Industrial park dynamic resource regulation and control method based on digital twinning

Abstract

The invention relates to the field of digital twinning technology and intelligent energy management, and discloses an industrial park dynamic resource regulation and control method based on digital twinning, which comprises the following steps of: constructing a digital twinning platform: acquiring operation states and environmental data of various devices in a park in real time through an Internet of Things sensor network, establishing a digital twin model of each device, and updating the virtual state of the device in real time through data transmission; real-time data of the devices are input into the platform, the state of each device is dynamically updated, synchronization of the digital twin model and the physical device is ensured, and accurate simulation of the current state of the device is provided. According to the invention, by integrating the park management system and the closed-loop control system, real-time collaborative scheduling of park equipment and energy management is realized, energy distribution is optimized, and balanced utilization of various energies is ensured; meanwhile, real-time feedback and error correction are realized through a closed-loop control mechanism.

Description

Technical Field

[0001] The present invention relates to the fields of digital twin technology and intelligent energy management, and specifically provides a method for dynamically regulating resources in industrial parks based on digital twins. Background Art

[0002] With the increasing scale of industrial parks and the complexity of their equipment and energy demands, traditional energy management and equipment scheduling methods can no longer meet the requirements of modern parks for efficient resource utilization and intelligent management. Traditional park management systems usually adopt static scheduling models, relying on manual intervention or prediction models based on historical data. This approach often fails to cope with fluctuations in equipment loads, changes in energy supply, and uncertainties in environmental conditions within the park. Therefore, the regulation efficiency of park resources is not high, resulting in energy waste and low equipment utilization. Even in case of emergencies, it may lead to equipment failures or insufficient energy supply, seriously affecting the operational efficiency and sustainable development of the park.

[0003] In recent years, as an emerging digital simulation technology, digital twin technology has begun to be widely used in the industrial field. Digital twin technology can simulate and optimize the performance of systems in the real world through real-time mapping and information exchange between virtual models and physical objects, thus providing accurate real-time data and decision-making support for park management. In the application of industrial parks, digital twin technology can reflect key indicators such as the operating status of equipment, energy consumption, and environmental changes within the park in real time, providing a comprehensive perspective for dynamic resource regulation.

[0004] However, despite the application of digital twin technology in some fields, there are still several major deficiencies in the existing technology. First, existing digital twin platforms often lack the ability to coordinate the scheduling of multiple energy types (such as electricity, solar energy, wind energy, energy storage, etc.), which results in the inability to flexibly adjust the park in case of unstable energy supply, causing energy waste. Second, the dynamic scheduling of equipment and resources within the park lacks an adaptive feedback mechanism and is difficult to cope with real-time changing load demands and external environmental disturbances, resulting in inaccurate system scheduling. Finally, existing park management systems often operate independently and cannot achieve the coordinated management of multi-dimensional data such as equipment, energy, and environment, leading to information silos and low scheduling efficiency.

[0005] Therefore, in view of the above problems, the present invention proposes a method for dynamically regulating resources in industrial parks based on digital twins. Summary of the Invention

[0006] Aiming at the deficiencies of the existing technology, the present invention provides a method for dynamically regulating resources in industrial parks based on digital twins, which solves the problems of inaccurate scheduling of equipment and energy resources and low management efficiency in industrial parks.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: A dynamic resource regulation method for industrial parks based on digital twins, comprising the following steps: Construct a digital twin platform: Real-time collect the operating status and environmental data of various devices in the park through the Internet of Things sensor network, establish a digital twin model for each device, and realize the real-time update of the virtual status of the device through data transmission; Real-time update the digital twin model: Input the real-time data of the device into the platform, dynamically update the status of each device, ensure the synchronization of the digital twin model and the physical device, and provide an accurate simulation of the current status of the device; Conduct energy demand prediction based on deep learning: Use a long short-term memory network to perform time series prediction on the future energy demand of the park, input historical energy consumption data, device status information, and external environmental data, and generate an energy demand prediction for future moments; Optimize energy scheduling: According to the predicted energy demand, use a multi-objective optimization algorithm to make decisions on the energy scheduling of each device in the park, minimize energy consumption and emissions, balance the load, and improve energy utilization efficiency; Implement adaptive feedback control: When there is a difference between the actual load demand and the predicted demand of the devices in the park, adjust the control input of the devices, and optimize the energy scheduling in real time through a closed-loop control mechanism; Conduct cross-device and multi-energy collaborative scheduling: Coordinate the scheduling between different types of energy and devices in the park to ensure the balance and efficient utilization of energy supply in the park. The different types of energy include electricity, solar energy, wind energy, and energy storage; Integrate the park management system and the closed-loop control system: Integrate the digital twin platform with the existing energy management system in the park to achieve unified scheduling control, and conduct long-term optimization and adjustment through the closed-loop control system.

[0008] Preferably, the real-time update of the digital twin model includes the following sub-steps: Device status evaluation and synchronization, input the real-time data into the virtual model of each device, and update the virtual status of the device using the formula; Real-time data feedback, collect the device data stream through sensors , input it into the digital twin platform, dynamically update the device status, and ensure real-time and accuracy.

[0009] Preferably, in the formula for updating the virtual status of the device, the formula is: ; Wherein, is the state vector of the device, is the control input, is the external disturbance, is the device Function of state change is a time variable is the change rate of the th device state vector at time

[0010] Preferably, the energy demand prediction based on deep learning includes the following sub-steps: Data input and training, using historical energy consumption data , device status information and external environment data as input data to train the LSTM deep learning model; Energy demand prediction, through the trained LSTM model, predict the energy demand at future time , and the prediction result is ; Optimize the prediction accuracy, during the training process, optimize the model parameters by minimizing the prediction error to improve the prediction accuracy.

[0011] Preferably, the energy scheduling optimization includes the following sub-steps: Define the optimization objective function, and the objective function is , where is the energy cost coefficient of the th device, is the energy consumption of the th device, is the emission amount, is the emission factor; Set the constraint conditions, by setting the maximum load of each device and the charge and discharge limits of the energy storage battery , constrain the scheduling to ensure that the device operates within a reasonable range; Optimize the scheduling strategy, through the multi-objective optimization algorithm, minimize the energy consumption and emissions, and at the same time meet the constraint conditions to optimize the scheduling strategy of each device in the park.

[0012] Preferably, the implementation of adaptive feedback control includes the following sub-steps: Error calculation, compare the predicted demand with the actual demand in real time, calculate the error and give feedback; Adjust the control input, according to the feedback error , adjust the control input of the device, and use the feedback control formula to ensure the optimization of energy scheduling.

[0013] Preferably, the feedback control formula is: ; Among them, is the control input, is the th device's control input at the previous time point , is the control gain, is the actual demand is the predicted demand.

[0014] Preferably, the cross-device and multi-energy collaborative scheduling includes the following sub-steps: Multi-energy collaborative scheduling, according to the prediction results and the scheduling objective function, coordinates the scheduling of various energies such as electricity, solar energy, wind energy, and energy storage in the park, and optimizes the use and emission of energy; Device start-stop scheduling, adjusts the start-stop timing of devices according to real-time demands, ensures the efficient use of the energy storage system and backup generators to meet the energy demands in the park.

[0015] Preferably, the integrated park management system and closed-loop control system includes the following sub-steps: System integration, integrates the digital twin platform with the park's existing energy management system, device monitoring system, and data processing platform to form a unified scheduling and control center; Closed-loop control and real-time optimization, through the closed-loop control mechanism, real-time optimizes and adjusts the energy scheduling strategy of devices in the park to ensure the accuracy and efficiency of energy use.

[0016] The present invention provides a method for dynamically regulating resources in an industrial park based on digital twins. It has the following beneficial effects: 1. The present invention adopts the technical solution of integrating the park management system and the closed-loop control system, achieving the real-time collaborative scheduling effect of devices and energy management in the park. Through the seamless integration of the digital twin platform and the existing management system, the system can collect and process the operation data and energy consumption information of park devices in real time. Compared with the dispersed scheduling systems in the prior art, it avoids the problems of data lag and device control disconnection, and greatly improves the efficiency and accuracy of energy scheduling.

[0017] 2. The present invention adopts the technical solution of cross-device and multi-energy collaborative scheduling, solving the instability of different types of energy supply in the park. Through real-time optimization of energy distribution, the system can flexibly schedule various energy types such as electricity, solar energy, wind energy, and energy storage to ensure energy supply balance and maximize the utilization efficiency of various energies. Compared with the scheduling methods that only rely on a single energy source in the prior art, the present invention effectively solves the problem that backup energy cannot be scheduled in time when the energy supply is insufficient.

[0018] 3. The present invention adopts a technical solution of a closed-loop control mechanism for dynamic adjustment, achieving a control effect of real-time feedback and error correction. When an error occurs between the actual load of the device and the predicted demand, the system can automatically adjust the control input, reduce energy waste, ensure the load balance of the device. Compared with the manual adjustment or predetermined strategy in the prior art, the present invention eliminates the blind spots of human intervention through real-time feedback, improving the adaptive ability and optimization level of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a schematic flow chart of the method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0020] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0021] Please refer to the attached Figure 1 , the embodiments of the present invention provide a method for dynamically regulating resources in an industrial park based on digital twin, including the following steps: S1. Build a digital twin platform: Real-time collect the operation status and environmental data of various devices in the park through the Internet of Things sensor network, establish a digital twin model for each device, and realize the real-time update of the virtual status of the device through data transmission; In the dynamic resource regulation of the industrial park, the construction of the digital twin platform is the foundation of the entire system, which realizes the virtual simulation and real-time monitoring of all devices, systems and environments in the park. The digital twin platform collects the operation status and environmental data of various devices in the park through the Internet of Things (IoT) sensor network, establishes a digital twin model for each device, and realizes the real-time update of the virtual status of the device through data transmission. This process provides accurate data support and decision-making basis for subsequent energy demand prediction, optimal scheduling and feedback control.

[0022] In this embodiment, the digital twin platform deploys an Internet of Things sensor network to collect the operation data of each device (such as power equipment, production lines, air conditioners, energy storage batteries, etc.) and the environment (such as temperature and humidity, light intensity, etc.) in the park in real time. The data is transmitted to the central control platform through wireless communication, and the platform processes the received data in real time. The digital twin platform inputs the real-time status and environmental information of the device into the virtual model, and establishes a digital twin model for each device in the platform. This model can not only reflect the physical state of the device, but also simulate the behavior and operation effect of the device under different working conditions in real time.

[0023] Each device is represented as a multi-dimensional state vector in the digital twin platform , where is the device number, is the time, representing multiple key metrics of the device. For example, the digital twin model of a certain power device may include the following items: : The power consumption of the device (kW); : The load of the device (kW); : The ambient temperature (°C); : The device efficiency (% ).

[0024] The dynamic changes of the device are described by the following differential equation: ; where: is the state vector of the device, containing the operating state information of the device, such as power consumption, load, etc.; is the control input, such as device start / stop, load adjustment, etc.; is the device function of the state change of, describing how the device changes over time, usually defined based on the physical characteristics and operating rules of the device; is the external disturbance, such as ambient temperature change, device failure, etc.

[0025] In some embodiments, the digital twin platform also combines the historical data of the device and the external environment data, and performs modeling through the fusion of physical models and data-driven models. Specifically, the operating behavior of the device can be accurately modeled jointly by physical models (such as the relationship between current and load, the energy efficiency function of electrical equipment, etc.) and data-driven machine learning models (such as regression models, time series prediction, etc.). In this way, the platform can dynamically update the state of the device according to real-time data and accurately simulate the actual operating conditions of the device.

[0026] The sensor network collects a large amount of real-time data by monitoring the working state of the device and environmental factors in real time. These data are transmitted to the central control platform through a wireless communication network, where data processing, analysis, and storage are performed. The data streams collected by the sensors include but are not limited to: Power consumption data (such as in kilowatts); Device load data (such as ); Environmental data (such as temperature and humidity and humidity ) such as temperature and humidity and humidity 。

[0027] In a typical embodiment, it is assumed that the real-time data of the th device is as follows: ; Wherein: represents power consumption, represents the load, represents temperature and humidity data, etc.

[0028] Based on the data from these sensors, the digital twin platform will update the digital model of the device in real time and perform dynamic regulation according to these data.

[0029] In a possible implementation, the external environmental data collected by the sensors, such as temperature, humidity, light intensity, etc., will be used as one of the input data of the digital twin platform. Assume that the external environmental data is represented as where contains the real-time data of all key environmental factors. These data directly affect the working state and energy consumption of the device, so they play an important role in device state assessment and energy scheduling optimization.

[0030] The working state of the device is not only affected by its own control input, but also closely related to external environmental factors. For example, the energy efficiency of solar photovoltaic panels is affected by light intensity, and the energy consumption of air conditioning equipment is closely related to the outdoor temperature. In these cases, the digital twin platform needs to combine the external environmental data with the device model in real time and dynamically adjust the scheduling strategy of the device.

[0031] In some embodiments, the digital twin platform will make decisions considering both environmental data and device status. For example, during the energy scheduling optimization process, if the light intensity changes, the platform will automatically adjust the power generation plan of the solar photovoltaic panels or start the backup generator in low light conditions to ensure the stability and efficiency of energy supply.

[0032] Through real-time data collection and virtual model update, the digital twin platform can accurately reflect the actual operating state of the devices in the park and provide real-time feedback on aspects such as energy consumption, load distribution, and environmental impact. The digital model in the platform can provide high-precision data support for subsequent energy demand prediction, scheduling optimization, and feedback control. Specifically, when the operating state of the device changes, the digital twin platform can reflect these changes in real time and adjust the control strategy to ensure the stable operation and efficient energy utilization of the system.

[0033] In addition, the construction of the digital twin platform also provides a basis for subsequent multi-objective optimal scheduling, ensuring that multiple objectives such as energy consumption, equipment load, and emission control can be optimized synchronously.

[0034] S2. Real-time update of the digital twin model: Input the real-time data of the equipment into the platform, dynamically update the status of each device, ensure the synchronization of the digital twin model with the physical device, and provide an accurate simulation of the current status of the device. In the park resource regulation system, one of the core functions of the digital twin platform is to achieve real-time update of the digital twin model. This process continuously inputs the real-time data of the equipment to update the virtual model of each device, thereby ensuring the synchronization of the virtual status of the device with the physical device status. This synchronization process is the basis of the park's intelligent scheduling and can provide accurate real-time data support for energy scheduling, load balancing, and feedback control.

[0035] In this embodiment, the digital twin platform uses the Internet of Things (IoT) sensor network to collect the operation status and environmental data of various devices in the park in real time. These sensors monitor multiple dimensions of parameters of the equipment, such as power consumption, load, temperature and humidity, wind speed, light intensity, etc. Through wireless communication technology, the data stream collected by the sensors is transmitted to the central control platform in real time, and the platform inputs these data into the digital twin model through the data processing and analysis module to dynamically update the virtual status of the device. This synchronization is not limited to the performance and efficiency of the device, but also includes the health status of the device and the impact of external environmental factors on its operation.

[0036] Through real-time data input, the digital twin platform can dynamically adjust the operation status of the device to ensure that the virtual model always reflects the actual situation of the physical device. Especially during the operation of the device, external factors such as environmental temperature and humidity, and the workload of the device will affect the performance of the device. The digital twin platform can quickly respond to these changes, adjust the model parameters, and ensure that the virtual status of the device is consistent with its actual status.

[0037] Specifically, the digital twin model of the device is represented by the state vector where is the device number, is the timestamp, indicating the status of the device at a certain moment. Assume that the state vector of the device contains multiple parameters, for example: : The power consumption of the device (unit: kilowatt); : The load of the device (unit: kilowatt); : The device temperature (unit: degree Celsius); : The device efficiency (unit: %).

[0038] The state change of the device can be described by the following differential equation: ; Where: is the state vector of the device, which contains the operating state information of the device, such as power consumption, load, etc.; is the control input, such as device start / stop, load adjustment, etc.; is the device function of the state change, which describes how the device changes over time and is usually defined based on the physical characteristics and operating rules of the device; is the external disturbance, such as environmental temperature change, device failure, etc.

[0039] In some embodiments, the digital twin platform not only reflects the immediate operating state of the device, but also can simulate the performance of the device under different operating conditions. For example, when the temperature in the park changes, the digital twin model can reflect in real time the impact of the temperature on the device performance, and then adjust the operating strategy of the device. If the temperature of the device rises, the platform may need to adjust the device load or start / stop the device to avoid device overload or damage.

[0040] The real-time data of the device is collected by sensors and input into the digital twin platform, and the device model in the platform will be dynamically adjusted according to the real-time input data. For example, when the external environmental temperature ( ) changes, the device model will adjust its efficiency parameters to reflect the impact of different temperatures on the device operation. This change will further affect the energy consumption and scheduling strategy of the device.

[0041] In a possible implementation, the device state is not only affected by the internal control input of the device (such as load regulation, start / stop control), but also affected by the changes of external environmental factors (such as temperature, humidity, etc.). These environmental factors have a significant impact on the operation of the device. The digital twin platform combines environmental data (such as light, wind speed, etc.) and device state data to accurately simulate the operating characteristics of the device. Assuming the temperature data is , when comprehensively evaluating the device state in the platform, the change of temperature will directly affect the operating efficiency of the device and reflect this impact through the adjustment of the device model.

[0042] Overall, the real-time update of the digital twin model not only helps to accurately simulate the device performance, but also can dynamically feedback on aspects such as the health status of the device, external environmental changes, energy consumption, etc. Whenever the sensor detects new data, the digital twin platform will update the state of the device based on the real-time data. This real-time update process helps the platform optimize the device operation strategy to ensure the best energy management and scheduling in the park.

[0043] As an option, the digital twin platform can also introduce more external sensor data, such as equipment health status information, load fluctuations, equipment aging, etc. Through comprehensive monitoring of the equipment, the platform can identify potential equipment failures or inefficiencies in advance, thereby optimizing scheduling plans and equipment maintenance strategies. This information can be used to predict equipment failures before they occur, providing the park with enough time to perform preventive maintenance or adjust scheduling plans.

[0044] By updating the digital twin model in real time, the equipment in the park can maintain a virtual state consistent with the physical equipment at all times. This process can not only accurately simulate the operating status, energy consumption and workload of the equipment, but also respond to the impact of external environmental changes on the equipment in real time, providing accurate data support for subsequent energy demand forecasting, scheduling optimization and feedback control. The digital twin platform ensures that the equipment status of the park always reflects the actual operating conditions, thereby improving the response speed and accuracy of the energy management system and providing stable and reliable support for the energy scheduling of the entire park.

[0045] S3. Energy demand forecasting based on deep learning: Use long short-term memory networks to make time series forecasts of the park’s future energy demand, input historical energy consumption data, equipment status information, and external environment data, and generate energy demand forecasts for the future. Based on the construction of the digital twin platform and the update of equipment status, energy demand forecasting is a key step to further optimize the resource scheduling of the park. Through the long short-term memory network (LSTM) based on deep learning, this step aims to make a time series forecast of the future energy demand of the park. The LSTM network inputs historical energy consumption data, equipment status information, and external environment data to learn the time series rules in historical data and generate energy demand forecasts for the future. This forecast result will provide key data support for subsequent energy scheduling, optimization decision-making, and feedback control.

[0046] In this embodiment, the LSTM network uses historical energy consumption data, equipment status information and external environment data to perform time series forecasting of future energy demand. Specifically, the LSTM model can capture the long-term dependency of the park's energy demand by learning from historical data, especially under the influence of production load fluctuations, seasonal changes and external environment (such as temperature, humidity, etc.), and can efficiently predict future energy demand.

[0047] In general, LSTM networks can effectively process time series data, especially in scenarios where energy consumption is affected by multiple factors. LSTM learns the long-term dependencies of data through its gating mechanism, and can make accurate predictions about future energy demand. For the park, the LSTM model can predict future energy demand based on input data (such as historical energy consumption, equipment status, external environmental factors, etc.), ensuring the rational allocation of park resources and load balance.

[0048] Specifically, the input data of the LSTM network includes: Historical energy consumption data: Historical energy consumption data of each device in the park ,This data includes the power load and energy consumption of the equipment; Equipment status information: Equipment operating status data , such as production load, equipment efficiency, equipment start and stop status, etc.; External environment data: data on the park's external environmental factors , such as temperature, humidity, light intensity and other environmental variables that affect device performance.

[0049] The LSTM model makes predictions based on the time series characteristics of these historical data. The moment is the current moment, predicting the future The energy demand at each moment is , the prediction formula of the LSTM model is: ; in: For the future moment energy demand forecasts; is the historical energy consumption data, at time Energy consumption value; The device status information indicates the device at time load, efficiency and other operating data; External environment data includes factors affecting the internal and external environment of the park, such as temperature and humidity.

[0050] By continuously optimizing the parameters of the LSTM model, we can effectively capture the long-term dependencies in the park’s energy demand, especially under the influence of factors such as long-term load fluctuations and seasonal changes, ensuring that the system can predict future energy demand in advance.

[0051] As an option, the LSTM network can also adopt a bidirectional LSTM structure to enhance the model's prediction ability. The bidirectional LSTM can not only learn information from past time steps but also from future time steps, which can further improve the prediction accuracy, especially for data with complex time dependencies. The bidirectional LSTM can comprehensively capture the laws of energy demand changes by propagating information forward and backward.

[0052] Specifically, the input data of the bidirectional LSTM model also includes historical energy consumption data , device status information and external environment data , and its output is the energy demand prediction at future time points . Through bidirectional learning, the model can take into account the context information of the predicted time series data, thus providing more accurate predictions.

[0053] When the LSTM model is used for energy demand prediction, it first needs to be trained with historical data. During the training process, the LSTM network optimizes the internal weights through the gradient descent method and the backpropagation algorithm, gradually approaching the optimal prediction value. The loss function during the training process usually uses the mean squared error (MSE) to measure the prediction error: ; where: is the predicted energy demand of the th device; is the actual energy demand of the th device; is the total number of devices.

[0054] The LSTM model gradually adjusts the model parameters by minimizing this loss function, thereby improving the prediction accuracy. When the training is completed, the model can make predictions based on the latest historical data and device status, providing accurate demand predictions for energy scheduling.

[0055] In a possible implementation, the LSTM network can not only predict the energy demand of the entire park but also predict each device individually. Through individual predictions of each device, the system can more precisely understand the energy demand of each device and optimize the energy scheduling at the device level according to the prediction results. For example, during the production peak period, the energy demand of some devices may increase sharply, and the LSTM model can predict this change and schedule backup power sources or energy storage devices in advance to avoid power shortages.

[0056] Through the energy demand prediction using a deep learning-based LSTM model, the park can achieve high-precision energy demand prediction. The LSTM model can process multi-dimensional input data, capture long-term dependencies in time-series data, and accurately predict the energy demand fluctuations of the equipment and systems in the park. This prediction result will provide an accurate basis for subsequent energy scheduling optimization, ensure the efficient allocation of resources in the park, improve energy utilization efficiency, and reduce unnecessary energy waste.

[0057] S4. Conduct energy scheduling optimization: According to the predicted energy demand, use a multi-objective optimization algorithm to make decisions on the energy scheduling of each device in the park, minimize energy consumption and emissions, balance the load, and improve energy utilization efficiency; After completing the construction of the digital twin platform and the real-time update of the device status, conducting energy scheduling optimization is the next key step. This step is based on the aforementioned energy demand prediction model and the real-time data of the device status, schedules various devices in the park, and optimizes energy utilization. The optimization goal is to minimize the total energy consumption and emissions of the park, while ensuring load balance and efficient energy utilization. In this embodiment, a multi-objective optimization algorithm is used to achieve this optimized scheduling task, ensuring the matching of device load and energy supply, and minimizing unnecessary energy waste and emissions to the greatest extent.

[0058] In this embodiment, the energy scheduling optimization is achieved through the following aspects: Minimization of energy consumption and emissions: Through a multi-objective optimization algorithm, combined with the energy demand and emission constraints during device operation, adjust the device scheduling strategy.

[0059] Load balance: Ensure the balance between the energy demand and energy supply of all devices in the park, and avoid overload or device idleness.

[0060] Efficient use of resources: Through real-time optimization of the scheduling strategy, make each device operate within its optimal working range to improve energy use efficiency.

[0061] The goal of energy scheduling optimization is to minimize energy consumption and emissions by adjusting the operation mode and load distribution of devices, while balancing the device load to ensure the adequacy of energy supply and the stability of the system.

[0062] In this embodiment, the objective function can be written as: ; where: is the energy cost coefficient of the th device, is the energy consumption of the th device, is the emission, Is the emission factor.

[0063] In the optimization of energy dispatch, the physical and operational constraints of equipment also need to be considered. The following are some common constraints: Load balance constraint: Since the total energy demand of all equipment in the park cannot exceed the energy supply capacity of the park, the load balance constraint is necessary. It can be expressed as: ; where: Is the total energy demand of all equipment in the park at time .

[0064] This constraint ensures that the energy demand of all equipment does not exceed the available energy supply.

[0065] Equipment operation constraint: Each equipment has a limit on its operating load, especially for adjustable equipment (such as air conditioners, heating equipment, energy storage equipment, etc.). Assuming that the load range of each equipment is To , the following constraints can be added: ; Where: Is the minimum load of the th equipment; Is the maximum load of the th equipment.

[0066] Energy storage equipment charge and discharge limit: For energy storage equipment, such as batteries, the charging and discharging capabilities also need to be restricted. Assuming that the charging capacity of the energy storage equipment is , the discharging capacity is , and the energy storage range of the equipment is And , the following constraints can be set:

[0067] Where: And Are the charging and discharging powers of the energy storage equipment at time (unit: kilowatt); And Are the maximum and minimum stored energies of the energy storage equipment (unit: kilowatt-hour); Represents the current energy level of the energy storage equipment.

[0068] To solve the above optimization problem, this embodiment adopts a multi-objective optimization algorithm, which includes: Genetic Algorithm (GA): It searches for the optimal solution by simulating the natural selection process, has strong global search ability, and is applicable to multi-objective and constrained optimization problems.

[0069] Particle Swarm Optimization (PSO): It simulates the search behavior of a particle swarm in the solution space, can quickly find an approximate optimal solution, and can handle multi-objective problems.

[0070] Heuristic algorithm: By designing problem-specific heuristic rules, it finds the optimal solution that meets the constraint conditions and is applicable to solving large-scale complex optimization problems.

[0071] For example, the steps of the genetic algorithm include: Initialization: Generate a number of individuals (i.e., device scheduling schemes), and each individual includes the load and operation strategy of the device.

[0072] Fitness evaluation: Calculate the fitness value of each individual, that is, the evaluation based on the objective function ZZZ.

[0073] Selection operation: Select excellent individuals according to the fitness value as the population of the next generation.

[0074] Crossover operation: Generate new individuals through crossover for information exchange.

[0075] Mutation operation: Make small changes to the individual to avoid local optimal solutions.

[0076] As an option, Particle Swarm Optimization (PSO) continuously updates the speed and position of each particle by simulating the flight behavior of a particle swarm, and quickly converges to the optimal solution of the objective function. The update rules of PSO are as follows: ; Where: is the speed of the th particle at time ; is the state vector of the device; is the historical best position of particle ; is the global best position of all particles; is the inertia weight, which controls the inertia of the particle; is the acceleration constant, which controls the gravity of the particle; is a random number, and its value range is [0,1], is the state vector of the device at time, is the th particle's speed at time .

[0077] S5. Implement adaptive feedback control: When there is a difference between the actual load demand and the predicted demand of the devices in the park, adjust the control input of the devices and optimize the energy scheduling in real time through a closed-loop control mechanism. In the above steps, the energy demand of the park has been predicted through a deep learning model (such as LSTM), and the energy scheduling has been optimized through a multi-objective optimization algorithm. However, since there may be errors between the actual load demand and the predicted load demand of the devices, especially in a dynamic and uncertain environment, it is necessary for us to correct the difference between the actual load demand and the predicted load demand in real time. The core role of the adaptive feedback control mechanism is to adjust the operation strategy of the devices according to these errors, so as to continuously optimize the energy scheduling system, ensure the minimum energy consumption, the lowest emissions, and maintain the balance between energy supply and demand.

[0078] In this embodiment, the goal of the adaptive feedback control is to adjust the control input of the devices in real time through a closed-loop feedback control mechanism, correct the difference between the actual load demand and the predicted load demand, and thus optimize the energy scheduling in the park. The specific steps are as follows: In the energy scheduling system, it is first necessary to monitor the actual load demand of the devices in real time and compare it with the predicted load demand to calculate the error. The error reflects the deviation between the actual load demand and the predicted load demand of the devices. This error is used to adjust the control input, so as to achieve scheduling optimization.

[0079] The error calculation formula is: ; where: is the actual load demand of the th device at time , in kilowatt-hours (kWh). This data usually comes from the Internet of Things sensors in the park, which collect the load status of the devices in real time.

[0080] is the predicted load demand of the th device at time , in kilowatt-hours (kWh), generated by the aforementioned energy demand prediction model (such as LSTM).

[0081] Through this formula, the system can calculate the load demand error of each device at a certain time, providing a basis for the adjustment of the control input in the next step.

[0082] Once the load error of the device is calculated, the system needs to adjust the control input of the device according to this error , minimizing the error as much as possible to ensure the smallest gap between the load demand and the predicted load. The control input adjustment follows the following formula: ; Where: is the control input, is the control input of the th device at the previous time point , is the control gain, and the gain reflects the sensitivity of the control system. A larger gain value means that the system responds more strongly to the load error, thus accelerating the speed of error correction. The value of the gain can be adjusted according to the actual load fluctuation situation and the device type, is the actual demand is the predicted demand. is the error of the th device at time , representing the difference between the actual load demand and the predicted load demand of the device.

[0083] Adjustment of the control input: When the error is large, the system will increase the control input of the device to correct the deviation of the device load.

[0084] Conversely, when the error is small, the system may reduce the adjustment amplitude of the control input to avoid unnecessary energy waste caused by over-regulation.

[0085] The gain coefficient is a key parameter of the adaptive feedback control, which determines the response degree of the system to the load error. Specifically, the gain coefficient can be dynamically adjusted according to the fluctuation characteristics of the device load and the energy supply situation in the park. Generally speaking: For devices with large load fluctuations, the gain coefficient can be increased so that the system can respond faster to changes in load demand; For devices with small load fluctuations and stable load demands, the gain coefficient can be reduced to avoid over-adjustment.

[0086] In some embodiments, the gain coefficient may be dynamically adjusted, automatically adjusted according to the historical load fluctuation situation of the device and the change of the prediction error to achieve more accurate control.

[0087] To achieve more efficient feedback control, the system can automatically adjust the control parameters according to the operation history of the device and the current environmental changes. For example, when the load of the device changes greatly, the system will automatically increase the control response intensity to accelerate the correction of the load demand deviation; while when the change in equipment load is small or the system is stable, the gain coefficient will be moderately reduced to reduce excessive interference with the equipment.

[0088] To further optimize the feedback control strategy, the system can incorporate a reinforcement learning (RL) algorithm. Through long-term operation experience, the system can automatically learn the optimal feedback gain coefficient and control strategy to adapt to complex and dynamic load demands. Reinforcement learning can be trained based on the historical operation data of the equipment, enabling the feedback control to continuously optimize under different working environments.

[0089] Specifically, the system can train the equipment load data through a reinforcement learning model to learn how to adjust the control input according to the error and gradually optimize the control gain parameters . This approach can achieve continuous improvement of system performance without manual intervention.

[0090] S6. Perform cross-device and multi-energy coordinated scheduling: Coordinate the scheduling among different types of energy and devices in the park to ensure the balance and efficient utilization of energy supply in the park. The different types of energy include electricity, solar energy, wind energy, and energy storage; In the foregoing steps, the energy demand of the park has been accurately predicted, and the operation of the equipment has been managed according to the optimized energy scheduling. However, the energy supply of the park not only depends on traditional power sources but also includes various energy types such as solar energy, wind energy, and energy storage. Therefore, how to efficiently coordinate the supply of these different types of energy to ensure their matching with the demand of equipment load is the key to achieving efficient energy management. The cross-device and multi-energy coordinated scheduling step is the core to solve this problem. The goal of this step is to achieve efficient utilization of energy, load balance, and maximum energy conservation in the park by coordinating different types of energy and devices.

[0091] In this embodiment, the goal of cross-device and multi-energy coordinated scheduling is to dynamically allocate various types of energy through a scheduling algorithm to minimize energy consumption, reduce emissions, and improve equipment operation efficiency while ensuring the load demand of the park. To achieve this goal, the system must comprehensively consider equipment requirements, energy supply capabilities, and various constraints to ensure the balance of equipment load and the efficient utilization of energy.

[0092] In the park, energy supply and demand are dynamically changing, especially for renewable energy such as solar energy and wind energy, whose supply is affected by external environments (such as weather, time, etc.). Therefore, how to formulate an optimal scheduling strategy based on the real-time load demand of the equipment, available energy types, and supply status is the key challenge in this step.

[0093] To address this challenge, the system requires an accurate optimization model to coordinate the scheduling relationships among all energies (electricity, solar, wind, energy storage, etc.) and devices within the park, avoid energy waste, and reasonably enable backup energy or adjust device loads when supply is insufficient.

[0094] To achieve coordinated scheduling across devices and multiple energies, an optimization model is defined in this embodiment with the aim of minimizing total energy consumption and emissions while ensuring balanced device loads. This model needs to consider the energy demands of devices and the supply capabilities of various energies, and allocate optimal energy usage amounts for each energy type and control input of the devices.

[0095] The objective function is as follows: ; Where: is the energy cost coefficient of the th device, is the energy consumption of the th device, is the emission amount, is the emission factor; is the usage cost coefficient of the th type of energy (such as solar, wind, energy storage, etc.), representing the unit usage cost of each energy type; is the usage amount of the th type of energy at time (unit: kilowatt-hour, kWh), including electricity, solar, wind, energy storage, etc.; is the weight coefficient for the th device load adjustment, representing the priority of this device in energy scheduling; is the load adjustment amount of the th device at time .

[0096] Through this objective function, the system comprehensively considers the energy consumption, emissions, costs, load adjustment of devices, and the usage efficiency of various energies, and finally obtains the optimal energy scheduling plan.

[0097] To ensure the rationality and feasibility of energy scheduling, the system also needs to consider a series of constraint conditions. The following are common constraint types: 1. Load balance constraint The energy load balance constraint ensures the matching between the load demands of devices and the energy supply. Assuming that the total energy demand of devices and the total energy supply must be equal at a certain moment , the load balance constraint can be expressed as: ; Where, is the The energy consumption of a device for the th type of energy at time (unit: kilowatt-hour, kWh). This constraint ensures that the energy demands of all devices can be met from different energy supplies.

[0098] 2. Device load constraint The load demand of each device must satisfy the maximum and minimum load limits for its operation. That is, the load of the device must be within its predetermined load range:

[0099] where and are the minimum and maximum load limits of the th device, respectively.

[0100] 3. Renewable energy production constraint For renewable energy such as solar and wind energy, the system needs to consider the impact of external environments (such as solar radiation, wind speed, etc.) on energy production. Therefore, the production volume is constrained by environmental factors and can be expressed as:

[0101] where represents the maximum available production volume of the th type of energy (such as solar or wind energy) at time , which is usually directly related to environmental conditions (such as weather, time period).

[0102] 4. Energy storage device constraint If the system includes energy storage devices (such as batteries), the charging and discharging volumes of the energy storage devices are limited by their capacities. The charging volume and discharging volume of the energy storage device must satisfy the following constraints: ; where and represent the charging and discharging powers of the energy storage device at time , respectively, and is the maximum charging and discharging capacity of the energy storage device.

[0103] To solve the above multi-objective optimization problem, the system can adopt various optimization algorithms, such as: Genetic Algorithm (GA): By simulating the process of natural selection, it evaluates the quality of solutions through a fitness function and generates new scheduling solutions through crossover and mutation operations. The genetic algorithm can effectively handle optimization problems with multiple objectives and complex constraints.

[0104] Particle Swarm Optimization (PSO): Particle Swarm Optimization simulates the behavior of particles flying in the search space. By updating the velocity and position of particles, it gradually finds the global optimal solution. Particle Swarm Optimization can efficiently handle multi-objective optimization problems and has strong global search ability.

[0105] Reinforcement Learning (RL): Through adaptive learning, Reinforcement Learning can be trained based on the historical operation data of the system, thereby optimizing control strategies and achieving long-term system self-optimization.

[0106] S7, Integrated Park Management System and Closed-loop Control System: Integrate the digital twin platform with the existing energy management system in the park to achieve unified scheduling control, and conduct long-term optimization and adjustment through the closed-loop control system.

[0107] In the foregoing steps, the energy scheduling system in the park has made reasonable scheduling decisions based on the energy demands of equipment, load adjustment, prediction, and optimization models to ensure the efficient use of energy. However, to maximize the effectiveness of the energy scheduling and management system in the park, it is necessary to integrate the park management system and the closed-loop control system to ensure that all systems are coordinated and optimized on a unified platform. This integration can not only enable the collaborative work of park equipment and energy systems but also continuously adjust and optimize energy management strategies through closed-loop control to ensure the long-term stable and efficient use of park energy.

[0108] In this embodiment, the goal of integrating the park management system and the closed-loop control system is to ensure the seamless collaboration of all equipment and energy management systems in the park through system integration, thereby achieving efficient energy scheduling, load balancing, real-time optimization, and long-term optimization. Specifically, the core of system integration lies in achieving the comprehensive real-time monitoring and scheduling of park equipment and energy through the feedback mechanism of the digital twin platform.

[0109] The integrated system mainly consists of the following components: Data Flow Integration: Uniformly collect and transmit the real-time data of various equipment and energy to the park management platform.

[0110] Scheduling Decision-making System: Formulate the optimal energy scheduling decision by integrating various data and prediction results.

[0111] Feedback and Optimization Mechanism: Dynamically adjust the scheduling decision based on real-time data feedback and error correction.

[0112] Long-term optimization ability: Through data accumulation and adaptive algorithms, improve the self-learning and optimization ability of the energy scheduling system.

[0113] During the integration process of the park management system, it is first necessary to effectively integrate real-time data from different devices, energy sources (such as electricity, solar energy, wind energy, energy storage), and the environment. Specifically, the system uses the Internet of Things sensor network (IoT) to collect the status data of park devices, environmental monitoring data, and energy supply data in real time, and transmits this data to the unified park management platform.

[0114] Generally, the required data includes: Device status data: including power consumption, temperature, humidity, pressure, etc. of the device; Environmental data: including weather conditions (such as wind speed, solar radiation, temperature); Energy consumption and supply data: including the supply volume, consumption volume, charge and discharge volume, etc. of each energy type (such as electricity, solar energy, wind energy, energy storage).

[0115] The system transmits this data to the park management platform through a standardized data protocol, ensuring that all data can be stored and processed in a unified format, and providing a basis for subsequent scheduling decisions and optimizations.

[0116] The park management system needs to comprehensively analyze various types of data collected in real time, and combine prediction models and optimization models (such as deep learning models like long short-term memory network (LSTM), etc.) to make decisions on energy scheduling. The core function of the scheduling decision system is to perform real-time scheduling of the device load demand and energy supply in the park, minimizing energy consumption, emissions, and ensuring load balance.

[0117] The objective function of the scheduling decision can be expressed as: ; Where: is the energy cost coefficient of the th device, is the energy consumption of the th device, is the emission volume, is the emission factor; is the usage cost coefficient of the th type of energy (such as solar energy, wind energy, energy storage, etc.), representing the unit usage cost of each energy type; is the usage volume of the th type of energy at time (unit: kilowatt-hour, kWh), including electricity, solar energy, wind energy, energy storage, etc.

[0118] The scheduling system will comprehensively consider factors such as equipment load requirements, energy type supply situations, emissions, and costs, and make energy allocation decisions through optimization algorithms (such as genetic algorithms, particle swarm optimization, etc.) to ensure efficient coordination among various energies and equipment within the park.

[0119] To ensure the real-time and accuracy of park energy management, the integrated system also introduces a feedback and optimization mechanism. The core of this mechanism is to make immediate adjustments to the scheduling results through closed-loop control based on real-time data collection.

[0120] Specifically, the system calculates the error between the predicted load demand and the actual load demand, and adjusts the control input of the equipment in real time. For example, when the system detects a deviation between the actual load demand and the predicted load demand, the control system will adjust the load of the equipment according to the error, and correct the error by increasing or decreasing the energy consumption.

[0121] The error calculation formula is: ; Where: is the actual load demand of the th device at time , in kilowatt-hours (kWh), is the predicted load demand of the th device at time , in kilowatt-hours (kWh), generated by the aforementioned energy demand prediction model (such as LSTM).

[0122] According to the error, the control input adjustment formula is as follows: ; Where: is the control input, is the control input of the th device at the previous time point , is the control gain, and the gain reflects the sensitivity of the control system. A larger gain value means that the system responds more strongly to the load error, thus accelerating the speed of error correction. The value of the gain can be adjusted according to the actual load fluctuation situation and equipment type, is the actual demand is the predicted demand.

[0123] Through this feedback adjustment mechanism, the system can real-time correct energy scheduling, ensure the effective utilization of energy, and maintain the balance between energy supply and demand in a dynamic environment.

[0124] In addition to real-time feedback control, the system also has long-term optimization and adaptive learning capabilities. During long-term operation, the park management system will accumulate a large amount of historical data, and optimize the system's decision-making model through adaptive learning algorithms (such as reinforcement learning, deep reinforcement learning, etc.), gradually improving the accuracy and stability of energy scheduling.

[0125] For example, the system can use reinforcement learning algorithms to learn the best scheduling strategy based on historical scheduling experience, and dynamically adjust the scheduling plan according to the operating characteristics of the equipment and changes in load demand. In the long-term optimization process, the system can achieve adaptive optimization of factors such as complex load demand, environmental changes, and equipment aging.

[0126] In order to achieve these goals, the system must have the following technical capabilities: Data synchronization and real-time transmission: The system needs to ensure that all device data can be synchronized to the management platform in real time and processed in a short time. At this time, high-speed data communication and efficient data processing algorithms are needed to reduce delays and improve response speed; Multimodal data fusion: The system must be able to process data from different sensors, including power, environment, equipment operation status and other types of data, and provide accurate scheduling decisions through data fusion technology; Decision support and optimization algorithms: The system optimizes scheduling strategies by integrating intelligent decision support tools (such as machine learning, deep learning, etc.) to improve the flexibility and intelligence of campus energy management.

[0127] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A dynamic resource regulation method for industrial parks based on digital twins, characterized in that Including the following steps: Construct a digital twin platform: Collect the operating status and environmental data of various devices in the park in real time through an Internet of Things sensor network, establish a digital twin model for each device, and achieve real-time update of the virtual status of the device through data transmission; Update the digital twin model in real time: Input the real-time data of the device into the platform, dynamically update the status of each device, ensure the synchronization of the digital twin model with the physical device, and provide an accurate simulation of the current status of the device; Conduct energy demand prediction based on deep learning: Use a long short-term memory network to perform time-series prediction on the future energy demand of the park, input historical energy consumption data, device status information, and external environmental data, and generate an energy demand prediction for future moments; Optimize energy scheduling: According to the predicted energy demand, use a multi-objective optimization algorithm to make decisions on the energy scheduling of each device in the park, minimize energy consumption and emissions, balance the load, and improve energy utilization efficiency; Implement adaptive feedback control: When there is a difference between the actual load demand and the predicted demand of the devices in the park, adjust the control input of the devices, and optimize the energy scheduling in real time through a closed-loop control mechanism; Conduct cross-device and multi-energy collaborative scheduling: Coordinate the scheduling between different types of energy and devices in the park to ensure the balance and efficient utilization of energy supply in the park. The different types of energy include electricity, solar energy, wind energy, and energy storage; Integrate the park management system and the closed-loop control system: Integrate the digital twin platform with the existing energy management system in the park to achieve unified scheduling control, and conduct long-term optimization and adjustment through the closed-loop control system; The real-time update of the digital twin model includes the following sub-steps: Device status evaluation and synchronization, input the real-time data into the virtual model of each device, and update the virtual status of the device using a formula; Real-time data feedback, collecting the device data stream through sensors , inputting it into the digital twin platform, dynamically updating the device status, and ensuring real-time and accuracy.

2. The dynamic resource regulation method for an industrial park based on digital twin according to claim 1, wherein, In the virtual state update of the device using the formula, the formula is as follows: , where is the state vector of the device, is the control input, is the external disturbance, is the device function of the state change, is the time variable, is the th device state vector at time change rate, indicating the speed at which the state of the device changes over time.

3. A method for dynamically regulating industrial park resources based on digital twin according to claim 1, characterized in that, The energy demand prediction based on deep learning includes the following sub-steps: Data input and training, using historical energy consumption data , equipment status information and external environment data as input data to train an LSTM deep learning model; Energy demand forecasting, through the trained LSTM model, forecasts the energy demand at future time and the forecasting result is ; Optimize the prediction accuracy. During the training process, minimize the prediction error Optimize the model parameters to improve the prediction accuracy.

4. A dynamic resource regulation method for industrial parks based on digital twins according to claim 1, characterized in that, The energy scheduling optimization includes the following sub-steps: Define the optimization objective function, which is , where is the energy cost coefficient of the -th device, is the energy consumption of the -th device, is the emission amount, is the emission factor; Set constraint conditions by setting the maximum load of each device and the charge and discharge limits of the energy storage battery to constrain the scheduling and ensure that the devices operate within a reasonable range; Optimize the scheduling strategy, through a multi-objective optimization algorithm, minimize energy consumption and emissions, while meeting the constraints, and optimize the scheduling strategy of each device in the park.

5. A method for dynamically regulating resources in an industrial park based on digital twin according to claim 1, characterized in that, The implementation of adaptive feedback control includes the following sub-steps: Error calculation, real-time comparison with predicted demand and actual demand , calculate the error and provide feedback; Adjust the control input according to the feedback error , adjust the control input of the device, and use the feedback control formula to ensure the optimization of energy scheduling.

6. The dynamic resource regulation method for industrial parks based on digital twins according to claim 5, characterized in that The feedback control formula is: ; wherein, is the control input, is the control input of the nth device at the previous time point, is the control gain, is the actual demand is the predicted demand.

7. A method for dynamically regulating industrial park resources based on digital twins according to claim 1, characterized in that The cross-device and multi-energy collaborative scheduling includes the following sub-steps: Multi-energy collaborative scheduling, according to the prediction results and the scheduling objective function, coordinate the scheduling of multiple energies such as electricity, solar energy, wind energy, and energy storage in the park, and optimize the use and emissions of energy; Device start-stop scheduling, adjust the start-stop timing of the devices according to the real-time demand, ensure the efficient use of the energy storage system and standby generators to meet the energy demand in the park.

8. A method for dynamically regulating resources in an industrial park based on digital twins according to claim 1, characterized in that, The integration of the park management system and the closed-loop control system includes the following sub-steps: System integration, integrate the digital twin platform with the existing energy management system, device monitoring system, and data processing platform in the park to form a unified scheduling control center; Closed-loop control and real-time optimization, through the closed-loop control mechanism, perform real-time optimization and adjustment on the energy scheduling strategy of the devices in the park to ensure the accuracy and efficiency of energy use.

Images