A dual-objective optimization control method for central air conditioning based on edge computing

Abstract

This invention discloses a dual-objective optimization control method for central air conditioning based on edge computing, belonging to the field of central air conditioning optimization control technology. The method includes the following steps: an edge control terminal acquires multi-source operating data and time-of-use electricity price data for the central air conditioning system, and establishes a thermal inertia model to characterize the thermal inertia of the building and pipe network. Within a preset time window before the electricity price peak, a set of candidate energy storage start-up times is generated. For each candidate energy storage start-up time, a corresponding energy storage strategy is set, including adjustments to the water supply temperature setpoint and water pump operating parameters. Based on the energy storage strategy and the thermal inertia model, the comfort trajectory of each control area from each candidate energy storage start-up time to the start of the electricity peak and the effective energy storage capacity that can be stored on the water supply side are predicted. This solves the problems in existing technologies where load transfer control based on medium capacity suffers from dynamic uncertainty in effective energy storage, leading to difficulty in determining the timing, and where prioritizing peak shaving while ignoring hard comfort constraints leads to overshooting or backfilling.

Description

Technical Field

[0001] This invention relates to the field of central air conditioning optimization control technology, and in particular to a dual-objective optimization control method for central air conditioning based on edge computing. Background Technology

[0002] In public buildings, industrial plants, and park-level building complexes, central air conditioning systems typically undertake the main task of supplying cooling and heating, and their operating energy consumption accounts for a significant portion of the building's total life-cycle energy consumption. With the widespread adoption of time-of-use pricing, demand-based pricing, and peak shaving and valley filling mechanisms, how to achieve load transfer and reduce electricity purchase costs during peak periods through more refined operating strategies without altering the existing system's main structure has become a crucial research and engineering direction in the HVAC control field. Simultaneously, edge computing is increasingly being applied in building automation and data center group control scenarios. Utilizing the proximity and real-time computing capabilities of the edge, it enables rapid closed-loop regulation under conditions of network latency, cloud dependence, and frequent on-site disturbances, gradually becoming one of the key technological paths for intelligent control of central air conditioning systems.

[0003] To achieve load transfer and peak shaving control, the most common approach in existing technologies is to introduce additional energy storage / cold storage devices or dedicated energy storage heat exchange structures to reduce the burden on the chiller during peak periods by "storing energy during low-price periods and releasing energy during high-price periods".

[0004] For example, Chinese invention patent CN104279667B discloses a phase change energy storage air conditioning system, the structure of which includes a cooling tower, a cooling water circulation pump group, a refrigeration unit, a terminal load and a chilled water circulation pump group. The refrigeration unit includes a condenser and an evaporator, and also includes a cold storage circulation system. The cold storage circulation system includes a cold storage tank, a variable frequency water pump and several electric valves. By controlling the opening and closing of several electric valves, it is possible to flexibly switch between various cold storage and cold release modes.

[0005] For example, Chinese invention patent CN108870598B discloses a split-type heat pipe energy storage air conditioning system, which includes a compressor, a four-way reversing valve, a gas-liquid separator, a finned heat exchanger, a shell-and-tube heat exchanger, an energy storage module, and an energy release module. The compressor and the gas-liquid separator are connected through a working fluid pipeline; the finned heat exchanger and the shell-and-tube heat exchanger are connected through a working fluid pipeline; an energy storage module is connected in parallel on one side of the shell-and-tube heat exchanger; and the energy release module is connected to both the shell-and-tube heat exchanger and the energy storage module.

[0006] The aforementioned solutions are effective in "enhancing peak shaving capacity through dedicated energy storage structures and achieving energy storage / release linkage through mode switching." However, their overall approach relies on adding new energy storage devices, dedicated heat exchange structures, and corresponding valve, pump, and pipeline modifications, often leading to increased equipment investment, server room space occupation, and increased complexity in system modification construction and operation and maintenance. In scenarios involving numerous existing building renovations or lightweight upgrades, owners and maintenance providers tend to leverage the inherent energy storage capacity of existing systems without adding (or with minimal addition) cold storage devices, achieving substantial load transfer benefits at a lower modification cost.

[0007] Specifically, the energy-carrying medium circulation system utilizes its own medium capacity as a transient energy buffer. The circulation network and terminal heat exchange components (which may include main / branch pipes, terminal coils or air conditioning unit heat exchangers, water tanks, and auxiliary water volumes) already contain a certain volume of energy-carrying medium (such as water or other heat exchange media). During periods of low price or low load, the system can adjust the operating parameters of the chiller / heat source unit and circulation pump to bring the medium to a target state more conducive to peak-load release (e.g., lowering the supply water temperature in cooling conditions and raising the supply water temperature in heating conditions). This allows the system to continuously provide the required cooling / heating to the terminals during peak periods, relying on the medium's temperature drop / rise process, until the medium's temperature approaches its limit before resuming main unit output. This achieves a certain degree of peak shaving and cost optimization. This approach requires less hardware modification and is more suitable for engineering implementation in existing buildings and complex terminal systems.

[0008] However, this application discovered, during the aforementioned process of medium capacity energy storage and load transfer, that the existing technology suffers from at least the following key technical defects, directly hindering its ability to achieve stable and reproducible optimization effects in real-world building scenarios. Firstly, because the energy-carrying medium continuously circulates within the system, while the system medium capacity is a fixed physical quantity, the effective energy storage capacity available for peak-segment reduction of the host output dynamically changes in real time with load, flow rate, and losses. On the one hand, continuous heat exchange at the terminal causes the medium state to continuously drift towards the return direction, and the contribution of the return medium to available energy storage diminishes over time. On the other hand, the coupling of load and flow rate causes changes in the medium's residence time and energy release rate in the pipeline network. Higher loads tend to result in stronger circulation, shorter residence times, and faster energy release, leading to a shorter peak-segment sustainable energy supply time; conversely, lower loads result in slower energy release and a longer duration. This makes it difficult to accurately set the pre-treatment / pre-start timing (i.e., the start-up moment to push the entire system medium to the target state before entering the peak segment), making it difficult to ensure that the system medium as a whole reaches the target state and maximizes energy storage utilization efficiency when the peak segment arrives.

[0009] Secondly, existing load transfer control systems generally set the timing and intensity of pre-processing based on cost / peak reduction, often neglecting comfort issues or treating comfort merely as a passive result of post-event correction. On the one hand, advancing or intensifying pre-processing in pursuit of peak reduction can cause comfort levels to exceed limits in some areas before the peak arrives, such as excessive cooling / heating, humidity imbalance, and condensation risks. This triggers reverse regulation or protection logic at the terminal, causing the system status to be reversed and effective energy storage to be ineffectively consumed, thus weakening the peak reduction capability. On the other hand, adopting a conservative advance amount or reducing the intensity of pre-processing to avoid comfort risks can result in the medium status not meeting standards or insufficient effective energy storage at the start of the peak. The main unit has to compensate for the output during high-price periods, making it difficult to achieve the expected cost optimization goals.

[0010] In existing technologies, load transfer control based on medium capacity has problems such as uncertainty in the dynamics of effective energy storage leading to difficulty in determining the timing, and neglecting the hard constraints of comfort due to peak shaving as the main focus, resulting in overshooting or backfilling. Summary of the Invention

[0011] To address the technical problems of existing technologies, such as the uncertainty of effective energy storage dynamics leading to difficulty in determining timing, and the neglect of hard comfort constraints due to peak shaving as the primary focus, resulting in overshooting or backfilling, this invention provides a dual-objective optimization control method for central air conditioning based on edge computing. The technical solution is as follows:

[0012] A dual-objective optimization control method for central air conditioning based on edge computing is provided, the method comprising:

[0013] S1. The edge control terminal acquires multi-source operation data and time-of-use electricity price data of the central air conditioning system, and establishes a thermal inertia model to characterize the thermal inertia of the building and pipeline network. A set of candidate times for energy storage startup is generated within a preset time window before the peak electricity price. A corresponding energy storage strategy is set for each candidate time for energy storage startup. The energy storage strategy includes the adjustment of the water supply temperature setpoint and the water pump operating parameters.

[0014] S2. Based on the energy storage strategy and thermal inertia model, predict the comfort trajectory of each control area from each energy storage start-up candidate moment to the start of the peak electricity demand and the effective energy storage capacity that can be stored on the water supply side.

[0015] S3. Based on the comfort trajectory and effective energy storage, apply dual constraints to the set of candidate energy storage start-up times to obtain a set of feasible start-up times, and determine the final energy storage start-up time from the set of feasible start-up times.

[0016] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0017] 1. This invention provides a dual-objective optimization control method for central air conditioning based on edge computing. This invention does not determine the start time before the peak based solely on the energy efficiency of the computer room or a single empirical time series. Instead, the edge control terminal integrates multiple operating status quantities such as the computer room layer, water system layer, indoor and outdoor environment, and time-of-use electricity price under the same time base, and constructs a thermal inertia model system of regional thermal inertia sub-model and water supply side network thermal inertia sub-model. The basic energy storage strategy (water supply temperature setpoint and water pump operating parameters) is simultaneously mapped to the terminal equivalent input (used to predict the regional temperature and humidity trajectory) and the segmented temperature status of the water supply side (used to predict the available energy storage during the peak). Compared to existing technologies that often only consider the power / temperature difference between supply and return water in the machine room or only make regional comfort predictions while ignoring the time lag of pipeline delivery, this invention incorporates end-point response and pipeline advancement into the same prediction framework. This allows the assessment of startup time to move beyond single-sided indicators and simultaneously answer two key engineering questions: whether the hard constraints of comfort will be triggered after startup, and whether the water supply side truly has low-temperature water available for peak shaving when peaks arrive. This significantly reduces failure modes caused by information fragmentation, such as comfort being met but peak shaving failing, or peak shaving being sufficient but comfort exceeding limits.

[0018] 2. This invention performs segmented parameterization on the water supply main and major branches along the water supply side flow path, and generates a set of segmented parameters containing geometric and heat transfer characteristics for each water supply pipe segment. This enables the water supply side network thermal inertia sub-model to express hysteresis, mixing, and friction-related heat exchange losses with the pipe segment as the smallest interpretable unit. Based on this, this invention not only calculates the theoretical energy increment relative to the normal water supply temperature baseline, but also introduces an energy storage retention coefficient to characterize the retention ratio of energy storage under cycling and load conditions, and outputs the water supply side coverage compliance rate for verifying low-temperature coverage during peak hours. Unlike existing technologies that only use total cooling capacity / total temperature difference to estimate energy storage, this invention forms a coupled closed loop of segmented temperature state prediction, retention coefficient correction, and coverage compliance rate verification: the theoretical energy increment ensures the calculability of the quantity, the retention coefficient solves the overestimation problem caused by cycling decay, and the coverage compliance rate addresses the structural defect of sufficient total energy but still high levels in distant branches. This avoids both artificially inflated energy dimensions and local mismatches in spatial dimensions, improving the certainty and feasibility of peak shaving effects.

[0019] 3. This invention upgrades the selection of candidate start-up times from empirical rules to a dual-condition constraint mechanism of full-process verification of comfort hard constraints and determination of energy demand satisfaction. It verifies the temperature and relative humidity of candidate start-up times at each moment within the prediction time domain at the edge end to form a set of feasible start-up times. Simultaneously, it clarifies the energy demand during peak periods based on load forecasting results and peak-shaving strategies, and performs energy satisfaction determination on the effective energy storage of each candidate start-up time. Compared to the common practice in existing technologies of treating comfort as a soft constraint / post-event correction or judging comfort only at a single point in time, this invention places comfort constraints in the form of hard constraints upfront, and provides structured output of reasons for exceeding limits, insufficient energy storage, and insufficient coverage, making scheduling decisions interpretable and diagnosable. Furthermore, when the feasible set is empty, this invention stipulates that the basic energy storage strategy should be adjusted first within the equipment safety boundary, and the forecasting and selection should be re-tested. If no solution is found, the peak-shaving target should be adjusted within the allowable range, and alarms and the closest feasible solution should be output. This ensures a bounded fallback path even under fluctuating actual operating conditions, avoiding the problem in existing technologies where manual guesswork or abandoning peak-shaving is the only option when no solution is found.

[0020] 4. Under the premise that the set of feasible start times already satisfies the hard constraints, this invention further proposes a point selection expansion mechanism that balances deterministic point selection and risk: On the one hand, under normal circumstances, the latest feasible start time principle is adopted to reduce the premature energy storage decay caused by starting too early; on the other hand, by quantifying the energy storage decay risk (characterized by both the duration of maintenance and the intensity of cyclic consumption) and the full storage risk (characterized by both the coverage compliance rate and the compliance status of key remote branches) of the candidate start times, a subset is first obtained by performing decay filtering, and then the final start time is determined by selecting the best option within the subset based on the full storage risk. Compared to existing technologies that rely solely on unidirectional rules such as the latest start or the earliest full storage, this invention incorporates both the physical attenuation law of continuous circulation on the water supply side and the coverage push law into a deterministic decision chain: attenuation filtration is used to suppress failures caused by early start but consumption before peak hours, while full storage selection is used to suppress failures caused by late start but insufficient coverage. The combination of the two makes the final selection point more robust to load and flow fluctuations, and can more accurately solve the engineering pain points of existing technologies, such as unstable energy storage availability in the circulation system, uncontrollable remote coverage, and the tendency for the start-up timing to fluctuate between the two types of failures. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1A flowchart of a dual-objective optimization control method for central air conditioning based on edge computing is provided for embodiments of this application;

[0023] Figure 2 A diagram of a central air conditioning room system provided for an embodiment of this application;

[0024] Figure 3 A schematic diagram of the chiller unit in the central air conditioning room system diagram provided in the embodiments of this application;

[0025] Figure 4 A schematic diagram of the chilled pump in the central air conditioning room system diagram provided in this application embodiment;

[0026] Figure 5 A schematic diagram of the cooling tower and cooling pump in the central air conditioning room system diagram provided in the embodiments of this application;

[0027] Figure 6 A schematic diagram of the heat recovery pump of the screw chiller unit in the central air conditioning room system diagram provided in this application embodiment. Detailed Implementation

[0028] Embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of the present disclosure are shown in the drawings, it should be understood that embodiments of the present disclosure may be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure.

[0029] It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure. In the description of the embodiments of this disclosure, the term "comprising" and similar terms should be understood as open-ended inclusion, i.e., "including but not limited to". The term "based on" should be understood as "at least partially based on". The term "one embodiment" or "this embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc., may refer to different or the same objects.

[0030] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0031] A specific embodiment of the present invention provides a dual-objective optimization control method for central air conditioning based on edge computing, such as... Figure 1 The diagram shown is a flowchart of a dual-objective optimization control method for central air conditioning based on edge computing, provided in an embodiment of this application. The method includes the following steps:

[0032] S1. The edge control terminal acquires multi-source operation data and time-of-use electricity price data of the central air conditioning system, and establishes a thermal inertia model to characterize the thermal inertia of the building and pipeline network. A set of candidate times for energy storage startup is generated within a preset time window before the peak electricity price. A corresponding energy storage strategy is set for each candidate time for energy storage startup. The energy storage strategy includes the adjustment of the water supply temperature setpoint and the water pump operating parameters.

[0033] The edge control terminal actively acquires data from the data center layer, water system layer, indoor data, outdoor environmental data, and time-of-use electricity price data. Data center layer data includes chiller start / stop, load rate, unit input power or meter power, supply and return water temperatures, chilled water pump frequency / flow rate, and key valve openings; water system layer data includes supply main or key branch temperature, key branch flow rate or differential pressure, and return water temperature; indoor data includes temperature, relative humidity, area setpoints, and occupancy status for each control area; outdoor environmental data includes outdoor temperature and humidity, solar radiation, and weather forecast characteristics; and time-of-use electricity price data includes off-peak, flat, and peak periods and their corresponding prices.

[0034] Data at the data center level is actively acquired by the edge control terminal through data interfaces with the data center group control system, unit controller, or field PLC; data at the water system level is actively acquired by the edge control terminal through measuring points and metering devices in the hydraulic loop; indoor data is actively acquired by the edge control terminal through the terminal controller network or indoor sensor network; outdoor environmental data is actively acquired by the edge control terminal through the local weather station or third-party weather services; and time-of-use electricity price data is actively acquired by the edge control terminal through the electricity price strategy library or external electricity price release interface.

[0035] In this invention, the edge control terminal is not only used to build the thermal inertia model, but also undertakes the full-link responsibilities of data aggregation, prediction and evaluation, constraint screening, strategy distribution, closed-loop verification, anomaly rollback, and interpretable trace retention. As an on-site computing and control unit deployed on the data center side or building side, it is used to complete the energy storage start-up decision and execution under the hard constraints of time-of-use electricity price and comfort without relying on real-time computing in the cloud.

[0036] The edge control terminal performs timestamp alignment, outlier removal (outside physical range, sudden jumps, sensor offline), and missing value imputation (linear interpolation or retaining the last valid value and marking the confidence level) on the above multi-source data at fixed time steps (e.g., 5 minutes). The processing result forms a unified state vector x(t0) for the current moment, where t0 is the current moment. The unified state vector x(t0) contains a set of basic state variables that characterize the current operating state, including at least: unit operating conditions, water system supply and return water temperatures and flow rates, temperature and humidity status of each control area, outdoor environmental status, and current electricity price segment identifiers. In cooling or heating mode, the signs and reference directions of the temperature state variables in the above state vector are adjusted accordingly, but the state vector structure remains consistent.

[0037] To ensure interpretability and feasibility, this embodiment employs a simplified thermal inertia model that can be updated online, including a regional thermal inertia sub-model and a water supply network thermal inertia sub-model. The regional and water supply network thermal inertia sub-models are computationally independent but work collaboratively in the prediction process: the former characterizes the thermal response of the end-region to changes in cooling conditions, while the latter characterizes the time lag and decay of water temperature changes on the supply side. The outputs of both serve as the basis for predictive evaluation of subsequent candidate energy storage schemes.

[0038] The regional thermal inertia sub-model is used to predict the dynamic response of regional temperature and humidity to changes in cooling conditions. Its inputs include: outdoor temperature and humidity, regional occupancy or internal load, and terminal cooling intensity (which can be obtained by mapping from supply water temperature, flow rate, or valve position). Model parameters (such as equivalent heat capacity, thermal resistance, and time delay) can be obtained at the edge using parameter identification or regression updates based on historical operating data, such as recursive least squares or moving regression, and other online update methods with equivalent effects are also allowed. The output of this sub-model is the temperature and relative humidity trajectory of each region over a future period, as well as conclusions derived directly from these trajectories, such as the remaining safety margin from the comfort boundary, whether the boundary is exceeded at any time, etc., for subsequent hard constraint verification.

[0039] The thermal inertia sub-model of the water supply network is used to describe the hysteresis, mixing, and frictional heat loss of chilled water pushed from the generator room to each pipe section in the water supply network. Its key interpretable parameters include: the volume V of each pipe section. iThe model's parameters are: 1) temperature distribution at the start of peak flow; 2) temperature distribution at the start of peak flow; 3) temperature distribution at the start of peak flow; 4) temperature distribution at the start of peak flow; 5) temperature distribution at the start of peak flow; 6) temperature distribution at the start of peak flow; 7) temperature distribution at the start of peak flow; 8) temperature distribution at the start of peak flow; 9) temperature distribution at the start of peak flow; 10) temperature distribution at the start of peak flow; 11) temperature distribution at the start of peak flow; 12) temperature distribution at the start of peak flow; 13) temperature distribution at the start of peak flow; 14) temperature distribution at the start of peak flow; 15) temperature distribution at the start of peak flow; 16) temperature distribution at the start of peak flow; 17) temperature distribution at the start of peak flow; 18) temperature distribution at the start of peak flow; 19) temperature distribution at the start of peak flow; 10 ...9) temperature distribution at the start of peak flow; 12) temperature distribution at the start of peak flow; 19) temperature distribution at the start of peak flow; 19) temperature distribution at the start of peak flow; 19) temperature distribution at the start of peak flow; 19) temperature distribution at the start of peak flow; 19) temperature distribution at the start of peak flow; 19) temperature distribution at the start of peak flow; 19) temperature

[0040] Through the two types of sub-models mentioned above, the thermal inertia model can not only explain why the same lead time has different effects under different loads / flow rates, but also provide a unified calculation framework for batch prediction of subsequent candidate schemes.

[0041] When establishing the thermal inertia model, the edge control terminal further segments the chilled water supply network. Specifically, it selects the main supply pipes and major branches as the segmentation objects, with major branches including at least those with a high proportion of remote load or complete temperature / flow monitoring points. Along the water flow direction, the network is divided into N supply pipe segments based on pipe diameter change points, tee branch points, valve / balancing valve nodes, temperature sensor locations, and pipe length thresholds (e.g., every 30-80 meters), each segment denoted as i (i=1, 2, ..., N, where N is the total number of supply pipe segments). During segmentation, the temperature sensor locations are prioritized as segment boundaries; if a temperature sensor exists on a branch, the segment is prioritized between two adjacent temperature measurement points; if a segment has no temperature measurement points, it is divided by pipe diameter and length, and the temperature of adjacent measurement points is used to calculate the heat loss using a heat loss model.

[0042] For each segment i, the edge control terminal generates a segmentation parameter table, which includes at least: segment length L. i Pipe diameter D i Segment volume V i Section insulation level, and hydraulic path markings from section to the far end. V i The parameters can be calculated from the design drawings (calculated based on the geometric volume of the pipeline and considering the equivalent correction of accessories); the above parameters are used together to characterize the temperature response speed and attenuation characteristics of different pipe sections under different flow rates and operating conditions, and serve as the basis for subsequent calculations to predict the energy storage capacity of each pipe section when the peak electricity price arrives.

[0043] In this example, the chilled water system serves an office building, with the supply side consisting of a main pipe and three main branches. The edge control terminal divides the supply side into N=5 supply pipe segments along the chilled water flow direction, based on branch nodes, pipe diameter change points, and temperature measurement point locations. The volumetric parameters of each supply pipe segment can be taken as follows, for example: Segment 1 (from the machine room to the first branch): V1 = 3.0m 3 Section 2 (Middle section of the main trunk): V2 = 2.5m 3 Section 3 (Branch A): V3 = 1.2m 3 Segment 4 (Branch B, far end): V4 = 1.5m 3 Section 5 (Branch C): V5 = 0.8m 3 The density ρ of water can be taken as 1000 kg / m³. 3 Specific heat of water c w The value can be taken as 4.2 kJ / (kg·℃).

[0044] If pipe segment i is equipped with a temperature sensor, the segment temperature is read directly; if there is no temperature sensor, it is calculated from the temperature of the nearest upstream measuring point combined with the friction loss model. The segment flow rate is preferentially obtained from the branch flow meter; if there is no flow meter, it can be estimated from the pump frequency, valve opening, pressure difference, and historical hydraulic calibration relationship. The edge control terminal solidifies the above segmentation results into a water supply side segmentation parameter table, which serves as the basic input for subsequent segment-by-segment temperature advancement prediction, effective energy storage calculation, and water supply side coverage compliance rate calculation in S2. Among them, the water supply side coverage compliance rate is used to characterize the proportion of water bodies or pipe segments in the water supply side network that meet the target energy storage temperature requirements at the beginning of the peak electricity price period, and can be obtained by weighting the pipe segment volume or by statistically analyzing the coverage of key branches. Segment-by-segment temperature advancement prediction is based on a time stepping method. In each time step, the advancement distance and lag of the water body between pipe segments are estimated based on the current flow rate, thereby updating the temperature status of each pipe segment.

[0045] Within each pipe segment, the edge control terminal uses a thermal inertia model to calculate the temperature change and energy storage capacity of that segment. For example, for pipe segment i, if energy storage is initiated at time t, the system will calculate the temperature change of that segment during peak hours based on the current temperature and flow rate. By tracking the temperature change of each pipe segment, the system can obtain the energy storage capacity of each segment during peak electricity demand.

[0046] The control method of this invention is described primarily using a cooling mode as an example, but it is equally applicable to a heating mode. Depending on the system's operating mode (cooling or heating), the corresponding control strategy, state variables (such as supply water temperature and flow rate), and thermal inertia model parameters will be adjusted. In heating mode, the supply water temperature is higher, and the thermal inertia model and energy storage strategy will be optimized according to different operating environments to ensure that the system can fully utilize low-electricity-price periods for heat storage or peak shaving without affecting comfort. In heating mode, the supply water temperature reference, the heat loss direction in the thermal inertia model, and the energy storage criterion are adjusted accordingly, but the segmentation method remains consistent with the prediction process. Therefore, the method and system provided by this invention can be widely applied to the dual-objective optimization control of central air conditioning systems, achieving both optimized electricity costs and ensured user comfort, whether used for cooling or heating.

[0047] Pre-set time window before peak electricity prices [t] peak -H max , t peak -H min Within a certain timeframe (e.g., 1–5 hours in advance), generate a set of candidate times for energy storage startup, T={t}, according to a step size Δt. s1 , t s2 , ..., t sk}, t peak The peak electricity price start time is determined by the time-of-use pricing table; where H... max Upper limit of lead time, H min t is the lower limit of the time lead. sk This indicates the k-th candidate energy storage start-up time, where k represents the k-th candidate point in the candidate set and is an index number used to distinguish different candidate times.

[0048] In this embodiment, the preset time window before the peak electricity price is not a fixed value, but is determined by the edge control terminal based on the water supply capacity, the achievable water propulsion capability, and comfort constraints. This ensures that the set of candidate start-up times covers an effective range that is both likely to be full and will not prematurely deplete. The water propulsion capability characterizes the ability of chilled water to be pushed to various pipe sections on the water supply side before the peak arrives, under given pump speed and flow conditions. It can be estimated by comprehensively considering the pump capacity curve, the current available flow range, and the hydraulic characteristics of the pipe network.

[0049] For each t sk The edge control terminal generates a basic energy storage strategy, which includes at least: a water supply temperature setpoint from t sk Starting with a limited slope, the slope decreases to the target T. tar And set a lower limit protection to avoid the risk of condensation at the end due to excessively low frequency; the water pump frequency is from t sk The curve was adjusted to a push coverage curve for use in tpeak The water supply side should reach the target energy storage temperature as much as possible, where T tar The upper limit of pump speed is determined by the equipment capacity, terminal heat exchange capacity and safety strategy (condensation margin), which can be given by commissioning parameters or operating experience database.

[0050] S2. Based on the energy storage strategy and thermal inertia model, predict the comfort trajectory of each control area from each energy storage start-up candidate moment to the start of the peak electricity demand and the effective energy storage capacity that can be stored on the water supply side.

[0051] This step targets each candidate time t in the set of candidate energy storage start-up times. sk Under its corresponding basic energy storage strategy, two types of prediction results are output: one is the comfort trajectory, which is used to characterize the comfort level of each control area at time t. sk The evolution of temperature and relative humidity over time up to the start of peak electricity price (optionally extended to the end of peak) provides a basis for the subsequent comfort hard constraint determination in S3; the second is effective energy storage, which is used to characterize the energy that the water supply side can retain and use for peak shaving at the start of peak electricity price, and is used to determine whether the peak shaving energy demand is met and to verify that the water supply side coverage meets the standards. The inputs of S2 include the unified state vector formed by S1, the segmented parameter table of the water supply side, the set of candidate energy storage start-up times and their corresponding basic energy storage strategies, and time-of-use electricity price information.

[0052] Edge terminals are grouped by end device ID and divided into several control areas according to the end control object, denoted as j={1,2,…,J}, where j is the control area index (the j-th area) and J is the total number of control areas.

[0053] For each candidate time t sk The edge control terminal maps the water supply temperature setpoint and pump speed strategy from the basic energy storage strategy to equivalent terminal cooling inputs, such as terminal coil heat exchange, supply air temperature changes, or valve position changes, based on the heat exchange characteristics and control logic of the terminal equipment. It then invokes the regional thermal inertia sub-model to predict the temperature changes from t... sk To t peak (If necessary) peak Extending to the end of the peak phase The temperature and humidity trajectory of the region: or ; where T j (t∣t sk ) indicates that during precooling from t sk Given that the system is started, the predicted temperature of the j-th region at time t; RH j (t∣t sk ) indicates that during precooling from t sk Given that the function is activated, the predicted relative humidity of the j-th region at time t.

[0054] For the same candidate time t sk The edge control terminal simulates the low-temperature water propulsion-mixing-friction heat loss process based on the thermal inertia sub-model of the water supply network, and obtains the predicted water supply temperature T of each pipe section at the beginning of the peak electricity demand. sk，i (t peak |t sk ).

[0055] Based on this, the energy storage capacity of each section relative to the normal water supply temperature reference is calculated. Normal water supply temperature reference T ref (t peak This refers to the water supply setpoint typically used for comfortable operation when no energy storage strategy is employed. This setpoint can be obtained statistically from historical data under similar weather / load conditions or directly read from the current BAS normal setpoint. Then, in the i-th segment at t... peak The predicted energy storage at any given time is: Where ρ is the density of water, c w For the specific heat of water, V i E represents the volume of that segment. i (t peak |t sk () represents the i-th segment of water on the water supply side, at the start time t of the peak. peak Because the water in this section is cooled by pre-storage cooling, its temperature is lower than the normal supply temperature; the extra energy in this section is the stored energy. When At that time, the corresponding pipe section does not form effective energy storage. .

[0056] To reflect that the effective energy storage will decrease with cycles and loads, an energy storage retention coefficient η is introduced for each segment. i (t sk →t peak The energy storage retention coefficient (EGRC) ∈ (0,1] can be obtained by regression or table lookup from the actual temperature rise or cooling capacity reduction ratio under the corresponding operating conditions in historical operating data, and can be dynamically updated according to operating conditions. The EGRC refers to the energy storage retention coefficient from the start-up time t of the i-th segment. sk At peak time t peak Between these values, the proportion of energy stored that can be retained until peak hours is considered important. A value closer to 1 indicates that this portion of cooling capacity is more easily retained (e.g., later pre-cooling, lower load, lower flow rate, better insulation); a smaller value indicates that more cooling capacity is consumed / lost before peak hours (e.g., too early pre-cooling, high load, fast circulation, high heat dissipation). The final effective energy storage is: .

[0057] Example B illustrates how to calculate the energy storage component of a single water supply segment at the start of a peak electricity price under segmented water supply conditions. This component serves as the basis for subsequent calculations of the total effective energy storage on the entire water supply side. It is assumed that for a candidate energy storage start-up time t... sk It is predicted that at the start of the peak electricity price period tpeak Taking the fourth section (remote branch) as an example, the normal water supply temperature reference T ref (t peak =7.0℃, predict the temperature T of this section after energy storage. sk，4 (t peak |t sk Temperature V = 5.5℃, pipe section volume V4 = 1.5m³ 3 Then the theoretical energy stored at the beginning of the peak in the fourth paragraph. If we further consider this segment from t sk To t peak Energy storage retention coefficient η4(t) sk →t peak If ) = 0.85, then its effective energy storage is: The effective energy storage component of the fourth section at the beginning of the peak is 9450 × 0.85 = 8032.5 kJ. Other water supply pipe sections can be calculated separately using the same method. Finally, the total effective energy storage on the water supply side at the beginning of the peak is obtained by summing them up. .

[0058] To meet the engineering target of using energy storage temperature water during peak electricity price periods, this embodiment further outputs a compliance rate indicator for the water supply side, assuming the target energy storage temperature is T. tar Then it is defined in t peak Water supply compliance rate at any time Where 1(·) is an indicator function, taking the value 1 if the condition is true and 0 if it is false. The condition here is whether the temperature of the i-th segment is ≤ the target energy storage temperature. In engineering, R can be taken as 1. cov ≥0.95 or configured according to project requirements. The water supply side compliance rate reflects the proportion of water volume reaching the target energy storage temperature to the total water volume on the water supply side at the beginning of the peak electricity price period. It directly reflects whether the entire water supply side is basically filled with water at the energy storage temperature, avoiding situations where only the total energy is met but some remote areas are still too high.

[0059] Example C, target energy storage temperature T tar =6.0℃, at the start of the peak t peak The predicted temperatures for each water supply pipe section are as follows: Section 1 (from the machine room to the first branch): V1 = 3.0m 3 The predicted temperature is 5.8℃; Section 2 (middle section of the main trunk): V2 = 2.5m 3 The predicted temperature is 5.9℃; Section 3 (Branch A): V3 = 1.2m 3 The predicted temperature is 6.2℃; Segment 4 (Branch B, far end): V4 = 1.5m 3 The predicted temperature is 5.5℃; Section 5 (Branch C): V5 = 0.8m 3The predicted temperature is 5.7℃, and the pipe sections that meet the requirements are sections 1, 2, 4, and 5; therefore, the water supply side coverage compliance rate is... If the minimum coverage compliance threshold R configured in the project min If the value is 0.90, then the candidate start time is deemed infeasible due to insufficient coverage.

[0060] S3. Based on the comfort trajectory and effective energy storage, apply dual constraints to the set of candidate energy storage start-up times to obtain a set of feasible start-up times, and determine the final energy storage start-up time from the set of feasible start-up times.

[0061] The edge control terminal first configures and reads the comfort hard constraint range for each control region j, including the temperature hard constraint range [T]. j，min T j，max [Relative humidity hard constraint range [RH]] j，min RH j，max The aforementioned upper and lower limits can be set by users, issued from the building automation system strategy library, or project debugging parameters, and stored as inviolable constraint parameters on the edge control terminal side.

[0062] The edge control terminal starts at time t for each candidate in the candidate set. sk Perform full validation in the prediction time domain t∈[t sk , t peak，end Within a given time step, the predicted values ​​of all control regions are checked hourly. Specifically, this includes checking if any control region j satisfies T at any prediction time t. j (t∣tsk) <T j,min or T j (t∣t sk )>T j,max Then determine the candidate start time t. sk Infeasible under temperature constraints; if there exists any control region j that satisfies RH at any prediction time t. j (t∣t sk ) <RH j,min or RH j (t∣t sk )>RH j,max Then determine the candidate start time t. sk This is not feasible under humidity constraints; the candidate start time t is only adopted if the above temperature and humidity conditions satisfy the constraint interval for the entire control area throughout the entire prediction time domain. sk Mark as feasible for comfort, denoted as ComfortOK(t) sk =1; otherwise, it is recorded as ComfortOK(t) sk )=0.

[0063] Based on the hard constraint verification of comfort, the edge control terminal further combines load forecasting and target peak shaving strategies to determine the cooling demand E required by energy storage during peak electricity demand periods. need Specifically, based on the outdoor environment forecast, occupancy plan, and historical operating data of similar days obtained from S1, the edge control terminal obtains the cooling load forecast curve for peak periods; and according to the target peak shaving strategy (such as limiting the peak unit load limit or limiting the peak input power limit), it determines the scale of cooling capacity that needs to be replaced by energy storage during peak periods, thereby obtaining the energy demand E. need The E need This serves as the basis for the second type of hard constraint when selecting candidates for subsequent screening.

[0064] Specifically, the edge control terminal compares the peak-hour cooling load forecast curve with the upper limit curve of the available cooling capacity of the data center under the peak-shaving strategy, and accumulates the load gap according to the peak time period to obtain the energy demand E. need When the peak shaving strategy is expressed as an upper limit of input power, the edge control terminal converts the upper limit of input power into the corresponding upper limit of available energy based on the unit performance curve or energy efficiency coefficient.

[0065] The edge control terminal for each candidate start time t sk Read the effective energy storage prediction value E output by S2 eff (t peak |t sk ), and perform energy satisfaction determination: when E eff (t peak |t sk )≥E need At that time, determine the candidate start time t. sk Feasible under the energy storage constraint, when E eff (t peak |t sk ) < E need At that time, determine the candidate start time t. sk Energy storage constraints are not feasible.

[0066] Optionally, this embodiment may also introduce a water supply side coverage compliance rate threshold R. min If R cov (t sk ) <R min If the energy requirement is met, it will be deemed infeasible to avoid a situation where the total energy is sufficient but the temperature at the far end or in some branches is still too high.

[0067] The edge control terminal will meet the following conditions at the candidate start time t sk Include the feasible start-up time set F: ComfortOK(t) sk ) = 1 (Comfort hard constraints are met throughout); E eff(t peak |t sk )≥E need (Effective energy storage at the start of peak hours to meet peak shaving demand); R cov (t sk )≥R min (Optional, low-temperature coverage on the water supply side meets the standard); Through the above cross-filtering, a set of feasible start-up times F is obtained. The edge control terminal can simultaneously output the infeasibility cause flag for each candidate time (comfort boundary violation, insufficient energy storage, insufficient coverage) for interpretation and operation and maintenance diagnosis.

[0068] When the feasible start-up set F is not empty, to reduce the risk of energy storage being consumed by end loads or prematurely decaying due to heat loss along the flow path before the peak electricity price, and to consider the economic efficiency of equipment operation, the edge control terminal sorts all feasible start-up times in the feasible start-up set F according to time sequence, and selects the feasible start-up time closest to the start time of the peak electricity price as the final energy storage start-up time. Finally, the edge control terminal outputs the final energy storage start-up time and its corresponding basic energy storage strategy, and sends the strategy to the unit controller and water pump inverter for execution. The unit controller receives and executes the supply water temperature setpoint (or outlet water temperature target), unit start / stop / load limits, and allowed down-adjustment rate control quantities in the strategy, so that the chiller unit operates according to the target outlet water conditions; the water pump inverter receives and executes the pump frequency / speed settings, upper and lower limits, adjustment slope or time-sharing given curves in the strategy, thereby changing the supply water flow rate and circulation intensity, and cooperating with the unit to achieve the propulsion and coverage of low-temperature water on the supply side.

[0069] If the feasible start-up set F is empty, it indicates that under the current equipment capacity and comfort hard constraints, the existing candidate strategies cannot simultaneously meet the energy storage requirements and comfort constraints. In this case, the edge control terminal executes according to the preset fallback sequence: slightly enhances the energy storage strategy within the equipment safety boundary (e.g., slightly reduces the target water supply temperature T). tar Alternatively, increase the pump speed limit), then re-execute step S2 and step S3 of this procedure; if no feasible solution is found, then reduce the peak shaving target (reduce E) within the user-allowed peak shaving strategy range. need If no feasible solution is found, an alarm is output, and the closest feasible candidate start time and its shortcomings (insufficient energy storage or comfort exceedance) are given for manual intervention or adjustment of the upper-level energy management strategy.

[0070] like Figure 2 , Figure 3 , Figure 4 , Figure 5 as well as Figure 6The diagram shown is a diagram of a central air conditioning room system provided by the present invention. The equipment that needs to be optimized in the central air conditioning room includes chillers, chilled water pumps, cooling pumps and cooling towers. If it is a heat pump system, it includes water-source heat pumps, circulating water pumps and ground source water pumps. If it is an air-cooled heat pump system, it includes air-cooled heat pumps (including air-cooled screw chillers and air-cooled modular chillers) and circulating water pumps.

[0071] The chiller unit includes centrifugal chillers (units 1 and 2) and screw chillers (unit 3). The left side of the chiller unit is the cooling water circuit, where the cooling water outlet delivers cooling water to cooling towers (units 1, 2, 3, 4, and 5). The cooling towers cool the cooling water from the chiller unit. The cooled water then returns to the chiller unit via cooling pumps (units 1, 2, 3, 4, and 5) to continue absorbing heat from the chiller unit. The right side of the chiller unit is the chilled water circuit, where the chilled water outlet delivers chilled water to a distributor. Chilled water is delivered to each air conditioning terminal. After heat exchange at each terminal, the chilled water returns to the water collector and then returns to the chiller unit via chilled water pumps (pumps #1, #2, #3, #4, and #5). The chiller unit then cools the returned chilled water again. The screw chiller unit is additionally connected to a heat recovery terminal and heat recovery pumps (pumps #1 and #2) to recover and utilize the condensation heat that would otherwise be dissipated into the environment by the screw chiller unit through the cooling water-cooling tower. This allows usable hot / warm water to be output to the heat recovery terminal (e.g., for domestic hot water preheating, air conditioning reheating, process heat, or heating) while the unit is operating in cooling mode.

[0072] In the provided central air conditioning system diagram Figure 3 Connecting line 1 in Figure 4 Connecting line 1 in the diagram is a single line; connecting line 1 will... Figure 3 The chilled water outlet of the chiller unit and Figure 4 The water distributors in the middle are connected together; Figure 3 Connecting line 2 in the middle and Figure 4 Connecting line 2 in the diagram is a single line; connecting line 2 will... Figure 3 The chilled water return end of the chiller unit and Figure 4 It is connected to the refrigeration pump in the system.

[0073] Figure 3 Connecting line 3 in the middle and Figure 5 Connecting line 3 in the diagram is a single line; connecting line 3 will... Figure 3 The cooling water outlet of the centrifuge unit in the middle and Figure 5 The cooling towers (cooling tower #1, cooling tower #2, cooling tower #3, and cooling tower #4) are connected together; Figure 3 Connecting line 4 in the middle Figure 5 The connecting line 4 in the diagram is a single line. Figure 3 Connecting line 5 in the middle Figure 5 Connecting line 5 is a single line, and connecting lines 4 and 5 will... Figure 3 The cooling water return end of the centrifuge unit in the middle and Figure 5 The cooling pumps (cooling pump #1, cooling pump #2, and cooling pump #3) are connected together. Figure 3 Connecting line 6 in the middle Figure 5 Connecting line 6 in the middle is a line, connecting line 6 will Figure 3 The cooling water outlet of the screw chiller unit and Figure 5 It is connected to the cooling tower (cooling tower #5) in the middle; Figure 3 Connecting line 7 in the middle Figure 5 Connecting line 7 in the diagram is a single line; connecting line 7 will... Figure 3 The cooling water return end of the screw turbine unit and Figure 5 The cooling pumps (cooling pump #4 and cooling pump #5) are connected in the middle.

[0074] Figure 3 Connecting line 8 in Figure 6 Connecting line 8 in the middle is a line, connecting line 8 will Figure 3 screw compressor units and Figure 6 The heat recovery terminal is connected in the middle; Figure 3 Connecting line 9 in Figure 6 Connecting line 9 in the middle is a line, connecting line 9 will Figure 3 screw compressor units and Figure 6 The heat recovery pumps (heat recovery pump #1 and heat recovery pump #2) are connected; the connecting line 10 is connected to the air conditioning terminal.

[0075] The provided central air conditioning system diagram shows that only the screw chiller unit is running, while the centrifugal chiller unit is not. Therefore, the chilled water circuit on the left side of the screw chiller unit cools the water temperature from 33.3℃ to 29.0℃ through the cooling tower. The chilled water circuit on the right side of the screw chiller unit, with an outlet temperature of 7.6℃, returns to the screw chiller unit at 9.8℃ after heat exchange at the air conditioning terminal. Since the centrifugal chiller unit is not running, the outlet and return water temperatures of the cooling water circuit on the left and chilled water circuit on the right side of the centrifugal chiller unit are basically the same.

[0076] In Specific Implementation Two, in Specific Implementation One, the edge control terminal, based on the basic energy storage strategy and thermal inertia model, predicted the comfort trajectory and effective energy storage on the water supply side for each energy storage start-up candidate time. Under the premise of satisfying the hard constraints of comfort and ensuring that the effective energy storage is not less than the energy storage demand, a set of feasible start-up times F was obtained. In this implementation, without changing the above-mentioned hard constraint screening logic, the point-based strategy for determining the final energy storage start-up time from the set of feasible start-up times is replaced by a balance selection of attenuation risk and full-storage risk, to better align with the physical laws of continuous chilled water circulation and the dynamic attenuation of effective energy storage with load and flow.

[0077] The edge control terminal summarizes the key factors affecting the rationality of the final start-up time into two mutually restrictive risks: energy storage decay risk and full storage risk.

[0078] The risk of energy storage degradation refers to the fact that when energy storage is activated too early, the low-temperature water on the supply side undergoes a longer circulation and terminal heat exchange process before the peak arrives. The stored cold energy is more easily consumed in advance through terminal consumption, heat dissipation in the pipeline network, and mixed return flow, resulting in a reduction of the effective stored energy at the beginning of the peak. This risk is manifested in the fact that the longer the energy storage is maintained and the easier it is to be consumed in advance under high load and high flow conditions.

[0079] The risk of over-storage occurs when water storage starts too late, meaning the supply side hasn't had time to fully push the water in the main pipes and branches to the target storage temperature. This results in sections of the supply side still having higher temperatures at the start of peak hours, leading to insufficient coverage. Consequently, during peak periods, the supply side cannot guarantee that the water is primarily at the storage temperature, resulting in inadequate peak-shaving effect. This risk manifests as insufficient coverage compliance on the supply side at the start of peak hours or failure of key remote branches to meet standards. Key remote branches can be identified based on the most unfavorable hydraulic principle or areas with frequent historical temperature deviations.

[0080] The two types of risks typically exhibit opposite trends at candidate start times: earlier start times result in higher decay risk and lower accumulation risk; later start times result in lower decay risk and higher accumulation risk. Therefore, this embodiment quantifies the two types of risks and performs balanced point selection to determine a more robust final start time.

[0081] For each candidate start time t in the feasible set F sk The edge control terminal extracts information to characterize the premature consumption of energy storage before the peak based on the prediction results of Example 1, and constructs a decay risk index R. decay (t skThe attenuation risk index can be constructed using a weighted combination, product, or normalized monotonic function of the storage duration and cyclic consumption intensity. Its numerical value is only used for relative comparison between different candidate start-up times within the same feasible set, and not as an absolute physical quantity. The energy storage storage duration is the time span Δt from the candidate start-up time to the start of the peak electricity price. s =t peak -t sk Characterization. The longer the retention time, the greater the chance that the stored energy will be consumed before the peak. The characterization of cycle consumption intensity refers to the period [t] sk , t peak Within a given time period, by combining load forecasting and pump speed / flow rate strategies, a characterization of circulation intensity (e.g., predicted flow rate, pump frequency, or the average number of water exchanges calculated from these values) is obtained. The greater the circulation intensity, the faster the low-temperature water is consumed by end-point heat exchange and recirculated for mixing, resulting in a higher risk of degradation.

[0082] The edge control terminal combines the aforementioned hold duration with the cycle consumption intensity to obtain each candidate t. sk The attenuation risk index R decay (t sk The duration of the cycle is directly determined by the peak electricity price and the candidate start-up time. The cycle intensity is characterized by the pump speed / flow rate curve and load prediction of the energy storage strategy in Example 1. Both of these can be obtained and interpreted in real time at the edge.

[0083] Example D, the feasible start-up time set F={t} is obtained by screening in Example 1. s1 , t s2 , t s3 Candidate start time t s1 The start-up time is 12:30, with a 90-minute interval from the peak electricity price, and an average water exchange frequency of 1.8 times; the candidate start-up time t s2 The start-up time is 13:00, with a 60-minute interval from the peak electricity price, and an average water exchange frequency of 1.2 times; the candidate start-up time t s3 The start-up time was 13:20, with a 40-minute interval from the peak electricity price, and the average number of water exchanges was 0.8. Based on the above data, the degradation risk ranking result, R, was obtained. decay (t s1 )>R decay (t s2 )>R decay (t s3 );t s1 The earliest start-up occurs when the low-temperature water undergoes more circulation and terminal heat exchange before the peak, making it most susceptible to degradation. s3 Latest start-up, lowest risk of degradation; t s2 It is at an intermediate level.

[0084] For the same candidate start time t sk The edge control terminal constructs a full-storage risk index R based on the segmented prediction results from the water supply side. fill (t sk It is based at least on the following quantities: the compliance rate of water supply side coverage and the compliance status of key remote branches. At the start of peak electricity prices, t... peak Statistics show that the temperature of each pipe section on the water supply side has reached the target energy storage temperature T. tar The volume fraction, as the coverage compliance rate R cov (t sk The lower the coverage compliance rate, the more pipe sections still have excessively high water temperatures, and the higher the risk of water filling. For pre-identified critical remote branches or sensitive terminal branches, check their performance at t... peak Whether the water supply temperature reaches the target storage temperature at any given time; if any critical branch fails to meet the target, the risk of full storage is deemed significantly increased. Specifically, R in Example 1... cov (t sk This is mainly used to determine whether the water supply side meets the hard or quasi-hard requirements for basic energy storage coverage, while R in this embodiment... fill (t sk The two are used to rank and evaluate the relative merits of different candidate schemes in terms of the degree of fullness within the set of feasible start-up times; their purposes are different.

[0085] Edge control terminals can combine coverage compliance rate and key remote branch compliance status to form R fill (t sk This indicator can directly explain whether the water supply side has basically switched to energy storage temperature water when the peak arrives.

[0086] After obtaining the feasible set F and the corresponding R for each candidate time step decay (t sk ) and R fill (t sk After that, the edge control terminal determines the final energy storage start-up time according to the following deterministic steps: Within the feasible set F, candidate start-up times with attenuation risk exceeding a preset threshold (the maximum value of the attenuation risk index within the specified range) are first eliminated, resulting in an acceptable attenuation subset F1. The aforementioned preset threshold is obtained from operational experience, pipeline insulation level, and historical energy storage attenuation statistics, and is used to avoid situations where premature start-up leads to a large amount of premature loss of energy storage before peak hours.

[0087] In some implementations, to facilitate interpretable sorting and parallel elimination of candidate solutions in the set of feasible start times, the edge control terminal can also calculate a comfort margin index based on the comfort prediction trajectory output by S2. This index characterizes the safety margin of candidate solutions without exceeding the hard comfort constraint boundary. The comfort margin index may, for example, include the minimum temperature margin M at the candidate time. T(t sk ) and minimum humidity margin M RH (t sk It is derived from the minimum safe distance of the entire control region relative to the upper and lower limits of the comfort constraints within the prediction time domain:

[0088] ;

[0089] ;

[0090] The aforementioned margin index is directly calculated from the predicted trajectory output by S2, representing the minimum safe distance from the comfort constraint boundary in all control regions across the entire prediction time domain.

[0091] If F1 is not an empty set, the final energy storage start-up time is determined first within F1 based on the full storage risk index to ensure that the water supply side meets the coverage requirements of the target energy storage temperature at the start of the electricity price peak. Furthermore, when the full storage risk index is difficult to distinguish among multiple candidate start-up times (e.g., the full storage risk is the same, or the difference is less than the preset distinction threshold), the edge control terminal can introduce a comfort margin index as a parallel resolution criterion, prioritizing the candidate scheme with a larger comfort margin to reduce the risk of operating close to the comfort boundary; when the comfort margin is still indistinguishable, the edge control terminal can further adopt a candidate start-up time with a later time position as the final start-up time to reduce the chance of premature energy storage decay before the peak.

[0092] In Example D, under the same operating condition and with the same set of feasible start times, F = {t} s1 , t s2 , t s3 The edge control terminal, based on the segmented prediction results from the water supply side, at t peak Continuously assess the risk of full storage. The target cold storage water supply temperature is 7℃; candidate start-up time t s1 The coverage compliance rate was 0.98; the candidate start time t s2 The coverage rate was 0.95; the candidate start time t s3 The coverage compliance rate was 0.88; the key remote branch inspection results were: candidate start time t s1 The critical remote branch water supply temperature is ≤T tar The target has been met; candidate start time t s2 The critical remote branch water supply temperature is approximately T. tar The target has been met; candidate start time t s3 There is at least one critical remote branch with a water supply temperature > T tar The coverage rate did not meet the standard; considering both the coverage compliance rate and the condition of key remote branches, the edge control terminal obtained R... fill (t s3 )>R fill (ts2 )>R fill (t s1 ), t s3 Starting too late means the low-temperature water hasn't had time to fully advance to the distant pipe section, posing the highest risk of overfilling. s1 The system started early, and the water supply side was almost completely covered by cold water storage, minimizing the risk of it becoming full. s2 Between the two. Candidate start time t s1 Those with excessively high attenuation risk are excluded. s2 and t s3 The attenuation risk is acceptable, resulting in a subset F1={t} s2 , t s3 In F1, the best option is chosen based on accumulating full risk; therefore, t is ultimately selected. s2 This marks the final moment for energy storage to begin.

[0093] If F1 is an empty set after attenuation filtering, it indicates that all feasible start-up times under the current operating conditions have a high risk of energy storage attenuation. In this case, the edge control terminal does not directly revert to the timing rules of Implementation Example 1, but switches to a safety net strategy that prioritizes minimizing the risk of full storage: it directly selects the time with the lowest risk of full storage from the feasible set F as the final start-up time, and adds conservative limits on the rate of water supply temperature reduction and pump speed limits at the execution layer to reduce the comfort and safety risks caused by premature precooling and ensure the certainty of peak shaving effect.

[0094] Finally, the edge control terminal outputs the final energy storage start-up time t. sk It also generates the corresponding basic energy storage strategy and generates interpretable decision records, including: the attenuation risk, full storage risk, coverage compliance rate, key branch compliance status, and the filtering and parallel elimination results when selecting points, which facilitates subsequent review and maintenance traceability.

[0095] Compared with the time-based location-based fixed-point strategy in Specific Implementation 1, this implementation explicitly constructs the energy storage attenuation risk and full-storage risk, and performs balanced point selection within the feasible set. This ensures that the final energy storage start-up time can simultaneously avoid two typical failure modes: premature start-up leading to early energy loss and late start-up leading to insufficient water supply coverage. Thus, under the real operating conditions of continuous chilled water circulation and dynamic changes in load and flow, a more robust peak-shaving effect and a more interpretable decision-making basis are obtained.

[0096] Through the above description of the implementation methods, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the above functions can be divided into different functional modules to complete all or part of the functions described above.

[0097] In the embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between devices or units, and may be electrical, mechanical, or other forms.

[0098] The units described as separate components may or may not be physically separate. A component shown as a unit can be one or more physical units, located in one place or distributed in multiple different locations. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0099] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0100] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, or the parts that contribute to the solution, or all or part of the technical solution, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0101] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A dual-objective optimization control method for central air conditioning based on edge computing, characterized in that, Includes the following steps: S1. The edge control terminal acquires multi-source operation data and time-of-use electricity price data for the central air conditioning system, and establishes a thermal inertia model to characterize the thermal inertia of the building and the pipe network. Specifically, it collects and organizes operating state variables to characterize the building's thermal response and the heat transfer process of the pipe network water. It then constructs regional thermal inertia sub-models to characterize the dynamic response relationship between temperature and humidity in each control area and changes in water supply-side operating parameters, as well as a water supply-side pipe network thermal inertia sub-model to characterize the transmission lag, mixing effect, and friction loss of water in each pipe section. The outputs of the regional thermal inertia sub-models and the pipe network thermal inertia sub-models are used as the statistical data for predicting the effectiveness of candidate energy storage strategies. The calculation basis, in the process of constructing the thermal inertia sub-model of the water supply network, also includes: when the edge control terminal establishes the thermal inertia sub-model of the water supply network, it actively divides the water supply network into segments along the water flow path of the water supply side, divides the water supply main pipe and main branches into segments according to the structural nodes and monitoring points of the network, and generates a set of segment parameters for each water supply pipe segment to characterize the geometry and heat transfer characteristics of the pipe segment, generates a set of energy storage start-up candidate times within a preset time window before the peak electricity price, and sets a corresponding energy storage strategy for each energy storage start-up candidate time, the energy storage strategy including the adjustment of the water supply temperature set value and the water pump operating parameters; S2. Based on the energy storage strategy and the thermal inertia model, predict the comfort trajectory of each control area from each energy storage start-up candidate moment to the start of the electricity peak and the effective energy storage on the water supply side. The specific process is as follows: When the edge control terminal executes the prediction process, for the energy storage start-up candidate moment in the set of energy storage start-up candidate moments, actively generate prediction results for the control area side and the water supply side based on the basic energy storage strategy corresponding to the energy storage start-up candidate moment. Specifically, it includes: mapping the water supply temperature setpoint and water pump operating parameters in the basic energy storage strategy to equivalent terminal input based on the heat exchange characteristics of the terminal equipment, and calling the regional thermal inertia sub-model to predict the comfort trajectory of temperature and relative humidity of each control area from the candidate start-up moment to the electricity peak moment. At the same time, it calls the water supply side pipeline thermal inertia sub-model to predict the temperature status of each water supply pipe section at the electricity peak moment, and calculates the effective energy storage on the water supply side based on the normal water supply temperature benchmark, pipe section volume and water body thermal properties. S3. Based on the comfort trajectory and effective energy storage, apply dual constraints to the set of candidate energy storage start-up times to obtain a set of feasible start-up times, and determine the final energy storage start-up time from the set of feasible start-up times.

2. The dual-objective optimization control method for central air conditioning based on edge computing as described in claim 1, characterized in that: The calculation of the effective energy storage on the water supply side also includes the following steps: When calculating the effective energy storage on the water supply side, the edge control terminal actively introduces an energy storage retention coefficient for each water supply pipe section to characterize the proportion of energy retained from the candidate start-up time to the peak electricity price time. The theoretical energy increment of each water supply pipe section relative to the normal water supply temperature benchmark is combined with the energy storage retention coefficient to obtain the effective energy storage on the water supply side at the peak electricity price time. The edge control terminal can also calculate the water supply side coverage compliance rate based on the relationship between the temperature of each water supply pipe section and the target water supply temperature during peak electricity price periods, and use the coverage compliance rate for coverage verification or screening at candidate start times.

3. The dual-objective optimization control method for central air conditioning based on edge computing as described in claim 1, characterized in that: The set of candidate energy storage start-up times is subject to dual constraints based on the comfort trajectory and effective energy storage to obtain a set of feasible start-up times. The specific process is as follows: When applying the dual condition constraints to the set of candidate energy storage start-up times, a comfort hard constraint interval is actively configured for each control area, and the predicted values ​​of temperature and relative humidity of each control area are fully verified in the corresponding prediction time domain for each candidate start-up time, so as to mark the candidate start-up times that meet the comfort hard constraints as comfortable and feasible. The edge control terminal combines load forecasting results with peak shaving strategies to determine the energy demand during peak periods, and performs energy satisfaction judgment on the effective energy storage corresponding to each candidate start-up time. Among them, the edge control terminal will include candidate start times that simultaneously meet the requirements of comfort feasibility and energy satisfaction into the set of feasible start times.

4. The dual-objective optimization control method for central air conditioning based on edge computing as described in claim 3, characterized in that: After obtaining the set of feasible start-up times, the specific process for determining the final energy storage start-up time is as follows: After obtaining the set of feasible start-up times, the edge control terminal determines the final energy storage start-up time from the set of feasible start-up times according to the preset deterministic point selection rules; When the set of feasible start times is empty, the edge control terminal adjusts the basic energy storage strategy parameters within the device safety boundary and re-executes prediction and screening. If the set of feasible start times is still empty after re-screening, the peak shaving target is adjusted within the allowed peak shaving strategy range to update the energy demand and the prediction and screening are re-executed. If the set of feasible start times is still empty, an alarm message is output, and the closest feasible candidate start time and its infeasibility reasons are output to support manual intervention.

5. The dual-objective optimization control method for central air conditioning based on edge computing as described in claim 4, characterized in that: The deterministic point selection rule is as follows: When the set of feasible start times is not empty, the edge control terminal selects the latest feasible start time from the set of feasible start times as the final energy storage start time. The basic energy storage strategy corresponding to the final energy storage start-up time will be sent to the unit controller and water pump frequency converter for execution.

6. The dual-objective optimization control method for central air conditioning based on edge computing as described in claim 5, characterized in that: The deterministic point selection rules also include: For each candidate start time in the set of feasible start times, the energy storage decay risk and the full storage risk are quantified respectively, and the final energy storage start time is determined based on the balance relationship between the energy storage decay risk and the full storage risk. Among them, the edge control terminal generates the attenuation risk index based on the duration of the hold from the candidate start time to the start time of the peak electricity price and the cyclic consumption intensity characterization during the hold time. The cyclic consumption intensity characterization is obtained from the load forecast results and the water pump operation parameter strategy. The edge control terminal generates the full storage risk index based on the water supply side coverage compliance rate at the start time of the peak electricity price and the compliance status of the pre-identified key remote branches. The attenuation risk index is used to characterize the risk that the energy storage cold energy is consumed or lost in advance before the start of the electricity price peak at the candidate start time, resulting in a decrease in effective energy storage. The full storage risk index is used to characterize the risk that the water supply side will not be fully switched to the energy storage temperature due to insufficient coverage of the target energy storage temperature or failure of key remote branches to meet the standard at the beginning of the peak electricity price. The feasible start-up time set is subjected to attenuation filtering based on the attenuation risk index to obtain a subset, and the final energy storage start-up time is determined based on the full-storage risk index on the subset.

7. The dual-objective optimization control method for central air conditioning based on edge computing as described in claim 6, characterized in that: The process of determining the final energy storage start-up time based on the risk index of full storage on the basis of the subset is as follows: When the subset is non-empty, the full-storage risk index corresponding to each candidate start time is compared within the subset, and the candidate start time with the minimum full-storage risk index is selected as the final energy storage start time. When the subset is empty, the candidate start time with the minimum value of the full energy storage risk index is selected from the set of feasible start times as the final energy storage start time.

Images