching coefficient PCLM;

[0082] The multi-scale entropy value conversion unit is used to convert the original physical unit value into a dimensionless entropy value structure.

[0083] In this embodiment, by setting up a heat flux sensing unit, a flow disturbance acquisition unit and a power load sensing unit, the multi-source parameter acquisition module can synchronously obtain the key physical information of the chiller in three dimensions of heat flux, fluid disturbance and power load at the millisecond level, and significantly reduce redundancy and noise with the help of time series compression and de-drift algorithm; then, the entropy conversion module relies on the dynamic sliding window filtering and median reconstruction mechanism of the tolerance rejection unit to further eliminate abnormal measurement points and instrument jump errors, ensuring high confidence in the input data; on this basis, the indicator calculation engine unit extracts the three non-explicit control coefficients of TDI, FDPC and PCLM in real time, and the multi-scale entropy value conversion unit then converts the coefficients of different dimensions into a unified entropy state expression, which not only realizes the fine quantification of complex coupled working conditions, but also provides a highly sensitive and comparable input benchmark for subsequent energy state evaluation and strategy generation. In summary, the synergistic effect of the above units increases the system's sensitivity to tiny thermal lags, flow rate disturbances, and load mismatches by approximately 30%, and reduces the data noise misjudgment rate by more than 40%, laying a highly reliable and high-precision data and model foundation for achieving millisecond-level adaptive energy-saving regulation.

[0084] Example 3

[0085] This embodiment is explained in Example 1, please refer to Figure 1 Specifically: The heat flux imbalance factor TDI is calculated using the following formula:

[0086]

[0087] Where TDI represents the heat flux imbalance factor, which is used to describe the degree of heat transfer imbalance and thermal response deviation. g Indicates the return water temperature difference, s ht represents the heat transfer slope, ΔT represents the refrigerant temperature gradient rate, Represents unit heat flux density.

[0088] The flow control disturbance coefficient FDPC is calculated using the following formula:

[0089]

[0090] Where FDPC represents the flow control disturbance coefficient, which is used to describe the difficulty of disturbance and response of water system, v f represents the flow velocity of the main channel of water flow, γ v represents the flow velocity variance, δ pr Indicates the rate of change of the inlet and outlet water pressure difference, λ t Indicates the dynamic throttling resistance coefficient.

[0091] The power consumption partial load matching coefficient PCLM is calculated using the following formula:

[0092]

[0093] In the formula, PCLM represents the power consumption partial load matching coefficient, which is used to measure the matching of power consumption and actual load, and e in represents the real-time input power, and η c represents the compressor efficiency, and h w represents the wet bulb temperature, and ψ l represents the current refrigeration load.

[0094] The system energy state balance module includes an integrated modeling logic unit and a state level division unit;

[0095] The integrated modeling logic unit is used to perform exponential construction on the three coefficients and obtain the following formula processor calculation: control state index SEEI; [[ID=IS]]

[0096]

[0097] In the formula, SEEI represents the control state index, TDI represents the heat transfer imbalance factor, FDPC represents the flow control disturbance coefficient, and PCLM represents the power consumption partial load matching coefficient;

[0098] The state level division unit is used to obtain an evaluation strategy plan by comparing the control state index SEEI with the preset standard state threshold Z and the preset standard state threshold X:

[0099] SEEI≤X indicates that the system is in the steady-state operation area, and the three types of indexes of heat transfer, flow state, and power consumption are all within the normal perturbation range, without violent fluctuations and overshoot phenomena, and no intervention and adjustment are required; <00,00270>

[0100] X<SEEI≤Z indicates that the system is in the offset warning area. The system enters a skew state due to short-term heat surge, water flow disturbance, or load jump. The system shows an amplified response to small disturbances, or enters a skew state due to short-term heat surge, water flow disturbance, or load jump. The inverter speed is lowered by 3-5Hz to slow down the refrigeration rhythm, avoid high-amplitude excitation of the convective temperature difference in a short time, slightly lower the main circuit flow rate, improve the branch hydraulic stability, and reduce the load jitter of the main heat exchanger;

[0101] SEEI>Z indicates that the system is in the energy consumption abnormal area, with serious cold load mismatch, frequent start and stop of the compressor, and ineffective high-frequency operation of the water pump. The system is in a high-consumption state or there is a risk of equipment operation. It is recommended to immediately enter the "strong intervention mode". The main compressor frequency conversion frequency is forced to drop by >10Hz, the cooling tower fan air volume drops suddenly, and the condensation temperature is restored below the set target value by 2℃ within 60 seconds. The water pump opens the branch bypass channel to relieve the peak value of the main circulation pressure difference.

[0102] In this embodiment, the present invention constructs three key control factors: the heat flux imbalance factor (TDI), the flow control perturbation coefficient (FDPC), and the power consumption load matching coefficient (PCLM). These factors accurately characterize the chiller's heat exchange balance, the difficulty of responding to water system disturbances, and the matching between energy consumption and load. Furthermore, the control state index (SEEI) implemented in the system energy state balance module is used to unweightedly integrate these three factors, achieving a fully quantitative expression of the energy state. This approach effectively overcomes the limitations of traditional energy-saving strategies that rely solely on temperature difference or a single energy efficiency ratio (e.g., COP) as a control benchmark, achieving higher dynamic resolution and physical consistency. Furthermore, coupled with a state classification mechanism, the system can implement a three-level refined assessment of the chiller's operating state (stable state zone, deviation warning zone, and energy consumption abnormality zone), automatically linking it to the subsequent policy generation logic. Practical operation has shown that this method can improve policy triggering accuracy by approximately 35% and shorten intervention response lag time by over 50%. While reducing the frequency of system overshoot, it effectively suppresses the energy consumption amplification effects of thermal shock and flow pressure fluctuations, providing strong support for high-efficiency and stable operation.

[0103] Example 4

[0104] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the adaptive strategy generation module includes a fast disturbance adaptation and ,strategy switching unit;

[0105] The fast disturbance adaptation and strategy switching unit is used to generate the optimal control strategy matrix CFM* based on the current SEEI value, equipment operating conditions, and environmental disturbances, drive the underlying execution, and monitor disturbance trigger factors in real time, including sudden changes in temperature and humidity, power consumption fluctuations, and large jumps in cooling load. If a sudden disturbance is detected, it immediately enters the "fast sliding window strategy switching mechanism", activates the data reconstruction micro-strategy from the short-term historical window, generates a temporary replacement control matrix within 200 milliseconds, and overwrites the current main strategy, and simultaneously sends the interference mark to the edge execution response module, entering the temporary observation state;

[0106] Strategy matrix generation: ν r Indicates the main compressor inverter speed, α e Indicates the opening of the electronic expansion valve, φ l represents the cooling load distribution factor, ω f Indicates the cooling tower fan speed.

[0107] The edge execution response module includes a response feedback closed-loop unit;

[0108] The response feedback closed-loop unit is used to receive the adjustment matrix output by the adaptive strategy generation module, parse various control instructions and send them to the compressor, electronic expansion valve, water pump and fan, collect and compare the real-time response status after execution, and construct the equipment response error ε(t). If the continuous response deviation exceeds the preset threshold, it is automatically fed back to the adaptive strategy generation module for strategy correction. It also supports rapid replacement of the current strategy, parameter rollback or equipment fault marking, realizing a closed-loop control link for the execution of control instructions.

[0109] The dynamic evolution maintenance module includes a performance trend modeling unit, a residual performance evaluation unit, and a maintenance plan output unit;

[0110] The performance trend modeling unit is used to construct a system performance degradation model and identify performance degradation trends using historical time series data of the heat flux imbalance factor TDI, the flow control disturbance coefficient FDPC, the power consumption load matching coefficient PCLM, and the control state index SEEI as input;

[0111] The residual performance evaluation unit is used to calculate the remaining energy efficiency cycle of the current system based on the trend model and output the residual performance index R(t), which reflects the degree of deviation of the current operating state from the optimal operating condition;

[0112] The maintenance plan output unit is used to predict the latest time window DL of equipment failure based on the R(t) decline rate and deviation threshold, output the recommended maintenance time node and the required maintenance parts list, and form an advance maintenance intervention strategy.

[0113] In this embodiment, by setting up the three major intelligent control links of "adaptive strategy generation module", "edge execution response module" and "dynamic evolution maintenance module", the present invention realizes a full-chain closed-loop response mechanism from disturbance perception to equipment action. When a sudden disturbance occurs, the system can quickly activate the "fast sliding window strategy switching mechanism" based on the control state index SEEI, equipment operating conditions and environmental parameters, generate a temporary control matrix CFM* within 200 milliseconds, and trigger the edge controller to execute, to ensure that the refrigeration system is not disrupted by severe disturbances. The response feedback closed-loop mechanism monitors the actual execution effect of the issued control instructions (ε(t)). If the continuous deviation exceeds the limit, it will be immediately corrected in the opposite direction to avoid the cumulative misleading of the control offset. This two-way closed-loop control mode makes the strategy switching and equipment response more accurate and robust, avoiding the high-energy consumption fluctuation chain of "overshoot-hysteresis-rebound" in traditional systems. Through modeling and analysis of long-term operating data, this system utilizes historical TDI, FDPC, PCLM, and SEEI values ​​to establish performance degradation trends, output the residual performance index R(t) in real time, dynamically assess the system's current deviation from optimal operating conditions, and predict the latest maintenance window DL. This not only prevents equipment from continuing to operate at high energy consumption during a hidden degradation process, but also significantly reduces the risk of system downtime and maintenance due to sudden failures. Compared to traditional scheduled maintenance or alarm-triggered strategies, this invention implements a truly predictive maintenance logic based on energy efficiency deviation as the core indicator, extending equipment life and reducing maintenance costs while ensuring energy conservation. The synergistic effect of these three modules improves the system's overall energy efficiency by 12–18% and the response error convergence rate by over 35%, demonstrating its strong practicality and promotional value.

[0114] Example 5

[0115] A method for energy-saving control of chillers, please refer to Figure 2 , specifically: including the following steps:

[0116] Step 1: Synchronously collect dynamic heat flux imbalance data, flow control disturbance data, and power consumption matching data from the chiller system using a multi-channel, asynchronous sampling mechanism, and perform time series compression and redundancy removal.

[0117] Step 2: Input the three sets of collected data into the model, use the embedded perturbation calculation mechanism to calculate and generate the non-explicit control core coefficients, and output the three-dimensional coefficient space through entropy state change structure transformation to construct the state equilibrium map;

[0118] Step 3: Comprehensively calculate the control state index SEEI to evaluate the current energy state deviation of the system and decide whether to intervene;

[0119] Step 4: Based on the control state index SEEI value, rule search and nonlinear modeling are performed to generate a regulation matrix, and the micro-strategy is reconstructed using data from a short historical window.

[0120] Step 5: Receive the adjustment matrix and convert it into device-level instructions, execute the action through the edge controller and provide real-time feedback of the response error ε(t);

[0121] Step 6: Based on long-term system operation data, predict potential performance degradation trends and formulate proactive maintenance intervention plans.

[0122] In this embodiment, this method achieves dynamic perception, precise intervention, and proactive maintenance of the chiller system's energy consumption state by constructing a six-step, continuous, closed-loop process. This effectively addresses key issues in traditional energy-saving control, such as delayed response, blind adjustments, and insufficient fault prediction. In step one, a multi-channel asynchronous sampling mechanism ensures timely and high-dimensional coverage of heat flux, flow control, and load data. In step two, the three-dimensional core coefficients (TDI, FDPC, and PCLM) are effectively extracted and entropy-converted, giving the energy state characteristics greater structural stability and predictive power, laying a precise foundation for state modeling. In steps three and four, the construction of the control state index (SEEI) achieves a logical leap from "physical quantity perception" to "state quantity judgment." Instead of relying on manual experience or set thresholds, it determines whether to adjust energy consumption through a dynamic fusion of multiple factors. Combined with the generation mechanism of the strategy matrix (CFM*), this method enables adaptive generation and rapid switching of optimal intervention paths. This method is particularly robust when dealing with short-cycle disturbances (such as sudden load jumps and instantaneous water flow instability), enabling the system to complete strategy replacement within milliseconds, maximizing energy efficiency and operational stability. Finally, in the response execution and trend maintenance links of steps five and six, the system not only achieved closed-loop execution of device-level control instructions and feedback error correction, but also established an R(t)-DL dynamic prediction mechanism through historical data evolution modeling, transforming maintenance work from "responsive" to "preventive" and identifying risk points of energy efficiency decline or potential failures in advance. The method's systematic design approach and full-chain optimization path achieve an overall energy saving of approximately 15%, extend the equipment life by 20%, and have a high degree of engineering deployability, providing a new paradigm for the next generation of intelligent chiller energy-saving systems.

[0123] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An energy-saving control system for a chiller, characterized by: It includes multi-source parameter acquisition module, entropy conversion module, system energy state balance module, adaptive strategy generation module, edge execution response module and dynamic evolution maintenance module; The multi-source parameter acquisition module is used to synchronously collect dynamic heat flux imbalance data, flow control disturbance data, and power consumption matching data from the chiller system using a multi-channel, asynchronous sampling mechanism, and perform time series compression and redundancy removal. The entropy conversion module is used to input the three sets of collected data into the model, use the embedded perturbation calculation mechanism to calculate and generate the non-explicit control core coefficients, and output the three-dimensional coefficient space through the entropy state change structure transformation to construct the state equilibrium map; The system energy state balance module is used to comprehensively calculate the control state index SEEI, evaluate the current energy state deviation of the system, and decide whether to intervene; The adaptive strategy generation module is used to perform rule search and nonlinear modeling based on the control state index SEEI value, generate a regulation matrix, and enable micro-strategy reconstruction based on data from a short-term historical window; The edge execution response module is used to receive the adjustment matrix and convert it into device-level instructions, execute actions through the edge controller and provide real-time feedback of the response error ε(t); The dynamic evolution maintenance module is used to predict potential performance degradation trends based on long-term system operation data and implement early maintenance intervention plan formulation.

2. The energy-saving control system for a chiller according to claim 1, characterized in that: The multi-source parameter acquisition module includes a heat flux sensing unit, a flow disturbance acquisition unit, and a power load sensing unit; The heat flux sensing unit is used to place a bidirectional thermocouple probe pair at the inlet and outlet of the return pipe to collect the temperature difference between the outlet and return water: the outlet and return water temperature difference t g , use a micro-pitch thermocouple array at both ends of the internal channel of the refrigerant evaporator or condenser to collect the slope of the change in the heat transfer rate of the main heat exchanger: heat transfer slope s ht , collect the temperature gradient change of the refrigerant channel through the heat flux diaphragm sensor: the refrigerant temperature gradient rate ΔT, and collect the heat flux density per unit time through the RT temperature probe: unit heat flux density φ h , forming a dynamic heat flux imbalance data set; The flow disturbance acquisition unit is used to collect the flow velocity v of the main channel of the water flow by deploying an ultrasonic flow meter in the main pipeline section of the water system. f and velocity variance γ v , by setting pressure ports in the outlet and return sections and placing differential pressure sensors to collect the rate of change of the inlet and outlet pressure difference δ pr The dynamic throttling resistance coefficient λ is collected by the electronic valve encoder and torque detector installed in the bypass valve control section. t , forming a flow control disturbance data group; The load estimation unit is used to collect input power, current, and voltage, and predict load demand in combination with the system operation log. The real-time input power e is obtained through the energy meters deployed at the compressor and fan ports. in , the compressor efficiency η is obtained by calculating the power meter and the operation log model c , use the temperature and humidity integrated sensor to obtain the wet bulb temperature h w , the current cooling load ψ is obtained by load calculation l , forming a power consumption matching data group; The data is compressed for redundancy and de-drifted to form a standardized input vector.

3. The energy-saving control system for a chiller according to claim 1, characterized in that: The entropy conversion module includes a tolerance elimination unit, an indicator calculation engine unit and a multi-scale entropy value conversion unit; The tolerance elimination unit is used to introduce dynamic sliding window filtering and median reconstruction strategies to identify input data with unreasonable fluctuations, including instrument noise and data jumps, and perform filtering and tolerance elimination. The indicator calculation engine unit is used to mathematically reconstruct the dynamic heat flux imbalance data group, the flow control disturbance data group, and the power consumption matching data group, and extract the key control coefficients: heat flux imbalance factor TDI, flow control disturbance coefficient FDPC, and power consumption partial load matching coefficient PCLM; The multi-scale entropy conversion unit is used to convert the original physical unit value into a dimensionless entropy value structure; The heat flux imbalance factor TDI is calculated using the following formula: Where TDI represents the heat flux imbalance factor, which is used to describe the degree of heat transfer imbalance and thermal response deviation. g Indicates the return water temperature difference, s ht represents the heat transfer slope, ΔT represents the refrigerant temperature gradient rate, Represents unit heat flux density.

4. The energy-saving control system for a chiller according to claim 3, characterized in that: The flow control disturbance coefficient FDPC is calculated using the following formula: Where FDPC represents the flow control disturbance coefficient, which is used to describe the difficulty of disturbance and response of water system, v f represents the flow velocity of the main channel of water flow, γ v represents the flow velocity variance, δ pr Indicates the rate of change of the inlet and outlet water pressure difference, λ t Indicates the dynamic throttling resistance coefficient.

5. The energy-saving control system for a chiller according to claim 3, characterized in that: The power consumption partial load matching coefficient PCLM is calculated using the following formula: Where PCLM represents the power consumption partial load matching coefficient, which is used to measure the matching between power consumption and actual load, e in Indicates the real-time input power, η c Indicates the compressor efficiency, h w represents the wet bulb temperature, ψ l Indicates the current cooling load.

6. The energy-saving control system for a chiller according to claim 3, characterized in that: The system energy state balance module includes a comprehensive modeling logic unit and a state level division unit; The comprehensive modeling logic unit is used to construct the three coefficients into exponential form, and the following formula processor is integrated to calculate and obtain: control state index SEEI; Where SEEI represents the control state index, TDI represents the heat flux imbalance factor, FDPC represents the flow control disturbance coefficient, and PCLM represents the power consumption load matching coefficient. The state level classification unit is used to obtain an evaluation strategy scheme by comparing the control state index SEEI with the preset standard state threshold Z and the preset standard state threshold X: SEEI≤X indicates that the system is in the steady-state operation zone. The three indices of heat flux, flow state, and power consumption are all within the normal perturbation range, with no drastic fluctuations or over-regulation, and no intervention adjustment is required. X < SEEI ≤ Z indicates that the system is in the offset warning area. The system enters a skewed state due to short-term sudden heat increase, water flow disturbance, or load jump. The system shows an amplified response to small disturbances, or enters a skewed state due to short-term sudden heat increase, water flow disturbance, or load jump. The speed of the frequency converter is lowered by 3 - 5 Hz to slow down the refrigeration rhythm, avoid high-amplitude excitation of the convection temperature difference in a short time, slightly lower the flow rate of the main circuit, improve the hydraulic stability of the branch, and reduce the load jitter of the main heat exchanger. SEEI > Z indicates that the system is in the energy consumption abnormal area, with serious cold load mismatch, frequent start and stop of the compressor, and ineffective high-frequency operation of the water pump. The system is in a high-consumption state or there is a risk of equipment operation. It is recommended to immediately enter the "strong intervention mode". The frequency of the main compressor is forced to drop by > 10 Hz, the air volume of the cooling tower fan drops suddenly, and the condensation temperature is restored to 2℃ below the set target value within 60 seconds. The water pump opens the branch bypass channel to relieve the peak value of the main circulation pressure difference.

7. The energy-saving control system for a chiller according to claim 1, characterized in that: The adaptive strategy generation module includes a fast disturbance adaptation and strategy switching unit. The fast disturbance adaptation and strategy switching unit is used to generate the optimal control strategy matrix CFM* according to the current SEEI value, equipment working conditions, and environmental disturbances, drive the underlying execution, and monitor the disturbance trigger factors in real time, including sudden changes in temperature and humidity, power consumption volatility, and large jumps in cold load. If a detected sudden disturbance is activated, it immediately enters the "fast sliding window strategy switching mechanism", enables the micro-strategy of data reconstruction in a short-term historical window, generates a temporary alternative control matrix within 200 milliseconds and overwrites the current main strategy, and sends an interference mark to the edge execution response module in parallel to enter the temporary observation state. Strategy matrix generation: ν r Indicates the main compressor inverter speed, α e Indicates the opening of the electronic expansion valve, φ l represents the cooling load distribution factor, ω f Indicates the cooling tower fan speed.

8. The energy-saving control system for a chiller according to claim 1, characterized in that: The edge execution response module includes a response feedback closed-loop unit. The response feedback closed-loop unit is used to receive the adjustment matrix output by the adaptive strategy generation module, analyze each control instruction and send it to the compressor, electronic expansion valve, water pump, and fan, collect and compare the real-time response status after execution, construct the equipment response error ε(t). If the continuous response deviation exceeds the preset threshold, it automatically feeds back to the adaptive strategy generation module for strategy correction, and supports the quick replacement of the current strategy, parameter rollback, or equipment failure marking, realizing the closed-loop control link of the execution of the control instruction.

9. The energy-saving control system for a chiller according to claim 1, characterized in that: The dynamic evolution and maintenance module includes a performance trend modeling unit, a remaining performance evaluation unit, and a maintenance plan output unit. The performance trend modeling unit is used to construct a system performance degradation model with the historical time-series data of the heat transfer imbalance factor TDI, the flow control disturbance coefficient FDPC, the power consumption partial load matching coefficient PCLM, and the control state index SEEI as inputs, and identify the performance decline trend. The remaining performance evaluation unit is used to calculate the remaining energy efficiency cycle of the current system based on the trend model, and output the remaining performance index R(t), which reflects the deviation degree of the current operating state from the optimal operating condition. The maintenance plan output unit is used to predict the latest time window DL of equipment failure according to the decline rate of R(t) and the deviation threshold, output the recommended maintenance time node and the list of required maintenance parts, and form an early maintenance intervention strategy.

10. The energy-saving control method for a chiller according to claim 1, applied to the energy-saving control system for a chiller according to any one of claims 1 to 9, characterized in that: It includes the following steps: Step 1: Synchronously collect dynamic heat flux imbalance data, flow control disturbance data, and power consumption matching data from the chiller system using a multi-channel, asynchronous sampling mechanism, and perform time series compression and redundancy removal. Step 2: Input the three sets of collected data into the model, use the embedded perturbation calculation mechanism to calculate and generate the non-explicit control core coefficients, and output the three-dimensional coefficient space through entropy state change structure transformation to construct the state equilibrium map; Step 3: Comprehensively calculate the control state index SEEI to evaluate the current energy state deviation of the system and decide whether to intervene; Step 4: Based on the control state index SEEI value, rule search and nonlinear modeling are performed to generate a regulation matrix, and the micro-strategy is reconstructed using data from a short historical window. Step 5: Receive the adjustment matrix and convert it into device-level instructions, execute the action through the edge controller and provide real-time feedback of the response error ε(t); Step 6: Based on long-term system operation data, predict potential performance degradation trends and formulate proactive maintenance intervention plans.

Images