A new energy control execution system and method based on multi-market coupling mechanism

Abstract

The present invention discloses a new energy control execution system and method based on a multi-market coupling mechanism, which relates to the technical field and comprises the following steps: obtaining control signals from multiple source markets, and performing synchronization verification on the control signals from the multiple source markets based on a time-sensitive network and a local alignment cache method; performing semantic consistency analysis, conflict discrimination and fusion on the verified control signals, and outputting a control instruction set based on a dynamic priority; constructing an interaction mapping matrix with the new energy equipment according to the control instruction set, and adaptively executing control actions on the new energy equipment based on the interaction mapping matrix and providing feedback; the present invention significantly improves the timeliness, consistency and accuracy of the new energy equipment in responding to the control signals from multiple source markets by constructing an integrated control mechanism of time synchronization guarantee, semantic conflict analysis and adaptive control fusion, thereby realizing accurate matching and adaptive execution of control instructions by the new energy equipment.

Description

Technical Field

[0001] The present invention relates to the field of new energy control technology, and more specifically, to a new energy control execution system and method based on a multi-market coupling mechanism. Background Art

[0002] As the penetration of new energy sources continues to increase, the integration of renewable energy sources like wind power and photovoltaics is placing higher demands on power system control. Due to the high volatility and uncontrollability of new energy sources, traditional energy control mechanisms based on centralized control struggle to meet the demands of distributed, flexible, and real-time control. To adapt to the development of new energy, current power systems are generally introducing multi-market mechanisms (cooperating with electricity markets, carbon markets, and green certificate markets) to coordinate the participation of various resources in system regulation. At the same time, new energy devices themselves are also equipped with greater intelligence and edge computing capabilities, laying the foundation for rapid response to multi-source control signals and autonomous decision-making.

[0003] For example, the invention patent with announcement number CN118778454A discloses a multimodal edge adaptive control system, specifically a multimodal edge adaptive control system, which includes a sensor interface configuration module, a data synchronization processing module, a resource allocation module, a dynamic response module, an environmental monitoring module, and a control strategy update module. In the present invention, by integrating multimodal data input and utilizing edge computing for data processing, the decision-making ability and response speed of the control configuration are improved. By synchronously receiving and processing multiple types of sensor data, efficient time alignment and format unification of data are achieved, ensuring the consistency and accuracy of data processing. By utilizing dynamic data stream docking and real-time feedback mechanisms, real-time data processing is optimized, data transmission delays are reduced, and the system can quickly adapt to environmental changes, thereby improving the operational flexibility and reliability of the system. By analyzing environmental data and adjusting the control strategy in a timely manner, more accurate environmental monitoring and control execution are achieved, thereby improving the overall performance and efficiency of the system.

[0004] For example, the invention patent announcement with the publication number CN115576278A discloses a multi-agent, multi-task hierarchical continuous control method based on temporal equilibrium analysis, comprising the following steps: S1, establishing a multi-robot dynamics model described by a general linear system, and constructing the optimal proportional consistency control problem of the multi-robot system through the Bellman optimality principle; S2, designing an adaptive dynamic programming algorithm to calculate the approximate solution of the discrete-time Hamilton-Jacobi-Bellman equation based on generalized policy iteration; S3, constructing an evaluation-execution neural network to fit the iterative control law and performance index respectively; S4, deploying the evaluation-execution network controller designed in step S3 to the robot group. The method of the present invention is aimed at a multi-robot system described by a general linear system, takes into account the optimal control situation with specific task objectives, and uses the evaluation-execution network to achieve online optimal collaborative control, which is applicable to the field of cluster collaborative control of multi-robot systems.

[0005] The above disclosed technical solutions have at least the following technical problems:

[0006] In the context of multiple market mechanisms operating in parallel, the control signals generated by different markets are heterogeneous, asynchronous, and conflicting. These control signals not only come from different sources and have varying granularity, but also face difficulties in time synchronization and differences in semantic understanding. Directly acting on new energy devices can easily lead to control logic conflicts, device execution failures, and even grid disruptions. This present invention proposes a solution to these problems. Summary of the Invention

[0007] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a new energy control execution system and method based on a multi-market coupling mechanism, which realizes the precise matching and adaptive execution of control instructions by new energy equipment by constructing an integrated control mechanism that integrates time synchronization guarantee, semantic conflict resolution and adaptive control.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A new energy control execution method based on a multi-market coupling mechanism includes the following steps: obtaining control signals from multiple source markets, and performing synchronization verification on the control signals from multiple source markets based on a time-sensitive network and a local alignment cache method; performing semantic consistency analysis, conflict discrimination, and fusion on the verified control signals, and outputting a control instruction set based on a dynamic priority; constructing an interaction mapping matrix with new energy equipment according to the control instruction set, and adaptively executing control actions on the new energy equipment based on the interaction mapping matrix and providing feedback.

[0010] In a preferred embodiment, the synchronization check of control signals of multi-source markets based on time-sensitive network and local aligned cache method is specifically as follows: constructing a delay matrix, decomposing the delay matrix into a low-rank matrix and a sparse matrix based on robust principal component analysis, and constructing an optimization objective function and constraints for solving; using the optimized low-rank matrix as a trusted delay matrix, performing synchronization calibration on the trusted delay matrix based on timestamps, and performing secondary correction on the data after synchronization calibration.

[0011] In a preferred embodiment, the secondary correction of the data after synchronization calibration specifically includes: obtaining asynchronous signals according to the trusted delay matrix, domain classification of the asynchronous signals according to the signal source and sliding window caching respectively, and clustering and sorting based on the time domain alignment threshold to obtain several time consistency clusters; constructing a synchronization deviation vector for each time consistency cluster and calculating its norm value; making a judgment based on a preset synchronization tolerance threshold through the norm value, and marking the instruction set to be processed, and performing differential correction and verification on the data after synchronization calibration through the instruction set to be processed.

[0012] In a preferred embodiment, the data after synchronization calibration is differentially corrected and verified through the instruction set to be processed, specifically: the instruction set to be processed is differentially corrected based on signal domain classification; the corrected instruction to be processed is newly added to the sliding window cache queue; the clustering and verification process is repeated until all time consistency clusters meet the synchronization verification.

[0013] In a preferred embodiment, the control signal after verification is subjected to semantic consistency analysis, conflict discrimination and fusion, and a control instruction set is output based on dynamic priority, specifically: the control signal after synchronization verification is obtained, parsed into several heterogeneous nodes, and a heterogeneous graph structure is constructed; based on a relational perception graph neural network model, each node of the heterogeneous graph structure is embedded to obtain a unified semantic vector for each control signal node; conflict detection is performed on the semantic vector to obtain a conflict matrix and obtain a strong conflict signal; based on Bayesian reasoning, the strong conflict signal is judged and fused to output a candidate set, and key attribute data is obtained to establish a priority evaluation model; the control instruction set is prioritized according to the priority evaluation model to obtain a priority order; the control output candidate set is formatted and mapped according to the priority order, and converted into a control instruction set that can be recognized and parsed by new energy equipment.

[0014] In a preferred embodiment, the conflict detection of the semantic vectors is performed to obtain a conflict matrix and a strong conflict signal, specifically: the unified semantic vectors are compared in pairs based on cosine similarity to obtain conflict features; a conflict matrix is ​​constructed based on the conflict features, and the signal corresponding to the maximum conflict intensity in the conflict matrix is ​​obtained as a strong conflict signal.

[0015] In a preferred embodiment, an interaction mapping matrix with the new energy equipment is constructed according to the control instruction set, and control actions are adaptively executed on the new energy equipment based on the interaction mapping matrix and feedback is provided. Specifically, an interaction mapping matrix between the new energy equipment and the control instructions is constructed, with the equipment operation status parameters as the row dimensions and the historical control instruction categories as the column dimensions, to form a multidimensional sparse matrix structure; the multidimensional sparse matrix structure is feature reconstructed based on a deep matrix factorization model to obtain a low-dimensional embedding vector of the equipment behavior preference and control responsiveness; the control instruction set is matched with the optimal control instruction vector in the low-dimensional embedding vector based on the control preference, and is screened based on the cosine similarity in combination with the current equipment status to determine the most suitable control action; the most suitable control action is sent to the target new energy equipment execution module, and real-time status monitoring is performed based on the edge computing node to achieve an execution closed loop.

[0016] In a preferred embodiment, the Bayesian reasoning-based adjudication and fusion output of the candidate set of strong conflicting signals is specifically as follows: based on the device topology and control process logic, a Bayesian causal network is constructed; based on the Bayesian causal network, strong conflicting signals from the same signal source are synthesized to output a joint confidence distribution; the posterior probability of each control signal is obtained through a confidence propagation algorithm, and the conflict-free signal combination with the largest posterior probability is selected as the control output candidate set.

[0017] In a preferred embodiment, the real-time status monitoring based on edge computing nodes is specifically as follows: deploying edge computing nodes, and after detecting that new energy equipment executes a control action, the edge computing nodes associate the unique identifier of the current control instruction and generate a status response code; based on the time window streaming cache, the status response codes are batched according to the preset time window to form a minimum feedback unit; and priority alarm feedback is performed by comparing the detection status of the minimum feedback unit with the abnormal threshold rule of the equipment operation status.

[0018] A new energy control execution system based on a multi-market coupling mechanism includes: a control signal synchronization verification module, which is used to obtain control signals from multiple source markets and perform synchronization verification on the control signals from multiple source markets based on a time-sensitive network and a local alignment cache method; a control signal synchronization processing module, which is used to perform semantic consistency analysis, conflict discrimination and fusion on the verified control signals, and output a control instruction set based on a dynamic priority; and an adaptive control module, which is used to construct an interaction mapping matrix with new energy equipment based on the control instruction set, and adaptively execute control actions on the new energy equipment based on the interaction mapping matrix and provide feedback.

[0019] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0020] 1. By introducing time-sensitive networking (TSN) and a local alignment cache mechanism, synchronization verification of multi-source market control signals is achieved under the premise of high real-time performance and high reliability, with the following advantages: First, it integrates multi-source heterogeneous data from the electricity market, carbon market, green certificate market, grid status, and equipment signals, while ensuring a unified time base while retaining their respective semantic characteristics; second, it effectively eliminates jitter and anomalies through robust principal component analysis to achieve reliable modeling of control delays; third, it adopts a sliding window cache and clustering strategy based on domain classification to accurately identify time-consistent clusters and significantly improve signal alignment efficiency; fourth, it ensures the dynamic availability and timing integrity of signals in various domains of the market, grid, and equipment through a differentiated correction mechanism, ultimately achieving highly robust collaborative control of new energy equipment in the context of multi-market coupling.

[0021] 2. By introducing heterogeneous graph structures and relationship-aware graph neural networks, semantic consistency analysis and conflict discrimination of control signals are achieved, and Bayesian reasoning is combined to adjudicate conflict signals. The three key attributes of scheduling real-time, grid impact and execution urgency are integrated to build a control priority model, and then an ordered and conflict-free control instruction set is output. At the same time, with the help of deep matrix factorization and context-aware mechanism, adaptive instruction adaptation and control response optimization of new energy equipment are achieved, and real-time monitoring and structured feedback of instruction execution status are achieved through edge computing nodes. The overall solution has the advantages of high semantic understanding ability, strong conflict resolution ability, high dynamic adaptability of control strategy, and strong equipment execution closed-loop capability, which significantly improves the safety, real-time performance and intelligence level of new energy control systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 A flowchart of a new energy control execution method based on a multi-market coupling mechanism is provided in an embodiment of the present application.

[0023] Figure 2 A schematic diagram of the structure of a new energy control execution system based on a multi-market coupling mechanism provided in an embodiment of the present application. DETAILED DESCRIPTION

[0024] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0025] Example 1, Figure 1 A schematic flow chart of a new energy control execution method based on a multi-market coupling mechanism provided in an embodiment of the present application includes the following steps:

[0026] S1, obtains control signals from multiple source markets and performs synchronization verification on the control signals from multiple source markets based on time-sensitive networking and local aligned cache method.

[0027] In this example, multi-source market control signals refer to a collection of multidimensional dynamic data signals collected and integrated in real time from different markets, power grids, devices, and environments to achieve adaptive control of new energy equipment within a multi-market coupling mechanism (such as the coordinated operation of the electricity market, carbon market, and green certificate market). The core goal is to provide a parseable control basis for new energy equipment, enabling it to dynamically optimize its operating strategy within the constraints of multi-market coordination, balancing grid stability and low-carbon goals.

[0028] Multi-source market control signals are inherently heterogeneous and asynchronous. Direct access to these signals can lead to inaccurate responses from new energy devices. Time-Sensitive Networking (TSN) combined with the PTP clock synchronization mechanism achieves a high-precision, unified time base for signals from different sources, preventing clock drift from interfering with control effectiveness.

[0029] The introduction of delay matrix analysis and robust principal component analysis (RPCA) can effectively identify and eliminate abnormal delay data, obtain a more reliable time domain data foundation, and thus improve the stability and accuracy of the device's response to multiple market signals.

[0030] In the signal clustering process, the time domain alignment threshold + density clustering algorithm is introduced to construct a dynamic "time consistency cluster" based on the control frequency, equipment tolerance and market response characteristics. This can maintain time domain consistency in scenarios with inconsistent control frequencies and large signal delay fluctuations, thereby supporting the generation of multi-market collaborative control strategies under complex constraints.

[0031] Obtain control signals from multiple source markets and perform synchronization verification on the control signals based on time-sensitive networking and local aligned caching. Specifically:

[0032] Obtaining control signals from multiple source markets, wherein the control signals include market dynamic signals, grid operation signals, and new energy equipment status signals. The market dynamic signals include power market signals, carbon market signals, and green certificate market signals. The grid operation signals include grid physical status and dispatch instructions.

[0033] Build a TSN communication network based on the IEEE802.1Qbv scheduling mechanism, connect control signals to the TSN communication network, perform unified timestamp processing on control instructions based on the time domain mapping model, unify the format and time base, and synchronize the control system based on the high-precision PTP synchronous clock;

[0034] Obtain the time difference between the actual arrival time of each control signal and the time after time synchronization in real time, and construct a delay matrix, in which each row of the delay matrix corresponds to a signal source and each column corresponds to the time difference of different sampling periods;

[0035] Based on the robust principal component analysis method, the delay matrix is ​​decomposed into a low-rank matrix and a sparse matrix, and an optimization objective function and constraints are constructed. The optimized low-rank matrix is ​​obtained by solving the optimization objective function and used as the trusted delay matrix. The trusted delay matrix is ​​synchronized and calibrated based on the timestamp, and the synchronized and calibrated data is then corrected twice.

[0036] The secondary correction of the synchronously calibrated data is specifically as follows:

[0037] Asynchronous signals are obtained based on a trusted delay matrix. These signals are domain-classified according to their source and cached in sliding windows. Clustering and sorting based on a time-domain alignment threshold yield several time-consistent clusters. The domain classifications include market, grid, and device domains. Asynchronous signals refer to instructions that have been synchronized but whose arrival times are still inconsistent. The sliding window cache uses an adjustable time window sliding mechanism to control the cache length.

[0038] For each time-consistent cluster, a synchronization deviation vector is constructed and its norm is calculated. The synchronization deviation vector is composed of the deviation between the timestamp of each instruction in the cluster and the cluster center time. The norm is calculated based on the Euclidean norm.

[0039] The norm value is used to determine the synchronization tolerance threshold. If the norm value is less than the synchronization tolerance threshold, the synchronization check is satisfied. Otherwise, it is not satisfied and marked as a pending instruction set.

[0040] Differentiated corrections are made to the pending instruction set based on signal domain classification. The corrected pending instructions are added to the sliding window cache queue. The clustering and verification process is repeated until all time-consistent clusters meet the synchronization check.

[0041] The output is finally input into the new energy equipment through the control instruction of synchronization verification.

[0042] The differential correction of the signal set to be processed based on signal domain classification is specifically as follows:

[0043] For market domain signals, a time domain stretching correction strategy is adopted to regress the timestamp to the cluster center through interpolation mapping;

[0044] For power grid domain signals, a precise time repackaging mechanism is used to retain the original timestamp but delay access to the control module;

[0045] For device domain signals, the device-side clock deviation correction method is used to adjust the device-side time base based on PTP return clock feedback.

[0046] The clustering sorting based on the time domain alignment threshold obtains several time consistency clusters, specifically:

[0047] Obtain the response sensitivity of multi-source control signals, device delay tolerance, power system control frequency, and set the maximum allowable timestamp offset threshold;

[0048] Extract the timestamp set of all synchronized calibration signals, construct the timestamp difference matrix, and set the time clustering radius;

[0049] The density clustering algorithm is used to divide the data and obtain several time-consistent clusters.

[0050] The specific calculation formula of the time domain mapping model is as follows:

[0051]

[0052] Where, To synchronize the reference timestamp, For the The local timestamp generated by the control instruction, For the The clock frequency deviation of each control instruction, For the The time difference between the two control instructions.

[0053] The specific calculation formula of the optimization objective function is as follows:

[0054]

[0055] The constraints are specifically:

[0056]

[0057] Where, To optimize the objective function, is the nuclear norm of the matrix, is the L1 norm of the matrix elements, A weight parameter that balances the effects of low-rank and sparse components. is the delay matrix, is a low-rank matrix, is a sparse matrix.

[0058] It should be noted that low-rank matrices represent normal structural components in delay data, while sparse matrices represent sparse anomalies caused by abnormal noise, sudden jitter, or measurement errors. A trusted delay matrix can be used to eliminate abnormal data introduced by network jitter, transient interference, or synchronization errors. Electricity market signals include time-of-use electricity prices at nodes / regions, and the prices and capacity requirements for services such as frequency regulation and reserve capacity. Carbon market signals include real-time or daily updated carbon trading prices and carbon emission intensity constraints. Green certificate market signals include green certificate prices. The physical state of the power grid includes frequency and voltage, and the remaining capacity of key transmission channels. Dispatch instructions include active / reactive power instructions issued by the superior dispatching agency and rate constraints on renewable energy output changes.

[0059] S2 performs semantic consistency analysis, conflict discrimination and fusion on the verified control signal, and outputs a control instruction set based on dynamic priority.

[0060] In this example, the control signal after synchronization verification is parsed into four types of heterogeneous nodes: instruction nodes, device nodes, parameter nodes, and intention nodes, and a relational heterogeneous graph is constructed to achieve formal modeling of the control signal from data to graph. A conflict matrix is ​​constructed based on the semantic vector space of the control signal to detect strong semantic conflicts between signals (such as target conflicts, parameter conflicts, and execution contradictions). This can: proactively identify and avoid scheduling failures such as "mutually exclusive control" and "reverse control"; provide accurate input for subsequent conflict resolution; and significantly reduce the control risks of new energy equipment caused by inconsistent instructions.

[0061] The verified control signals are analyzed for semantic consistency, conflict identification and fusion, and a control instruction set is output based on dynamic priority, specifically:

[0062] Obtain the control signal and its control elements after synchronization verification, parse each control signal into several heterogeneous nodes, and construct a heterogeneous graph structure. The control elements include control instructions, target objects, control parameters and control intentions.

[0063] The heterogeneous graph structure is specifically: defining control instructions as instruction nodes, defining control objects as device nodes, defining control parameters as parameter nodes, and defining control intentions as intention nodes, and establishing edge relationships between the nodes;

[0064] Based on the relationship-aware graph neural network model, each node of the heterogeneous graph structure is embedded to obtain a unified semantic vector for each control signal node;

[0065] Conflict detection is performed on the unified semantic vectors of different signals to obtain a conflict matrix and obtain strong conflict signals;

[0066] Based on Bayesian reasoning, strong conflicting signals are fused and adjudicated to output a semantically consistent and conflict-free control output candidate set.

[0067] Acquiring key attribute data based on the control output candidate set, wherein the key attribute data includes scheduling real-time performance, power grid impact degree, and execution urgency;

[0068] Build a priority assessment model based on key attribute data based on machine learning;

[0069] Based on the analysis of historical operation data, three types of control priority intervals are obtained. According to the calculation results of the priority evaluation model, the control instruction set is prioritized and each type of signal is assigned a unique identifier and number. The historical operation data includes control instruction response logs, operation event records and alarm logs, market prices and response behavior data, new energy equipment operation records and power grid operation status curves. The three types of control priority intervals are high priority, medium priority and low priority, and each interval corresponds to a corresponding numerical range;

[0070] Based on the priority order, the control output candidate set is formatted and mapped through a unified communication interface standard and converted into a control instruction set that can be recognized and parsed by new energy equipment.

[0071] The specific calculation formula for scheduling real-time performance is as follows:

[0072]

[0073] Where, To ensure real-time scheduling, For the The update cycle of the control signal, is the variance of the historical control response rate, is the response inertia coefficient of new energy equipment.

[0074] The specific calculation formula for the degree of grid impact is as follows:

[0075]

[0076] Where, is the impact degree of the power grid, For the The coupling degree of the control signal to the frequency control, For the The coupling degree of a control signal to the voltage control, It is the load proportion of the key transmission channel on the control path.

[0077] The specific calculation formula for the execution urgency is as follows:

[0078]

[0079] Where, To implement urgency, is the state residual of the device corresponding to the current signal, is the residual threshold, To regulate the sudden increase in urgency when the signal approaches the constraint, is the historical risk weighting factor, A weighted score for the consequences of execution delays caused by such historical signals.

[0080] The specific calculation formula of the priority evaluation model is as follows:

[0081]

[0082] Where, is the priority evaluation value, To ensure real-time scheduling, is the impact degree of the power grid, To implement urgency, 、 、 is the weight parameter.

[0083] It should be noted that dispatch real-time performance evaluates the real-time processing requirements of a control signal for the dispatch system, taking into account factors such as update time granularity, device state fluctuation frequency, and control response inertia. The variance of the historical control response rate measures the sensitivity of signal changes to fluctuations in execution feedback. The degree of grid impact measures the direct effect of the control signal on the physical state and stability of the grid, constructing multi-factor indicators such as frequency-voltage coupling and critical channel occupancy. Execution urgency measures the urgency with which a signal must be executed under the dynamic constraints of the device, taking into account the convergence speed of the device's adjustable range, rate constraints, and historical over-limit risks.

[0084] The conflict detection is performed on the unified semantic vectors of different signals to obtain a conflict matrix and acquire a strong conflict signal, specifically:

[0085] All unified semantic vectors are compared pairwise based on cosine similarity to obtain conflict features, including semantic similarity conflict, object coupling conflict, and parameter domain overlap conflict;

[0086] The semantic similarity conflict is determined to be a semantic conflict if the cosine similarity of the two semantic vectors is greater than a set threshold and the control intentions they represent have direction conflicts;

[0087] The object coupling conflict is detected by increasing the conflict weight if the control object nodes of the two signals have a parent-child, aggregation or electrical adjacency relationship in the graph and have similar cosine similarity;

[0088] The parameter domain overlap conflict is marked as a potential conflict when two signal control parameters belong to the same measurement domain and their action ranges overlap;

[0089] A conflict matrix is ​​constructed based on the conflict characteristics, and the signal corresponding to the maximum conflict intensity in the conflict matrix is ​​obtained as the strong conflict signal.

[0090] The specific calculation formula of the conflict matrix is ​​as follows:

[0091]

[0092] Where, is the conflict intensity, For the Signal semantic vector, No. Signal semantic vector, For the Signal and Object coupling conflict of signals, For the Signal and The parameter domains of the signals overlap and conflict, 、 、 is the impact ratio of each conflict feature.

[0093] The Bayesian reasoning-based adjudication fusion of strong conflicting signals is specifically as follows:

[0094] Based on the device topology and control process logic, a Bayesian causal network consisting of state nodes, operation nodes, and result nodes is constructed;

[0095] The strong conflict signal is used as the input observation node, the target control instruction is used as the decision node, and the control result is used as the output evaluation node;

[0096] Based on the DST evidence theory, strong conflicting signals from the same signal source are synthesized and the joint confidence distribution is output;

[0097] Perform forward reasoning in the Bayesian network and obtain the posterior probability of each control signal through the belief propagation algorithm;

[0098] According to the posterior probability of each control signal in the Bayesian network, the conflict-free signal combination with the highest probability is selected as the control output candidate set.

[0099] It should be noted that a conflict-free signal combination refers to a set of control instructions in which all signals do not contradict each other at the semantics, object and parameter levels, can coexist and can be safely executed.

[0100] S3, constructing an interaction mapping matrix with the new energy device according to the control instruction set, and adaptively executing control actions on the new energy device based on the interaction mapping matrix and providing feedback.

[0101] The state of new energy equipment represents the operating conditions of the new energy power generation unit at a specific moment, including but not limited to the current output power, active / reactive power regulation capability, voltage and current status, control response capability, equipment communication status, resource constraints, and control boundary parameters of the power grid area in which it is located; the equipment state is used to establish a mapping relationship with the control instructions as an input feature of the control adaptation model.

[0102] An interaction mapping matrix with the new energy device is constructed according to the control instruction set, and control actions are adaptively executed on the new energy device based on the interaction mapping matrix and feedback is provided, specifically:

[0103] Obtain the historical control instruction categories and equipment operating status of new energy equipment, and construct an interactive mapping matrix between new energy equipment and control instructions, with equipment operating status parameters as row dimensions and historical control instruction categories as column dimensions, forming a multi-dimensional sparse matrix structure;

[0104] Performing feature reconstruction and latent space representation learning on the multi-dimensional sparse matrix structure based on a deep matrix factorization model to obtain a low-dimensional embedding vector representing the device behavior preference and control responsiveness;

[0105] Acquire equipment operation context features to build a context-aware control preference model. The equipment operation context features include control period, voltage level, equipment type, and load fluctuation level in the area.

[0106] Based on the context-aware control preference model, the control instruction set is matched with the optimal control instruction vector in the low-dimensional embedding vector, and the optimal control action is determined by screening based on cosine similarity in combination with the current device status.

[0107] The most suitable control action is sent to the target new energy equipment execution module, and real-time status monitoring is performed based on the edge computing node to achieve a closed execution loop and ensure timeliness and safety.

[0108] It should be noted that deep matrix factorization methods have the ability to process sparse matrices and mine implicit preference patterns. This makes them suitable for practical situations where the association information between new energy devices and control instructions is incomplete, labels are scarce, and behaviors vary widely. By constructing a device-instruction interaction matrix and extracting device behavior representations using low-rank embedding vectors, rapid screening of control actions and instruction set adaptation can be effectively achieved. Furthermore, a context-aware modeling mechanism that integrates control context information enables control strategies to have stronger dynamic responsiveness to time-varying loads and fluctuations in resource intensity. This mechanism significantly improves the system's real-time performance and adaptability while ensuring execution accuracy.

[0109] The real-time status monitoring by edge computing nodes is specifically as follows:

[0110] Deploy an edge computing node on each new energy device. After detecting that the new energy device executes a control action, the edge computing node associates the unique identifier of the current control instruction and generates a status response code. The status response code includes whether the instruction execution is successful, the response delay level, the control deviation level, and an abnormality flag.

[0111] The time-window-based streaming cache aggregates status response codes in batches according to a preset time window to form a minimum feedback unit, thereby reducing status return jitter and improving feedback efficiency and bandwidth adaptability.

[0112] Based on the abnormal threshold rules for equipment operating status, when the status detected by the minimum feedback unit matches any rule, a high-priority alarm feedback is automatically triggered to avoid the master station discovering faults late due to low-frequency polling. The abnormal threshold rules for equipment operating status include long response time, control deviation offset, and current reversal.

[0113] It should be noted that the status response code is a structured feedback identifier generated by the edge computing node based on the device's execution status after the control command is executed. It is used to quantify the success of the control action and the presence and severity of abnormalities during the execution process, providing a unified parsing interface for the main control system or platform. This code is designed to enable status confirmation, result identification, and rapid feedback for closed-loop command execution, and is a key component of achieving refined control closed-loop control and intelligent operations and maintenance.

[0114] It should be noted that by constructing a sparse interaction matrix with device operating status parameters as rows and historical control instruction categories as columns, a structured representation of historical control behaviors is achieved, and device behavior preferences are explicitly modeled: different devices have behavioral differences in responding to different categories of instructions in different states; implicit adaptation relationships are mined: even if historical data is sparse, "similar response trajectories" of control actions can be discovered through potential factors; subsequent deep embedding and instruction screening provide a learnable data foundation.

[0115] Example 2, Figure 2 This is a schematic diagram of the structure of a new energy control execution system based on a multi-market coupling mechanism provided in an embodiment of the present application, including a control signal synchronization verification module, a control signal synchronization processing module, and an adaptive control module. There are connections between the modules:

[0116] The control signal synchronization verification module is used to obtain control signals from multiple source markets and perform synchronization verification on the control signals from multiple source markets based on time-sensitive networking and local alignment caching method;

[0117] The control signal synchronization processing module is used to perform semantic consistency analysis, conflict identification and fusion on the verified control signals, and output the control instruction set based on dynamic priority;

[0118] The adaptive control module is used to construct an interactive mapping matrix with the new energy device according to the control instruction set, and adaptively execute control actions on the new energy device based on the interactive mapping matrix and provide feedback.

[0119] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0120] The above embodiments may be implemented in whole or in part through software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0121] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0122] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0123] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0124] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A new energy control execution method based on a multi-market coupling mechanism, characterized in that: The steps include: Obtain control signals from multiple markets and verify their synchronization based on time-sensitive networking and local aligned caching. Perform semantic consistency analysis, conflict identification and fusion on the verified control signals, and output control instruction sets based on dynamic priorities; The verified control signal is subjected to semantic consistency analysis, conflict discrimination and fusion, and a control instruction set is output based on a dynamic priority, specifically: Obtain the control signal after synchronization verification, parse it into several heterogeneous nodes, and build a heterogeneous graph structure; Based on the relationship-aware graph neural network model, each node of the heterogeneous graph structure is embedded to obtain a unified semantic vector for each control signal node; Performing conflict detection on the semantic vector to obtain a conflict matrix and acquire a strong conflict signal; Based on Bayesian reasoning, strong conflict signals are judged and fused to output candidate sets, and key attribute data is obtained to establish a priority evaluation model; The control instruction set is prioritized according to the priority evaluation model to obtain a priority order; Format and map the control output candidate set according to the priority order and convert it into a control instruction set that can be recognized and parsed by the new energy equipment; An interaction mapping matrix with the new energy device is constructed according to the control instruction set, and control actions are adaptively executed on the new energy device based on the interaction mapping matrix and feedback is provided.

2. The new energy control execution method based on the multi-market coupling mechanism according to claim 1 is characterized in that: The synchronization check of control signals of multi-source markets based on time-sensitive networking and local aligned caching is specifically as follows: Construct a delay matrix, decompose it into a low-rank matrix and a sparse matrix based on the robust principal component analysis method, and construct an optimization objective function and constraints for solving it; The optimized low-rank matrix is ​​used as the trusted delay matrix, the trusted delay matrix is ​​synchronously calibrated based on the timestamp, and the synchronized and calibrated data is corrected twice.

3. The new energy control execution method based on the multi-market coupling mechanism according to claim 2 is characterized in that: The secondary correction of the synchronously calibrated data specifically includes: Asynchronous signals are obtained based on the trusted delay matrix, and domain classification is performed on the asynchronous signals according to their sources. The signals are cached in sliding windows and clustered based on the time domain alignment threshold to obtain several time-consistent clusters. Construct a synchronization deviation vector for each time-consistent cluster and calculate its norm value; The norm value is used to make a judgment based on a preset synchronization tolerance threshold, and the instruction set to be processed is marked. The data after synchronization calibration is differentially corrected and verified through the instruction set to be processed.

4. The new energy control execution method based on the multi-market coupling mechanism according to claim 3 is characterized in that: The differential correction and verification of the synchronously calibrated data is performed by the instruction set to be processed, specifically: Differentiated corrections are made to the instruction set to be processed based on signal domain classification; Add the revised pending instructions to the sliding window cache queue; The clustering and verification process is repeated until all time-consistent clusters meet the synchronization check.

5. The new energy control execution method based on the multi-market coupling mechanism according to claim 1 is characterized in that: The conflict detection is performed on the semantic vector to obtain a conflict matrix and acquire a strong conflict signal, specifically: Performing pairwise comparison on the unified semantic vectors based on cosine similarity to obtain conflict features; A conflict matrix is ​​constructed based on the conflict characteristics, and the signal corresponding to the maximum conflict intensity in the conflict matrix is ​​obtained as the strong conflict signal.

6. The new energy control execution method based on the multi-market coupling mechanism according to claim 1 is characterized in that: An interaction mapping matrix with the new energy device is constructed according to the control instruction set, and control actions are adaptively executed on the new energy device based on the interaction mapping matrix and feedback is provided, specifically: Construct an interactive mapping matrix between new energy equipment and control instructions, with equipment operating status parameters as row dimensions and historical control instruction categories as column dimensions, forming a multi-dimensional sparse matrix structure; Based on the deep matrix factorization model, the multi-dimensional sparse matrix structure is reconstructed to obtain the low-dimensional embedding vector of the device behavior preference and control response capability. Based on the control preference, the optimal control instruction vector is matched to the control instruction set in the low-dimensional embedding vector, and the optimal control action is determined by screening based on the cosine similarity in combination with the current device status; The most suitable control action is sent to the target new energy equipment execution module, and real-time status monitoring is performed based on the edge computing node to achieve a closed execution loop.

7. The new energy control execution method based on the multi-market coupling mechanism according to claim 1 is characterized in that: The Bayesian reasoning-based decision-making and fusion output candidate set for strong conflict signals is as follows: Construct a Bayesian causal network based on the device topology and control process logic; Based on the Bayesian causal network, strong conflicting signals from the same signal source are synthesized and the joint confidence distribution is output; The posterior probability of each control signal is obtained through the belief propagation algorithm, and the conflict-free signal combination with the largest posterior probability is selected as the control output candidate set.

8. The new energy control execution method based on the multi-market coupling mechanism according to claim 6 is characterized in that: The real-time status monitoring based on edge computing nodes is specifically as follows: Deploy edge computing nodes. After detecting that new energy equipment is executing a control action, the edge computing nodes associate the unique identifier of the current control instruction and generate a status response code. The time-window-based streaming cache aggregates status response codes in batches according to a preset time window to form a minimum feedback unit; Priority alarm feedback is provided by comparing the detection status of the minimum feedback unit with the abnormal threshold rules of the equipment operation status.

9. A system using the new energy control execution method based on a multi-market coupling mechanism as described in any one of claims 1 to 8, characterized in that: include: The control signal synchronization verification module is used to obtain control signals from multiple source markets and perform synchronization verification on the control signals from multiple source markets based on time-sensitive networking and local alignment caching method; The control signal synchronization processing module is used to perform semantic consistency analysis, conflict identification and fusion on the verified control signals, and output the control instruction set based on dynamic priority; The adaptive control module is used to construct an interactive mapping matrix with the new energy device according to the control instruction set, and adaptively execute control actions on the new energy device based on the interactive mapping matrix and provide feedback.

Images