Industrial air conditioning unit flexible grid-connected control method and system based on power prediction

Abstract

The application provides a power prediction-based flexible grid-connected control method and system for an industrial air conditioning unit, which comprises collecting unit operation data, plant environment data, grid flexible regulation demand and production plan scheduling data; through the fusion of the production plan and a thermal inertia model, multi-time-scale power prediction is performed to generate a basic power prediction curve and an adjustable potential curve; multi-objective rolling optimization is performed with the tracking of grid demand as the main target and the production process environment requirement as a hard constraint, combined with the thermal inertia model to generate a refined control instruction sequence; the instruction is safely executed and real-time monitoring is performed, and local protection is started in an abnormal situation; finally, effect evaluation and model self-learning are performed, which relates to the technical fields of industrial energy saving and smart grid, wherein the industrial air conditioning unit is transformed into a flexible grid-connected resource that can be accurately predicted and flexibly regulated, and safe, efficient and standardized interaction with the grid is realized on the premise of strict guarantee of the industrial production environment.

Description

Technical Field

[0001] This invention relates to the field of industrial energy conservation and smart grid technology, and in particular to a flexible grid-connected control method and system for industrial air conditioning units based on power prediction. Background Technology

[0002] With the deepening of the construction of new power systems, the demand for flexible load resources in the power grid is becoming increasingly urgent. Industrial air conditioning, as a typical temperature-controlled load, is characterized by high power output and large adjustability potential, making it an ideal flexible load resource. However, existing technologies have significant shortcomings in practice:

[0003] Single control objective: Existing technologies mostly focus on local energy-saving control based on time-of-use pricing, only considering the economic benefits on the user side, without actively incorporating air conditioning load into the power grid dispatch system, and thus failing to respond to the flexible adjustment needs of the power grid such as peak shaving and frequency regulation.

[0004] Insufficient forecast accuracy: Some equipment attempts to forecast air conditioning load, but mainly relies on outdoor meteorological data and does not take into account production scheduling and plant thermal inertia, which play a decisive role in industrial load. This results in a serious disconnect between the forecast curve and grid dispatch requirements (such as adjustability potential and response speed).

[0005] Rigid control methods: Most control methods involve simple start / stop or setpoint adjustment, failing to consider the stringent requirements of the industrial environment. Aggressive power regulation can easily lead to excessive internal temperatures in the plant, affecting production process stability and product quality, and making it impossible to achieve flexible regulation while ensuring production safety.

[0006] Lack of systematic grid connection architecture: Virtual power plants aggregate air conditioning loads as black box resources, lack detailed characterization of the internal operating characteristics and safety constraints of individual industrial air conditioning units, issue general instructions, and lack reliable, standardized grid-load interaction interfaces and safety fallback mechanisms with local autonomy capabilities.

[0007] Therefore, there is an urgent need for a flexible grid-connected control technology for industrial air conditioning that can accurately predict, optimize decision-making, execute safely, and achieve standardized interaction with the power grid. Summary of the Invention

[0008] This invention aims to overcome the aforementioned deficiencies of the prior art and provide a flexible grid-connected control method and system for industrial air conditioning units. Its core objective is to transform industrial air conditioning units from traditional rigid electrical loads into a high-quality, flexible grid-connected resource that is predictable, plannable, precisely controllable, and safely interactive, achieving bidirectional and friendly interaction with the power grid while strictly ensuring the requirements of the industrial production environment.

[0009] The technical solution adopted by the embodiments of this application to solve its technical problem is:

[0010] A flexible grid-connected control method and system for industrial air conditioning units based on power prediction includes the following steps:

[0011] Step 1: Data Collection and Requirements Acquisition

[0012] Real-time collection of operational data from industrial air conditioning units and environmental data from the plant where they are located;

[0013] Step 2, Multi-timescale power prediction steps:

[0014] By integrating prediction models, the baseline power prediction curves and corresponding adjustable potential curves of industrial air conditioning units are generated for multiple time scales in the future.

[0015] Step 3: Multi-objective optimization decision-making steps:

[0016] With the optimization objectives of system operation economy and energy efficiency, and combined with the thermal inertia model of the plant, the basic power prediction curve is solved by rolling optimization to generate a sequence of refined control commands for the unit in the future.

[0017] Step 4: Safe Implementation and Interactive Feedback:

[0018] The control command sequence is sent to the industrial air conditioning unit for execution. When the risk of environmental parameters exceeding the limit or communication abnormality is detected, the local protection control strategy is executed first.

[0019] Step 5: Effect Evaluation and Model Self-Learning Steps:

[0020] After a complete power grid regulation service is completed, the deviation between the actual regulation effect and the expected target is compared and the reasons for the deviation are analyzed. Using the deviation data, the parameters of the plant thermal inertia model are automatically fine-tuned.

[0021] The plant thermal inertia model quantifies the dynamic relationship between internal temperature changes, air conditioning power, production heat load, and outdoor environment. Its core is characterized by the following thermal dynamic equation:

[0022] ;

[0023] In the formula, The equivalent heat capacity of the factory building The temperature inside the factory building. The cooling / heating power provided by the air conditioning system This refers to the heat exchange capacity between the factory and the outside environment. The internal heat load of the plant is determined by production planning and scheduling data.

[0024] Preferably, the adjustable potential curve in step two specifically includes:

[0025] Lower potential Indicates a future time window Maximum air conditioning power reduction:

[0026] ;

[0027] Upward potential Indicates a future time window Maximum increase in air conditioning power:

[0028] ;

[0029] In the formula, Based on the power prediction curve, This is the minimum operating power limit for the air conditioner. This is the maximum operating power limit for the air conditioner. The minimum safe temperature limit required by the production process. The maximum safe temperature limit required by the production process. For air conditioner energy efficiency ratio, As an empirical safety factor, For time window, To predict heat load.

[0030] The multi-timescale power prediction in step two includes day-ahead prediction, intraday rolling prediction, and real-time ultra-short-term prediction.

[0031] The day-ahead forecast is based on the production plan, weather forecast and time-of-use electricity price information for the next 24-48 hours, and uses a long short-term memory network model to predict the day-ahead base power curve of the unit.

[0032] The intraday rolling forecast uses a 4-hour cycle and, based on updated short-term meteorological data and actual load, performs rolling corrections on the day-ahead forecast curve to generate a more accurate intraday planning curve.

[0033] The real-time ultra-short-term forecast uses a 15-minute cycle and, based on the real-time operating status of the unit, employs a state-space model to predict the instantaneous power for the next 15-30 minutes, which is used for real-time control calibration.

[0034] Preferably, the plant thermal inertia model is used to quantify the plant's heat capacity and insulation characteristics to achieve time-shifted power adjustment, and its optimization process is as follows:

[0035] When it is necessary to reduce the power, increase the cooling capacity in advance within the allowable time window and store the cooling capacity in the building structure and air. When it is necessary to reduce the power, reduce the cooling capacity output and use the stored cooling capacity to maintain the ambient temperature.

[0036] When it is necessary to increase the power, reduce the cooling capacity in advance, allowing the temperature to rise reasonably, provided that the temperature does not exceed the upper limit. Then, start the unit to provide cooling capacity when it is necessary to increase the power.

[0037] Preferably, the rolling optimization solution in step three adopts a model predictive control framework. In each control cycle, based on the latest measurement value and prediction information, an optimization problem in a finite time domain is solved, and the first element of the optimization solution sequence is issued as the current control command.

[0038] Preferably, the local protection control strategy in step four includes: when the temperature and humidity parameters of the plant environment approach the preset safety limit, or when the communication interruption with the power grid exceeds the set threshold, the system switches to local autonomous mode and controls according to the preset safety operation curve to ensure the safety of the production process.

[0039] Preferably, the model self-learning in step five specifically includes: based on the input and output data of the actual adjustment process, using parameter identification or gradient descent method to perform online or offline correction on the equivalent heat capacity, heat loss coefficient and parameters of the fusion prediction model in the plant thermal inertia model.

[0040] It includes a data acquisition and communication module, a multi-scale power prediction module, a multi-objective optimization decision-making module, a safety control and execution module, and a system platform;

[0041] The data acquisition and communication module is used to collect operating data of industrial air conditioning units and environmental data of the plant, and to interact with the power grid dispatching system through standard communication protocols, receive flexible adjustment requirements, and send feedback information.

[0042] The multi-scale power prediction module is used to perform power prediction steps at multiple time scales to generate a basic power prediction curve and an adjustable potential curve.

[0043] The multi-objective optimization decision module has a built-in plant thermal inertia model, which is used to execute multi-objective optimization decision steps and generate a refined control instruction sequence.

[0044] The safety control and execution module is used to send control command sequences to the industrial air conditioning unit for execution, monitor key parameters, and execute local protection control strategies in abnormal situations.

[0045] The system platform is used to integrate and manage the operation of the above modules, provide a human-computer interaction interface, and perform parameter configuration, status monitoring and system management.

[0046] Preferably, the objective function of the multi-objective optimization decision module includes at least two of the following: minimizing the power grid regulation demand tracking deviation, minimizing the energy consumption cost of air conditioning system operation, and minimizing the frequency / amplitude of equipment actions.

[0047] Preferably, the safety control and execution module has instruction verification and hardware / software interlock functions to ensure that the issued control instructions are within the safe operating range of the equipment.

[0048] The advantages of the embodiments of this application are:

[0049] A flexible grid-connected technology system for industrial air conditioning that actively serves the power grid has been constructed, transforming it from a passive power-consuming device into an active resource that can participate in grid regulation, thereby improving the flexibility and stability of the power grid.

[0050] By integrating production planning with multi-scale prediction using thermal inertia models, the generated adjustable potential curve is accurate and reliable, enabling the power grid to plan and utilize this resource in advance, much like scheduling generators. This solves the problem of unknown and uncontrollable loads in existing technologies. At the same time, based on optimized control using thermal inertia models, it enables smooth and flexible adjustment of air conditioning power without affecting the core temperature and humidity requirements of the production process, thus resolving the contradiction between load regulation and production assurance in industrial scenarios. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the flexible grid-connected control method for industrial air conditioning units based on power prediction according to the present invention.

[0052] Figure 2 This is a schematic diagram of the flexible grid-connected control system architecture for industrial air conditioning units based on power prediction, as described in this invention.

[0053] Figure 3 This is a schematic diagram illustrating the principle of time-shift control of thermal inertia power in the industrial air conditioning unit flexible grid-connected control method and system based on power prediction, as described in this invention. Detailed Implementation

[0054] 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 a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention. In addition, for the sake of convenience, the terms "upper," "lower," "left," and "right" are equivalent to the upper, lower, left, and right directions of the accompanying drawings themselves, and the terms "first," "second," etc., are used for descriptive purposes and have no other special meaning.

[0055] This application provides a flexible grid-connected control method and system for industrial air conditioning units based on power prediction, which solves the problems in the prior art and constructs a flexible grid-connected technology system for industrial air conditioning that actively serves the power grid. This transforms the unit from a passive power-consuming device into an active resource that can participate in grid regulation, thereby improving the flexibility and stability of the power grid.

[0056] By integrating production planning with multi-scale prediction using thermal inertia models, the generated adjustable potential curve is accurate and reliable, enabling the power grid to plan and utilize this resource in advance, much like scheduling generators. This solves the problem of unknown and uncontrollable loads in existing technologies. At the same time, based on optimized control using thermal inertia models, it enables smooth and flexible adjustment of air conditioning power without affecting the core temperature and humidity requirements of the production process, thus resolving the contradiction between load regulation and production assurance in industrial scenarios.

[0057] The technical solution in this application is to solve the above problems, and the overall approach is as follows:

[0058] Example

[0059] This embodiment presents a flexible grid-connected control method and system for industrial air conditioning units based on power prediction, such as... Figure 1-3 As shown, it includes the following steps:

[0060] Step 1: Data Collection and Requirements Acquisition

[0061] Real-time acquisition of operating data of industrial air conditioning units, environmental data of the plant, and acquisition of multi-timescale flexible adjustment demand signals and factory production planning and scheduling data from the power grid side;

[0062] Furthermore, the system collects the following data in real time:

[0063] Unit operating data: Start-up and shutdown status, operating power, current, supply and return water temperature / pressure, compressor frequency, etc. of each chiller, water pump, and cooling tower are obtained through the unit's built-in controller or smart meter;

[0064] Factory environment data: Real-time indoor temperature and humidity data are collected through temperature and humidity sensors distributed in key process areas and representative areas of the factory; external temperature, humidity, solar radiation intensity and other data are obtained through weather stations.

[0065] Production planning data: Obtain the production schedule for the next 24 hours, the start and stop plans for each production line, and the equipment load rate forecast from the factory's MES system to calculate the internal heat load;

[0066] Receive multi-timescale flexible regulation demand signals from the power grid dispatching system through standard communication protocols, for example:

[0067] Pre-holiday demand: Expected load curves or adjustment capacity requests for the following day, issued one day in advance, for different time periods (such as peak, off-peak, and valley periods);

[0068] Intraday demand: Adjustment instructions issued several hours in advance, such as increasing or decreasing power during specific periods;

[0069] Real-time requirements: Automatic generation control (AGC) or fast frequency response (FFR) adjustment commands issued on a minute or second basis.

[0070] Step 2, Multi-timescale power prediction steps:

[0071] Based on production planning and scheduling data, environmental data, and historical operating data, a fusion prediction model is used to generate the basic power prediction curves and corresponding adjustable potential curves of industrial air conditioning units at multiple time scales in the future. The adjustable potential curves include upward and downward adjustment potential.

[0072] Among them, multi-timescale power forecasting includes day-ahead forecasting, intraday rolling forecasting, and real-time ultra-short-term forecasting;

[0073] The day-ahead forecast is based on production plans, weather forecasts and time-of-use electricity price information for the next 24-48 hours, and uses a long short-term memory network model to predict the day-ahead base power curve of the unit.

[0074] The intraday rolling forecast uses a 4-hour cycle and is based on updated short-term meteorological data and actual loads to continuously correct the day-ahead forecast curve, generating a more accurate intraday planning curve.

[0075] Real-time ultra-short-term forecasts are based on a 15-minute cycle and use a state-space model to predict the instantaneous power in the next 15-30 minutes, based on the real-time operating status of the unit, for real-time control calibration.

[0076] The adjustable potential curve specifically includes:

[0077] Lower potential Indicates a future time window Maximum air conditioning power reduction:

[0078] ;

[0079] Upward potential Indicates a future time window Maximum increase in air conditioning power:

[0080] ;

[0081] In the formula, Based on the power prediction curve, This is the minimum operating power limit for the air conditioner. This is the maximum operating power limit for the air conditioner. The minimum safe temperature limit required by the production process. The maximum safe temperature limit required by the production process. For air conditioner energy efficiency ratio, As an empirical safety factor, For time window, To predict heat load;

[0082] Furthermore, the system incorporates a fusion prediction model that combines a Long Short-Term Memory (LSTM) network with a state-space model.

[0083] Forecast (time scale: 24-48 hours): At a fixed time each day, based on the production plan for the next 48 hours, weather forecast (temperature, humidity, radiation), and historical air conditioning power data for the same period, the LSTM model is used to predict and generate the basic power curve of the air conditioning system for the next 24 hours, with a time resolution of 15 minutes.

[0084] Intraday rolling forecast (time scale: next 4 hours, cycle 4 hours): Every 4 hours, based on the latest short-term weather forecast, actual power data from the past few hours, and possibly updated production plans, the curve for the next 4 hours of the previous day's forecast is rolled over to generate a more accurate intraday planning curve.

[0085] Real-time ultra-short-term forecast (time scale: 15-30 minutes in the future, 15-minute cycle): Every 15 minutes, based on the current instantaneous operating status of the air conditioning system (such as compressor frequency and valve opening) and real-time indoor and outdoor temperature and humidity, the instantaneous power change trend in the next 15-30 minutes is predicted using a state-space model for real-time calibration of subsequent optimization control.

[0086] Step 3: Multi-objective optimization decision-making steps:

[0087] With the main objective of tracking the flexible regulation needs of the power grid, the hard constraint of meeting the temperature and humidity requirements of the production process environment, and the optimization objectives of system operation economy and energy efficiency, the basic power prediction curve is solved by rolling optimization in combination with the thermal inertia model of the plant, and the refined control command sequence of the unit for a period of time in the future is generated.

[0088] Among them, the rolling optimization solution adopts the model predictive control framework. In each control cycle, based on the latest measurement value and prediction information, an optimization problem in a finite time domain is solved, and the first element of the optimization solution sequence is issued as the current control command.

[0089] Step 4: Safe Implementation and Interactive Feedback:

[0090] The control command sequence is sent to the industrial air conditioning unit for execution; at the same time, the actual power of the unit, the plant environmental parameters and the communication status with the power grid are monitored in real time; the actual power and adjustable capacity information are fed back to the power grid side through the standard communication interface; when the risk of environmental parameters exceeding the limit or communication abnormality is detected, the local protection control strategy is executed first.

[0091] The local protection and control strategy includes: when the temperature and humidity parameters of the plant environment approach the preset safety limit, or when the communication interruption with the power grid exceeds the set threshold, the system switches to local autonomous mode and controls according to the preset safety operation curve to ensure the safety of the production process.

[0092] Furthermore, the safety control and execution module will optimize the control command sequence generated by the decision-making module and send it safely and reliably to the local controllers of each air conditioning subsystem for execution via the fieldbus.

[0093] Continuous system monitoring:

[0094] Actual power of the unit: fed back by smart meters and compared with the commanded value.

[0095] Factory environmental parameters: temperature and humidity at key measuring points to determine if they are close to safety limits.

[0096] Communication status: Whether the communication link with the power grid dispatching master station is normal.

[0097] Information such as the actual operating power of the air conditioning system and the current available upward / downward adjustment potential is fed back to the grid side in real time or periodically through a standard communication interface, providing adjustable resource information for grid dispatch.

[0098] When an anomaly is detected, the local protection and control strategy is immediately triggered. After triggering, the system automatically switches to local autonomous mode, ignores the grid regulation requirements, controls according to the pre-set basic operating curve that ensures process safety, and issues an alarm to the operation and maintenance personnel.

[0099] Step 5: Effect Evaluation and Model Self-Learning Steps:

[0100] After a complete power grid regulation service is completed, the deviation between the actual regulation effect and the expected target is compared and the reasons for the deviation are analyzed. Using the deviation data, the parameters of the plant thermal inertia model are automatically fine-tuned.

[0101] Specifically, this includes: using parameter identification or gradient descent methods to perform online or offline correction of the equivalent heat capacity, heat loss coefficient, and parameters of the fusion prediction model in the plant thermal inertia model based on the input and output data of the actual adjustment process.

[0102] The thermal inertia model of a factory building is used to quantify the utilization of the building's heat capacity and insulation characteristics to achieve time-shifted power regulation. The optimization process is as follows:

[0103] When it is necessary to reduce the power, increase the cooling capacity in advance within the allowable time window and store the cooling capacity in the building structure and air. When it is necessary to reduce the power, reduce the cooling capacity output and use the stored cooling capacity to maintain the ambient temperature.

[0104] When it is necessary to increase the power, reduce the cooling capacity in advance, allowing the temperature to rise reasonably, provided that the temperature does not exceed the upper limit. Then, start the unit to provide cooling capacity when it is necessary to increase the power.

[0105] The thermal inertia model of the factory building quantitatively describes the dynamic relationship between internal temperature changes, air conditioning power, production heat load, and outdoor environment. Its core is characterized by the following thermal dynamic equation:

[0106] ;

[0107] In the formula, The equivalent heat capacity of the factory building The temperature inside the factory building. The cooling / heating power provided by the air conditioning system This refers to the heat exchange capacity between the factory and the outside environment. The internal heat load of the plant is determined by production planning and scheduling data.

[0108] Furthermore, after a complete grid regulation service (such as a two-hour peak-shaving demand response), the system automatically generates an evaluation report. The report compares the deviation between the actual regulated power curve and the grid demand curve, and analyzes indicators such as total regulation, response speed, and tracking accuracy. Simultaneously, it compares the differences between actual plant temperature changes and model predictions.

[0109] Using the actual input (control commands, environmental data, production data) and output (actual power, actual temperature) data accumulated during this adjustment process, the key model parameters were fine-tuned:

[0110] By employing parameter identification algorithms such as recursive least squares, parameters such as equivalent heat capacity and heat loss coefficient in the model are fine-tuned online or offline, making the model's predicted temperature closer to reality.

[0111] By using new operational data, the parameters of the LSTM and state-space prediction models are periodically (e.g., weekly) trained and updated offline to improve future prediction accuracy.

[0112] It includes a data acquisition and communication module, a multi-scale power prediction module, a multi-objective optimization decision-making module, a safety control and execution module, and a system platform;

[0113] The data acquisition and communication module is used to collect operating data of industrial air conditioning units and environmental data of the plant, and to interact with the power grid dispatching system through standard communication protocols, receive flexible adjustment requirements, and send feedback information.

[0114] The multi-scale power prediction module is used to perform power prediction steps at multiple time scales, generating a basic power prediction curve and an adjustable potential curve.

[0115] The multi-objective optimization decision module has a built-in plant thermal inertia model, which is used to execute multi-objective optimization decision steps and generate a refined control instruction sequence;

[0116] The optimization objective function of the multi-objective optimization decision module includes at least two of the following: minimizing the power grid regulation demand tracking deviation, minimizing the energy consumption cost of air conditioning system operation, and minimizing the frequency / amplitude of equipment actions.

[0117] The safety control and execution module is used to send control command sequences to the industrial air conditioning unit for execution, monitor key parameters, and execute local protection control strategies in abnormal situations;

[0118] The safety control and execution module has instruction verification and hardware / software interlock functions to ensure that the issued control instructions are within the safe operating range of the equipment.

[0119] The system platform is used to integrate and manage the operation of the above modules, providing a human-computer interaction interface for parameter configuration, status monitoring and system management.

[0120] Through the implementation of the above specific embodiments, the system of the present invention can safely, economically and efficiently utilize the flexible adjustment capability of industrial air conditioning units, and reliably respond to the multi-timescale adjustment needs of the power grid while ensuring the requirements of the production process environment, thus realizing the friendly interaction between industrial load and power grid.

[0121] Finally, it should be noted that the above embodiments are merely examples for clearly illustrating the present invention and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A flexible grid-connected control method for industrial air conditioning units based on power prediction, characterized in that, Includes the following steps: Step 1: Data Collection and Requirements Acquisition Real-time collection of operational data from industrial air conditioning units and environmental data from the plant where they are located; Step 2, Multi-timescale power prediction steps: By integrating prediction models, the baseline power prediction curves and corresponding adjustable potential curves of industrial air conditioning units are generated for multiple time scales in the future. Step 3: Multi-objective optimization decision-making steps: With the optimization objectives of system operation economy and energy efficiency, and combined with the thermal inertia model of the plant, the basic power prediction curve is solved by rolling optimization to generate a sequence of refined control commands for the unit in the future. Step 4: Safe Implementation and Interactive Feedback: The control command sequence is sent to the industrial air conditioning unit for execution. When the risk of environmental parameters exceeding the limit or communication abnormality is detected, the local protection control strategy is executed first. Step 5: Effect Evaluation and Model Self-Learning Steps: After a complete power grid regulation service is completed, the deviation between the actual regulation effect and the expected target is compared and the reasons for the deviation are analyzed. Using the deviation data, the parameters of the plant thermal inertia model are automatically fine-tuned. The plant thermal inertia model quantifies the dynamic relationship between internal temperature changes, air conditioning power, production heat load, and outdoor environment. Its core is characterized by the following thermal dynamic equation: ； In the formula, The equivalent heat capacity of the factory building The temperature inside the factory building. The cooling / heating power provided by the air conditioning system This refers to the heat exchange capacity between the factory and the outside environment. The internal heat load of the plant is determined by production planning and scheduling data.

2. The flexible grid-connected control method for industrial air conditioning units based on power prediction according to claim 1, characterized in that, The adjustable potential curve in step two specifically includes: Lower potential Indicates a future time window Maximum air conditioning power reduction: ； Upward potential Indicates a future time window Maximum increase in air conditioning power: ； In the formula, Based on the power prediction curve, This is the minimum operating power limit for the air conditioner. This is the maximum operating power limit for the air conditioner. The minimum safe temperature limit required by the production process. The maximum safe temperature limit required by the production process. For air conditioner energy efficiency ratio, As an empirical safety factor, For time window, To predict heat load.

3. The flexible grid-connected control method for industrial air conditioning units based on power prediction according to claim 1, characterized in that, The multi-timescale power prediction in step two includes day-ahead prediction, intraday rolling prediction, and real-time ultra-short-term prediction. The day-ahead forecast is based on the production plan, weather forecast and time-of-use electricity price information for the next 24-48 hours, and uses a long short-term memory network model to predict the day-ahead base power curve of the unit. The intraday rolling forecast uses a 4-hour cycle and, based on updated short-term meteorological data and actual load, performs rolling corrections on the day-ahead forecast curve to generate a more accurate intraday planning curve. The real-time ultra-short-term forecast uses a 15-minute cycle and, based on the real-time operating status of the unit, employs a state-space model to predict the instantaneous power for the next 15-30 minutes, which is used for real-time control calibration.

4. The flexible grid-connected control method for industrial air conditioning units based on power prediction according to claim 1, characterized in that, The plant thermal inertia model is used to quantify the plant's heat capacity and insulation characteristics to achieve time-shifted power regulation. Its optimization process is as follows: When it is necessary to reduce the power, increase the cooling capacity in advance within the allowable time window and store the cooling capacity in the building structure and air. When it is necessary to reduce the power, reduce the cooling capacity output and use the stored cooling capacity to maintain the ambient temperature. When it is necessary to increase the power, reduce the cooling capacity in advance, allowing the temperature to rise reasonably, provided that the temperature does not exceed the upper limit. Then, start the unit to provide cooling capacity when it is necessary to increase the power.

5. The flexible grid-connected control method for industrial air conditioning units based on power prediction according to claim 1, characterized in that, The rolling optimization solution in step three adopts a model predictive control framework. In each control cycle, based on the latest measurement value and prediction information, an optimization problem in a finite time domain is solved, and the first element of the optimization solution sequence is issued as the current control command.

6. The flexible grid-connected control method for industrial air conditioning units based on power prediction according to claim 1, characterized in that, The local protection and control strategy in step four includes: when the temperature and humidity parameters of the plant environment approach the preset safety limit, or when the communication interruption with the power grid exceeds the set threshold, the system switches to local autonomous mode and controls according to the preset safety operation curve to ensure the safety of the production process.

7. The flexible grid-connected control method for industrial air conditioning units based on power prediction according to claim 1, characterized in that, The model self-learning in step five specifically includes: based on the input and output data of the actual adjustment process, using parameter identification or gradient descent methods to perform online or offline correction on the equivalent heat capacity, heat loss coefficient, and parameters of the fusion prediction model in the plant thermal inertia model.

8. A flexible grid-connected control system for industrial air conditioning units based on power prediction, used to implement the flexible grid-connected control method for industrial air conditioning units based on power prediction as described in any one of claims 1-7, characterized in that, It includes a data acquisition and communication module, a multi-scale power prediction module, a multi-objective optimization decision-making module, a safety control and execution module, and a system platform; The data acquisition and communication module is used to collect operating data of industrial air conditioning units and environmental data of the plant, and to interact with the power grid dispatching system through standard communication protocols, receive flexible adjustment requirements, and send feedback information. The multi-scale power prediction module is used to perform power prediction steps at multiple time scales to generate a basic power prediction curve and an adjustable potential curve. The multi-objective optimization decision module has a built-in plant thermal inertia model, which is used to execute multi-objective optimization decision steps and generate a refined control instruction sequence. The safety control and execution module is used to send control command sequences to the industrial air conditioning unit for execution, monitor key parameters, and execute local protection control strategies in abnormal situations. The system platform is used to integrate and manage the operation of the above modules, provide a human-computer interaction interface, and perform parameter configuration, status monitoring and system management.

9. The flexible grid-connected control system for industrial air conditioning units based on power prediction according to claim 8, characterized in that, The objective function of the multi-objective optimization decision module includes at least two of the following: minimizing the power grid regulation demand tracking deviation, minimizing the energy consumption cost of air conditioning system operation, and minimizing the frequency / amplitude of equipment actions.

10. The flexible grid-connected control system for industrial air conditioning units based on power prediction according to claim 8, characterized in that, The safety control and execution module has instruction verification and hardware / software interlock functions to ensure that the issued control instructions are within the safe operating range of the equipment.

Images