Energy-saving intelligent control and peak load shifting and valley filling collaborative operation method for water-cooled central air conditioner

Abstract

This invention discloses a method for energy-saving intelligent control and peak-shaving / valley-filling coordinated operation of water-cooled central air conditioning systems, relating to the field of water-cooled central air conditioning control technology. The method collects data on chiller equipment, building environment, chilled water storage status, and grid-side operation to form a joint state variable. Based on the joint state variable, it predicts future cooling load, regional temperature, and baseline power, and establishes a peak-shaving cooling capacity account consisting of building thermal inertia cooling capacity, inherent chilled water network chilled water storage capacity, and physical chilled water storage capacity. Based on the peak-shaving cooling capacity account, comfort boundary, equipment safety boundary, and power limitation boundary, it generates a reduceable power envelope and a target power curve. Based on the target power curve, it forms an execution command frame and corrects the peak-shaving cooling capacity account according to the actual cooling load, actual system power, and actual regional temperature, forming a cooling capacity debt and a recovery power upper limit. This method is applicable to energy-saving operation, peak shaving / valley filling, and demand response control of water-cooled central air conditioning systems.

Description

Technical Field

[0001] This invention relates to the field of water-cooled central air conditioning control technology, specifically to a method for energy-saving intelligent regulation and peak-shaving and valley-filling coordinated operation of water-cooled central air conditioning. Background Technology

[0002] Water-cooled central air conditioning systems are commonly used in commercial complexes, hospitals, hotels, and industrial park chiller plants. The main working principle of this type of central air conditioning system is to deliver the produced cooling capacity to terminal units such as air handling units and fan coil units via chilled water pumps and chilled water pipe networks, and then discharge the condensation heat through cooling water pumps and cooling towers. Its operating load is affected by outdoor weather, personnel density, operating hours, and equipment heat dissipation; there is a certain energy consumption coupling between the chiller, water pumps, and cooling towers. The mainstream control method relies on an edge intelligent controller to collect data on supply and return water temperature, flow rate, pressure difference, equipment frequency, power, and alarm status, and determines operating parameters based on setpoint reset, equipment group control, and energy efficiency model optimization.

[0003] Chinese patent document CN118361827A discloses an energy-saving optimization control method, system, and network-side server for a water-cooled central air conditioning system. This technology collects historical and real-time operating data, preprocesses operating parameters to form chiller datasets, chilled water pump datasets, cooling water pump datasets, cooling tower datasets, and system operating status datasets. Using historical outdoor temperature and humidity as input to the load prediction model and system cooling capacity as output, an RNN (Recurrent Neural Network) is used to train the load prediction model. During operation, real-time outdoor temperature and humidity are input to the load prediction model to calculate real-time cooling capacity and determine the current load. Constraints are set for chilled water supply temperature, cooling water return temperature, chilled water supply and return pressure difference, cooling water temperature difference, chilled water pump frequency, cooling water pump frequency, and cooling tower frequency. These constraints are then combined with chiller COP models, chilled water pump power models, chilled water flow models, cooling water pump power models, cooling water flow models, cooling tower power models, and cooling tower proximity temperature models to iterate through control parameter combinations and select the parameter setpoint corresponding to the lowest total system power.

[0004] The aforementioned existing technologies can establish prediction and optimization processes around the energy consumption of chiller equipment, making them suitable for comparing energy consumption under different parameter combinations under the same load. However, when time-of-use pricing, demand control, and grid demand response are involved in building operation, central air conditioning not only needs to reduce immediate energy consumption but also needs to reduce power consumption during specified periods and resume cooling after the response ends. Most existing technologies rely on real-time load and equipment energy efficiency models to find low-energy operating points, with their controllable boundaries concentrated on chillers, pumps, and cooling towers. When the system experiences peak-hour load reduction, adjusting chilled water temperature, pump frequency, and chiller load solely based on the current load or equipment efficiency can easily overlook factors such as the temperature rise rate in the terminal area, building heat storage characteristics, network water lag, and differences in the remaining capacity of cold storage equipment.

[0005] The principle is that temperature changes in a building area have thermal inertia, the return temperature of chilled water is delayed relative to the terminal load, and the power changes of chillers, water pumps, and cooling towers have different response speeds. If peak shaving and recovery actions are not uniformly constrained, it will cause regional comfort to exceed limits, frequent start-ups and shutdowns of chillers, rebound of peak demand, and fluctuations in operating costs. Therefore, we must solve this technical problem: how to determine the adjustable power amplitude, duration, and recovery boundary of water-cooled central air conditioning when participating in peak-valley regulation while meeting indoor comfort and equipment safety. Summary of the Invention

[0006] (a) Technical problems to be solved

[0007] To address the shortcomings of existing technologies, this invention provides a method for energy-saving intelligent control and peak-shaving / valley-filling coordinated operation of water-cooled central air conditioning. It establishes a peak-shaving cooling capacity account consisting of building thermal inertia cooling capacity, inherent chilled water network cooling capacity, and physical cooling capacity. Based on the peak-shaving cooling capacity account, comfort boundary, equipment safety boundary, and power limitation boundary, it generates a power reduction envelope and a target power curve. Based on the target power curve, it forms an execution command frame and corrects the peak-shaving cooling capacity account according to the actual cooling load, actual system power, and actual regional temperature, thus establishing cooling capacity debt and a power recovery limit. This solves the technical problems described in the background art.

[0008] (II) Technical Solution

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] The energy-saving intelligent control and peak-shaving and valley-filling collaborative operation method of water-cooled central air conditioning is executed by the cold station IoT cloud platform, including: acquiring the operation data of cold station equipment, terminal area, cold storage status and grid interface, calculating the current cooling load and current system power, and forming a joint state quantity;

[0011] Based on the joint state variables, the future cooling load, regional temperature and baseline power are predicted, and the building thermal inertia cooling capacity, the inherent cooling capacity of the chilled water network and the physical cooling capacity are summarized into a peak-shifting cooling capacity account. When no physical cooling capacity is configured, the physical cooling capacity is zero. Based on the peak-shifting cooling capacity account, comfort boundary, equipment safety boundary and power limitation boundary, the reduceable power envelope and target power curve are generated.

[0012] Based on the target power curve, an execution command frame containing chilled water temperature, water pump differential pressure, cooling water temperature, cooling tower frequency, chiller combination, regional temperature deviation, and cold storage power is generated. The peak-shifting cold capacity account is then corrected according to the actual cooling load, actual system power, and actual regional temperature to form a cold capacity debt and a recovery power limit.

[0013] Furthermore, the formation of the peak-shifting cooling capacity account includes:

[0014] The building's thermal inertia cooling capacity is determined based on the regional comfort upper limit, regional temperature, regional equivalent heat capacity, and regional availability factor in the comfort boundary; the inherent cooling capacity of the chilled water network is determined based on the chilled water network volume, network water temperature, cooling reference water temperature, and network availability factor; and the physical cooling capacity is determined based on the cooling water temperature, cooling water volume, remaining ice volume, and cooling branch status.

[0015] Write the building’s thermal inertia, chilled water network’s inherent cold storage capacity, and physical cold storage capacity into the same account balance field.

[0016] Furthermore, the generation of the power reduction envelope includes: forming a device load reduction boundary based on the difference between the baseline power and the minimum operable power; forming a power constraint boundary based on the difference between the baseline power and the power limit boundary; and forming a cooling capacity support boundary based on the account balance field of the peak-shifting cooling capacity account.

[0017] The restricted values ​​in the equipment load reduction boundary, power constraint boundary, and cooling capacity support boundary are written into the reduceable power envelope, and the target power curve is constrained by the reduceable power envelope.

[0018] Furthermore, the formation of cooling debt includes: during the execution of the target power curve, the accumulation of cooling deficit based on the positive difference between the predicted cooling load and the actual cooling supply; and deducting the account balance field of the peak-shifting cooling account based on the cooling deficit.

[0019] After the target power curve is completed, the upper limit of the recovery power is formed based on the cooling capacity gap, the actual system power, and the power limit boundary, and the upper limit of the recovery power is written into the constraint field of the next control cycle.

[0020] Furthermore, the formation of the joint state variables includes reliable processing of the sampled fields:

[0021] When a sampled field passes the range check, rate of change check, and physical consistency check, the sampled field is written into the joint state quantity; when a sampled field fails at least one of the range check, rate of change check, and physical consistency check, the sampled field is marked as an abnormal field and excluded from the current cooling load calculation.

[0022] When the number of missing sampled fields does not exceed the hold window, the previous reliable sampled field is used.

[0023] When a sampled field is missing for more than the hold window, write the sampled field to an unavailable flag.

[0024] Furthermore, the preprocessing before issuing the execution instruction frame includes frame sequence number verification, execution time verification, and verification field verification:

[0025] When an execution instruction frame passes the frame sequence number check, execution time check, and check field check, the execution instruction frame is written to the field control queue; when an execution instruction frame fails at least one of the frame sequence number check, execution time check, and check field check, the previous valid execution instruction frame is retained; when the retention window is reached, a safety control mode flag is generated and the writing of new load reduction instructions is stopped.

[0026] Further modifications to the peak-shifting cooling capacity account include security interception:

[0027] When the actual zone temperature exceeds the comfort boundary, cancel the available field for the zone temperature offset and reduce the building thermal inertia cooling capacity; when the chilled water flow rate is lower than the chiller protection flow rate in the equipment safety boundary, freeze the derating field for the chiller combination and water pump differential pressure.

[0028] When the actual system power does not exceed the target power curve and there are no instances of actual regional temperature exceeding the limit or chilled water flow exceeding the limit, the current account balance field of the peak-shifting cooling capacity account is used.

[0029] Furthermore, when the cold storage state characterization does not include physical cold storage equipment, the physical cold storage capacity is zero, and the peak-shifting cold capacity account is formed by the building's thermal inertia cold capacity and the inherent cold storage capacity of the chilled water network.

[0030] Write a zero value to the cold storage power field in the execution instruction frame;

[0031] The target power curve forms the field control object through fields corresponding to chilled water temperature, water pump differential pressure, chiller combination, and zone temperature offset.

[0032] Furthermore, when configuring physical cold storage equipment, the physical cold storage capacity is formed based on the cold storage water temperature, cold storage water volume, remaining ice volume, and cold storage branch status.

[0033] When the target power curve is in the peak-shaving section, the cold storage power field is written with the cold release value; when the target power curve is in the recovery section, the cold storage power field is written with the cold preservation value, and the supplementary cooling command is limited according to the upper limit of the recovery power.

[0034] Furthermore, the mode labels for the target power curve are formed according to the following priority:

[0035] When a grid demand response command is received, a power tracking mode flag is written; when no grid demand response command is received, the off-peak electricity state is established, and the available peak-shaving cooling capacity account is lower than the future peak-shaving demand, a pre-cooling storage mode flag is written; when neither the power tracking mode flag nor the pre-cooling storage mode flag is written, and at least one of the peak electricity state or the demand close state is established, a peak-shaving mode flag is written.

[0036] If none of the above mode flags are written, write the normal energy-saving mode flag.

[0037] Furthermore, the execution instruction frame includes a frame sequence number field, an execution time field, a mode flag field, a target power field, a chilled water temperature field, a water pump differential pressure field, a cooling water temperature field, a cooling tower frequency field, a chiller combination field, a zone temperature offset field, a cold storage power field, a peak-shifting cold capacity account field, a cold capacity debt field, a recovery power limit field, and a verification field; the execution instruction frame forms one frame in each control cycle.

[0038] Furthermore, the cooling station IoT cloud platform writes the execution instruction frame into the operation log record. The operation log record includes the frame sequence number field, execution time field, mode flag field, target power field, peak shiftable cooling capacity account field, cooling capacity debt field, recovery power limit field, and actual system power field.

[0039] When the actual system power field is higher than the target power field, the operation log records and writes a power deviation flag.

[0040] When the actual system power field is not higher than the target power field, the operation log records the power tracking flag.

[0041] Furthermore, the cooling station IoT cloud platform generates monitoring interface status data, which includes mode flag field, target power field, actual system power field, peak shiftable cooling capacity account field, cooling capacity debt field, recovery power limit field, and safety control mode flag field.

[0042] When a valid value is written to the safety control mode flag field, the monitoring interface status data retains the previous valid target power field and refreshes the safety control mode flag field.

[0043] When an invalid value is written to the safety control mode flag field, the target power field of the monitoring interface status data is refreshed.

[0044] (III) Beneficial Effects

[0045] This invention provides a method for energy-saving intelligent control and peak-shaving / valley-filling coordinated operation of water-cooled central air conditioning, which has the following beneficial effects:

[0046] By constructing a joint state variable for air conditioning, buildings, and the power grid, data from chilled water, cooling water, chillers, terminal areas, electricity prices, and demand response are incorporated into the same control entry point. This provides a unified basis for load forecasting, power constraint judgment, and equipment control, reducing deviations caused by adjustments of individual devices.

[0047] By constructing a peak-shifting cooling capacity account, the building's thermal inertia releaseable cooling capacity, the inherent cooling capacity of the chilled water network, and the releaseable cooling capacity of physical cooling storage are metered on this basis. This allows the system to clearly identify the required source of available cooling capacity before precooling, cooling storage, peak shaving, and cooling release, avoiding situations such as excessive precooling based solely on peak and off-peak periods, insufficient peak shaving, or exceeding comfort limits. By combining the established peak-shifting cooling capacity account with baseline power, minimum operable power, comfort constraints, equipment safety constraints, and power limit constraints to generate a target power curve, the chiller, water pump, cooling tower, zone temperature setpoint, and cooling storage device are all aligned around the same power target.

[0048] During peak power and demand response periods, instead of simply shutting down the system to reduce power, we take into account both indoor comfort boundaries and equipment safety boundaries by increasing the chilled water supply temperature, limiting the chilled water flow, utilizing peak-shifting cooling capacity, adjusting the zone temperature setpoint, and allocating chiller load.

[0049] The system compares and adjusts the peak-shaving cooling capacity account and control parameters based on actual cooling load, system power, and actual indoor temperature to prevent prediction deviations from propagating into subsequent cycles. If anomalies or exceeding limits occur, it switches to a safety control mode to reduce the risk to equipment. After peak shaving or demand response ends, the system records the cooling capacity debt. Based on the cooling capacity debt, indoor temperature recovery demand, and power recovery slope, it gradually restores the chiller load, water pump frequency, and cooling tower fan frequency to suppress new peak values ​​formed by concentrated supplemental cooling, achieving a closed-loop synergy of energy-saving operation, peak shaving and valley filling, and rebound control. Attached Figure Description

[0050] Figure 1 A schematic diagram of the field scene and topology of the water-cooled central air conditioning energy-saving intelligent control system provided in the embodiments of the present invention;

[0051] Figure 2 A closed-loop flowchart of the water-cooled central air conditioning energy-saving intelligent control and peak-shifting and valley-filling coordinated operation method provided in the embodiments of the present invention;

[0052] Figure 3 This is a schematic diagram of the joint state variable generation and reliable sampling field processing flow provided in an embodiment of the present invention;

[0053] Figure 4 This is a schematic diagram of load forecasting and peak-shifting cooling capacity account aggregation provided in an embodiment of the present invention;

[0054] Figure 5 This is a timing diagram illustrating the target power curve and the power reduction envelope provided in an embodiment of the present invention.

[0055] Figure 6 This is a schematic diagram of the decomposition of execution instruction frames and device collaborative execution provided in an embodiment of the present invention;

[0056] Figure 7 This is a schematic diagram illustrating the closed-loop process of on-site execution, account correction, and peak shaving recovery provided in this embodiment of the invention. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] Please see Figures 1-7 This invention provides a method for energy-saving intelligent control and peak-shaving / valley-filling coordinated operation of water-cooled central air conditioning, including:

[0059] Step 1: The edge controller converts the water-cooled central air conditioning system, building usage status, and power grid constraint information into a joint state quantity under the same time, the same naming rule, and the same quality boundary. This enables Step 2 to directly predict future cooling load and peak-shifting cooling capacity based on this joint state quantity.

[0060] In water-cooled central air conditioning systems, chillers, chilled water pumps, cooling water pumps, and cooling towers are managed separately. Indoor temperature, personnel density, and terminal valve openings are distributed across edge intelligent controllers. Electricity prices and demand response commands originate from the system or grid demand response interface. If this data is directly fed into subsequent models, common problems are not data gaps, but rather inconsistencies in data timing, units, names of the same physical quantity, and unmarked instrument anomalies. This leads to subsequent predictions misinterpreting short-term communication jitter as load changes. The processing logic in step one involves establishing a closed-loop relationship between data source, physical meaning, timestamp, trust boundary, and output fields on the field side, and then concatenating cooling capacity, power, regional comfort margin, and grid constraints into a joint state quantity. This design prevents the subsequent step two from repeatedly determining whether a temperature value originates from the supply or return water side, and also avoids using distorted chilled water flow rates to calculate peak-shifting cooling capacity.

[0061] Step one is executed by the edge controller, with field instruments, chillers, water pumps, and cooling all acting as data providers. The edge controller receives data according to a uniform sampling period, preferably 30 to 300 seconds, and more preferably 60 to 180 seconds. For chiller start / stop status, demand response commands, and equipment alarm status, event triggering and periodic acquisition are used in parallel to write data into the same data frame.

[0062] The temperature difference and flow rate of chilled water supply and return water should belong to the same operating segment; otherwise, the calculated cooling capacity will mix up the pump frequency conversion delay, chiller loading delay, and terminal valve action.

[0063] The preferred engineering implementation involves installing temperature sensors on the main supply and return water pipes of the chiller plant, a flow meter on the main chilled water pipe, and power metering devices on the power distribution circuits of the chiller, chilled water pumps, cooling water pumps, and cooling tower. The edge controller reads data via BACnet communication, Modbus communication, or equivalent industrial communication methods. Each data frame includes the device number, field name, acquisition time, value, unit, alarm flag, and source flag. When communication fails, the edge controller retains the value from the previous reliable sampling period and adds a hold flag. If communication is not restored for three consecutive sampling periods, the field becomes unavailable and is not involved in the generation of the current effective cooling load and current effective system power. This data frame format allows subsequent steps to determine whether a value is a newly acquired value, a hold value, or an unavailable value, avoiding the misinterpretation of communication interruption as equipment downtime.

[0064] The edge controller first aggregates the operating data of the water-cooled central air conditioning system. For the chilled water side, five categories of names are consistently used: chilled water supply temperature, chilled water return temperature, chilled water flow rate, chilled water supply and return pressure difference, and chilled water pump operating frequency. For the cooling water side, six categories of names are consistently used: cooling water supply temperature, cooling water return temperature, cooling water flow rate, cooling water pump operating frequency, cooling tower fan frequency, and outdoor wet-bulb temperature. For the chiller side, five categories of names are consistently used: chiller start / stop status, chiller load rate, chiller power, chiller energy efficiency coefficient, and chiller alarm status. These names are not changed after being generated in step one, and the same names are used in steps two through four.

[0065] For example, in a commercial complex, there are three chillers, four chilled water pumps, three cooling water pumps, and two cooling towers. The edge controller does not directly upload the original location names of each device. Instead, it unifies different location names such as CHWS_T1 and chilled water supply temperature into a single category called chilled water supply temperature, while retaining the source device number and acquisition time. The on-site operation of this implementation is as follows: maintenance personnel bind the physical meaning only once when importing the point table; subsequently, the edge controller outputs a fixed field, and the model reads the same field instead of multiple aliases.

[0066] By merging the above steps, the equipment operation, regional status, and grid restrictions are placed into the same data frame, so that the subsequent judgment on whether to enter pre-cooling or peak shaving has the same time basis.

[0067] After field merging is completed, the edge controller does not immediately calculate the cooling load. Instead, it generates data reliability coefficients for the chilled water supply temperature, chilled water return temperature, and chilled water flow rate involved in the calculation. The data reliability coefficient characterizes whether the data within the sampling period can represent the same physical process. The closer its value is to 1, the more suitable the sensor state, time synchronization, and physical trend are for participating in the prediction input in step two. Preferably, the data reliability coefficient consists of a sensor state coefficient, a time synchronization coefficient, and a physical consistency coefficient.

[0068]

[0069] Among them, the data credibility coefficient : Indicates the reliability of the cooling load calculation input within the current sampling period, with a value ranging from 0 to 1, used to determine whether the cooling load calculation result is directly written into the joint state variable; Sensor state coefficient : Indicates the working status of the thermometer and flow meter, with a value range of 0 to 1. When the sensor is normal and passes the range check, the value is close to 1; when the sensor alarms or exceeds the range, the value is 0.

[0070] Allowable time deviation : Indicates the maximum allowed timestamp lag by the edge controller, ranging from one-tenth to one-fifth of the sampling period; actual time deviation : Represents the difference between the current data timestamp and the master clock, with a value not less than 0; Physical consistency coefficient This indicates the consistency between the temperature difference between the chilled water supply and return water, the flow direction, and the operating status of the chiller. The value ranges from 0 to 1. When the chiller is running and the return water temperature is higher than the supply water temperature, a value close to 1 is used. When the chiller is stopped and the temperature difference suddenly increases, a lower value is used.

[0071]

[0072] Operating consistency coefficient This indicates the degree of consistency between the chiller operating status, the chilled water pump operating status, and the cooling capacity calculation scenario; when at least one chiller is operating and the chilled water pump frequency is not lower than the minimum allowable frequency, Take 1; when the chiller stops and the chilled water pump stops, Set to 0; when the chiller is stopped but the chilled water pump is in residual cooling circulation mode, Take a value between 0.5 and 1, the specific value of which is determined by the opening degree of the residual cooling circulation valve.

[0073] Temperature difference uniformity coefficient This indicates the reasonableness of the chilled water return temperature being higher than the chilled water supply temperature; when When the value is greater than 0 and does not exceed the upper limit of the design temperature difference, Take 1; when Less than or equal to hour, Take 0; when When the design temperature difference limit is exceeded, Reduced based on the ratio of the design temperature difference upper limit to the measured temperature difference. Flow consistency coefficient. This indicates whether the chilled water flow rate meets the chiller's minimum protection flow rate; when When the flow rate is not lower than the minimum protection flow rate of the chiller, Take 1; when When the flow rate is lower than the minimum protection flow rate of the chiller, according to The ratio to the minimum protection flow rate of the chiller is reduced.

[0074] For example, if the communication delay of a chilled water flow meter exceeds the allowable time deviation, the edge controller retains the flow value for trend display, but does not multiply the flow value with the current temperature difference to generate the current effective cooling load. Instead, it uses the flow hold value from the previous reliable sampling period with an added delay flag. If the chilled water supply temperature and chilled water return temperature show an inverse relationship, the edge controller first checks the chiller start / stop status and water pump operating frequency. When both the water pump and the chiller are stopped, the edge controller identifies the temperature difference as a mixed state of the stopped water and does not output it as a sudden load drop. This processing does not simply eliminate outliers, but simultaneously writes the sensor status, communication delay, and physical relationship into the data reliability coefficient, enabling step two to distinguish between real load fluctuations and acquisition chain anomalies.

[0075] In one implementation, the IoT cloud platform establishes range boundaries, rate of change boundaries, and physical consistency boundaries for each sampling field. If a sampled value exceeds the sensor's range, or if the change in a value within two adjacent sampling periods exceeds the allowable rate of change for that physical quantity, the sampled value is marked as an outlier and is not included in the calculation of the current effective cooling load and the current effective system power. When the duration of a missing value does not exceed three sampling periods, the previous reliable sampled value is used to maintain it; when the duration of a missing value exceeds three sampling periods but there are reliable sampled values ​​before and after it, linear interpolation is used; when the duration of a missing value exceeds a preset missing window, the field is written to an unavailable state. Time synchronization is based on the master clock, and any field's timestamp lags behind by more than the allowable time deviation. When this happens, the field is added to the delay marker queue. Normalization uses the device nameplate range or historical valid operating boundaries as the upper and lower limits of normalization. The same normalization boundary is used for the same field during the training, prediction, and control phases.

[0076] After the data reliability coefficient is generated, the edge controller calculates the current effective cooling load. The calculation uses chilled water flow rate, chilled water return temperature, and chilled water supply temperature, and the data reliability coefficient is written as the cooling capacity into the gate to prevent abnormal sampling from directly changing the peak-shifting cooling capacity account.

[0077]

[0078] Among them, the current effective cooling load : This represents the actual cooling capacity taken up by the building side from step one to step two, with a value not less than 0, used as the current benchmark for future cooling load forecasting; medium density. : Indicates the density of chilled water or ethylene glycol aqueous solution. The value range is determined by the medium detection results. For clean water conditions, it is preferably 990 kg / m³ to 1005 kg / m³.

[0079] Specific heat at constant pressure of medium This represents the amount of heat required to raise the temperature per unit mass of medium. The value range is determined by the type of medium; for clean water, it is preferably 4.0 kJ / kg°C to 4.3 kJ / kg°C; chilled water flow rate. : Indicates the volumetric flow rate through the evaporator or primary side main pipe during the current sampling period. The value range is greater than 0 and does not exceed the on-site flow metering range.

[0080] Chilled water return temperature This indicates the temperature entering the evaporator or the main return water pipe of the chiller plant. The range of this value is jointly defined by the chiller manufacturer's allowable range and the system's set range; chilled water supply temperature. This indicates the temperature leaving the evaporator or the main water supply pipe of the chiller plant. The range of values ​​is jointly limited by the chiller manufacturer's allowable range and the cooling requirements of the terminal coils; data reliability coefficient. The meaning, range of values, and function of are the same as those defined above.

[0081] Furthermore, the edge controller also calculates the current effective system power as a benchmark for demand control and power envelope generation:

[0082]

[0083] Among them, the current effective system power This represents the total power consumption of the water-cooled central air conditioning system during the current sampling period, with a value not less than 0, used for comparison with the demand threshold and the target power limit; chiller power. : Represents the sum of electrical power of all operating chillers, with a value not less than 0 and not exceeding the sum of the rated power of the chiller distribution circuits; chilled water pump power. This represents the sum of the electrical power of all operating chilled water pumps, with a value not less than 0 and not exceeding the sum of the rated power of the chilled water pump distribution circuits; cooling water pump power. : Represents the sum of the electrical power of all operating cooling water pumps, with a value not less than 0 and not exceeding the sum of the rated power of the cooling water pump power distribution circuits; cooling tower fan power. : Represents the sum of the electrical power of all operating cooling tower fans, with a value range of not less than 0 and not exceeding the sum of the rated power of the cooling tower fan power distribution circuit.

[0084] Furthermore, the edge controllers generate joint state variables:

[0085]

[0086] Among them, joint state variables : Represents the unique state object delivered from step one to step two, used to carry all the inputs required for subsequent load forecasting and peak-shifting cooling capacity calculation; current effective cooling load. The meaning, range of values, and function of the current effective system power follow the aforementioned definitions; The meaning, range of values, and function of are the same as those defined above;

[0087] Area Status Group This represents the set of factors including zone indoor temperature, zone indoor humidity, zone carbon dioxide concentration, zone personnel density, zone temperature setpoint, terminal valve opening, zone importance level, and zone comfort allowable range. The value range is jointly limited by the sensor range and system configuration.

[0088] Equipment Status Group This represents the set of chiller start / stop status, chiller load rate, chiller energy efficiency coefficient, chiller alarm status, water pump operating frequency, and cooling tower fan frequency. The value range is limited by feedback from the equipment controller.

[0089] Power grid constraint group : Represents the set of time-of-use pricing, peak pricing, real-time pricing, demand threshold, target power reduction, and target power ceiling. The value range is provided by the system or demand response interface.

[0090] Cold storage state group : This represents the set of cold storage water temperature, cold storage water volume, inlet and outlet water temperatures of the cold storage device, remaining ice volume of the ice storage, cold storage status, cooling power, and remaining releaseable cold energy. When no physical cold storage device is configured, an empty set is used as the marker. This is used to ensure that the same interface is used in both systems with and without cold storage in step two.

[0091] For example, in an office building without physical cold storage devices, the cold storage state group Write the empty set marker, joint state variable The output remains complete; step two calculates only the building's thermal inertia-released cooling capacity and the inherent cooling capacity of the chilled water network. In regional chilled stations equipped with chilled water storage tanks, the cooling capacity is grouped into... Step two involves recording the chilled water temperature, chilled water volume, and cooling capacity, and then incorporating the release capacity of the physical chilled water storage device into the peak-shifting cooling capacity account. This parallel implementation method allows the same set of steps one to be adapted for both existing building renovations and new chilled water storage stations, rather than being limited to a single chilled water storage device.

[0092] Furthermore, Step 1 unifies the previously scattered field data from chillers, water pumps, cooling towers, and the system into fixed fields, eliminating input ambiguity caused by multiple names for the same physical quantity. Step 1 generates a data reliability coefficient before calculating the cooling load, ensuring that communication delays, sensor alarms, and physical trend conflicts are not mistakenly interpreted as actual load changes.

[0093] Step 2: The IoT cloud platform bases its decisions on the joint state variables. Projecting future loads and including the building thermal inertia, chilled water network water volume, and physical cold storage devices that can support peak shaving in the same peak-shaving cold capacity account. This is for use in step three.

[0094] The peak-shaving capacity of a water-cooled central air conditioning system is not equivalent to its current power output. If the office area has already been pre-cooled, the interior walls, floors, furniture, and air can absorb external heat for a period of time. If the water temperature in the chilled water network is lower than the allowable cooling reference temperature, the network water itself can also release some cooling capacity before the return water at the terminals heats up. If the system is equipped with a chilled water storage tank or ice storage tank, the chilled water storage device can also handle a portion of the terminal load. The physical location, response speed, and comfort boundary of these three types of cooling sources are different. Treating them separately will result in double-counting or omission of peak-shaving capacity.

[0095] Step two first identifies future loads using the same forecast time domain, then uses the same account rules to convert the three types of cooling capacity into available balances and provides a degree of confidence, so that subsequent power curves are no longer set based on experience.

[0096] IoT cloud platform receives joint state variables Then, historical state segments are extracted according to time windows. These historical state segments consist of the current effective cooling load from the most recent several sampling periods. Current effective system power Regional Status Group Equipment Status Group and power grid constraint group The system is structured by adding outdoor weather forecasts, date type, and future electricity prices as external quantities. The IoT cloud platform outputs a predicted cooling load sequence within the same prediction time domain. Predicting indoor temperature sequences Predicting baseline power sequences and account availability coefficient Among them, the predicted cooling load sequence This indicates how much cooling capacity will be needed at the terminal in the future, and predicts the indoor temperature series. Explain the comfort margins for each region and predict the baseline power sequence. Explain the typical chiller plant power and account availability factor. Indicate whether the account balance can be released.

[0097] The IoT cloud platform does not write the prediction model as an isolated mathematical black box, but instead uses joint state variables. The field inputs are grouped according to their physical meaning. Chilled water supply temperature, chilled water return temperature, and chilled water flow rate are entered into the load-side input group; chiller start / stop status, chiller load rate, water pump frequency, and cooling tower fan frequency are entered into the equipment-side input group; area indoor temperature, area temperature setpoint, terminal valve opening degree, and area importance level are entered into the building-side input group; time-of-use electricity price, demand threshold, and response time are entered into the grid-side input group.

[0098] In a preferred embodiment, the prediction model employs a combination of a gradient boosting tree model and a first-order thermal inertia correction term: the former identifies the impact of weekdays, weather, population density, and terminal opening on the load, while the latter incorporates the current effective cooling load. The difference between the model's back-calculated values ​​and future fragments is propagated. The calculation relationship is as follows:

[0099]

[0100] Where: Predicted cooling load : indicates the first The estimated cooling load for each future control segment, with a value not less than 0 and not exceeding the rated cooling capacity of the chiller plant, is used as the peak-shifting cooling capacity account. Consumption reference; load function : Indicates a fragment of historical state The cold load mapping function obtained through training preferably adopts a gradient boosting tree model, and the training samples are historical joint state variables. Compared with the measured cooling load, it is used to output an uncorrected cooling load estimate;

[0101] Historical state fragments : indicates the joint state quantity The continuous input segment extracted from the text, with values ​​ranging from the most recent... Each sampling period preferably covers the time from the action of the end valve to the response of the return water temperature, in order to retain the antecedent of load changes; future segment sequence number. : Indicates the segment location within the predicted time domain, with a value ranging from 1 to , used to distinguish different future moments;

[0102] Load correction factor : Indicates the current deviation towards the first The proportion of each future segment transmitted, ranging from 0 to 1, depends on the future segment number. It increases and decreases sequentially to prevent the current instantaneous deviation from being amplified indefinitely;

[0103]

[0104] Load correction factor Indicates the current load deviation towards the first The proportion of future segments transmitted, ranging from 0 to 1; controls the segment duration. Indicates the time length between adjacent control segments. Load response time constant. This represents the equivalent lag time from a change in end load to a change in chilled water return temperature in a building and water system, and the preferred value is... to It is obtained by fitting the hysteresis relationship between the change in the opening degree of the terminal valve and the change in the return water temperature in the historical operation data.

[0105] Current effective cooling load : Represents the current load baseline output in step one, with a value range of not less than And it does not exceed the rated cooling capacity of the chiller plant, and is used to incorporate the actual cooling capacity taken on site into the prediction results.

[0106] In one implementation, the IoT cloud platform uses continuous historical state fragments. Using the measured cooling load of future control segments as input samples, the load function is trained. . From the recent Current effective cooling load within each sampling period Current effective system power Regional Status Group Equipment Status Group and power grid constraint group Composition. Training samples are labeled according to weekdays, non-weekdays, peak electricity hours, average electricity hours, and off-peak electricity hours. Each regression tree in the gradient boosting tree model uses the load residual as the fitting target, with a tree depth of 3 to 8 layers, a number of trees of 50 to 500, and a learning rate of 0.01 to 0.2. After training, the model outputs... Then, based on the load correction factor Current effective cooling load The deviation between the model's back-calculated values ​​and the predicted values ​​is propagated to future prediction segments.

[0107] If LSTM, GRU, or Transformer are used as alternative models, their inputs should be kept the same. The output remains the same predicted cooling load sequence. Predicting indoor temperature sequences and predicted baseline power sequence The subsequent account calculation and power envelope interface must not be changed.

[0108]

[0109] Number of regression trees The number of regression trees in the gradient boosting tree model is represented by a positive integer, and its function is to control the model's expressive power; learning weights. Indicates the first The output weights of each regression tree, with values ​​greater than 0 and not exceeding 1, are used to limit the influence of a single tree on the prediction results; regression tree function. Indicates the first A regression tree based on historical state fragments And future fragment number The output cooling load component has a value range of not less than 0; historical state segments. Represents the joint state variables of step one. Continuous segments; future segment sequence number This indicates the location of the segment within the predicted time domain.

[0110] For example, some chillers in a commercial complex were already running before the complex opened in the morning, and the terminal valves were open at a low degree, but the population density was about to increase. The IoT cloud platform detected this information from the joint state data. The data includes planned personnel density, historical door opening times, and current effective cooling load. Predicting cooling load sequences It initially rises slowly, then forms a higher load platform after people enter.

[0111] Furthermore, predict the cooling load sequence. At the same time, constrained by historical patterns and current on-site cooling capacity, the pre-cooling status will not be ignored due to schedule changes, nor will the upcoming personnel load be underestimated due to the current low load.

[0112] Obtain the predicted cooling load sequence Subsequently, the IoT cloud platform established a peak-shifting cooling capacity account. This account does not consider all low-temperature conditions as available cooling capacity; instead, it first checks the release boundaries of each type of cooling source. For building thermal inertia, the IoT cloud platform reads the regional indoor temperature, the regional comfort limit, and the regional importance level zone by zone. Only when the regional indoor temperature is below the regional comfort limit and the region is not a temperature-sensitive area is the remaining temperature rise space converted into available cooling capacity for building thermal inertia. For chilled water pipe networks, the IoT cloud platform reads the average temperature of the pipe network water and the pipe network volume. Only when the average temperature of the pipe network water is below the allowable cooling release reference temperature and the chilled water pump is in adjustable speed operation is the water temperature difference converted into the inherent cold storage capacity of the chilled water pipe network. For physical cold storage devices, the IoT cloud platform reads the cold storage status group. If the cold storage state group If the set is empty, then the entity's cold storage can release cold energy. If the cold storage state group If the stored chilled water temperature or remaining ice volume is included, it is converted into the physical cold storage release capacity based on the water outlet boundary. The aggregation relationship is as follows:

[0113]

[0114] Where: Account cooling capacity : Indicates the current available peak cooling capacity account The remaining cooling capacity, with a value not less than 0, is used to limit the upper limit of cooling capacity that can be used for peak shaving in step three; region number. : Indicates the air-conditioned zone number, with a value range of 1 to Used to calculate building thermal inertia zone by zone; total number of zones : Indicates the number of air-conditioned zones included in the account calculation, with a value range of positive integers, determined by the system point table;

[0115] Equivalent heat capacity of the region : indicates the first The comprehensive heat capacity of each region, taking a positive value, is used to convert temperature margin into cooling capacity; the equivalent heat capacity of the region. It is obtained by superimposing the equivalent heat capacity of the area envelope, indoor air, furniture and equipment; when the material details of the existing building are not available, it can be obtained by back-calculating the cooling capacity during the pre-cooling stage and the temperature drop of the area.

[0116] Maximum permissible area comfort : indicates the first The maximum permitted indoor temperature for each zone during peak shaving periods, with the range determined by the zone's intended use and comfort settings; zone indoor temperature : indicates the first The current indoor temperature of each area, with a value range limited by the temperature sensor's range, is used to determine the remaining temperature rise potential; positive truncation operator. : This means that the value inside the parentheses is retained when it is greater than 0, and is taken as 0 when it is less than or equal to 0. This is used to avoid including areas that have exceeded the comfort limit in the account.

[0117] Regional availability factor : indicates the first The permissible level of participation in peak shifting in each region is determined, ranging from 0 to 1, with higher values ​​for ordinary regions and 0 for temperature-sensitive regions; the regional availability coefficient. The value is determined based on the importance level of the area, population density, and temperature margin of the area; the value is higher for ordinary areas than for densely populated areas, and 0 for temperature-sensitive areas.

[0118] Dielectric density : This represents the density of chilled water or ethylene glycol aqueous solution. The value range is determined according to the type of medium and is used in the calculation of cooling capacity of water in the pipe network; Specific heat at constant pressure of the medium. : Represents the specific heat at constant pressure of chilled water or ethylene glycol aqueous solution. The range of values ​​is determined according to the medium type and is used in the conversion of cooling capacity of water in the pipe network; Pipe section number : Indicates the segment number of the chilled water pipe network or water system, with a value range from 1 to It is used to calculate the cooling capacity of the pipe network segment by segment;

[0119] Total number of pipe sections : Indicates the number of network segments included in the account calculation, with a value ranging from positive integers, determined by chilled water network zoning or hydraulic balance zoning; segment volume. : indicates the first The water volume of a section of the pipeline network or water system, taking a positive value, is determined from the as-built drawings or water system commissioning data; the reference temperature of the pipeline section. : indicates the first The reference temperature of the water body after cooling in the pipe section shall not exceed the upper limit of the allowable cooling water temperature at the end of the pipe section; water temperature of the pipe section : indicates the first The current average temperature of the water body in the section is determined by the range of the pipeline temperature sensor.

[0120] Pipeline availability factor : indicates the first The permissible level of cooling capacity participation in peak shifting within a pipe section, ranging from 0 to 1, with a value of 0 when valves in that section are closed or flow is uncontrollable; pipe section availability coefficient. Determined based on valve status, chilled water pump automatic status, and pipe section flow status; 0 is used when pipe section valves are closed or pumps are manually locked. Physical cold storage capacity. : Indicates the amount of cold energy that can be released from the cold water tank or ice storage tank. The value range is not less than 0. If there is no physical cold storage device, take 0.

[0121] For example, in an office building without a cold storage device, before the midday peak electricity consumption, the IoT cloud platform reads the indoor temperature and terminal valve opening of each floor area. Ordinary office areas still have room for temperature rise, while meeting rooms are in use and have a higher area importance. The IoT cloud platform writes the thermal inertia of some ordinary office areas into the account's cooling capacity. The availability factor of the meeting room area Take the lower value. For regional chilled water stations equipped with chilled water storage tanks, the outlet water temperature and chilled water volume of the storage tanks are recorded in the physical chilled water storage capacity. .

[0122] Furthermore, account coldness The source is clear, there is no duplicate metering of buildings, pipelines and cold storage devices, and temperature-sensitive areas are not used as peak shaving buffer zones.

[0123] Peak-shifting cooling capacity account Once formed, the IoT cloud platform does not directly allocate account cold storage. Instead of relying entirely on step three, this involves combining the predicted indoor temperature sequence. Predicting baseline power sequences and device status group Generate account availability coefficient Account availability factor The setting principle is: if the predicted indoor temperature sequence is within a certain future segment... If the area is nearing its comfort limit, reduce the account availability for the corresponding segment; if chilled water pumps, cooling towers, or cold storage valves are in an alarm or locked state, reduce the corresponding source of cooling capacity from the account's cooling capacity. Freeze; if the demand response start time has not yet arrived, the account availability factor will be frozen. Retain the account balance, but do not trigger a release.

[0124] Ultimately, a normalized account status is formed. :

[0125]

[0126] Where: Normalized account status : indicates the first The peak-shifting cooling capacity account in the future control segment The callability level, ranging from 0 to 1, is used as input for step three to determine the peak reduction intensity and duration; account availability coefficient. : indicates the first Account cooling capacity in a future control segment The available proportion, which is limited by comfort, equipment status and power grid time, ranges from 0 to 1 and is used to freeze cold energy that cannot be safely released;

[0127]

[0128] Account comfort level Indicates the first The temperature margin in each region within a future segment restricts the account's release, with values ​​ranging from 0 to 1; the account equipment coefficient. Indicates whether the chilled water pump, cold storage valve, chiller, and cooling tower are in an executable state, with a value ranging from 0 to 1; Account Power Grid Coefficient Indicates the first Whether a future segment is within a peak power, demand control, or demand response release window, with a value ranging from 0 to 1;

[0129] Account comfort rating can be calculated using:

[0130]

[0131] Predicted regional temperature Indicates the first The region in the first Predicted temperatures for a future segment; upper limits for regional comfort. Indicates the highest permissible temperature during peak shaving; lower limit of permissible regional comfort. This indicates the lowest temperature that is allowed to be reached during precooling;

[0132] Minimum value operator This indicates that the minimum comfort margin is applied to all areas included in peak shaving, and its purpose is to prevent a critical area from exceeding its limits prematurely. Account Cooling Capacity The meaning, range of values, and function of the upper limit of cold capacity for accounts follow the aforementioned definitions; : Indicates the peak-shifting cooling capacity account The maximum cooling capacity allowed to be recorded under the current engineering configuration, with a positive value range, is determined by the building area, pipe network volume, and cold storage device capacity, and is used to convert the account balances of different projects to the same scale.

[0133] In one implementation, a building freeze marker is written as frozen when any of the following conditions are met: predicted area temperature. The upper limit of regional comfort has been reached. The area is classified as a temperature-sensitive area; the population density in the area exceeds the population density threshold and the terminal valves are nearly fully open; the network freeze indicator is set to freeze when any of the following conditions occur: the chilled water pump is in manual lockout; the pipe section valve is closed; the water temperature in the pipe section is not lower than the reference temperature of the pipe section; the chilled water flow rate is lower than the minimum protection flow rate of the chiller; the cold storage freeze indicator is set to freeze when any of the following conditions occur: the cold storage device's cold release valve is faulty; the remaining cold storage device has zero cold capacity; the cold storage device's branch pump is shut down; the cold storage device is in maintenance lockout.

[0134] Subsequently, the IoT cloud platform generated the predicted cooling load sequence. Predicting indoor temperature sequences Predicting baseline power sequences Account coldness Normalized account status The frozen source marker is output to step three. The frozen source marker uses fixed fields: building frozen marker, pipeline frozen marker, and cold storage frozen marker correspond to three types of cooling sources, and the field value is available or frozen.

[0135] For example, if the chilled water pump is in a manually locked state, the IoT cloud platform retains the network water temperature record but marks the network as frozen. Therefore, step three no longer calls the inherent chilled water network storage capacity when generating the target power curve. Consequently, step three receives an input object with source boundaries, availability, and freezing status, thus enabling peak shaving actions to be applied to the specific control of the chiller, water pump, cooling tower, zone temperature setpoint, and chilled water storage valve.

[0136] When using it, predict the cooling load sequence. Based on historical conditions and current effective cooling load This, combined with the status of on-site equipment, reduces the risk of directly interpreting schedule changes or short-term data collection deviations as load trends. (Account cooling capacity) Including building thermal inertia, inherent cold storage capacity of chilled water pipe network, and physical cold storage capacity. Unified aggregation, and based on regional availability coefficients Pipe section availability factor Avoid ignoring comfort and hydraulic boundaries.

[0137] Step 3: Based on the load forecast results from Step 2 and the peak-shifting cooling capacity account, the IoT cloud platform generates a target power curve that does not exceed the comfort boundary, equipment safety boundary, and grid power boundary, and converts this target power curve into a set of control commands that can be sent to Step 4. .

[0138] Peak shaving for central air conditioning cannot be determined solely based on the current power level, because a high current power level does not necessarily mean a high power reduction is possible. If the indoor temperature in a region is already close to the upper limit of comfort, further increasing the chilled water supply temperature will cause the temperature in the terminal area to exceed the limit. If the source of the chilled water network freezes is marked as frozen, even if the water temperature in the network is low, this portion of the cooling capacity cannot be used for peak shaving. If the physical cold storage device is in a locked state with the cold release valve, the remaining cold storage capacity cannot be called up by the target power curve.

[0139] Therefore, step three will take the account coldness output from step two. and normalized account status Instead of directly converting peak power periods or demand response commands into shutdown orders, this approach uses a peak-shaving cap. This allows for controlled and coordinated equipment load reduction, regional heating, pipeline cooling, and cold storage cooling under the same power curve, preventing hydraulic imbalance or rebound after peak shaving caused by the action of a single piece of equipment.

[0140] The IoT cloud platform first reads the predicted baseline power sequence. Based on the power upper limit in the power grid constraint group, construct the allowable power boundary for future control segments; then read the predicted indoor temperature sequence. Freeze source tags and device status groups to construct minimum operable power; then, based on account cooling capacity... With normalized account status Calculate the power reduction envelope; finally, solve for the control instruction set within the envelope. The control cycle is preferably 5 to 15 minutes, and the prediction time domain preferably covers 2 to 24 hours. This range can encompass both peak and off-peak electricity price switching and demand response duration, while also avoiding frequent changes to chiller start-up and shutdown combinations by the field controller within a short period. The IoT cloud platform can employ a mixed-integer quadratic programming solver, a sequential quadratic programming solver, a dynamic programming enumeration solver, or a particle swarm search solver. It is preferable to first enumerate the chiller start-up and shutdown combinations, and then perform quadratic programming on the continuous variables, ensuring that both discrete equipment selection and continuous setpoint adjustments produce definite outputs.

[0141] Minimum operating power Defined as:

[0142]

[0143] Minimum operating power Indicates the first A future fragment in the feasible control set Minimum system power within; feasible control set This includes chilled water temperature boundaries, chilled water flow rate boundaries, cooling water temperature boundaries, minimum chiller start-up time, minimum chiller shutdown time, chiller start-up / shutdown frequency, zone comfort boundaries, and chilled water storage capacity boundaries; predicting the execution power function. Represents control variables Predicted system power after action.

[0144] The IoT cloud platform first puts the account cold volume The conversion moves from remaining cooling capacity to a power reduction threshold. Instead of recalling all account cooling capacity at once, the conversion uses the normalized account status based on future control segments. The allocation is performed, and the freezing source marker is superimposed. If the building freezing marker is frozen, the corresponding part of the building's thermal inertia does not participate in the allocation; if the pipe network freezing marker is frozen, the inherent cold storage capacity of the chilled water pipe network does not participate in the allocation; if the cold storage is frozen, the cold release capacity of the physical cold storage device does not participate in the allocation.

[0145] Therefore, the target power curve only uses the cooling capacity that can be released on-site without exceeding the comfort boundary.

[0146]

[0147] In the formula: power envelope can be reduced : indicates the first A future control segment allows for the extraction of power from the predicted baseline power sequence. The power reduction, with a value not less than 0, is used to limit the reduction magnitude of the target power curve; the predicted baseline power sequence. : indicates the first The central air conditioning power when peak shifting and valley filling are not performed in a future control segment, with a value not less than 0, is used as the power baseline before reduction; minimum operable power. This represents the minimum power output under the comfort boundary, minimum chiller flow rate, cooling water temperature boundary, and equipment start-up / shutdown limits. Its value is not less than 0 and not higher than the predicted baseline power sequence. This is to prevent excessive reduction; power grid capacity limit. : indicates the first The demand threshold or upper limit of the demand response target power for each future control segment, with a value range provided by the system or demand response interface, is used to limit peak power; account cooling capacity. : Represents the remaining peak-shifting cooling capacity formed in step two, with a value range not less than 0, used to limit the peak-shaving sustainability;

[0148] Normalized account status : indicates the first The account callability level of each future control segment, ranging from 0 to 1, is used to control account cooldown. Allocated to future segments; Account release factor : Indicates the release ratio after the combined effect of the frozen source marker and the regional comfort margin, with a value ranging from 0 to 1, used to deduct cold sources that cannot be safely released; controls segment duration. : Indicates the time length between adjacent control commands, preferably ranging from 5 to 15 minutes, used to convert cooling capacity to a power scale; cooling-electricity conversion factor. : indicates the first The equivalent coefficient used when converting a reduction in cooling capacity to a reduction in electrical power in a future control segment. The coefficient, with a value greater than 0, is determined by the predicted chiller efficiency, pump power, and cooling tower power. This is used to avoid directly equating the remaining cooling capacity with electrical power. (Forward truncation operator) : This means that if the value inside the parentheses is greater than 0, the value is retained; if the value inside the parentheses is less than or equal to 0, it is taken as 0, which is used to avoid negative power reduction.

[0149] For example, before the afternoon peak power generation begins in the commercial complex, the IoT cloud platform reads the available cooling capacity for the general office area and the frozen status of the meeting area given in step two. The meeting area does not participate in the account release, while the general office area does. At the same time, the chilled water pumps are in automatic mode, and the pipe network frozen status is available. Therefore, the cooling capacity of the pipe network water enters the account release coefficient. .

[0150] Furthermore, the power envelope is generated by combining the actual available cooling capacity and equipment status, avoiding the miscalculation of peak-shaving capacity by including unadjustable areas, locked equipment, or stagnant water bodies.

[0151] After completing the power envelope, the IoT cloud platform transforms the electricity price status, demand response commands, and predicted load status into a target power curve. If the current period is a flat power period and there are no demand response commands, the target power curve closely approximates the predicted baseline power sequence. Only the total system energy consumption optimization space is retained; if the current power period is off-peak and the account cooling capacity for future peak power periods is insufficient, the target power curve allows for higher power during off-peak periods so that pre-cooling or cold storage can be performed in step four; if the current power period is peak or demand response period, the target power curve has a power envelope that can be reduced. The internal power should decrease, but not fall below the minimum operating power. .

[0152]

[0153] Where: target power curve : indicates the first The future control segment aims to track a power value for the central air conditioning system that is not less than the minimum operating power. Used as a set of subsequent control instructions Power target; minimum operational power The meaning, range of values, and function of the prediction baseline power sequence follow the aforementioned definitions; The meaning, range of values, and function follow the aforementioned definitions; it can reduce the power envelope. The meaning, range of values, and function of are the same as those defined above;

[0154] Expected power reduction : Represents the power reduction demand given by peak-valley pricing strategies, demand control strategies, or demand response instructions. Its value is not less than 0 and is used to express external peak-shaving targets; function : Indicates taking the smaller value among multiple inputs to achieve the desired power reduction. No more than the power envelope that can be reduced ;function : Indicates taking the larger value among multiple inputs to optimize the target power curve. Not less than the minimum operating power ;

[0155] In one implementation, when the first When a future segment is in the demand response period Reduce power to meet demand response targets; during demand control periods. Take the positive difference between the predicted baseline power and the demand threshold; when the power supply is flat and there are no power limitations. Set to 0; when it is during the off-peak electricity pre-cooling period, Take the negative reduction flag and switch from step three to the pre-cooling or cold storage command, without entering the peak reduction calculation.

[0156] For example, the management system issues a target power limit at noon, and the IoT cloud platform first reads the desired power reduction. Then, with a power reduction envelope Comparison; Target power curve when the account's cooling capacity is insufficient to support the entire reduction. Instead of forcibly reducing it, the actual responsive power is returned to the management system and included in the control instruction set. Cooling is retained in temperature-sensitive areas.

[0157] Furthermore, the target power curve is generated from the executable boundary, which can take into account both external power requirements and field comfort boundaries, reducing the risk of demand response execution exiting midway.

[0158] After the target power curve is formed, the IoT cloud platform decomposes the power target into chillers, chilled water pumps, cooling water pumps, cooling towers, zone temperature setpoints, and cold storage devices. A preferred implementation method is to first fix the zone temperature setpoints for temperature-sensitive areas, then allocate zone temperature setpoint offsets to ordinary areas; subsequently, based on the predicted cooling load sequence... and target power curve Select the chiller start / stop combination; finally, under this combination, adjust the chilled water supply temperature setting, chilled water flow setting, cooling water supply temperature setting, cooling tower fan frequency, cold storage power, and cold storage release power.

[0159] The chilled water supply temperature setting shall not be lower than the chiller's lower limit and not higher than the terminal heat exchange capacity's upper limit; the chilled water flow rate setting shall not be lower than the chiller's minimum protection flow rate; and the cooling water supply temperature setting shall not be lower than the sum of the outdoor wet-bulb temperature and the cooling tower's minimum approximation temperature difference.

[0160]

[0161] Where: objective function : Represents the control instruction set in the future prediction time domain The overall cost, with a value not less than 0, is used to select control commands that meet power targets, comfort boundaries, and equipment safety boundaries; the predicted time domain length... : Indicates the number of future control segments participating in the solution, with a value range of positive integers, used to cover off-peak electricity, peak electricity, or demand response processes; electricity cost weight : Indicates the electricity cost item in the objective function The weights in the formula, with values ​​not less than 0, are used to adjust the impact on electricity prices;

[0162] Electricity Price Series : indicates the first The electricity price for each future control segment, with a value not less than 0, is provided by the management system to guide off-peak electricity pre-cooling and peak electricity peak shaving; predicting the execution power. : Indicates the control instruction set The expected power of the central air conditioning system after operation, with a value not less than 0, is used to compare with the target power curve. Compare;

[0163] Controlling segment duration The meaning, range of values, and function of power tracking weights follow the previous definitions; : Represents the weight of the power deviation term, with a value not less than 0, used to limit the predicted execution power. Above the target power curve Positive deviation operator : This indicates that the value inside the parentheses is retained when it is greater than 0, and is taken as 0 when it is less than or equal to 0. It is used to focus on constraining the part that exceeds the target power.

[0164] Comfort Deviation This represents the sum of deviations from the allowable comfort range in predicted temperatures for all air-conditioned zones. The value must be at least 0 and is used to protect the indoor environment of the zone; start / stop penalty amount. : Represents the constraint quantity for the start-stop changes of chillers, water pumps, and cooling towers in adjacent control segments. Its value range is not less than 0, used to reduce frequent start-stops; setpoint change quantity. This represents the sum of changes in chilled water supply temperature, chilled water flow rate, cooling water supply temperature, and zone temperature setpoint. Its value is not less than 0 and is used to limit sudden changes in control commands; comfort weight. Start and stop weights and set value weight Both indicate that the corresponding constraints are in the objective function. The weights in the equation, with values ​​not less than 0, are used to ensure that comfort, equipment lifespan, and control stability participate in the same solution process.

[0165] In the preferred implementation, the IoT cloud platform runs on the local industrial control computer of the cooling plant or the energy server of the park, with inputs being the output objects of step two and the device status group of step one, and outputs being a set of control instructions. Control instruction set This includes chilled water supply temperature setpoints, chilled water flow rate setpoints, chilled water pump differential pressure setpoints, cooling water supply temperature setpoints, cooling water pump frequency, cooling tower fan frequency, chiller start / stop combinations, chiller load distribution ratio, zone temperature setpoint offset, cold storage device cold storage power, cold storage device cold release power, and mode markings. In office buildings without physical cold storage devices, the cold storage power and cold storage cold release power are marked as unavailable, and the IoT cloud platform only utilizes the building's thermal inertia and the inherent cold storage capacity of the chilled water network. In chiller plants equipped with ice storage tanks, the cold storage device's cold release power prioritizes handling a portion of the cooling capacity during peak power periods, and the chiller start / stop combinations are kept within the range required to meet minimum operating time.

[0166] Furthermore, the target power curve is broken down into specific set values ​​that the field actuators can identify. The algorithm results do not remain at the mathematical level, but are incorporated into the control object that can be executed by step four.

[0167] Preferably, step three employs a two-stage solution; the first stage enumerates chiller start-up and shutdown combinations that satisfy the minimum start-up time, minimum shutdown time, and start-up / shutdown count constraints, eliminating those that cannot meet the predicted cooling load. The second stage involves continuously solving for chilled water supply temperature, chilled water flow rate, cooling water supply temperature, cooling tower fan frequency, zone temperature setpoint offset, and cold storage power under the retained chiller combination. If the objective function is quadratic and the constraints are linear or piecewise linear, mixed integer quadratic programming is used. If the equipment performance curve is nonlinear, sequential quadratic programming is used and re-linearized based on actual feedback after each control segment. If the solution fails, the feasible solution of the previous control segment is used, and the zone temperature setpoint offset and cold storage power are reduced.

[0168] Furthermore, step three transforms the peak-shaving cooling capacity account into a scalable power envelope, ensuring that the peak-shaving intensity is constrained by the actual release of cooling capacity from buildings, piping networks, and cold storage devices. Step three generates a target power curve within the power envelope, ensuring consistency between external demand response commands, demand thresholds, and on-site comfort boundaries. The target power curve is then decomposed into a set of control commands. This enables the chiller, water pump, cooling tower, zone temperature setpoint, and cold storage device to operate in coordination according to the same power target, and provides clear input for the execution feedback and rebound suppression in step four.

[0169] Step 4: The IoT cloud platform and the field controller jointly transmit the control command set. This is converted into on-site equipment actions, and the account cooling capacity is adjusted based on the actual cooling load, actual system power, and actual indoor temperature. Normalized account status And the subsequent recovery process.

[0170] The control instruction set given in step three The planned values ​​fall within the prediction time domain. During on-site execution, they are affected by factors such as the minimum chiller start-up time, pump frequency converter response, terminal valve lag, chilled water storage valve stroke, and changes in the number of people in the area. If the IoT cloud platform only sends the chilled water supply temperature setpoint, chilled water flow setpoint, and area temperature setpoint offset all at once, the chiller station may temporarily reduce its power output. However, the temperature in some areas will rise along the building's thermal inertia, requiring centralized supplemental cooling from the chiller after peak shaving. The approach in step four is to... It is divided into three categories: executable instructions, observation instructions, and recovery instructions. The field controller first executes the executable instructions, the IoT cloud platform then determines the source of the deviation, and finally uses the recovery instructions to distribute the cooling load debt.

[0171] The following actions are performed by the IoT cloud platform and the field controller, respectively. The IoT cloud platform receives the control command set. Generate execution data frames Execute data frame This includes chilled water supply temperature setpoint, chilled water pump differential pressure setpoint, chilled water flow rate setpoint, cooling water supply temperature setpoint, cooling water pump frequency, cooling tower fan frequency, chiller start / stop combination, chiller load distribution ratio, zone temperature setpoint offset, cold storage device cold storage power, cold storage device cold release power, mode flag, execution time, frame number, and verification fields. After the field controller completes the verification field check, it writes the continuous setpoints into the proportional-integral control loop, writes the chiller start / stop combination into the chiller group control queue, and writes the cold storage device cold storage power and cold storage device cold release power into the cold storage valve control loop. If executing a data frame... If a data frame arrives repeatedly within a control segment, the field controller will only retain the first executed data frame that passes verification. If executing data frames If lost, the field controller retains the previously verified setpoint and returns a hold flag.

[0172] In one implementation, the data frame is executed. This includes frame sequence number, execution time, mode flag, device number, control field, target value, hold flag, freeze flag, and verification field; the field controller only executes data frames with incrementing frame sequence number and passed verification field; for repeated data frames arriving within the same control segment, only the first data frame that passes verification is retained; if an executed data frame is lost, the field controller retains the previous valid set value and returns a hold flag to the IoT cloud platform; when more than three control segments are lost consecutively, the field controller switches to safety control mode.

[0173] The IoT cloud platform issues execution data frames First, the control instruction set... Implement equipment safety gating. Equipment safety gating does not change the target power curve in step three, but rather rewrites individual execution instructions within the permissible action boundaries of the equipment. For example, a chiller start-stop combination requires shutting down a chiller, but if the chiller has not reached its minimum operating time, the IoT cloud platform retains the chiller's operating status and transfers the reduction amount to the chilled water pump differential pressure setpoint, the area temperature setpoint offset, or the cooling power released by the cold storage device; if the cooling tower fan is in manual mode, the cooling tower fan frequency is not written, and the cooling water supply temperature setpoint enters a hold state.

[0174] In the preferred embodiment, the field controller and the chiller controller are interlocked by hard contacts and communicate in parallel with each other. The hard contacts transmit start and stop permissions, and the communication commands transmit set values ​​and operation feedback. Building automation gateways or chiller plant local control cabinets with equivalent functions can perform this action.

[0175] For example, when the commercial complex is in peak power shaving mode, the IoT cloud platform sends execution data frames to increase the temperature setpoint offset of the general office area, increase the chilled water supply temperature setpoint, and limit the chilled water pump differential pressure setpoint. The field controller first adjusts the end area setting value, and then smoothly changes the chilled water side setting value.

[0176] To express the relationship between equipment safety gating and execution deviation, the IoT cloud platform generates an execution deviation gating value. :

[0177]

[0178] Where: Execution deviation gate quantity : indicates the first The degree to which the execution result within a control segment deviates from the predicted result and the target power curve, with a value not less than 0, is used to trigger account correction or safety control modes; cooling output. : Indicates that the field controller executes the control instruction set The actual cooling load supplied to the building, with a value not less than 0, is used to compare with the predicted cooling load sequence. Comparison; Predicting cooling load sequences Following the definition in step two, the value range is not less than 0, and it is used as a reference for cooling deviation; actual system power : Indicates the execution of a data frame After taking effect, the system power read by the electricity metering device will have a value not less than 0, which will be used to compare with the target power curve. contrast;

[0179] Target power curve Following the definition in step three, the value range is not less than the minimum operable power, and is used as a reference for power deviation; exceeding the limit in certain areas... : Represents the normalized cumulative value after the temperature of each air-conditioned zone exceeds the allowable comfort range for that zone. The value is not less than 0 and is used to incorporate comfort deviations into the execution judgment; Load weight Power weight Temperature weighting All represent the corresponding deviation percentage, with values ​​ranging from 0 to 1; stability constant : Indicates preventing positive numbers with a denominator of 0, with values ​​greater than 0 and less than the lower limit of the corresponding range; absolute value operator : Indicates the non-negative amplitude of the value within the parentheses, used to identify the deviation amplitude;

[0180] Furthermore, the control instruction set The solution first passes through equipment safety gating before entering the field loop, preventing the predicted solution from directly exceeding the chiller and water pump boundaries; deviation gating is executed. Cooling output Actual system power and the more limited the area Bind to the same trigger value.

[0181] Execute data frame After taking effect, the IoT cloud platform reads field feedback data at the end of each control segment. This field feedback data includes chilled water supply temperature, chilled water return temperature, chilled water flow rate, chiller power, chilled water pump power, cooling water pump power, cooling tower fan power, indoor temperature of each air-conditioned area, terminal valve opening, and cold storage device cooling power. The IoT cloud platform first determines the execution deviation threshold. Does it exceed the execution deviation threshold? If it does not exceed the execution deviation threshold, then deduct the account's cold storage. And normalize the account status Write the prediction input for the next round; if it exceeds the execution bias threshold, perform bias attribution.

[0182] Actual system power Above the target power curve And the more limited the area, the more limited the quantity. If the temperature is not high, first check the minimum operating time of the chiller and the manual status of the water pump; cooling output. Lower than the predicted cooling load sequence And the more limited the area, the more limited the quantity. When the temperature rises, the availability factor of the corresponding area decreases; the cooling power of the cold storage device has reached the capacity boundary of the cold storage device, while the actual system power... Still above the target power curve When the frozen storage is in progress, the frozen storage mark will be written as frozen in the next round.

[0183] Among them, account coldness The correction relationship is as follows:

[0184]

[0185] Where: Account cooling capacity : indicates the first At the start of each control segment, the remaining cooling capacity of the peak-shaving cooling capacity account, with a value not less than 0, is used to pass to steps two and three as the boundary for the next round of peak shaving; account cooling capacity. : indicates the first At the start of each control segment, the remaining cooling capacity in the peak-shifting cooling capacity account, with a value not less than 0, is used as the deduction benchmark; the account releases cooling capacity. : indicates the first The actual cooling capacity undertaken by buildings, pipe networks, and physical cold storage devices within a control segment, with a value not less than 0, is used to reflect the peak-shifting resources already utilized; temperature deduction factor. : Indicates that the area exceeds the limit Account coldness The reduced strength, with a value not less than 0; deviation deduction coefficient. : Indicates the execution deviation gating value Account coldness The reduction strength, with a value not less than 0; positive truncation operator : indicates that the result inside the parentheses is greater than The result is retained, and if the result within the parentheses is less than or equal to 0, it is taken as 0, to ensure the account's cooling capacity. Negative values ​​are not allowed.

[0186] Furthermore, account coldness The deduction is not only based on the planned release amount, but also takes into account the amount of funds released from the account. The area exceeds the limit and execution deviation gate quantity When the on-site comfort or equipment operation is inconsistent with the prediction, the target power curve will narrow in the next round to avoid continuing to use cooling capacity that can no longer be safely released.

[0187] When peak shaving mode or power point tracking mode ends, the IoT cloud platform does not directly restore the settings before step three, but instead first generates a cooling capacity debt. Cold energy debt The demand for cooling comes from the reduced cooling supply to buildings during peak shaving periods, the comfort recovery demand caused by the deviation of the regional temperature setpoint, and the replenishment demand caused by the release of cooling capacity by the cold storage device.

[0188] In terms of on-site actions, the system first cancels the temperature setpoint offset of the normal area, but does not immediately reduce the chilled water supply temperature setpoint back to the normal value; the chilled water pump differential pressure setpoint is increased segment by segment according to the power recovery slope; the chiller group control queue is adjusted according to the minimum chiller downtime and actual system power. Decide whether to add chillers. For chilled water stations equipped with chilled water storage tanks, the storage unit will prioritize entering the cooling-preservation state during the recovery phase, pending the actual system power... Below the grid power limit Cooling replenishment will be arranged later; for office buildings without physical cold storage devices, the recovery phase will be completed in stages by adjusting the chilled water supply temperature setpoint and the area temperature setpoint.

[0189] Among them, cold energy debt The generation and deduction relationship is as follows:

[0190]

[0191] Where: Cold energy debt : indicates the first At the start of each control segment, the cooling capacity shortfall still needs to be eliminated by restoring cooling supply, with a value not less than 0, used to determine the recovery speed; cooling capacity debt. : indicates the first The remaining cooling capacity gap that needs to be recovered at the start of each control segment, with a value not less than 0, is used as the recovery baseline; the predicted cooling load sequence. and cooling output The meaning, range of values, and function of the control segment duration follow the aforementioned definitions; : Indicates the time length between adjacent control segments, with a preferred value range of 5 to 15 minutes, used to convert the cooling gap into cooling capacity;

[0192] Cooling capacity : indicates the first Within each control segment, the power limit of the power grid shall not be exceeded. The completed cooling capacity under the given conditions, with a value not less than 0, is used to deduct the cooling capacity debt. Positive truncation operator The meaning, range of values, and function of are the same as those defined above;

[0193] Subsequently, the IoT cloud platform adopted the power limit recovery method. Limit the loading range during the recovery phase:

[0194]

[0195] Where: Upper limit of recovery power : indicates the first Each control segment allows the system to reach a maximum recovery power, with a value not less than 0, used to limit power rebound after peak shaving; grid power limit. : indicates the first The demand threshold or upper limit of the demand response target power for each control segment, with a value range provided by the management system, serves as the external power boundary for the recovery phase; actual system power. The meaning, range of values, and function of are the same as those defined above;

[0196] Recovery slope : indicates the first Each control segment allows for actual system power The rate of increase, with a value not less than 0, is used to account for the amount of cold energy debt. Distributed to subsequent control segments; control segment duration The meaning, range of values, and function of the minimum function follow the previous definition; This indicates that the smaller value in the input is used to ensure the upper limit of the recovery power. Not exceeding the grid power limit ;

[0197] Furthermore, after peak shaving, the amount of cold energy debt will recover. and recovery power limit It is jointly decided that the chiller, water pumps, cooling tower, and terminal areas will not simultaneously return to high-load conditions; both cold storage systems and non-cold storage systems can use the same recovery logic. During use, the control instruction set... Safety gating is performed before entering the equipment, and the field controller receives executable execution data frames. Account coldness After execution, deductions are made based on on-site feedback, and the account status is normalized. No longer stuck on predicted values. The recovery process after peak shaving is affected by cold energy debt. and recovery power limit Segmented constraints reduce power rebound caused by concentrated cooling and keep temperature recovery synchronized with power boundaries.

[0198] The algorithm models in this method all run on an IoT cloud platform, which is connected to the controller; Step 1 outputs the joint state variables. Step Two Historical fragments As input, a gradient boosting tree model and a first-order thermal inertia correction term are used to output the predicted cooling load sequence. Predicting indoor temperature sequences and predicted baseline power sequence Step two further generates the accounted cooling capacity based on the regional temperature margin, chilled water network temperature, and the status of the cold storage device. Normalized account status And freeze the source mark; Step 3 is based on , Freeze source markers and device feasible control sets to generate a power reduction envelope. and target power curve And optimize the objective function by rolling. Solving the control instruction set Step four will Write it into the field controller and adjust it according to the actual cooling output. Actual system power and the more limited the area Correct account coldness Cold energy debt and recovery power limit .

[0199] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0200] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0201] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of 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 system, 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; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0202] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0203] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology 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 method for energy-saving intelligent control and peak-shaving / valley-filling coordinated operation of water-cooled central air conditioning, executed by a cooling plant IoT cloud platform, characterized in that: include: Acquire operational data of chiller plant equipment, terminal areas, chilled water storage status, and grid interfaces; calculate current cooling load and current system power; and generate joint state variables. Based on the joint state variables, the future cooling load, regional temperature and baseline power are predicted, and the building thermal inertia cooling capacity, the inherent cooling capacity of the chilled water network and the physical cooling capacity are summarized into a peak-shifting cooling capacity account. When no physical cooling capacity is configured, the physical cooling capacity is taken as zero. Generate a reduceable power envelope and target power curve based on the peak-shifting cooling capacity account, comfort boundary, equipment safety boundary, and power limitation boundary; Based on the target power curve, an execution command frame containing chilled water temperature, water pump differential pressure, cooling water temperature, cooling tower frequency, chiller combination, area temperature deviation, and cold storage power is generated. The peak-shifting cold capacity account is then corrected according to the actual cooling load, actual system power, and actual area temperature to form a cold capacity debt and a recovery power limit. The generation of the power reduction envelope includes: forming the equipment derating boundary based on the difference between the baseline power and the minimum operable power; forming the power constraint boundary based on the difference between the baseline power and the power limit boundary; and forming the cooling capacity support boundary based on the account balance field of the peak-shifting cooling capacity account. The restricted values ​​in the equipment load reduction boundary, power constraint boundary, and cooling capacity support boundary are written into the reduceable power envelope, and the target power curve is constrained by the reduceable power envelope. The formation of cooling load debt includes: during the execution of the target power curve, the accumulation of cooling load deficit based on the positive difference between the predicted cooling load and the actual cooling load; and deducting the account balance field of the peak-shifting cooling load account based on the cooling load deficit. After the target power curve is completed, the upper limit of the recovery power is formed based on the cooling capacity gap, the actual system power, and the power limit boundary, and the upper limit of the recovery power is written into the constraint field of the next control cycle.

2. The energy-saving intelligent control and peak-shifting and valley-filling coordinated operation method according to claim 1, characterized in that: The formation of the peak-shifting cooling capacity account includes: The building's thermal inertia cooling capacity is determined based on the regional comfort upper limit, regional temperature, regional equivalent heat capacity, and regional availability factor in the comfort boundary; the inherent cooling capacity of the chilled water network is determined based on the chilled water network volume, network water temperature, cooling reference water temperature, and network availability factor; and the physical cooling capacity is determined based on the cooling water temperature, cooling water volume, remaining ice volume, and cooling branch status. Write the building’s thermal inertia, chilled water network’s inherent cold storage capacity, and physical cold storage capacity into the same account balance field.

3. The energy-saving intelligent control and peak-shifting and valley-filling coordinated operation method according to claim 1, characterized in that: Corrections to the peak-shifting cooling capacity account include security interception: When the actual zone temperature exceeds the comfort boundary, cancel the available field for the zone temperature offset and reduce the building thermal inertia cooling capacity; when the chilled water flow rate is lower than the chiller protection flow rate in the equipment safety boundary, freeze the derating field for the chiller combination and water pump differential pressure. When the actual system power does not exceed the target power curve and there are no instances of actual regional temperature exceeding the limit or chilled water flow exceeding the limit, the current account balance field of the peak-shifting cooling capacity account is used.

4. The energy-saving intelligent control and peak-shifting and valley-filling coordinated operation method according to claim 1, characterized in that: When the cold storage status characterization does not have physical cold storage equipment, the physical cold storage capacity is zero, and the peak-shifting cold capacity account is formed by the building's thermal inertia cold capacity and the inherent cold storage capacity of the chilled water network. Write a zero value to the cold storage power field in the execution instruction frame; The target power curve forms the field control object through fields corresponding to chilled water temperature, water pump differential pressure, chiller combination, and zone temperature offset.

5. The energy-saving intelligent control and peak-shifting and valley-filling coordinated operation method according to claim 1, characterized in that: When the cold storage state characterizes the configuration of physical cold storage equipment, the physical cold storage capacity is formed based on the cold storage water temperature, cold storage water volume, remaining ice volume, and the status of the cold storage branch. When the target power curve is in the peak clipping section, the cold storage power field is written to the cold release value; When the target power curve is in the recovery phase, the cold storage power field is written to the cold preservation value, and the supplementary cooling command is limited according to the upper limit of the recovery power.

6. The energy-saving intelligent control and peak-shifting and valley-filling coordinated operation method according to claim 1, characterized in that: The mode labeling of the target power curve is formed according to the following priority: When a grid demand response command is received, a power tracking mode flag is written; when no grid demand response command is received, the off-peak electricity state is established, and the available peak-shaving cooling capacity account is lower than the future peak-shaving demand, a pre-cooling storage mode flag is written; when neither the power tracking mode flag nor the pre-cooling storage mode flag is written, and at least one of the peak electricity state or the demand close state is established, a peak-shaving mode flag is written. If none of the above mode flags are written, write the normal energy-saving mode flag.

7. The energy-saving intelligent control and peak-shifting and valley-filling coordinated operation method according to claim 6, characterized in that: The execution instruction frame includes a frame sequence number field, an execution time field, a mode flag field, a target power field, a chilled water temperature field, a water pump differential pressure field, a cooling water temperature field, a cooling tower frequency field, a chiller combination field, a zone temperature offset field, a cold storage power field, a peak-shifting cold capacity account field, a cold capacity debt field, a recovery power limit field, and a verification field; the execution instruction frame is formed in each control cycle.

8. The energy-saving intelligent control and peak-shifting and valley-filling coordinated operation method according to claim 7, characterized in that: The cooling plant IoT cloud platform writes the execution instruction frames into the operation log. The operation log includes fields such as frame sequence number, execution time, mode flag, target power, peak-shifting cooling capacity account, cooling capacity debt, recovery power limit, and actual system power. When the actual system power field is higher than the target power field, the operation log records and writes a power deviation flag. When the actual system power field is not higher than the target power field, the operation log records the power tracking flag.

9. The energy-saving intelligent regulation and peak-shifting and valley-filling coordinated operation method according to claim 8, characterized in that: The cooling plant IoT cloud platform generates monitoring interface status data, which includes the mode flag field, target power field, actual system power field, peak-shifting cooling capacity account field, cooling capacity debt field, recovery power limit field, and safety control mode flag field. When a valid value is written to the safety control mode flag field, the monitoring interface status data retains the previous valid target power field and refreshes the safety control mode flag field. When an invalid value is written to the safety control mode flag field, the target power field of the monitoring interface status data is refreshed.

Images