Distributed energy collaborative operation method based on industrial internet

Abstract

The invention discloses a distributed energy collaborative operation method based on the industrial internet, and relates to the technical field of distributed control, and the method comprises the steps: collecting the original physical quantity data of a distributed energy terminal, and carrying out the local event discrimination and formatting processing, and obtaining multi-source information; the edge intelligent gateway performs data fusion and autonomous collaborative decision making on the multi-source information through a local AI prediction and event triggering mechanism to generate a local decision making instruction set; and the cloud platform receives the local decision instruction set, constructs a global AI optimization model, performs multi-target calculation on the local decision instruction set by using the global AI optimization model, generates a preliminary scheduling plan, and performs security check and strategy fusion on the preliminary scheduling plan to obtain an authoritative collaborative instruction set. According to the method, the instruction version identifier and the preset execution constraint condition are uniformly bound and are subjected to uplink evidence storage, so that the collaborative scheduling instruction has clear version traceability and execution consistency in the transmission and execution process, and instruction drifting or cross-border execution is avoided.

Description

Technical Field

[0001] This invention relates to the field of distributed control technology, and in particular to a method for the coordinated operation of distributed energy resources based on the Industrial Internet. Background Technology

[0002] With the large-scale integration of new energy power generation, energy storage devices, and controllable loads, the power system is gradually evolving from a centralized dispatch mode to a multi-source collaborative operation mode. Distributed energy systems are increasingly widely used in distribution networks and microgrids. Regarding the operation and management of distributed energy, related technologies are gradually evolving from single-point control to a multi-level collaborative control architecture. Through edge computing, the Industrial Internet, and artificial intelligence technologies, rapid response at the terminal side and global optimization at the cloud side are achieved. In this process, the concept of distributed control has been widely introduced, becoming an important development direction for the operation and control of new energy systems.

[0003] While existing distributed control technologies offer advantages in response speed and local autonomy, they still face challenges in achieving globally consistent scheduling while ensuring terminal autonomy in multi-terminal collaborative operation scenarios. This is particularly true regarding the secure transmission of scheduling instructions, the reliability of execution, and the feedback of execution results to the global strategy, where a unified closed-loop technical mechanism is lacking. Existing solutions often focus on a single dimension of local decision-making and centralized optimization, failing to simultaneously consider global optimization, execution constraints, and dynamic feedback within a distributed control structure. This limits the collaborative efficiency and operational reliability of distributed energy systems in complex operating environments. Summary of the Invention

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

[0005] Therefore, this invention provides a distributed energy collaborative operation method based on the Industrial Internet to solve the problem that it is difficult to form a secure, reliable and dynamically optimizable closed loop between autonomous decision-making of distributed energy terminals and global collaborative scheduling under a distributed control architecture.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a method for the collaborative operation of distributed energy resources based on the Industrial Internet. The method includes: collecting raw physical quantity data from distributed energy terminals and performing local event discrimination and formatting processing to obtain multi-source information; an edge intelligent gateway performing data fusion and autonomous collaborative decision-making on the multi-source information through a local AI prediction and event triggering mechanism to generate a local decision instruction set; a cloud platform receiving the local decision instruction set, constructing a global AI optimization model, using the global AI optimization model to perform multi-objective calculations on the local decision instruction set to generate a preliminary scheduling plan, and performing security verification and strategy fusion on the preliminary scheduling plan to obtain an authoritative collaborative instruction set; distributing the authoritative collaborative instruction set to the blockchain network, and binding instruction version identifiers and execution constraints to the authoritative collaborative instruction set during the distribution process to form an authoritative collaborative instruction record, and using a smart contract to perform trusted storage and execution condition verification of the authoritative collaborative instruction record to generate device-level execution instructions; each distributed energy terminal receiving and executing the device-level execution instructions, using end-to-end sensors to collect execution feedback data, and dynamically optimizing the parameters of the global AI optimization model based on the execution feedback data.

[0007] As a preferred embodiment of the distributed energy collaborative operation method based on the Industrial Internet described in this invention, the steps for obtaining multi-source information are as follows: The raw three-phase voltage and current data of each distributed energy terminal are collected and high-precision synchronous sampling and time-scale alignment are performed to obtain the synchronous sampling sequence; Real-time filtering and denoising of the synchronous sampling sequence yields an electrical characteristic sequence. Perform morphological gradient-based mutation detection on electrical feature sequences to generate local event alerts; The electrical characteristic sequence and local event alarm are encapsulated in a predefined format to generate multi-source information.

[0008] As a preferred embodiment of the distributed energy collaborative operation method based on the Industrial Internet described in this invention, the steps for generating a local decision instruction set are as follows: Based on multi-source information, the edge intelligent gateway uses a local AI prediction model to analyze electrical characteristic quantity sequences in real time, extract key information, and generate a fused information set. Based on the fused information set, the edge intelligent gateway uses an event-triggered mechanism to automatically trigger collaborative decision-making and generate a set of candidate decision solutions. The candidate decision scheme set is optimized by multi-objective optimization using deep reinforcement learning algorithm to determine the optimal decision strategy, and the candidate decision scheme is converted into a candidate decision instruction set under the constraint of the optimal decision strategy. The candidate decision instruction sets are subjected to security verification and policy fusion to generate a local decision instruction set.

[0009] As a preferred embodiment of the distributed energy collaborative operation method based on the Industrial Internet described in this invention, the steps for constructing the global AI optimization model are as follows: The local decision instruction set is cleaned and normalized to generate a spatiotemporally correlated sample set; By performing multi-layer convolution and non-linear mapping on the spatiotemporally correlated sample set through a convolutional neural network, spatiotemporal correlation features are extracted to form a high-dimensional feature vector. Based on high-dimensional feature vectors, a global optimization problem is constructed, and a genetic algorithm is used to solve the global optimization problem in multiple objectives, determine the optimal parameter configuration, and generate a global AI optimization model.

[0010] As a preferred embodiment of the distributed energy collaborative operation method based on the Industrial Internet described in this invention, the steps for generating a preliminary scheduling plan are as follows: The execution time of the local decision instruction set is aligned, the scale is unified, and the spatiotemporal correlation is established to form a spatiotemporally correlated input set. The spatiotemporal coupling features in the spatiotemporal correlation input set are extracted by a global AI optimization model to determine the target weights and operational constraint parameters required for multi-objective optimization. Based on the target weights and operational constraint parameters, a global AI optimization model is used to perform multi-objective optimization calculations on the local decision instruction set to generate a preliminary scheduling plan.

[0011] As a preferred embodiment of the distributed energy collaborative operation method based on the Industrial Internet described in this invention, the steps for obtaining the authoritative collaborative instruction set are as follows: The scheduling parameters of each distributed energy terminal in the preliminary scheduling plan are extracted, and a scheduling status dataset is constructed by combining the operating status of the distributed energy terminals. Perform runtime constraint checks on the scheduling parameters in the scheduling status dataset and generate safety verification information; For scheduling parameters with constraint violations in the security verification information, targeted correction is performed, and policy consistency fusion is performed on the scheduling parameters during the correction to generate a fused scheduling parameter set; The scheduling parameters in the fusion scheduling parameter set are verified for integrity and uniformly encapsulated to obtain the authoritative collaborative instruction set.

[0012] As a preferred embodiment of the distributed energy collaborative operation method based on the Industrial Internet described in this invention, the steps for forming an authoritative collaborative instruction record are as follows: A unique instruction version identifier is added to the scheduling parameters of each distributed energy terminal in the authoritative collaborative instruction set, and the instruction version identifier is bound to the corresponding scheduling parameter to form the content of the authoritative collaborative instruction. The scheduling parameters in the authoritative collaborative instruction are bound to corresponding preset execution constraints, and the scheduling parameters, instruction version identifiers and execution constraints are structurally associated to form authoritative collaborative instruction data. Authoritative collaborative instruction data is sent to the blockchain network for consistency confirmation, forming an authoritative collaborative instruction record.

[0013] As a preferred embodiment of the distributed energy collaborative operation method based on the Industrial Internet described in this invention, the steps for generating device-level execution instructions are as follows: Based on authoritative collaborative instruction records, the corresponding smart contract in the blockchain network is triggered, and scheduling parameters, instruction version identifiers and execution constraints are loaded into the smart contract to form an instruction verification context; Based on the instruction verification context, the smart contract verifies the execution conditions of each scheduling parameter corresponding to each distributed energy terminal in the authoritative collaborative instruction record and generates a set of verification status identifiers. The smart contract filters and verifies the scheduling parameters that meet the execution constraints in the status identifier set, and then encapsulates the filtered scheduling parameters under the corresponding execution constraints to generate device-level execution instructions.

[0014] As a preferred embodiment of the distributed energy collaborative operation method based on the Industrial Internet described in this invention, the steps for collecting execution feedback data using end-to-end sensors are as follows: Each distributed energy terminal receives the device-level execution instruction and performs a consistency check on the instruction version identifier in the device-level execution instruction to confirm that the device-level execution instruction is the current valid instruction; Update the control parameters of the terminal's current operating status according to the scheduling parameters in the current valid instructions, and continuously collect the operating status of the terminal during the execution process using end-to-end sensors to form execution feedback data.

[0015] As a preferred embodiment of the distributed energy collaborative operation method based on the Industrial Internet described in this invention, the step of dynamically optimizing the global AI optimization model parameters based on execution feedback data includes the following steps: Based on the execution feedback data, the execution deviation between the actual operating state of the terminal and the corresponding scheduling parameters in the device-level execution instructions is calculated, and execution deviation data is generated. Based on execution deviation data, the parameters of the global AI optimization model are optimized.

[0016] The beneficial effects of this invention are as follows: By using smart contracts to reliably store and verify the execution conditions of authoritative collaborative instruction records and generate device-level execution instructions, a reliable transition and constraint implementation of scheduling instructions from global generation to terminal execution can be achieved in a distributed control architecture; by uniformly binding instruction version identifiers with preset execution constraints and storing them on the blockchain, collaborative scheduling instructions have clear version traceability and execution consistency during transmission and execution, avoiding instruction drift or out-of-bounds execution; at the same time, the automatic verification mechanism based on smart contracts ensures that the generation of device-level execution instructions does not rely on manual intervention, ensuring the objective implementation of scheduling strategies on the distributed terminal side, and providing a reliable data foundation for the collection of execution feedback data and the dynamic optimization of global model parameters, thereby enhancing the stability and controllability of collaborative operation under the distributed control system. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a distributed energy collaborative operation method based on the Industrial Internet.

[0019] Figure 2 A flowchart for generating a local decision instruction set.

[0020] Figure 3 A flowchart for generating an authoritative collaborative instruction set.

[0021] Figure 4 A flowchart for optimizing global AI model parameters. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used 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 different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4 This is one embodiment of the present invention, which provides a method for the coordinated operation of distributed energy resources based on the Industrial Internet, including the following steps: S1. Collect raw physical quantity data from distributed energy terminals, and perform local event discrimination and formatting processing to obtain multi-source information.

[0026] S1.1: Collect the raw three-phase voltage and current data of each distributed energy terminal, and perform high-precision synchronous sampling and time-scale alignment to obtain the synchronous sampling sequence; Specifically, the raw three-phase voltage and current are synchronously collected at each distributed energy terminal, and a unified time identifier is attached to each set of raw three-phase voltage and current data during the collection process. Based on the time identifier, the raw three-phase voltage and current data of different distributed energy terminals are synchronously sampled and organized to keep the raw three-phase voltage and current data corresponding to the same time identifier consistent in time. The raw three-phase voltage and current data with unified time identifiers are time-aligned so that the three-phase voltage and current data of different distributed energy terminals at the same time position form a corresponding relationship, resulting in a synchronous sampling sequence arranged in chronological order.

[0027] S1.2: Perform real-time filtering and denoising on the synchronous sampling sequence to obtain the electrical characteristic quantity sequence; Specifically, real-time filtering is continuously applied to the original three-phase voltage and current data in the synchronous sampling sequence in chronological order to suppress high-frequency interference and random fluctuations superimposed during the sampling process. Based on the real-time filtering, the original three-phase voltage and current data is denoised to keep the original three-phase voltage and current data in the synchronous sampling sequence smooth and continuous. Based on the original three-phase voltage and current data after real-time filtering and denoising, the voltage amplitude, current amplitude, and phase difference between voltage and current are calculated and summarized into electrical characteristic quantities. The electrical characteristic quantities are then arranged continuously in chronological order to form an electrical characteristic quantity sequence.

[0028] The expressions for calculating the voltage amplitude, current amplitude, and phase difference between voltage and current in the original three-phase voltage and current data are as follows:

[0029] Wherein, U represents the voltage amplitude; u1 represents the instantaneous sampled value of the first phase voltage in the original three-phase voltage and current data at the synchronous sampling moment; u2 represents the instantaneous sampled value of the second phase voltage in the original three-phase voltage and current data at the synchronous sampling moment; u3 represents the instantaneous sampled value of the third phase voltage in the original three-phase voltage and current data at the synchronous sampling moment; i1 represents the instantaneous sampled value of the first phase current in the original three-phase voltage and current data at the synchronous sampling moment; i2 represents the instantaneous sampled value of the second phase current in the original three-phase voltage and current data at the synchronous sampling moment; i3 represents the instantaneous sampled value of the third phase current in the original three-phase voltage and current data at the synchronous sampling moment; I represents the current amplitude; φ represents the phase difference between voltage and current.

[0030] S1.3: Perform morphological gradient-based mutation detection on the electrical feature sequence to generate local event alerts; Specifically, in the electrical feature quantity sequence, morphological dilation and morphological erosion operations are used to compare electrical feature quantities at the same time position, so that the amplitude differences caused by changes in operating state in the electrical feature quantity sequence can be presented; the time positions in the electrical feature quantity sequence where the amplitude difference is continuously prominent are marked, and the time positions are associated with the corresponding electrical feature quantity sequence states; when the amplitude difference exceeds the preset abrupt change judgment threshold at the corresponding time position and persists in the adjacent time positions, the corresponding time position and associated state information are summarized and marked with a local event alarm mark to form a local event alarm.

[0031] It should be noted that the mutation judgment threshold is set based on the distinction between the normal fluctuation amplitude of the electrical characteristic quantity sequence under stable operating conditions and the morphological gradient difference level when the operating state of the distributed energy terminal undergoes substantial changes. The specific setting steps include: under long-term stable operating conditions of the distributed energy terminal, statistically analyzing the morphological gradient difference level corresponding to the electrical characteristic quantity sequence to determine the difference distribution interval reflecting normal operating condition fluctuations; summarizing the morphological gradient difference levels corresponding to operating state switching, fault triggering, and load mutations during historical operation to form a state mutation difference distribution interval; and selecting a threshold that can stably distinguish the difference distribution interval from the state mutation difference interval. The minimum boundary difference between stable fluctuation differences and state abrupt changes is used as the abrupt change judgment threshold. An exemplary value range can be set within the transition range between the upper limit of the normal fluctuation difference distribution range and the lower limit of the state abrupt change difference distribution range. For example, the difference level corresponding to the region with the lowest overlap between the two difference distribution ranges can be used as a benchmark. When the morphological gradient difference level is lower than the abrupt change judgment threshold, it indicates that the change in electrical characteristic quantity is still within the normal fluctuation range of the stable operation state and does not trigger a local event alarm. When the morphological gradient difference level is higher than the abrupt change judgment threshold and persists in adjacent time positions, it indicates that the operating state of the distributed energy terminal has undergone a substantial change and triggers a local event alarm.

[0032] S1.4: Encapsulate the electrical characteristic sequence and local event alarms according to a predefined format to generate multi-source information.

[0033] Specifically, according to the time sequence of electrical characteristic quantities, the electrical characteristic quantity sequence is matched one-to-one with the local event alarm at the same time position, and the matched electrical characteristic quantity and local event alarm are sequentially written into fixed fields in a predefined format, so that the continuous change information of the electrical characteristic quantity sequence and the local event alarm identifier remain consistent and related, forming multi-source information.

[0034] It should be noted that the predefined format refers to the field arrangement and information correspondence rules uniformly determined based on the time identifier structure of the electrical characteristic quantity sequence in the distributed energy terminal and the identifier content of the local event alarm.

[0035] S2, the edge intelligent gateway uses local AI prediction and event triggering mechanisms to perform data fusion and autonomous collaborative decision-making on multi-source information, generating a local decision instruction set.

[0036] S2.1: Based on multi-source information, the edge intelligent gateway uses a local AI prediction model to perform real-time analysis of electrical characteristic quantity sequences, extract key information, and generate a fused information set; Specifically, the edge intelligent gateway uses a local AI prediction model to continuously discriminate and assess the state of the electrical feature quantity sequence based on the aligned electrical feature quantity sequence and local event alarm identifiers from multi-source information. It identifies key segments in the electrical feature quantity sequence that reflect changes in operating state and associates these key segments with the local event alarm identifiers at the corresponding time locations. After associating and organizing, the edge intelligent gateway aggregates the state change information in the electrical feature quantity sequence and the local event alarm identifiers in chronological order to form a fused information set containing the state characteristics of the electrical feature quantity sequence and event identifier information.

[0037] It should be noted that the training process of the local AI prediction model is as follows: Before the edge smart gateway is put into operation, the pre-training of the local AI prediction model is based on the synchronous sampling sequence, electrical characteristic quantity sequence and corresponding local event alarm identifier collected and formed during historical operation. During the pre-training process, the electrical characteristic quantity sequence and the local event alarm identifier are aligned and organized in chronological order to form a stable correspondence between changes in electrical characteristic quantity sequence and changes in operating status. The internal parameters of the local AI prediction model are repeatedly adjusted according to the stable correspondence. The consistency of the local AI prediction model's response to changes in the state of electrical characteristic quantity sequence is verified through multiple rounds of historical data, thus completing the pre-training of the local AI prediction model.

[0038] S2.2: Based on the fused information set, the edge intelligent gateway automatically triggers collaborative decision-making using an event-triggered mechanism to generate a set of candidate decision schemes; Specifically, the edge intelligent gateway checks the fused information set item by item in chronological order based on the electrical characteristic quantity sequence state characteristics and local event alarm identifiers contained in the fused information set. When the local event alarm identifier in the fused information set meets the conditions set by the event triggering mechanism, the event triggering mechanism automatically starts the collaborative decision-making process. After the collaborative decision-making is started, the edge intelligent gateway expands the feasible scheduling adjustment directions based on the electrical characteristic quantity sequence state characteristics and event identifier information in the fused information set, and forms a variety of executable scheduling schemes around different combinations of operating states. The various scheduling schemes corresponding to the fused information set are sorted and collected in sequence to form a candidate decision scheme set.

[0039] It should be noted that the event triggering mechanism is set based on the appearance, persistence, and disappearance of local event alarm identifiers in the fusion information set in chronological order. This is used to determine whether the operating status of the distributed energy terminal has entered a state that requires the initiation of collaborative decision-making.

[0040] S2.3: Perform multi-objective optimization on the candidate decision scheme set through deep reinforcement learning algorithm to determine the optimal decision strategy, and convert the candidate decision scheme into a candidate decision instruction set under the constraint of the optimal decision strategy; Specifically, based on the scheduling parameter combinations corresponding to each scheduling scheme in the candidate decision scheme set, as well as the corresponding electrical characteristic quantity sequence state features and local event alarm identifiers, a deep reinforcement learning algorithm is used to compare and evaluate the scheduling adaptability of the candidate decision scheme set under multi-objective constraints. This allows the scheduling orientation that can simultaneously adapt to multiple operational objectives to be gradually strengthened during the iteration process, forming the optimal decision strategy. The edge intelligent gateway then filters and constrains each scheduling parameter combination in the candidate decision scheme set according to the optimal decision strategy, and organizes the scheduling parameter combinations that conform to the optimal decision strategy into a unified instruction form, forming a candidate decision instruction set.

[0041] It should be noted that multi-objective constraints refer to the operational boundary requirements formed on the range of values, direction of change, and mutual combination relationships of scheduling parameters based on the state characteristics of electrical characteristic quantity sequences, local event alarm identifiers, and the operating status of distributed energy terminals in the fused information set.

[0042] S2.4: Perform security verification and policy fusion on the candidate decision instruction set to generate a local decision instruction set.

[0043] Specifically, based on the scheduling parameter combinations and corresponding electrical characteristic quantity sequence state characteristics and local event alarm identifiers in the candidate decision instruction set, a safety check is performed on each candidate decision instruction set; the scheduling parameters in the candidate decision instruction set are compared with multi-objective constraints, scheduling parameters that do not meet the multi-objective constraints are identified and adjusted accordingly, and the scheduling orientations reflected by different scheduling parameters in the candidate decision instruction set are unified and coordinated during the adjustment process to ensure that the adjusted scheduling parameters remain consistent under multi-objective constraints; the scheduling parameters that meet the multi-objective constraints and have completed strategy consistency fusion are uniformly organized to form a local decision instruction set.

[0044] S3: The cloud platform receives the local decision instruction set, builds a global AI optimization model, uses the global AI optimization model to perform multi-objective calculations on the local decision instruction set, generates a preliminary scheduling plan, and performs security verification and policy fusion on the preliminary scheduling plan to obtain an authoritative collaborative instruction set.

[0045] S3.1: Clean and normalize the local decision instruction set to generate a spatiotemporally correlated sample set; Specifically, the scheduling parameters in the local decision instruction set are sequentially arranged according to the time order contained therein, and incomplete or non-executable scheduling parameters are removed to ensure that the scheduling parameters in the local decision instruction set remain consistent and complete. The retained scheduling parameters are then arranged in a unified form according to a uniform time order to ensure that the scheduling parameters corresponding to different distributed energy terminals are consistent in both time order and scheduling parameter form. Finally, the scheduling parameters of each distributed energy terminal at continuous time locations are correlated to form a spatiotemporal correlation sample set that simultaneously reflects time correlation and terminal correlation.

[0046] S3.2: Multi-layer convolution and non-linear mapping are performed on the spatiotemporal correlated sample set through convolutional neural networks to extract spatiotemporal correlated features and form high-dimensional feature vectors; Specifically, the spatiotemporal correlated sample set is unfolded according to the order of continuous time positions. The convolutional neural network uses multiple layers of convolution to extract the change patterns of scheduling parameters at adjacent time positions and the correlation distribution features between different distributed energy terminals. Under the action of nonlinear mapping, the change patterns and correlation distribution features are combined and enhanced hierarchically, so that the time correlation relationship of scheduling parameters and the terminal correlation relationship in the spatiotemporal correlated sample set are gradually converged into stable spatiotemporal correlated features. The spatiotemporal correlated features are summarized according to the feature arrangement order corresponding to continuous time positions to form a high-dimensional feature vector.

[0047] It should be noted that the training of the convolutional neural network is based on historical scheduling samples that are consistent with the structure of the local decision instruction set during historical operation (historical scheduling samples are obtained by organizing the temporal arrangement relationship of the scheduling behavior of distributed energy terminals and the terminal association relationship within a historical period). During the training process, the convolutional neural network iterates through multiple rounds of historical scheduling samples, so that the internal parameters of the convolutional neural network gradually adapt to the changing pattern of scheduling parameters at continuous time positions and the association relationship between different distributed energy terminals. After multiple rounds of iteration, the consistency of the feature output of the convolutional neural network on the historical scheduling samples is checked, and the training of the convolutional neural network is completed.

[0048] S3.3: Based on high-dimensional feature vectors, construct a global optimization problem, and use a genetic algorithm to solve the global optimization problem in multiple objectives, determine the optimal parameter configuration, and generate a global AI optimization model.

[0049] Specifically, based on the spatiotemporal correlation features in high-dimensional feature vectors, the combination of scheduling parameters contained in the local decision instruction set and the multi-objective constraints are organized into the objective and constraint contents of the global optimization problem, thus constructing the global optimization problem. A genetic algorithm is used to solve the global optimization problem in multiple objectives. Using the scheduling parameter combinations in the local decision instruction set as the initial parameter configuration source, an iterative process of recombining, filtering, and retaining scheduling parameter combinations is employed to gradually retain scheduling parameter configurations that meet the multi-objective constraints and match the spatiotemporal correlation features. The retained scheduling parameter configurations are determined as the optimal parameter configurations and solidified into the parameter settings of the global AI optimization model, generating the global AI optimization model.

[0050] S3.4: Align the execution time, unify the scale, and correlate the spatiotemporal information of the local decision instruction set to form a spatiotemporally correlated input set; Specifically, according to the time sequence of the local decision instruction set, the scheduling parameters corresponding to each distributed energy terminal are uniformly organized in terms of time position, so that scheduling parameters generated at different times form a corresponding relationship at the same time position; the scheduling parameters in the local decision instruction set are organized in a uniform expression form, so that the scheduling parameters of different distributed energy terminals are consistent in expression and scheduling meaning; after uniform organization, the scheduling parameters of each distributed energy terminal at the same time position are correlated, so that the scheduling parameters simultaneously reflect the time correlation relationship and the terminal correlation relationship, forming a spatiotemporal correlated input set.

[0051] S3.5: Extract the spatiotemporal coupling features from the spatiotemporal correlation input set through the global AI optimization model, and determine the target weights and running constraint parameters required for multi-objective optimization; Specifically, a global AI optimization model is used to map the temporal and terminal relationships of the scheduling parameters of each distributed energy terminal in the spatiotemporal input set. This ensures that the linkage between different time locations and different distributed energy terminals is uniformly reflected in the global AI optimization model, forming a spatiotemporal coupling feature. Based on this spatiotemporal coupling feature, the manifestation of multi-objective constraints in the overall scheduling is distinguished and organized. This clarifies the emphasis of multi-objective constraints in different scheduling scenarios and determines the target weights that reflect the emphasis of scheduling objectives and the operational constraint parameters that reflect the boundaries of scheduling behavior.

[0052] It should be noted that the scheduling objective refers to the operational focus that is defined by multi-objective constraints and needs to be taken into account simultaneously during the collaborative operation of distributed energy resources.

[0053] S3.6: Based on the target weights and operational constraint parameters, a global AI optimization model is used to perform multi-objective optimization calculations on the local decision instruction set to generate a preliminary scheduling plan.

[0054] Specifically, a global AI optimization model is used to uniformly simulate the scheduling parameters in the local decision instruction set. This ensures that the scheduling parameters reflect the order of selection among different scheduling objectives under the weighting of the objectives, and maintain the executability of the scheduling behavior within the boundaries defined by the operational constraint parameters. During the scheduling simulation, the scheduling effects of different combinations of scheduling parameters in the local decision instruction set are comprehensively weighed, and scheduling parameter combinations that simultaneously meet multiple objective constraints are gradually retained. After the weighing is completed, the scheduling parameter combinations that conform to the weighting of objectives and are subject to the operational constraint parameters are uniformly organized to form a preliminary scheduling plan.

[0055] S3.7: Extract the scheduling parameters of each distributed energy terminal in the preliminary scheduling plan, and construct a scheduling status dataset by combining the operating status of the distributed energy terminals; Specifically, the scheduling parameters corresponding to different distributed energy terminals are organized and collected item by item according to the terminal association relationship in the preliminary scheduling plan, and the scheduling parameters are associated with the operating status of the distributed energy terminals in the current time period, so that the scheduling parameters and operating status form a complete description under the same terminal dimension; the scheduling parameters and operating status of each distributed energy terminal are summarized and organized in a unified structural order to form a scheduling status dataset that simultaneously reflects the scheduling arrangement and operating status characteristics.

[0056] It should be noted that the unified structural sequence refers to arranging the scheduling parameters and operating status in a consistent order according to the arrangement order of the distributed energy terminals in the preliminary scheduling plan and the arrangement order of the scheduling parameters corresponding to each distributed energy terminal.

[0057] S3.8: Perform runtime constraint determination on the scheduling parameters in the scheduling status dataset and generate safety verification information; Specifically, based on the multi-objective constraints, the scheduling parameters corresponding to each distributed energy terminal in the scheduling status dataset are checked item by item to ensure that the scheduling parameters and the multi-objective constraints are clearly associated under the same terminal and the same scheduling scenario. During the check, the status of whether the scheduling parameters meet the multi-objective constraints is marked, so that each scheduling parameter corresponds to a clear constraint compliance status mark and constraint deviation status mark. The constraint compliance status marks, constraint deviation status marks and corresponding scheduling parameters are collected in a unified manner to generate safety verification information.

[0058] S3.9: Perform targeted correction on scheduling parameters with constraint violations in the security verification information, and perform policy consistency fusion on the scheduling parameters during the correction to generate a fused scheduling parameter set; Specifically, based on the scheduling parameters and multi-objective constraints that have constraint deviation status indicators in the safety verification information, the corresponding scheduling parameters are adjusted in a targeted manner to bring the scheduling parameters back to the scheduling boundaries defined by the multi-objective constraints; according to the scheduling orientation requirements, multiple scheduling parameters under the same scheduling scenario are coordinated in a unified manner to ensure that the adjusted scheduling parameters are consistent in terms of the emphasis relationship of scheduling objectives; the scheduling parameters that have completed constraint regression and scheduling orientation coordination are organized in a unified manner to generate a fusion scheduling parameter set.

[0059] It should be noted that the scheduling orientation requirement is determined by the emphasis relationship of each scheduling objective based on the multi-objective constraints and objective weights. Specifically, it refers to the unified specification of the priority order and emphasis direction of different scheduling objectives under the same scheduling scenario, guiding multiple scheduling parameters to maintain a consistent objective emphasis relationship when adjusting their values.

[0060] S3.10: Perform integrity verification and unified encapsulation on the scheduling parameters in the fusion scheduling parameter set to obtain the authoritative collaborative instruction set.

[0061] Specifically, the integrity of each scheduling parameter corresponding to each distributed energy terminal in the fusion scheduling parameter set is verified to ensure that each scheduling parameter contains complete scheduling parameter information, and scheduling parameters with missing or contradictory information are removed. The scheduling parameters that pass the integrity verification are then uniformly organized according to the order of distributed energy terminals and the order of scheduling parameters, and encapsulated in a consistent structural form to ensure that the scheduling parameters are consistent in both the terminal and scheduling dimensions, forming an authoritative collaborative instruction set.

[0062] S4. The authoritative collaborative instruction set is distributed to the blockchain network. During the distribution process, the instruction version identifier and execution constraints are bound to the authoritative collaborative instruction set to form an authoritative collaborative instruction record. The authoritative collaborative instruction record is then verified for trusted storage and execution conditions through a smart contract to generate device-level execution instructions.

[0063] S4.1: Add a unique instruction version identifier to the scheduling parameters of each distributed energy terminal in the authoritative collaborative instruction set, and bind the instruction version identifier with the corresponding scheduling parameter to form the content of the authoritative collaborative instruction; Specifically, following the order of distributed energy terminals in the authoritative collaborative instruction set, the scheduling parameters corresponding to each distributed energy terminal are identified and added item by item. A unique instruction version identifier is configured for each scheduling parameter to distinguish different scheduling cycles and different scheduling contents. The instruction version identifier is bound one-to-one with the corresponding scheduling parameter, so that each scheduling parameter is associated with a clear instruction version identifier. The content containing scheduling parameters and corresponding instruction version identifiers is uniformly organized to form the authoritative collaborative instruction content.

[0064] S4.2: Bind the corresponding preset execution constraints to each scheduling parameter in the authoritative collaborative instruction content, and structure the scheduling parameters, instruction version identifier and execution constraints to form authoritative collaborative instruction data; Specifically, based on the correspondence between distributed energy terminals and scheduling parameters, each scheduling parameter in the authoritative collaborative instruction content is bound with a corresponding preset execution constraint. The scheduling parameters, the instruction version identifier corresponding to each scheduling parameter, and the preset execution constraints are combined and organized in a unified field order, so that the three form a one-to-one and inseparable relationship in structural position, thus forming authoritative collaborative instruction data.

[0065] It should be noted that execution constraints are the fixed constraints in multi-objective constraints on the range of values ​​of scheduling parameters, restrictions on scheduling order, and the scope of application of scheduling scenarios. They are used to constrain the specific execution boundaries of scheduling parameters in the actual issuance and execution phases. The range of values ​​for scheduling parameters is defined by the state characteristics of electrical characteristic quantity sequences, local event alarm indicators, and the operating status of distributed energy terminals. The scope of application of scheduling scenarios is the applicable boundary formed by the allowed combination of values ​​and the direction of change of scheduling parameters under different combinations of operating states.

[0066] S4.3: The authoritative collaborative instruction data is sent to the blockchain network for consistency confirmation, forming an authoritative collaborative instruction record.

[0067] Specifically, authoritative collaborative instruction data is submitted to the blockchain network in a predetermined order, and the blockchain network uniformly confirms the integrity and consistency of the authoritative collaborative instruction data, so that the authoritative collaborative instruction data forms an unalterable unified record in the blockchain network; the authoritative collaborative instruction data confirmed by the blockchain network is fixedly saved as a traceable record, forming the authoritative collaborative instruction record under the corresponding scheduling cycle.

[0068] It should be noted that the predetermined distribution order refers to the order in which distributed energy terminals are arranged according to the order of their corresponding scheduling parameters in the authoritative collaborative instruction set.

[0069] S4.4: Based on the authoritative collaborative instruction record, trigger the corresponding smart contract in the blockchain network, and load scheduling parameters, instruction version identifier and execution constraints into the smart contract to form an instruction verification context; Specifically, based on the correspondence between the authoritative collaborative instruction record and the blockchain network, the smart contract pre-deployed in the blockchain network and associated with the authoritative collaborative instruction record is triggered, causing the smart contract to enter the instruction verification and execution state. After the smart contract is triggered, the scheduling parameters in the authoritative collaborative instruction record, the instruction version identifier corresponding to the scheduling parameters, and the bound execution constraints are loaded into the internal verification space of the smart contract. This ensures that the scheduling parameters, instruction version identifier, and execution constraints form a clear correspondence in the same contract environment, constituting the instruction verification context.

[0070] It should be noted that a smart contract refers to an automatically executed contract logic that is pre-written into the blockchain network and published based on the structure and execution constraints of the authoritative collaborative instruction record. During the initialization phase of the blockchain network, it is deployed by the blockchain network administrator to fix the rules for verifying scheduling parameters, matching instruction versions, and checking execution constraints. When the authoritative collaborative instruction record is written into the blockchain network, the blockchain network automatically locates and triggers the corresponding smart contract based on the established relationship between the authoritative collaborative instruction record and the contract.

[0071] S4.5: Based on the instruction verification context, the smart contract verifies the execution conditions of each scheduling parameter corresponding to each distributed energy terminal in the authoritative collaborative instruction record and generates a set of verification status identifiers. Specifically, the smart contract verifies the scheduling parameters, instruction version identifier, and corresponding execution constraints loaded in the instruction verification context. It then checks each scheduling parameter in the authoritative collaborative instruction record against the corresponding distributed energy terminal to confirm that the scheduling parameters are consistent with the bound execution constraints. During the verification process, the smart contract assigns an execution condition satisfied flag and an execution condition dissatisfied flag to the corresponding scheduling parameter based on the verification status of each scheduling parameter. The execution condition satisfied flags and execution condition dissatisfied flags corresponding to each scheduling parameter are then collected to form a set of verification status flags.

[0072] S4.6: The smart contract filters and verifies the scheduling parameters that meet the execution constraints in the status identifier set, and encapsulates the filtered scheduling parameters under the corresponding execution constraints to generate device-level execution instructions.

[0073] Specifically, the smart contract checks each scheduling parameter in the authoritative collaborative instruction record according to the one-to-one correspondence between the scheduling parameters and the execution condition satisfaction and non-satisfaction indicators in the verification status identifier set, and retains only the scheduling parameters whose verification status identifier is the execution condition satisfaction indicator. The smart contract organizes the retained scheduling parameters, corresponding instruction version identifiers, and bound execution constraints in a unified manner, and encapsulates them according to the arrangement order of the distributed energy terminals, so that each encapsulated content clearly defines the executable scheduling parameters and corresponding execution constraints, forming a device-level execution instruction.

[0074] S5. Each distributed energy terminal receives and executes device-level execution instructions, collects execution feedback data using end-to-end sensors, and dynamically optimizes the global AI optimization model parameters based on the execution feedback data.

[0075] S5.1: Each distributed energy terminal receives the device-level execution instruction and performs a consistency check on the instruction version identifier in the device-level execution instruction to confirm that the device-level execution instruction is the current valid instruction; Specifically, after receiving a device-level execution instruction, each distributed energy terminal reads the instruction version identifier carried in the instruction and compares it item by item with the latest instruction version identifier currently stored in the distributed energy terminal. If the instruction version identifier in the device-level execution instruction is lower than the latest instruction version identifier already stored, the device-level execution instruction is determined to be invalid. If the instruction version identifier in the device-level execution instruction matches the latest instruction version identifier already stored in the distributed energy terminal, the device-level execution instruction is confirmed to be a valid instruction within the current scheduling cycle.

[0076] It should be noted that the latest instruction version identifier currently stored in the distributed energy terminal is derived from the instruction version identifier carried in the device-level execution instruction that the distributed energy terminal successfully received and executed in the previous scheduling cycle, and is retained as the current baseline identifier for the distributed energy terminal to use for subsequent instruction consistency verification after the device-level execution instruction has been completed and confirmed.

[0077] S5.2: Update the control parameters of the terminal's current operating status according to the scheduling parameters in the current valid instructions, and continuously collect the operating status of the terminal during the execution process using end-to-end sensors to form execution feedback data.

[0078] Specifically, the scheduling parameters corresponding to the distributed energy terminal are read from the currently valid instructions, and the corresponding scheduling parameters are mapped to the operation control parameter items of the distributed energy terminal's operation control interface, so that the operation status of the distributed energy terminal is consistent with the scheduling parameters in the currently valid instructions. During the continuous operation of the distributed energy terminal according to the scheduling parameters, the port voltage operation status information, active power operation status information, reactive power operation status information and operation condition status information of the distributed energy terminal are continuously collected using full-link sensors. The collected operation status information is then collected and organized in chronological order to form execution feedback data.

[0079] S5.3: Based on the execution feedback data, calculate the execution deviation value between the actual operating state of the terminal and the corresponding scheduling parameters in the device-level execution instructions, and generate execution deviation data; Specifically, based on the distributed energy terminal operating status information collected in chronological order in the execution feedback data, the port voltage operating status information, active power operating status information, reactive power operating status information, and operating condition status information in the execution feedback data are matched one-to-one with the scheduling parameters of the corresponding distributed energy terminal in the device-level execution instructions at the same time position. The execution deviation value between the actual operating status of the terminal and the corresponding scheduling parameters in the device-level execution instructions is calculated, so that each scheduling parameter corresponds to a clear execution deviation value, and the data is collected to form execution deviation data.

[0080] The expression for the execution deviation between the actual operating state of the computing terminal and the corresponding scheduling parameters in the device-level execution instructions is as follows:

[0081] Wherein, ΔD represents the execution deviation value; P1 represents the actual active power value in the active power operation status information; P2 represents the active power scheduling parameters issued to the distributed energy terminal in the device-level execution command; P3 represents the rated active power in the distributed energy terminal's operation file; Q1 represents the actual reactive power value in the reactive power operation status information; Q2 represents the reactive power scheduling parameters corresponding to the distributed energy terminal in the device-level execution command; Q3 represents the rated reactive power parameters of the distributed energy terminal; V1 represents the actual port voltage value in the voltage operation status information; V2 represents the voltage scheduling parameters issued to the distributed energy terminal in the device-level execution command; and V3 represents the rated voltage parameters of the distributed energy terminal.

[0082] S5.4: Optimize the parameters of the global AI optimization model based on execution deviation data.

[0083] Specifically, the execution deviation data is correlated and organized with the corresponding device-level execution instructions and the operating status of distributed energy terminals, so that the execution deviation data can reflect the deviation of different scheduling parameters in actual operation; the execution deviation data is collected and organized according to the order of scheduling parameters in the device-level execution instructions, so that the global AI optimization model can perceive the deviation changes of different scheduling parameters in the continuous scheduling process; the parameter configuration used to characterize the effect of scheduling parameters in the global AI optimization model is synchronously adjusted, so that the global AI optimization model can automatically enhance the correction of scheduling parameters with persistent deviations during the scheduling process, and complete the optimization and update of the global AI optimization model parameters.

[0084] In summary, this invention achieves a reliable transition and constraint implementation of scheduling instructions from global generation to terminal execution within a distributed control architecture by: using smart contracts to reliably store and verify the execution conditions of authoritative collaborative instruction records and generate device-level execution instructions; by uniformly binding instruction version identifiers with preset execution constraints and storing them on the blockchain, collaborative scheduling instructions possess clear version traceability and execution consistency during transmission and execution, preventing instruction drift or out-of-bounds execution; simultaneously, the automatic verification mechanism based on smart contracts ensures that the generation of device-level execution instructions does not rely on manual intervention, guaranteeing the objective implementation of scheduling strategies on the distributed terminal side, and providing a reliable data foundation for the collection of execution feedback data and the dynamic optimization of global model parameters, thereby enhancing the stability and controllability of collaborative operation under the distributed control system.

[0085] 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 it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for coordinated operation of distributed energy resources based on the Industrial Internet, characterized in that: include, The raw physical quantity data of distributed energy terminals are collected, and local event discrimination and formatting are performed to obtain multi-source information; The edge intelligent gateway uses local AI prediction and event triggering mechanisms to perform data fusion and autonomous collaborative decision-making on multi-source information, generating a local decision instruction set. The cloud platform receives the local decision instruction set, builds a global AI optimization model, uses the global AI optimization model to perform multi-objective calculations on the local decision instruction set, generates a preliminary scheduling plan, and performs security verification and policy fusion on the preliminary scheduling plan to obtain an authoritative collaborative instruction set. The authoritative collaborative instruction set is distributed to the blockchain network. During the distribution process, the instruction version identifier and execution constraints are bound to the authoritative collaborative instruction set to form an authoritative collaborative instruction record. The authoritative collaborative instruction record is then trusted and its execution conditions are verified through a smart contract to generate device-level execution instructions. Each distributed energy terminal receives and executes device-level execution instructions, collects execution feedback data using end-to-end sensors, and dynamically optimizes the parameters of the global AI optimization model based on the execution feedback data.

2. The distributed energy collaborative operation method based on the Industrial Internet as described in claim 1, characterized in that: The steps to obtain multi-source information are as follows: The raw three-phase voltage and current data of each distributed energy terminal are collected and high-precision synchronous sampling and time-scale alignment are performed to obtain the synchronous sampling sequence; Real-time filtering and denoising of the synchronous sampling sequence yields an electrical characteristic sequence. Perform morphological gradient-based mutation detection on electrical feature sequences to generate local event alerts; The electrical characteristic sequence and local event alarm are encapsulated in a predefined format to generate multi-source information.

3. The distributed energy collaborative operation method based on the Industrial Internet as described in claim 2, characterized in that: The steps for generating the local decision instruction set are as follows: Based on multi-source information, the edge intelligent gateway uses a local AI prediction model to analyze electrical characteristic quantity sequences in real time, extract key information, and generate a fused information set. Based on the fused information set, the edge intelligent gateway uses an event-triggered mechanism to automatically trigger collaborative decision-making and generate a set of candidate decision solutions. The candidate decision scheme set is optimized by multi-objective optimization using deep reinforcement learning algorithm to determine the optimal decision strategy, and the candidate decision scheme is converted into a candidate decision instruction set under the constraint of the optimal decision strategy. The candidate decision instruction sets are subjected to security verification and policy fusion to generate a local decision instruction set.

4. The distributed energy collaborative operation method based on the Industrial Internet as described in claim 3, characterized in that: The steps for constructing the global AI optimization model are as follows: The local decision instruction set is cleaned and normalized to generate a spatiotemporally correlated sample set; By performing multi-layer convolution and non-linear mapping on the spatiotemporally correlated sample set through a convolutional neural network, spatiotemporal correlation features are extracted to form a high-dimensional feature vector. Based on high-dimensional feature vectors, a global optimization problem is constructed, and a genetic algorithm is used to solve the global optimization problem in multiple objectives, determine the optimal parameter configuration, and generate a global AI optimization model.

5. The distributed energy collaborative operation method based on the Industrial Internet as described in claim 4, characterized in that: The steps for generating a preliminary scheduling plan are as follows: The execution time of the local decision instruction set is aligned, the scale is unified, and the spatiotemporal correlation is established to form a spatiotemporally correlated input set. The spatiotemporal coupling features in the spatiotemporal correlation input set are extracted by a global AI optimization model to determine the target weights and operational constraint parameters required for multi-objective optimization. Based on the target weights and operational constraint parameters, a global AI optimization model is used to perform multi-objective optimization calculations on the local decision instruction set to generate a preliminary scheduling plan.

6. The distributed energy collaborative operation method based on the Industrial Internet as described in claim 5, characterized in that: The steps to obtain the authoritative collaborative instruction set are as follows: The scheduling parameters of each distributed energy terminal in the preliminary scheduling plan are extracted, and a scheduling status dataset is constructed by combining the operating status of the distributed energy terminals. Perform runtime constraint checks on the scheduling parameters in the scheduling status dataset and generate safety verification information; For scheduling parameters with constraint violations in the security verification information, targeted correction is performed, and policy consistency fusion is performed on the scheduling parameters during the correction to generate a fused scheduling parameter set; The scheduling parameters in the fusion scheduling parameter set are verified for integrity and uniformly encapsulated to obtain the authoritative collaborative instruction set.

7. The distributed energy collaborative operation method based on the Industrial Internet as described in claim 6, characterized in that: The steps for creating an authoritative collaborative instruction record are as follows: A unique instruction version identifier is added to the scheduling parameters of each distributed energy terminal in the authoritative collaborative instruction set, and the instruction version identifier is bound to the corresponding scheduling parameter to form the content of the authoritative collaborative instruction. The scheduling parameters in the authoritative collaborative instruction are bound to corresponding preset execution constraints, and the scheduling parameters, instruction version identifiers and execution constraints are structurally associated to form authoritative collaborative instruction data. Authoritative collaborative instruction data is sent to the blockchain network for consistency confirmation, forming an authoritative collaborative instruction record.

8. The distributed energy collaborative operation method based on the Industrial Internet as described in claim 7, characterized in that: The steps for generating device-level execution instructions are as follows: Based on authoritative collaborative instruction records, the corresponding smart contract in the blockchain network is triggered, and scheduling parameters, instruction version identifiers and execution constraints are loaded into the smart contract to form an instruction verification context; Based on the instruction verification context, the smart contract verifies the execution conditions of each scheduling parameter corresponding to each distributed energy terminal in the authoritative collaborative instruction record and generates a set of verification status identifiers. The smart contract filters and verifies the scheduling parameters that meet the execution constraints in the status identifier set, and then encapsulates the filtered scheduling parameters under the corresponding execution constraints to generate device-level execution instructions.

9. The distributed energy collaborative operation method based on the Industrial Internet as described in claim 8, characterized in that: The steps for collecting execution feedback data using end-to-end sensors are as follows: Each distributed energy terminal receives the device-level execution instruction and performs a consistency check on the instruction version identifier in the device-level execution instruction to confirm that the device-level execution instruction is the current valid instruction; Update the control parameters of the terminal's current operating status according to the scheduling parameters in the current valid instructions, and continuously collect the operating status of the terminal during the execution process using end-to-end sensors to form execution feedback data.

10. The distributed energy collaborative operation method based on the Industrial Internet as described in claim 9, characterized in that: The steps for dynamically optimizing the global AI optimization model parameters based on execution feedback data are as follows: Based on the execution feedback data, the execution deviation between the actual operating state of the terminal and the corresponding scheduling parameters in the device-level execution instructions is calculated, and execution deviation data is generated. Based on execution deviation data, the parameters of the global AI optimization model are optimized.

Images