A distributed energy resource intelligent scheduling method based on an internet of things

Abstract

The application discloses a kind of distributed energy resource intelligent scheduling methods based on Internet of Things, it is related to Internet of Things technical field, including, the operating data of distributed energy equipment is collected and pretreated, obtain equipment operating dataset, and the group intelligence initialization of equipment operating dataset is carried out, generates initial pheromone matrix;Equipment operating dataset and initial pheromone matrix are locally optimized, generate updated pheromone matrix and equipment candidate scheduling scheme, and obtain local optimal scheduling scheme by screening and verification;Quantum computing node receives local optimal scheduling scheme, and local optimal scheduling scheme is used as initial solution, constructs global optimization problem, outputs global optimal scheduling scheme, while converting into global scheduling instruction;The application captures equipment interaction, economic factors and multidimensional constraints within network range by formulating global optimization problem, realizes coordination between equipment and constraint compliance, prevents the problem of unstable power grid.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things, and in particular to an intelligent scheduling method for distributed energy resources based on the Internet of Things. Background Art

[0002] Intelligent dispatching technology for distributed energy resources (DERs) has rapidly advanced with the widespread adoption of smart grids and renewable energy. In recent years, the convergence of the Internet of Things (IoT), edge computing, and artificial intelligence (AI) has significantly improved the operational efficiency and coordination of distributed energy devices (DERs), such as photovoltaic panels, energy storage batteries, and electric vehicle charging stations. Traditional dispatching methods often employ centralized optimization models, allocating the output of DERs through linear programming or heuristic algorithms (such as genetic algorithms) to achieve supply-demand balance and minimize costs. Recent technological advances have introduced swarm intelligence algorithms, which optimize local dispatch by simulating the behavior of biological swarms. These algorithms, combined with IoT sensors for real-time data collection, further enhance dispatch accuracy. Furthermore, the rise of quantum computing technology offers new approaches for solving complex multidimensional optimization problems. These technologies have made significant progress in improving energy utilization, reducing dispatch costs, and enhancing grid stability, particularly in scenarios such as microgrids and virtual power plants.

[0003] However, the current technical solutions still have some areas that need improvement. For example, although swarm intelligence algorithms can effectively solve complex optimization problems, in practical applications, how to accurately initialize the pheromone matrix to reflect the interactions between different distributed energy devices remains a challenge. In addition, the ability to globally optimize is limited, especially when considering multidimensional constraints. It is difficult to find a global optimal solution that satisfies all constraints and minimizes costs. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides an intelligent scheduling method for distributed energy resources based on the Internet of Things to solve the problem of how to accurately reflect the interaction between distributed energy devices and find a global optimal solution that meets constraints and minimizes costs.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a distributed energy resource intelligent scheduling method based on the Internet of Things, which includes:

[0008] Collect and pre-process the operating data of distributed energy devices to obtain the device operation data set, and perform swarm intelligence initialization on the device operation data set to generate the initial pheromone matrix;

[0009] Perform local optimization on the equipment operation data set and the initial pheromone matrix to generate an updated pheromone matrix and candidate equipment scheduling solutions, and obtain the local optimal scheduling solution through screening and verification;

[0010] The quantum computing node receives the local optimal scheduling solution and uses it as the initial solution to construct a global optimization problem. The global optimal scheduling solution is output and converted into a global scheduling instruction.

[0011] Global scheduling instructions are sent to distributed energy devices through the Internet of Things and executed, and the device execution status data set is obtained and evaluated to generate a scheduling feedback report.

[0012] As a preferred solution of the distributed energy resource intelligent scheduling method based on the Internet of Things described in the present invention, the obtaining of the equipment operation data set refers to using the Internet of Things sensor to collect real-time operation data from the distributed energy equipment, and performing denoising and integrity check operations on the collected real-time operation data to obtain the equipment operation data set.

[0013] As a preferred solution of the distributed energy resource intelligent scheduling method based on the Internet of Things described in the present invention, the generating of the initial pheromone matrix refers to extracting parameters based on the equipment operation data set, establishing the initial pheromone matrix, calculating the initial pheromone concentration of each pair in the initial pheromone matrix and assigning the roles of the distributed energy devices, and completing the establishment of the initial pheromone matrix.

[0014] As a preferred solution of the distributed energy resource intelligent scheduling method based on the Internet of Things of the present invention, wherein: generating the updated pheromone matrix and the device candidate scheduling scheme refers to using an improved ant colony algorithm to optimize the device operation data set and the initial pheromone matrix to generate candidate output values;

[0015] The edge computing node receives the candidate output values ​​and uses the reverse learning mechanism to update the initial pheromone matrix to generate an updated pheromone matrix and device candidate scheduling schemes.

[0016] As a preferred solution of the distributed energy resource intelligent scheduling method based on the Internet of Things of the present invention, wherein: the generating of the local optimal scheduling solution by screening and verification refers to selecting the scheduling path with the highest pheromone concentration from the updated pheromone matrix, and extracting the candidate output value and operating status corresponding to the selected scheduling path from the local optimal scheduling solution to form a preliminary optimal solution;

[0017] Perform constraint verification on the preliminary optimal solution and use the preliminary optimal solution that passes the constraint verification as the local optimal scheduling solution.

[0018] As a preferred solution of the distributed energy resource intelligent scheduling method based on the Internet of Things described in the present invention, wherein: the quantum computing node is used to receive the local optimal scheduling solution, and the local optimal scheduling solution is used as the initial solution. Constructing the global optimization problem means mapping the local optimal scheduling solution to the initial spin state of the quantum Ising model and defining the Hamiltonian of the quantum Ising model;

[0019] The Lagrange multiplier method is used to embed constraints into the Hamiltonian and complete the construction of the global optimization problem.

[0020] As a preferred solution of the IoT-based distributed energy resource intelligent scheduling method of the present invention, the output of the global optimal scheduling solution and the conversion into a global scheduling instruction at the same time refer to performing quantum annealing within a set time and iteration, and each iteration generates a candidate solution;

[0021] Evaluate the Hamiltonian energy of each candidate solution, find the lowest energy state of the Hamiltonian, and obtain the initial global optimal scheduling solution;

[0022] The initial global optimal scheduling solution is optimized and constraint-verified using classical gradient descent to generate a global optimal scheduling solution, which includes the device IDs, output status, and power allocation of all distributed energy devices.

[0023] The output status and power allocation are mapped into executable commands according to the type of distributed energy equipment, and priorities are assigned to distributed energy equipment.

[0024] As a preferred solution of the distributed energy resource intelligent scheduling method based on the Internet of Things described in the present invention, the global scheduling instructions are sent to the distributed energy devices through the Internet of Things and executed, the device execution status data set is obtained, and the device execution status data set is evaluated to generate a scheduling feedback report, which specifically includes the following steps:

[0025] Perform constraint verification on the global scheduling instructions, parse and execute the verified global scheduling instructions, and obtain the device execution status data set;

[0026] The device execution status data is evaluated. If it is less than the expected value, dynamic optimization is performed. The dynamic optimization refers to using the trained LSTM to output the supply and demand trend of the next scheduling cycle, adjusting the quantum Ising model according to the supply and demand trend, and obtaining a scheduling feedback report.

[0027] In a second aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the intelligent scheduling method for distributed energy resources based on the Internet of Things as described in the first aspect of the present invention.

[0028] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the intelligent scheduling method for distributed energy resources based on the Internet of Things as described in the first aspect of the present invention.

[0029] The beneficial effects of the present invention are: creating a customized initial pheromone matrix reflecting economic (electricity price / load demand) and operational factors, while using role allocation to align optimization with device functions, then guiding local update optimization by prioritizing cost-effective device interactions and ensuring compliance with role-specific constraints, and finally embedding real conditions into the swarm intelligence framework to improve optimization efficiency, accelerate convergence to low-cost plans and ensure functional alignment; in addition, by formulating a global optimization problem, capturing network-wide device interactions, economic factors and multi-dimensional constraints, the overall optimization of large-scale energy networks is achieved, ensuring coordination between devices and constraint compliance, and preventing problems such as grid instability. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0031] Figure 1 This is a flow chart of the intelligent scheduling method for distributed energy resources based on the Internet of Things.

[0032] Figure 2 This is the architecture diagram for intelligent scheduling of distributed energy resources.

[0033] Figure 3 Diagram of the local optimization and pheromone matrix updating mechanism.

[0034] Figure 4 Constructing a graph with the quantum annealing process for a global optimization problem. DETAILED DESCRIPTION

[0035] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0036] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0037] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0038] Reference Figures 1 to 4 , is an embodiment of the present invention, which provides a distributed energy resource intelligent scheduling method based on the Internet of Things, comprising the following steps:

[0039] S1. Collect and pre-process the operating data of distributed energy equipment to obtain the equipment operation data set, and perform swarm intelligence initialization on the equipment operation data set to generate an initial pheromone matrix.

[0040] The specific steps include:

[0041] S1.1. Equip distributed energy devices (e.g., photovoltaic panels, energy storage batteries, and electric vehicle charging stations) with IoT sensors. IoT sensors collect big data, gathering real-time operational data from distributed energy devices and generating massive, high-frequency data streams. Big data collection covers a wide range of topics.

[0042] For example, information includes photovoltaic power generation, energy storage battery state of charge, electric vehicle charging requirements, user-side load requirements, and real-time grid electricity prices. This reflects the scale of big data collection. IoT sensors encapsulate the collected raw data into data packets and transmit this data to local edge computing nodes via low-power wide area networks.

[0043] S1.2. Edge computing nodes preprocess the raw data generated by big data collection and perform denoising. This denoising process uses a Kalman filter algorithm. For example, this algorithm removes the effects of sensor noise and environmental interference (such as light fluctuations and grid voltage fluctuations) on data such as photovoltaic power generation, energy storage battery state of charge, electric vehicle charging requirements, and user-side load requirements. The Kalman filter algorithm, based on a state-space model, takes as input the raw data and the noise covariance (set to 0.01) to generate denoised data.

[0044] The denoised data is integrity checked to ensure the quality of the collected data. For example, the following rules apply: PV power generation must not be negative (if less than 0, it is marked as an anomaly); the energy storage battery state of charge must range from 0% to 100% (if it exceeds this, it is marked as an anomaly); EV charging demand and user-side load demand must not be negative; and the real-time grid electricity price must not be less than 0 yuan / kWh. The edge computing node traverses each data packet and checks the above rules. If an abnormal data point is found (for example, PV power generation of -0.1kW), it is removed and the data is supplemented by linear interpolation from the last 5 seconds of cached data. Data that passes the integrity check is formatted and stored in a standardized format, resulting in a device operation dataset containing the device ID, data type, value, and timestamp.

[0045] Preferably, high-precision and real-time data collection is achieved through big data collection. Deploying IoT sensors on distributed energy equipment can collect big data and generate high-frequency, large-scale data streams to capture operating parameters. This high-precision, real-time data collection provides a reliable foundation for subsequent intelligent scheduling, and can make accurate decisions in dynamic energy environments where conditions such as light intensity or grid demand fluctuate rapidly.

[0046] S1.2. Extract initial electricity prices and load demands from the device operating data set to initialize the pheromone matrix for the ant colony algorithm used in swarm intelligence. The pheromone matrix initialization process calculates the initial pheromone concentration for each pair of relationships between distributed energy devices. The pheromone matrix is ​​a two-dimensional array whose dimensions are the number of distributed energy devices multiplied by the number of distributed energy devices. For example, 1000 distributed energy devices correspond to a 1000-row by 1000-column matrix. Each matrix element represents the pheromone concentration from distributed energy device 1 to distributed energy device 2.

[0047] The initial pheromone concentration is calculated using the following steps: First, a constant (e.g., 10, used to normalize the pheromone concentration) is multiplied by the communication delay (in milliseconds) between distributed energy devices i and j to obtain a product. Then, this product is divided by 1 to obtain the base pheromone concentration. Communication delay is measured by measuring the round-trip time of MQTT (Message Queuing Telemetry Transport) messages at the edge computing node. Specifically, the edge computing node sends a test message to distributed energy devices i and j, records the difference between the message sending time and the message receiving time, repeats this process, and takes the average of these time differences.

[0048] To reflect the economic and load impact of the initial electricity price and load demand, the base pheromone concentration is further weighted and adjusted. The adjustment process is as follows: take the initial electricity price (unit: yuan per kilowatt-hour) and divide it by the load demand (unit: kilowatt) to obtain the ratio of electricity price to load; multiply this ratio by the weighting factor of 0.1 to obtain the weighted term; add 1 to the weighted term to obtain the adjustment factor; multiply the base pheromone concentration by the adjustment factor to obtain the final initial pheromone concentration, complete the initial pheromone matrix, and store it.

[0049] Distributed energy device roles are assigned based on device operation datasets. Specifically, the device ID and data type are extracted from the standardized device operation dataset to assign a unique identifier and role to each distributed energy device. The unique identifier directly uses the device ID in the standardized device operation dataset to ensure consistency. Role assignment is based on device type, with the following rules: photovoltaic panels are assigned the "generation" role, energy storage batteries are assigned the "energy storage" role, and electric vehicle charging stations and user-side loads are assigned the "load" role.

[0050] Preferably, the pheromone matrix of the ant colony algorithm of swarm intelligence is initialized using data extracted from the equipment operation data set, ensuring that the initial pheromone matrix accurately reflects the economic and operating environment of the distributed energy equipment.

[0051] S2. Use the improved ant colony algorithm to perform local optimization on the equipment operation data set and the initial pheromone matrix to generate an updated pheromone matrix and equipment candidate scheduling scheme, and obtain the local optimal scheduling scheme through screening and verification.

[0052] The specific steps include:

[0053] S2.1 extracts electricity prices, load demands, and energy storage battery state of charge from the device operating data set as external inputs. The edge computing node drives an improved ant colony algorithm (ACA) to perform local optimization calculations, treating each distributed energy device as an "ant." Each distributed energy device calculates an output adjustment strategy based on the initial pheromone matrix and external inputs. The ACA aims to minimize the dispatch cost in the local area. The dispatch cost is calculated as follows: for all distributed energy devices, multiply the electricity price by the output (kilowatts) of the distributed energy device, then sum all these products to obtain the total cost (yuan). The optimization process must meet the supply and demand balance requirement. This means that the absolute value of the difference between the sum of the outputs of all distributed energy devices and the load demand, divided by the load demand, must be less than 0.001 (with an error of less than 0.1%). Furthermore, the device capacity constraint must be met. The output of each distributed energy device must not exceed its rated power. Distributed energy devices exchange pheromone updates through communication, and the edge computing node executes the algorithm iterations in real time, calculating the candidate output value and operating status (e.g., whether the storage battery is charging, discharging, or in idle mode) for each distributed energy device. The candidate output values ​​select the highest probability path based on the pheromone concentration, and the path probability is determined by the pheromone concentration and the heuristic factor (the ratio of electricity price to load demand).

[0054] S2.2. The edge computing node receives the candidate output value and the initial pheromone matrix, executes the pheromone update of the improved ant colony algorithm, and generates a local optimal scheduling solution:

[0055] Pheromone updates utilize a reverse learning mechanism. The update process is as follows: For each element of the pheromone matrix (representing the pheromone concentration from distributed energy device i to distributed energy device j), the current pheromone concentration is first multiplied by 0.9 (1 minus the volatility factor of 0.1) to obtain the volatilized concentration. The optimal path increment is then added, along with a reverse learning term. The reverse learning term is calculated by dividing 1 by the absolute value of the worst path increment, then multiplying by the reverse learning weight of 0.05. The optimal path increment is calculated by dividing the constant 1 by the minimum dispatch cost to obtain the optimal path increment. The minimum dispatch cost is based on the lowest-cost path among the candidate output values ​​(the cost is the sum of the product of the electricity price and the output). The worst path increment is calculated by dividing the constant 1 by the highest dispatch cost to obtain the increment. The highest dispatch cost is based on the highest-cost path among the candidate output values. This results in the updated pheromone matrix and candidate device dispatch solutions (device ID, candidate output value, operating status, and timestamp).

[0056] The updated pheromone matrix and candidate device scheduling solutions are screened and verified to generate a locally optimal scheduling solution. The updated pheromone matrix is ​​verified to ensure that all elements are non-negative. Then, candidate device scheduling solutions are screened based on the updated pheromone matrix. Specifically, for each distributed energy device, the path with the highest pheromone concentration is selected, and its candidate output value and operating status are determined as the preliminary optimal solution. This preliminary optimal solution is then subjected to constraint verification to verify supply-demand balance constraints and device capacity constraints. The preliminary optimal solution that passes verification is confirmed as the locally optimal scheduling solution.

[0057] Ideally, an improved ant colony algorithm optimizes local dispatch to minimize dispatch costs, ensuring cost-effective energy distribution within the local area. It dynamically adapts to economic and operational conditions to reduce energy costs (for example, prioritizing low-cost distributed energy devices during periods of high electricity prices).

[0058] S3. Use quantum computing nodes to receive the local optimal scheduling plan, and use the local optimal scheduling plan as the initial solution to construct a global optimization problem, output the global optimal scheduling plan, and convert it into a global scheduling instruction at the same time.

[0059] The specific steps include:

[0060] S3.1. The cloud-based quantum computing node receives the locally optimal scheduling solution uploaded by the edge computing node. The locally optimal scheduling solution includes the output value and status (e.g., whether the storage battery is charging, discharging, or idle) of each distributed energy device (e.g., photovoltaic panels, energy storage batteries, and electric vehicle charging stations). The locally optimal scheduling solution is used as the initial solution to construct the global optimization problem.

[0061] Cloud-based quantum computing nodes map the distributed energy resource scheduling problem to a quantum Ising model for global optimization. The Hamiltonian of the quantum Ising model is defined as the sum of two components: the first component represents the interaction between all pairs of distributed energy devices (DERs). This component is calculated as the coupling coefficient for each pair (DERs i and j) multiplied by the product of the output state of DERs i and j, negated and summed. The second component represents the external field of each DER. This component is calculated as the external field coefficient multiplied by the output state of each DER, negated and summed. The output state of a DER is represented as -1 (off) or +1 (on). The coupling coefficient is dynamically generated based on the energy complementarity ratio and the synergy between photovoltaics and energy storage. It is calculated as: a constant (e.g., 0.1) multiplied by the product of the output of DERs i and j, divided by the communication latency between them (in milliseconds, measured by measuring the MQTT message round-trip time at the edge computing node).

[0062] For example, if distributed energy device i has an output of 10 kilowatts, distributed energy device j has an output of 5 kilowatts, and the communication delay is 2 milliseconds, then the coupling coefficient is 0.1 multiplied by 10 multiplied by 5 divided by 2, which equals 2.5. The external field coefficient is determined by the device capacity and electricity price and is calculated by multiplying a constant (e.g., 0.05) by the product of the device capacity of distributed energy device i and the electricity price. For example, if distributed energy device i has a capacity of 20 kilowatts and the electricity price is 0.5 yuan per kilowatt-hour, then the external field coefficient is 0.05 multiplied by 20 multiplied by 0.5, which equals 0.5.

[0063] The global optimization problem must satisfy multidimensional constraints, including supply-demand balance, distributed energy device capacity, and grid stability (for example, voltage fluctuations less than ±5%). These constraints are embedded in the Hamiltonian of the quantum Ising model using the Lagrange multiplier method. A penalty function is calculated by multiplying a constant (for example, 10) by the square of the difference between the sum of all distributed energy device outputs and the load demand. This penalty function is used to embed the supply-demand balance constraints and specifically to address the deviation between the sum of all distributed energy device outputs and the load demand in the global optimization problem.

[0064] For example, if the total output of distributed energy devices is 100 kilowatts and the load demand is 102 kilowatts, the penalty function is 10 times (100 minus 102) squared, which equals 40. The cloud-based quantum computing node maps the output value and state of the local optimal scheduling solution to the initial spin state of the quantum Ising model (-1 or +1) and adjusts the Hamiltonian parameters based on the constraints to complete the construction of the global optimization problem.

[0065] It is further shown that integrating multidimensional constraints into the Hamiltonian of the quantum Ising model via Lagrangian ensures that the globally optimal scheduling scheme is both feasible and stable, thus preventing problems such as power shortages or grid instability in large-scale energy networks.

[0066] S3.2. The quantum annealing processor begins with the initial spin state of the locally optimal schedule and performs an annealing process, gradually reducing the energy of the quantum system to find the lowest energy state of the Hamiltonian. This lowest energy state corresponds to the globally optimal schedule, representing the output state and power distribution of all distributed energy resources. The annealing time and number of iterations (e.g., 500 microseconds, repeated 100 times) are set to obtain the lowest energy solution. Each annealing step generates a candidate solution containing the spin state (-1 or +1) and the corresponding output value of each distributed energy resource (based on the capacity and state mapping of the distributed energy resources; for example, a +1 state corresponds to 80% of the rated power). The Hamiltonian energy of each candidate solution is evaluated. The energy calculation includes the interaction term (the sum of the product of the coupling coefficient and the spin state), the external field term (the sum of the external field coefficient and the spin state), and the penalty function term. For example, if the supply-demand deviation of a candidate solution is 2 kilowatts, the Hamiltonian energy includes a penalty function term of 10 times 2 squared, which equals 40. Increasing the energy value decreases the probability of the solution being selected. The solution with the lowest energy is selected from the number of annealing times as the initial global optimal scheduling solution for quantum annealing.

[0067] Preferably, high-precision optimization using quantum annealing significantly improves the accuracy and speed of finding the lowest energy state of the Hamiltonian, which corresponds to the globally optimal scheduling solution.

[0068] To address quantum computing noise (e.g., qubit decoherence or thermal noise), a hybrid quantum-classical algorithm is used to further optimize the initial solution from quantum annealing. Using the initial global optimal scheduling solution provided by quantum annealing as a starting point, a classical gradient descent algorithm iteratively adjusts the output to minimize the local error (defined as the deviation of the Hamiltonian energy from the ideal minimum energy). Classical gradient descent adjusts the output of each distributed energy resource (DER) based on the Hamiltonian's gradient (partial derivative with respect to the output). The number of iterations is set, for example 10, and the output adjustment for all DERs is calculated each iteration. The adjustment is determined by multiplying the gradient by a learning rate (e.g., 0.01). The optimized solution includes the device ID, output status (on or off), and power allocation for each DER. Constraint verification (supply-demand balance constraints, DER capacity constraints, and grid stability constraints) is performed on the optimized solution. Solutions that pass verification are considered the global optimal scheduling solution, including the device ID, output status, and power allocation for all DERs.

[0069] S3.3. Convert the global optimal scheduling plan into a global scheduling instruction, traverse each distributed energy device in the global optimal scheduling plan, retain the distributed energy device ID and output value, and assign a priority based on the device type and output value. The priority is an integer ranging from 1 to 10. For example, when the output of the photovoltaic panel is greater than 50% of its rated power, the priority is set to 8; when the energy storage battery is discharging, the priority is set to 6; and the priority of the electric vehicle charging pile is set to 4.

[0070] S4. Send global dispatch instructions to distributed energy devices through the Internet of Things and execute them to obtain a device execution status data set;

[0071] The specific steps include:

[0072] S4.1. To ensure the feasibility of global dispatch instructions under local conditions in the real world, the feasibility of global dispatch instructions needs to be verified. The verification uses linear constraint checks, which include two constraints: first, the capacity constraint of distributed energy devices, which ensures that the output value of each distributed energy device does not exceed its rated power; second, the supply and demand balance constraint, which ensures that the absolute difference between the sum of the outputs of all distributed energy devices and the load demand divided by the load demand is less than 0.001 (the error is less than 0.1%). If the global dispatch instruction is not feasible (for example, overload or the supply and demand error exceeds the standard), a backup instruction is generated from the candidate device dispatch plan. The backup instruction selects the path with the second highest pheromone concentration from the candidate device dispatch plan and adjusts the output value to meet the device capacity constraint and the supply and demand balance constraint.

[0073] Verified global dispatch instructions or backup instructions are issued to distributed energy devices. The distributed energy devices receive and interpret the instructions and execute the corresponding actions. The photovoltaic panel inverter adjusts its output power based on the instructions, for example, from 7 kW to 8.5 kW. The energy storage battery switches its operating mode (charging, discharging, idle) based on the instructions, for example, from charging to discharging. Distributed energy devices upload their execution status once per second. Edge computing nodes receive and calculate dispatch efficiency in real time, including energy utilization and cost savings. Energy utilization is calculated by dividing actual total output by load demand, with a target of >95%. Cost savings is calculated by comparing the cost of electricity prices with actual output, with a target of >10%. The execution status of all distributed energy devices is aggregated to generate a device execution status dataset, which includes device ID, actual output, status, and timestamp.

[0074] S5. Evaluate the equipment execution status data set and generate a scheduling feedback report.

[0075] The specific steps include:

[0076] S5.1. Calculate the objective function value based on the device execution status dataset to evaluate the scheduling effectiveness, including three indicators: cost, energy utilization, and grid stability. Compare the objective function value with the expected value: the expected value is: cost below budget, energy utilization above 90%, and voltage fluctuation within ±5%. If any indicator deviates by more than 0.1%, dynamic optimization is triggered.

[0077] Dynamic optimization uses an LSTM (Long Short-Term Memory) network to predict supply and demand trends for the next scheduling period (e.g., one hour). The LSTM network input is a 24-hour data set of device execution status, including the actual output, status, load demand, and electricity price of each distributed energy resource (DER) device. This data is normalized to a range of 0 to 1. The LSTM network is configured with three layers and uses the tanh activation function. Training weights are based on historical data, such as seven days of data. The LSTM network output is the predicted power generation and load demand for the next hour. Based on these predicted power generation and load demand, the cloud-based quantum computing nodes adjust the swarm intelligence parameters and the quantum Ising model to optimize the next scheduling period.

[0078] The adjusted parameters are sent to the edge computing node, updating the pheromone matrix and the Hamiltonian of the quantum Ising model. The cloud-based quantum computing node generates a feedback report to evaluate scheduling performance. Scheduling efficiency includes energy utilization, cost savings, and prediction error. The prediction error is the deviation between the long-short-term memory network's predicted value and the actual value. The feedback report contains the following fields: energy utilization, cost savings, prediction error, and timestamp. This completes the scheduling task.

[0079] This embodiment also provides a computer device, which is suitable for the case of an intelligent scheduling method for distributed energy resources based on the Internet of Things, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the intelligent scheduling method for distributed energy resources based on the Internet of Things proposed in the above embodiment.

[0080] The computer device may be a terminal, comprising a processor, memory, a communication interface, a display, and an input device connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system and computer programs. The internal memory provides an environment for the operating system and computer programs stored in the non-volatile storage media. The communication interface of the computer device is used to communicate with external terminals via wired or wireless communication. Wireless communication may be achieved via Wi-Fi, a carrier network, NFC (near-field communication), or other technologies. The display of the computer device may be a liquid crystal display or an electronic ink display. The input device may be a touchscreen overlay on the display, buttons, a trackball, or a touchpad on the computer device housing, or an external keyboard, touchpad, or mouse.

[0081] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for intelligent scheduling of distributed energy resources based on the Internet of Things as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0082] In summary, the present invention creates a customized initial pheromone matrix that reflects economic (electricity price / load demand) and operational factors, while utilizing role assignment to align optimization with device functionality. It then guides local update optimization by prioritizing cost-effective device interactions and ensuring compliance with role-specific constraints. Finally, it embeds real-world conditions into a swarm intelligence framework to improve optimization efficiency, accelerate convergence to low-cost plans, and ensure functional alignment. Furthermore, by formulating a global optimization problem that captures network-wide device interactions, economic factors, and multidimensional constraints, it achieves holistic optimization of large-scale energy networks, ensures coordination between devices and constraint compliance, and prevents problems such as grid instability.

[0083] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A distributed energy resource intelligent scheduling method based on the Internet of Things, characterized by: include, Collect and pre-process the operating data of distributed energy devices to obtain the device operation data set, and perform swarm intelligence initialization on the device operation data set to generate the initial pheromone matrix; Perform local optimization on the equipment operation data set and the initial pheromone matrix to generate an updated pheromone matrix and candidate equipment scheduling solutions, and obtain the local optimal scheduling solution through screening and verification; The quantum computing node receives the local optimal scheduling solution and uses it as the initial solution to construct a global optimization problem. The global optimal scheduling solution is output and converted into a global scheduling instruction. The cloud-based quantum computing node receives the local optimal scheduling solution uploaded by the edge computing node. The local optimal scheduling solution includes the output value and status of each distributed energy device. The local optimal scheduling solution is used as the initial solution to construct the global optimization problem. Cloud-based quantum computing nodes map the distributed energy resource scheduling problem to a quantum Ising model to achieve global optimization. The Hamiltonian of the quantum Ising model is defined as the sum of two parts: the first part is the interaction between all pairs of distributed energy devices, calculated as the product of the coupling coefficient of distributed energy device i and distributed energy device j for each device pair, multiplied by the output state of distributed energy device i and the output state of distributed energy device j, and then summed up after taking the negative value; the second part is the external field of each distributed energy device, calculated as the external field coefficient of each distributed energy device multiplied by its output state, and then summed up after taking the negative value; The output status of distributed energy devices is represented as -1 off or +1 on; The coupling coefficient is calculated as follows: take a constant multiplied by the product of the output value of distributed energy device i and the output value of distributed energy device j, and then divide it by the communication delay between the two; The external field coefficient is determined by the device capacity and electricity price and is calculated as follows: take a constant and multiply the product of the device capacity and electricity price of distributed energy device i; The global optimization problem must satisfy multidimensional constraints, including supply and demand balance, distributed energy device capacity, and grid stability. These constraints are embedded in the Hamiltonian of the quantum Ising model using the Lagrange multiplier method. The penalty function is calculated as: taking a constant multiplied by the square of the difference between the sum of the outputs of all distributed energy devices and the load demand. The penalty function is used to embed the supply and demand balance constraints. The cloud-based quantum computing node maps the output value and state of the local optimal scheduling solution to the initial spin state of the quantum Ising model, and adjusts the Hamiltonian parameters according to the constraints to complete the construction of the global optimization problem. The quantum annealing processor starts with the initial spin state of the local optimal scheduling scheme and performs an annealing process to find the lowest energy state of the Hamiltonian by gradually reducing the energy of the quantum system. Set the annealing time and number to obtain the lowest energy solution. Each annealing generates a candidate solution, which includes the spin state and corresponding output value of each distributed energy device. Evaluate the Hamiltonian energy of each candidate solution. The energy calculation includes interaction terms, external field terms, and penalty function terms. The solution with the lowest energy is selected from the number of annealing times as the initial global optimal scheduling solution for quantum annealing; Global scheduling instructions are sent to distributed energy devices through the Internet of Things and executed, and the device execution status data set is obtained and evaluated to generate a scheduling feedback report.

2. The method for intelligent scheduling of distributed energy resources based on the Internet of Things according to claim 1, characterized in that: Obtaining the equipment operation data set refers to using IoT sensors to collect real-time operation data from distributed energy devices, and performing denoising and integrity check operations on the collected real-time operation data to obtain the equipment operation data set.

3. The method for intelligent scheduling of distributed energy resources based on the Internet of Things according to claim 2, characterized in that: Generating the initial pheromone matrix refers to extracting parameters from the equipment operation data set, establishing the initial pheromone matrix, calculating the initial pheromone concentration of each pair in the initial pheromone matrix and assigning the roles of distributed energy devices to complete the establishment of the initial pheromone matrix.

4. The method for intelligent scheduling of distributed energy resources based on the Internet of Things according to claim 3, characterized in that: Generating the updated pheromone matrix and the candidate equipment scheduling scheme refers to optimizing the equipment operation data set and the initial pheromone matrix using the improved ant colony algorithm to generate candidate output values; The edge computing node receives the candidate output values ​​and uses the reverse learning mechanism to update the initial pheromone matrix to generate the updated pheromone matrix and the candidate device scheduling scheme.

5. The method for intelligent scheduling of distributed energy resources based on the Internet of Things according to claim 4, characterized in that: The generating of the local optimal scheduling solution by screening and verification refers to verifying the updated pheromone matrix to ensure that all elements are non-negative, and then screening the candidate device scheduling solutions according to the updated pheromone matrix; For each distributed energy device, the path with the highest pheromone concentration is selected, and its candidate output value and operating status are determined as the preliminary optimal solution; Perform constraint verification on the preliminary optimal solution and use the preliminary optimal solution that passes the constraint verification as the local optimal scheduling solution.

6. The method for intelligent scheduling of distributed energy resources based on the Internet of Things according to claim 5, characterized in that: The outputting of the global optimal scheduling solution and converting it into a global scheduling instruction also include: The initial global optimal scheduling solution is optimized and constraint-verified using classical gradient descent to generate a global optimal scheduling solution, which includes the device IDs, output status, and power allocation of all distributed energy devices. The output status and power allocation are mapped into executable commands according to the type of distributed energy equipment, and priorities are assigned to distributed energy equipment.

7. The method for intelligent scheduling of distributed energy resources based on the Internet of Things according to claim 6, characterized in that: Send global dispatch instructions to distributed energy devices through the Internet of Things and execute them, obtain the device execution status data set, evaluate it, and generate a dispatch feedback report. The specific steps include the following: Perform constraint verification on the global scheduling instructions, parse and execute the verified global scheduling instructions, and obtain the device execution status data set; The device execution status data is evaluated. If it is less than the expected value, dynamic optimization is performed. The dynamic optimization refers to using the trained LSTM to output the supply and demand trend of the next scheduling cycle, adjusting the quantum Ising model according to the supply and demand trend, and obtaining a scheduling feedback report.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the distributed energy resource intelligent scheduling method based on the Internet of Things according to any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the distributed energy resource intelligent scheduling method based on the Internet of Things according to any one of claims 1 to 7 are implemented.

Images