A method for coordinated regulation of a building energy system

Abstract

The application belongs to the technical field of building energy management, and provides a kind of building energy system's collaborative regulation method, comprising: obtaining the dynamic response characteristics of heterogeneous energy subsystem, constructing feature matrix and determining the response system group with the highest coupling contribution degree through cross-correlation analysis;Secondly, based on historical data and genetic algorithm, a load mutation model is constructed to simulate periodic step load scenarios and evaluate the degree of mismatch in the collaborative response between subsystems;Then, correlation analysis of communication protocol heterogeneity is performed, and the dominant mismatch factor is identified and classified using a random forest algorithm;Finally, for physical property dominant mismatch, a compensation strategy is dynamically constructed to perceive and suppress the ripple diffusion effect of load mutation through state equation and recursive filtering.The application realizes the collaborative regulation between the subsystems of building energy system, and improves the robustness and energy utilization efficiency of the system in response to load mutation.

Description

Technical Field

[0001] This invention belongs to the field of building energy management technology, specifically a method for the coordinated control of building energy systems. Background Technology

[0002] In the field of building energy systems, as the integration of heterogeneous subsystems such as power, HVAC, and lighting continues to improve, the coordinated control under load change scenarios (such as the switching of large venue events and the peak-valley load fluctuations of commercial buildings) faces many challenges.

[0003] Existing technologies rely heavily on empirical parameter settings in mutation simulation, without combining historical data and genetic algorithms to construct constraints, thus failing to realistically reproduce periodic step load scenarios (such as sudden load changes during halftime in sports stadiums). At the same time, mismatch assessment lacks quantitative indicators (such as mismatch coverage ratio and coupling deviation ratio), making it difficult to pinpoint the specific time periods and extent of subsystem response asynchrony.

[0004] Faced with the chain reaction of subsystems caused by sudden load changes, existing technologies lack dynamic time-series compensation mechanisms, cannot cut off the chain of change propagation in real time, and are unable to improve the stability of energy systems.

[0005] Therefore, the present invention provides a method for coordinated control of building energy systems. Summary of the Invention

[0006] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0007] The technical solution adopted by this invention to solve its technical problem is: a method for coordinated control of building energy systems, comprising:

[0008] The dynamic response characteristics of heterogeneous energy subsystems within a building are obtained, a dynamic response characteristic matrix is ​​constructed, and a coupled analysis of the dynamic response characteristics of any two subsystems is performed based on the dynamic response characteristic matrix to determine the response system group.

[0009] By acquiring the load data of the subsystems and combining it with the dynamic response feature matrix, load mutation simulation is performed on the periodic step load scenario of the response system group to evaluate the degree of mismatch in the coordinated response between subsystems.

[0010] Based on the degree of response mismatch between subsystems, a correlation analysis is performed on the heterogeneity of subsystem communication protocols and physical dynamic characteristics to identify the dominant mismatch factor and determine whether the mismatch type corresponding to the mismatch factor is the dominant type.

[0011] If the mismatch is dominated by physical characteristics, a timing compensation strategy is dynamically constructed to generate a targeted instruction pre-compilation and dynamic timing alignment scheme, resulting in a dynamic timing alignment instruction sequence. The cooperative state of the subsystem after compensation is tracked to detect and suppress the ripple effect of load mutation.

[0012] As a further technical solution of the present invention: the method for determining the response system group is as follows:

[0013] Extract the response features of any two subsystems from the dynamic response feature matrix, and construct the corresponding response feature sequences respectively;

[0014] Cross-correlation analysis is performed on the response characteristic sequences of any two subsystems, and the correlation coefficient corresponding to the maximum value of the delay hysteresis is obtained as the coupling contribution.

[0015] Identify the two subsystems with the highest coupling contribution and construct the response system group.

[0016] As a further technical solution of the present invention: the method for evaluating the degree of mismatch in the cooperative response between the subsystems is as follows:

[0017] Construct a load mutation model, obtain historical load constraints, input the historical load constraints into the load mutation model, set a simulation period to perform load mutation simulation on the response system group, and extract the response characteristics of the response system group after the load mutation simulation.

[0018] The response characteristics of two subsystems in the response system group are obtained in multiple simulation cycles. The response characteristic sequence of each subsystem simulation output is constructed respectively to obtain the characteristic simulation sequence.

[0019] Response mismatch analysis was performed on the characteristic simulation sequence to extract the simulation mismatch segments across all simulation cycles:

[0020] Extract the mismatch coverage ratio and coupling deviation ratio of the simulated mismatch segment, and calculate the cooperative mismatch coefficient of the response system group based on the mismatch coverage ratio and coupling deviation ratio;

[0021] The degree of mismatch in the cooperative response among subsystems within a response system group is assessed based on the cooperative mismatch coefficient.

[0022] As a further technical solution of the present invention: the method for obtaining the historical load constraint conditions is as follows:

[0023] Acquire historical load data of building energy systems and identify the characteristics of step load abrupt changes in periodic step load scenarios;

[0024] Based on the periodic step load mutation characteristics, historical load constraints for historical step load mutation characteristics are constructed using a genetic algorithm.

[0025] As a further technical solution of the present invention: the method for extracting the mismatch coverage ratio and coupling deviation ratio of the simulated mismatch segment is as follows:

[0026] Obtain the single duration period of the simulated mismatch segment, as well as the single duration period of the periodic step load scenario;

[0027] The coverage ratio of the single duration of the simulated mismatch segment and the single duration of the periodic step load scenario is calculated to obtain the mismatch coverage ratio.

[0028] The peak value of coupling contribution of the characteristic simulation sequence in a single duration of the simulation mismatch segment is obtained. The deviation ratio between the peak value of coupling contribution and the target boundary of coupling contribution is calculated to obtain the coupling deviation ratio.

[0029] As a further technical solution of the present invention: the method for identifying the dominant mismatch factor is as follows:

[0030] If there is a risk of mismatch in the coordinated response among subsystems within the response system group, the historical communication characteristics of each subsystem in the response system group are obtained from the logs of the building energy system and input into the load mutation model.

[0031] Historical communication characteristics include peak communication latency and protocol compatibility index.

[0032] Obtain the communication feature sequences of all simulated mismatch segments and the feature simulation sequences of subsystems, construct the mismatch factor matrix, and classify the mismatch factors by mismatch type;

[0033] A collaborative factor model is constructed based on the random forest algorithm. The mismatch factor matrix is ​​input into the collaborative factor model to extract the dominant mismatch factor.

[0034] Matching correspondence analysis is performed based on the mismatch dominant factor to determine whether the mismatch type corresponding to the mismatch factor is the dominant type.

[0035] As a further technical solution of the present invention: the matching and correspondence analysis is performed in the following manner:

[0036] Simulation verification is performed based on the mismatch dominant factor and the corresponding mismatch type. The mismatch dominant factor is optimized, and the optimized cooperative mismatch coefficient is output.

[0037] Calculate the deviation of the collaborative mismatch coefficient before and after optimization, perform type matching evaluation on the optimization results, and determine whether the mismatch type corresponding to the mismatch factor is the dominant type.

[0038] As a further technical solution of the present invention: the method for dynamically constructing the timing compensation strategy is as follows:

[0039] Based on the load mutation model of the building energy subsystem, control commands are pre-compiled;

[0040] Dynamic timing alignment is performed based on the timing deviation characteristics of the simulated mismatch segment.

[0041] For cases where the magnitude of mutation exceeds the load, load smoothing is performed, and dynamic timing alignment instructions are constructed.

[0042] As a further technical solution of the present invention: the method for sensing and suppressing the ripple diffusion effect of the load mutation is as follows:

[0043] After tracking compensation, obtain the load change and coupling contribution of the subsystems within the response system group, and construct the load mutation vector and coupling strength matrix;

[0044] A ripple effect propagation equation is constructed, and recursive filtering is performed using historical parameters to realize the ripple effect triggering criterion.

[0045] Once a ripple effect is detected, the instruction timing is adjusted in the subsystem to suppress the ripple effect.

[0046] As a further technical solution of the present invention, the method for implementing the ripple effect triggering criterion is as follows:

[0047] Establish the system state vector and sudden change influence coefficient vector of the response system group, and construct the ripple effect propagation equation by combining the load sudden change vector and the coupling strength matrix;

[0048] The ripple effect propagation equation is recursively filtered, and the Euclidean distance deviation of the system state vector is calculated.

[0049] When the Euclidean distance deviation value exceeds a preset Euclidean distance deviation threshold, a ripple effect is determined to exist between the subsystems. The beneficial effects of this invention are as follows:

[0050] 1. By acquiring dynamic characteristics such as the transient response ratio of the power subsystem, the energy efficiency fluctuation rate of the HVAC subsystem, and the response delay ratio of the lighting subsystem, a multi-dimensional feature matrix is ​​constructed. Based on cross-correlation analysis, the coupling contribution between subsystems is quantified, which is beneficial for accurately identifying response system groups with close synergistic effects. This improves the targeting of system analysis, reduces ineffective regulation of weakly correlated subsystems, and reduces the deviation of regulation strategies caused by ambiguity in the coupling relationship of factor systems, providing a basis for object selection for subsequent load change simulation.

[0051] 2. By identifying the characteristics of periodic step loads and constructing historical load constraints using a genetic algorithm, the mismatch coverage ratio, coupling deviation ratio, and cooperative mismatch coefficient are simulated and calculated using a load mutation model. This helps to recreate real load mutation scenarios (such as the step load during halftime in a stadium), improving the accuracy of mismatch risk assessment. Quantifying the mismatch coverage ratio and coupling deviation ratio helps to pinpoint the timing and extent of subsystem response asynchrony, reducing assessment errors caused by traditional experience-based judgments and providing data support for mismatch type analysis.

[0052] 3. By associating historical communication characteristics with simulated mismatch segments, a communication characteristic sequence is constructed. This sequence is then used to classify mismatch types and extract dominant factors using a random forest algorithm. This facilitates the analysis of the root causes of mismatches. For example, through cross-analysis of communication delays and physical response parameters, the specific problem of protocol conversion delays causing HVAC response lags can be identified. Classification processing improves the targeting of control strategies, while simulation verification optimizes mismatch factors, enhancing the reliability of dominant factor identification.

[0053] 4. Based on the load mutation model, pre-compile control instructions and dynamically adjust timing alignment to smooth overload mutations. Furthermore, state equations and recursive filtering are used to suppress ripple propagation. This helps reduce the impact of load mutations on the system and improves the synchronization of subsystem responses. By tracking the coordinated state in real time and adjusting instructions, it helps to cut off the mutation propagation chain, reduce the global chain reaction caused by load mutations, and improve the operational stability and energy utilization efficiency of the building energy system. Attached Figure Description

[0054] The invention will now be further described with reference to the accompanying drawings.

[0055] Figure 1 This is a flowchart illustrating the steps of a collaborative control method for a building energy system according to an embodiment of the present invention;

[0056] Figure 2 This is a flowchart illustrating the steps of the collaborative mismatch analysis described in the embodiments of the present invention;

[0057] Figure 3 This is a module architecture diagram of a collaborative control system for a building energy system according to an embodiment of the present invention. Detailed Implementation

[0058] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0059] Example 1

[0060] Please see Figure 1 As shown in the embodiment of the present invention, a method for coordinated control of a building energy system includes the following steps:

[0061] Step 1: Obtain the dynamic response characteristics of heterogeneous energy subsystems within the building, construct a dynamic response characteristic matrix, and perform coupled analysis on the dynamic response characteristics of any two subsystems based on the dynamic response characteristic matrix to determine the response system group;

[0062] Preferably, the heterogeneous energy subsystems within the building include electrical, HVAC, and lighting subsystems;

[0063] The method for quantifying the dynamic response characteristics of each subsystem is as follows:

[0064] Preferably, the heterogeneous energy subsystems within the building include electrical, HVAC, and lighting subsystems;

[0065] The transient response ratio of the power subsystem, the energy efficiency fluctuation rate of the HVAC subsystem, and the response delay ratio of the lighting subsystem were obtained during the historical monitoring period and used as the response characteristics of each subsystem.

[0066] Understandably, during the historical monitoring period, the transient response data of voltage and current of the power subsystem during load changes are collected in real time by sensors, and compared with the reference values ​​under rated operating conditions to calculate the transient response ratio; the input power and output energy efficiency data of the HVAC subsystem are acquired simultaneously, and the energy efficiency fluctuation rate is calculated based on the fluctuation range of energy efficiency indicators per unit time through standard deviation or coefficient of variation; for the lighting subsystem, the time difference between the control command sending time and the actual time that the lamps reach the target state is recorded, and the ratio of this delay time to the standard response time is used as the response delay ratio;

[0067] Construct a dynamic response feature matrix that includes the transient response ratio of the power subsystem, the energy efficiency fluctuation rate of the HVAC subsystem, and the response delay ratio of the lighting subsystem within multiple monitoring periods;

[0068] Based on the dynamic response feature matrix, a coupling analysis of heterogeneous energy subsystems within a building is conducted to determine the coupling contribution between different subsystems.

[0069] The method for coupling analysis of heterogeneous energy subsystems within buildings is as follows:

[0070] Extract the response features of any two subsystems from the dynamic response feature matrix, and construct the corresponding response feature sequences respectively;

[0071] Cross-correlation analysis is performed on the response characteristic sequences of any two subsystems, and the correlation coefficient corresponding to the maximum value of the delay hysteresis is obtained as the coupling contribution.

[0072] Those skilled in the art will understand that cross-correlation analysis calculates the correlation between the response feature sequences of two subsystems under different time delay lags to obtain a correlation coefficient sequence. The correlation coefficient corresponding to the largest delay lag value indicates that the response feature sequences of the two subsystems have the highest matching degree under that time delay lag.

[0073] It should be explained that the coupling contribution metric characterizes the strength of the temporal correlation between the dynamic responses of subsystems, reflecting the degree of influence of a sudden load change or response deviation in one subsystem on the other. A higher coupling contribution value indicates a stronger degree of synergistic response coupling between the two subsystems under load change scenarios (e.g., a more significant mutual influence between power fluctuations and the start-up and shutdown of HVAC equipment). It is a core quantitative indicator for identifying key synergistic pathways in building energy systems, providing a physical-level correlation basis for screening response system groups (e.g., the power-HVAC subsystem with the highest coupling contribution).

[0074] Identify the two subsystems with the highest coupling contribution and construct the response system group.

[0075] Step 2: Obtain the load data of the subsystems and combine it with the dynamic response characteristic matrix to simulate the periodic step load scenario of the response system group and evaluate the degree of mismatch in the coordinated response between subsystems.

[0076] The method for simulating load mutations is as follows:

[0077] Acquire historical load data of building energy systems and identify the characteristics of step load abrupt changes in periodic step load scenarios;

[0078] For example, a periodic step load scenario includes halftime at a game in a large stadium;

[0079] It is understandable that historical load data of the power, HVAC, and lighting subsystems are extracted from the building energy system. The historical load data is then used to identify similar abrupt change patterns of the power, HVAC, and lighting subsystems through the DBSCAN clustering algorithm, thereby identifying the step load change characteristics of periodic step load scenarios.

[0080] Historical load data includes: time series of power and HVAC loads; step load abrupt change characteristics include: load abrupt change amplitude P(t) and abrupt change time t.

[0081] The coupling contribution sequence of the step load mutation characteristics and the response system group is correlated and screened to extract the periodic step load scenarios that are associated with it.

[0082] Based on the periodic step load mutation characteristics, historical load constraints for historical step load mutation characteristics are constructed using a genetic algorithm.

[0083] Preferably, the method for constructing historical load constraints for historical step load mutation characteristics using a genetic algorithm is as follows:

[0084] S201, Chromosome coding and parameter definition;

[0085] Preferably, the load mutation amplitude, mutation rate, and duration of each subsystem are mapped to the chromosome structure of the genetic algorithm, and the physical meaning of each gene position is defined by real number encoding (for example, the first position represents the mutation amplitude of the power subsystem, with a range of 0 to 120% of the rated value).

[0086] Meanwhile, the mutation rate does not exceed the maximum slope allowed by the device as a limiting parameter boundary, so that the chromosome directly corresponds to the quantifiable characteristics of the load mutation, providing an optimization object for subsequent genetic iteration;

[0087] S202. Construct a comprehensive fitness function based on the historical load data fitting degree function and the coupling contribution matching degree function;

[0088] Through formula one: Construct the F1 fit function for historical load data;

[0089] Through formula two: Construct the coupling contribution matching degree function F2;

[0090] Through formula three: Construct the comprehensive fitness function F;

[0091] in, These represent the simulated load fluctuation magnitude and the historical load fluctuation magnitude at time t, respectively. , where is the preset fitness control factor, T is the total length of the time series for monitoring load mutation characteristics, m is the total number of pairwise combinations of subsystems in the building energy system, and i and j are the index identifiers of subsystems.

[0092] Simulate the coupling contribution between the i-th subsystem and the j-th subsystem under the load mutation scenario and the coupling contribution between the i-th subsystem and the j-th subsystem under the historical load mutation scenario, respectively;

[0093] S203, Genetic Algorithm Training and Iteration;

[0094] Preferably, an initial population containing several chromosomes (e.g., 50-100 chromosomes, with randomly generated parameter combinations) is established, and the quality of each chromosome is calculated using the comprehensive fitness function of S202. High-quality individuals are selected using roulette wheel selection and elite retention strategies, and offspring are generated through linear combination of parent parameters and Gaussian mutation to produce a new population.

[0095] Repeat the fitness calculation, selection, crossover, and mutation process until the maximum number of iterations of 200 generations is reached, and finally output the chromosome with the highest fitness, corresponding to the optimal combination of load mutation parameters.

[0096] S204. Generation of historical load constraints;

[0097] Preferably, the load mutation constraint is extracted from the optimal chromosome output by the genetic algorithm;

[0098] The constraints include: target boundaries for load mutation magnitude, rate, duration, and coupling contribution.

[0099] A load mutation model is constructed, historical load constraints are input into the load mutation model, a simulation period is set to perform load mutation simulation on the response system group, and the response characteristics of the response system group after the load mutation simulation are extracted.

[0100] Those skilled in the art will understand that the transient circuit equations of the power subsystem, the thermal balance equations of the HVAC subsystem, and the control logic of the lighting subsystem are integrated to form a load mutation model coupled in multiple physical domains.

[0101] Input historical load constraints into the load mutation model, use historical load data optimized by genetic algorithm as the excitation source boundary conditions of the model, and set the simulation period to 30 minutes and refine the time step to the second level according to the time scale of historical load mutation (such as the 30-minute load change scenario during halftime in a stadium).

[0102] For highly collaborative subsystem groups selected by coupling contribution, load mutation excitation is injected into the model to simulate the dynamic interactive response of the subsystems under the impact of mutation. Through the data acquisition module built into the model, the characteristic sequences of power transient response ratio, HVAC energy efficiency fluctuation rate and lighting response delay ratio in the simulation time series are tracked in real time, providing dynamic data support for subsequent mismatch assessment.

[0103] The response characteristics of two subsystems in the response system group are obtained in multiple simulation cycles. The response characteristic sequence of each subsystem simulation output is constructed respectively to obtain the characteristic simulation sequence.

[0104] The method for assessing the mismatch in collaborative response among response system groups is as follows:

[0105] Response mismatch analysis is performed on the characteristic simulation sequence to extract the simulation mismatch segments within all simulation cycles;

[0106] Obtain the single duration period of the simulated mismatch segment, as well as the single duration period of the periodic step load scenario;

[0107] The coverage ratio of the single duration of the simulated mismatch segment and the single duration of the periodic step load scenario is calculated to obtain the mismatch coverage ratio.

[0108] The peak value of coupling contribution of the characteristic simulation sequence in a single duration period of the simulation mismatch segment is obtained, and the deviation ratio between the peak value of coupling contribution and the target boundary of coupling contribution is calculated to obtain the coupling deviation ratio.

[0109] The cooperative mismatch coefficient of the response system group is calculated based on the mismatch coverage ratio and coupling deviation ratio.

[0110] Preferably, by formula: Obtain the coefficient of mismatch M;

[0111] in, Here, N represents the number of the simulated mismatch segment, and N is the total number of simulated mismatch segments. The first The mismatch coverage ratio of the simulated mismatch segment, the first The coupling deviation ratio of each simulated mismatched segment;

[0112] It is understandable that the coordination mismatch coefficient quantifies the overall coordination deviation of the response system group under load change scenarios;

[0113] The purpose of calculating the coordination mismatch coefficient is:

[0114] Objective 1: To verify the effectiveness of simulation scenarios. By comparing the deviation between the calculated values ​​of the simulation scenarios and the measured values ​​of historical scenarios, the cooperative mismatch coefficient is used to determine the ability of the load mutation model to reproduce the cooperative mismatch patterns of real multi-subsystem systems.

[0115] Objective 2: As a basis for formulating the timing of control strategies, the timing characteristics of the coordination mismatch coefficient (such as the peak occurrence time and fluctuation period) can guide the timing arrangement of control commands: if the coefficient reaches its peak 10 seconds after a sudden load change, it indicates that the coordination mismatch of the subsystem is most severe during that period, and the timing logic of issuing control commands 1 second in advance should be preset in the control strategy.

[0116] Objective 3: Locating the root cause of mismatch and using the coordination mismatch coefficient to support the quantitative decomposition of the subsystem's contribution to coupling. By calculating the coupling mismatch weights of different subsystem pairs (such as power-HVAC, power-lighting), the coordination deviation between subsystems can be directly identified as the dominant factor in overall mismatch, improving the efficiency of problem localization.

[0117] like Figure 2 As shown, the degree of mismatch between subsystems within the response system group is evaluated based on the mismatch coefficient. If the mismatch coefficient is not within the preset range, it is considered that there is a risk of mismatch between the subsystems within the response system group.

[0118] If the coordination mismatch coefficient is within the preset range, the change in the coordination mismatch coefficient will be continuously monitored.

[0119] Example 2

[0120] like Figure 1 As shown, a method for coordinated control of a building energy system further includes the following steps:

[0121] Step 3: Based on the degree of response mismatch between subsystems, conduct correlation analysis on the heterogeneity of subsystem communication protocols and physical dynamic characteristics, identify the dominant mismatch factors, and determine whether the mismatch type corresponding to the mismatch factor is the dominant type.

[0122] The method for correlation analysis between the heterogeneity of subsystem communication protocols and physical dynamic characteristics is as follows:

[0123] Preferably, if there is a risk of mismatch in the coordinated response among subsystems within the response system group, the historical communication characteristics of each subsystem in the response system group are obtained from the logs of the building energy system and input into the load mutation model;

[0124] Historical communication characteristics include peak communication latency and protocol compatibility index.

[0125] The load mutation model correlates communication characteristics with all simulated mismatch segments to construct a communication characteristic sequence;

[0126] It should be explained that the protocol compatibility index is calculated by summing the matching load of data format, instruction set, timing rules, and error handling dimensions through parsing or packet capture analysis.

[0127] The mutation model associates communication features with all simulation mismatch segments in the following way: it extracts communication features and their corresponding timestamps from the building energy system logs, and then aligns them spatiotemporally with the simulation-generated mismatch segments according to the timestamps.

[0128] A subset of communication features corresponding to each simulated mismatch segment is extracted using a sliding window. The delay peak value is standardized by quantiles, and the protocol compatibility index is converted into a numerical feature in the 0-1 range. At the same time, a dynamic time warping algorithm is used to match the temporal correlation between communication features and mismatch segments, and invalid features with a time overlap rate of less than 30% with the mismatch segment are removed. The processed communication feature vectors are arranged in the order of occurrence of mismatch segments to construct a three-dimensional communication feature sequence containing timestamps, delay peak values, and protocol indices, providing standardized input for the subsequent construction of the mismatch factor matrix.

[0129] Obtain the communication feature sequences of all simulated mismatch segments and the feature simulation sequences of subsystems, construct the mismatch factor matrix, and classify the mismatch factors by mismatch type;

[0130] For example: if the peak communication delay in the simulated mismatch segment continuously exceeds 200ms and the protocol compatibility index is less than 0.4, and the mismatch coverage reaches 40%, then it is classified as a communication-dominated mismatch.

[0131] If the energy efficiency fluctuation rate of the HVAC system corresponding to the simulated mismatch segment exceeds the rated value by 25% and the power transient response ratio deviation reaches 30%, but the communication delay is only 80ms, then it is classified as a physical characteristic-driven mismatch.

[0132] If the communication delay is 180ms and the HVAC equipment response time is 5 seconds, and both contribute more than 35% to the mismatch coverage ratio, it is judged as a mixed-effect mismatch.

[0133] During classification, the communication delay ≥150ms and the physical parameter deviation ≥20% are set as thresholds, and the type division is achieved by combining the importance of random forest features;

[0134] A collaborative factor model is constructed based on the random forest algorithm. The mismatch factor matrix is ​​input into the collaborative factor model to extract the dominant mismatch factor.

[0135] Those skilled in the art will understand that, by integrating multiple dimensions such as power transient response ratio, HVAC energy efficiency fluctuation rate, lighting response delay ratio, coupling contribution, communication delay peak, and protocol compatibility index, each sample corresponds to characteristic data under a certain load change scenario.

[0136] The mismatch factor matrix is ​​divided into training and test sets. A collaborative factor model is constructed based on the random forest algorithm. Multiple sets of training samples are generated by Bootstrap sampling. A feature subset is randomly selected from each set of samples to train the decision tree.

[0137] By calculating the contribution of each mismatch factor to the collaborative mismatch state in the decision tree split, the dominant factors are screened, sorted in descending order of feature importance weight, and the factors with higher weights (such as peak communication delay and power transient response ratio) are selected as the dominant mismatch factors.

[0138] Simulation verification is performed based on the mismatch dominant factor and the corresponding mismatch type. The mismatch dominant factor is optimized, and the optimized cooperative mismatch coefficient is output.

[0139] Calculate the deviation of the collaborative mismatch coefficient before and after optimization, perform type matching evaluation on the optimization results, and determine whether the mismatch type corresponding to the mismatch factor is the dominant type;

[0140] It should be explained that if the deviation value exceeds the preset threshold, the mismatch factor is determined to have a significant impact on the collaborative mismatch. Combined with the mismatch type corresponding to the mismatch factor (such as communication-dominant type or physical characteristic-dominant type), if it has the highest contribution to the reduction of the collaborative mismatch coefficient, then the mismatch type is determined to be the dominant type, thereby completing the type matching evaluation of the optimization results.

[0141] The purpose of identifying whether a mismatch type is dominant is to provide an optimization target for compensation strategies. If it is communication-dominant, adjustments should focus on communication protocol adaptation; if it is physical characteristic-dominant, physical layer parameters need to be adjusted. By distinguishing the root causes, ineffective adjustments due to strategy mismatches can be reduced, thus compressing trial-and-error costs.

[0142] Step 4: If the mismatch is dominated by physical characteristics, a timing compensation strategy is dynamically constructed to generate a targeted instruction pre-compilation and dynamic timing alignment scheme, obtain a dynamic timing alignment instruction sequence, and track the cooperative state of the subsystem after compensation to sense and suppress the ripple diffusion effect of load mutation.

[0143] The method for constructing a dynamic construction timing compensation strategy is as follows:

[0144] S401. Based on the load mutation model of the building energy subsystem, pre-compile control commands;

[0145] Preferably, based on the load mutation amplitude, response delay, and duration in the simulation mismatch parameters output by the load mutation model, combined with the response threshold, the optimal control parameters for compensating for the mismatch are pre-calculated from the transient equations of the power, HVAC, and lighting subsystems. Standardized instruction templates containing adjustment stages, time intervals, and adjustment target parameters are generated according to equipment type (e.g., staged frequency regulation instructions for HVAC water pumps). The effectiveness of the instructions is verified through closed-loop simulation of the model. Finally, the templates are stored as standardized files that can be quickly retrieved and called according to mismatch characteristics, thus realizing the pre-compilation of control instructions.

[0146] S402. Based on the simulation of timing deviation characteristics of mismatch segments, perform dynamic timing alignment;

[0147] Preferably, the subsystem response delay time difference and abrupt trigger time offset are extracted from the simulation mismatch segment as timing deviation features. The deviation trend under the current operating condition is predicted by the LSTM algorithm, and the control command trigger time is dynamically adjusted according to the predicted deviation plus the safety margin.

[0148] A rigid time window is set after a sudden load change occurs, and the actual start-up time and adjustment range are compared with the simulation expectation through real-time acquisition equipment. If the deviation exceeds the threshold, the timing of subsequent instructions is corrected, forming a closed-loop alignment mechanism of mismatch feature extraction, deviation prediction, timing compensation, and real-time calibration.

[0149] S403. Perform load smoothing for excessive mutation amplitude and construct dynamic timing alignment instructions;

[0150] Preferably, when the load change amplitude exceeds the standard in real time, the original step change command is split into a multi-level step adjustment sequence, for example, the 100% power change is split into 3 levels, with each level spaced 5 seconds apart;

[0151] The optimal adjustment slope for each stage is calculated using the equipment physical model, and buffer time is inserted between stages to reduce system impact. At the same time, the number of steps and time intervals are dynamically adjusted based on the simulation mismatch segment data, and finally a dynamic instruction sequence containing adjustment stages, target parameters and timestamps is generated, which is sent to the subsystem controller in real time to achieve gradual load adjustment.

[0152] The method for tracking the coordinated state of the subsystems after compensation and sensing and suppressing the ripple effect of load mutations is as follows:

[0153] S411. Obtain the load change and coupling contribution of the subsystems within the response system group after tracking compensation, and construct the load mutation vector and coupling strength matrix.

[0154] Preferably, the load mutation vector of the two subsystems is defined as follows: ;

[0155] in, , , Indicates that subsystem i is in time interval Load variation within, This indicates that subsystem j is in time interval Load variation within;

[0156] It should be explained that subsystems i and j are the subsystem numbers of the response system group;

[0157] The coupling strength matrix is ​​defined as , where c is the coupling contribution of the two subsystems;

[0158] S412. Construct the ripple effect propagation equation and perform recursive filtering based on historical parameters to realize the ripple effect triggering criterion.

[0159] Through the formula: Construct the ripple effect propagation equation;

[0160] Understandably, constructing the ripple effect propagation equation quantitatively analyzes the propagation path, intensity attenuation, and temporal correlation of the coupled chain reaction caused by load changes between subsystems in a building energy system through mathematical relationships. This transforms the dynamic process of energy signal interaction into a computable state evolution law; it reflects the physical mechanism of how changes spread across systems (such as coupling strength determining propagation amplitude and response delay shaping timing differences), providing model support for predicting the impact boundary and peak time of ripples, as well as designing advance compensation.

[0161] in, The system state vector of the response system group is given by S1 and S2, which are the response characteristics of two subsystems within the response system group, respectively.

[0162] This is the mutation impact coefficient matrix. These represent the mismatch coverage ratios of the two subsystems within the response system group, respectively.

[0163] It should be explained that the mismatch coverage ratio of the two subsystems within the response system group is obtained by breaking down the overall mismatch coverage ratio of the response system group.

[0164] The recursive filtering method is as follows:

[0165] Through the system of equations: The state is updated recursively, where c is the coupling contribution between subsystem 1 and subsystem 2;

[0166] β is a preset smoothing coefficient; for example, when the physical characteristics dominate the mismatch, β is 0.7.

[0167] in, These are the response characteristics of subsystem 1 and subsystem 2 at time t, respectively.

[0168] The response characteristics of subsystem 1 and subsystem 2 at time t-1 are respectively.

[0169] These represent the load changes of subsystem 1 and subsystem 2 at time t, respectively.

[0170] The timing compensation parameters for each subsystem are obtained by linearly mapping the coupling deviation ratio to the reference compensation value;

[0171] It is understandable that; through the coupling bias ratio The degree of mismatch coupling in quantified subsystem 1 is determined by the coupling deviation ratio, which is the difference in mismatch between the power and HVAC subsystems, relative to the normalized proportion of their average mismatch. Then, the average compensation parameters of the power subsystem in historical fully coordinated scenarios are extracted as the benchmark compensation value. Represents the basic compensation amount;

[0172] Finally, a preset simulation scaling factor, optimized by simulation, is introduced. ,For example =0.1, used to describe the impact of coupling mismatch on the compensation strength, establishing a linear relationship: ,make The compensation intensity is dynamically adjusted as the coupling mismatch worsens, so as to match the degree of mismatch with the intensity of compensation.

[0173] It should be noted that the linear mapping method used in subsystem 2 is the same as that used in subsystem 1;

[0174] Through the formula: Calculate the Euclidean distance deviation of the system state vector. ;

[0175] When the Euclidean distance deviation value is higher than the preset Euclidean distance deviation threshold, it is determined that there is a ripple effect between the subsystems.

[0176] Those skilled in the art will understand that when a ripple effect is detected, the timing of instructions to the subsystem is adjusted (e.g., HVAC adjustment is delayed by 300ms) to cut off the mutation propagation chain; finally, the suppression effect is verified by the cooperative mismatch coefficient, requiring the coefficient to decrease by more than 25% after compensation, thereby achieving closed-loop control of the entire process from mutation monitoring and propagation analysis to dynamic suppression.

[0177] Example 3

[0178] like Figure 3 As shown, a collaborative control system for a building energy system further includes the following modules:

[0179] Response filtering module: used to obtain the dynamic response characteristics of heterogeneous energy subsystems within a building, construct a dynamic response characteristic matrix, and perform coupled analysis on the dynamic response characteristics of any two subsystems based on the dynamic response characteristic matrix to determine the response system group;

[0180] Mismatch assessment module: Acquires load data of subsystems and combines it with dynamic response feature matrix to simulate load mutations in periodic step load scenarios of response system group and assess the degree of mismatch in cooperative response between subsystems;

[0181] Type Analysis Module: Based on the degree of response mismatch between subsystems, it performs correlation analysis on the heterogeneity of subsystem communication protocols and physical dynamic characteristics, identifies the dominant mismatch factor, and determines whether the mismatch type corresponding to the mismatch factor is the dominant type;

[0182] Control and analysis module: If the mismatch is dominated by physical characteristics, a timing compensation strategy is dynamically constructed to generate a targeted instruction pre-compilation and dynamic timing alignment scheme, obtain a dynamic timing alignment instruction sequence, and track the cooperative state of the subsystem after compensation, so as to sense and suppress the ripple diffusion effect of load mutation.

[0183] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for coordinated control of a building energy system, characterized in that: Includes the following steps: The dynamic response characteristics of heterogeneous energy subsystems within a building are obtained, a dynamic response characteristic matrix is ​​constructed, and a coupled analysis of the dynamic response characteristics of any two subsystems is performed based on the dynamic response characteristic matrix to determine the response system group. By acquiring the load data of the subsystems and combining it with the dynamic response feature matrix, load mutation simulation is performed on the periodic step load scenario of the response system group to evaluate the degree of mismatch in the coordinated response between subsystems. Based on the degree of response mismatch between subsystems, a correlation analysis is performed on the heterogeneity of subsystem communication protocols and physical dynamic characteristics to identify the dominant mismatch factor and determine whether the mismatch type corresponding to the mismatch factor is the dominant type. If the mismatch is dominated by physical characteristics, a timing compensation strategy is dynamically constructed to generate a targeted instruction pre-compilation and dynamic timing alignment scheme, resulting in a dynamic timing alignment instruction sequence. The cooperative state of the subsystem after compensation is tracked to detect and suppress the ripple effect of load mutation.

2. The method for coordinated control of a building energy system according to claim 1, characterized in that: The method for determining the response system group is as follows: Extract the response features of any two subsystems from the dynamic response feature matrix, and construct the corresponding response feature sequences respectively; Cross-correlation analysis is performed on the response characteristic sequences of any two subsystems, and the correlation coefficient corresponding to the maximum value of the delay hysteresis is obtained as the coupling contribution. Identify the two subsystems with the highest coupling contribution and construct the response system group.

3. The method for coordinated control of a building energy system according to claim 1, characterized in that: The method for assessing the degree of mismatch in the cooperative response between the subsystems is as follows: Construct a load mutation model, obtain historical load constraints, input the historical load constraints into the load mutation model, set a simulation period to perform load mutation simulation on the response system group, and extract the response characteristics of the response system group after the load mutation simulation. The response characteristics of two subsystems in the response system group are obtained in multiple simulation cycles. The response characteristic sequence of each subsystem simulation output is constructed respectively to obtain the characteristic simulation sequence. Response mismatch analysis was performed on the characteristic simulation sequence to extract the simulation mismatch segments across all simulation cycles: Extract the mismatch coverage ratio and coupling deviation ratio of the simulated mismatch segment, and calculate the cooperative mismatch coefficient of the response system group based on the mismatch coverage ratio and coupling deviation ratio; The degree of mismatch in the cooperative response among subsystems within a response system group is assessed based on the cooperative mismatch coefficient.

4. The method for coordinated control of a building energy system according to claim 3, characterized in that: The historical load constraints are obtained as follows: Acquire historical load data of building energy systems and identify the characteristics of step load abrupt changes in periodic step load scenarios; Based on the periodic step load mutation characteristics, historical load constraints for historical step load mutation characteristics are constructed using a genetic algorithm.

5. The method for coordinated control of a building energy system according to claim 3, characterized in that: The method for extracting the mismatch coverage ratio and coupling deviation ratio of the simulated mismatch segment is as follows: Obtain the single duration period of the simulated mismatch segment, as well as the single duration period of the periodic step load scenario; The coverage ratio of the single duration of the simulated mismatch segment and the single duration of the periodic step load scenario is calculated to obtain the mismatch coverage ratio. The peak value of coupling contribution of the characteristic simulation sequence in a single duration of the simulation mismatch segment is obtained. The deviation ratio between the peak value of coupling contribution and the target boundary of coupling contribution is calculated to obtain the coupling deviation ratio.

6. The method for coordinated control of a building energy system according to claim 1, characterized in that: The method for identifying the dominant mismatch factor is as follows: If there is a risk of mismatch in the coordinated response among subsystems within the response system group, the historical communication characteristics of each subsystem in the response system group are obtained from the logs of the building energy system and input into the load mutation model. Historical communication characteristics include peak communication latency and protocol compatibility index. Obtain the communication feature sequences of all simulated mismatch segments and the feature simulation sequences of subsystems, construct the mismatch factor matrix, and classify the mismatch factors by mismatch type; A collaborative factor model is constructed based on the random forest algorithm. The mismatch factor matrix is ​​input into the collaborative factor model to extract the dominant mismatch factor. Matching correspondence analysis is performed based on the dominant mismatch factor to determine whether the mismatch type corresponding to the mismatch factor is the dominant type.

7. The method for coordinated control of a building energy system according to claim 6, characterized in that: The matching and correspondence analysis is performed as follows: Simulation verification is performed based on the mismatch dominant factor and the corresponding mismatch type. The mismatch dominant factor is optimized, and the optimized cooperative mismatch coefficient is output. Calculate the deviation of the collaborative mismatch coefficient before and after optimization, perform type matching evaluation on the optimization results, and determine whether the mismatch type corresponding to the mismatch factor is the dominant type.

8. The method for coordinated control of a building energy system according to claim 1, characterized in that: The timing compensation strategy is constructed dynamically as follows: Based on the load mutation model of the building energy subsystem, control commands are pre-compiled; Dynamic timing alignment is performed based on the timing deviation characteristics of the simulated mismatch segment. For cases where the magnitude of mutation exceeds the load, load smoothing is performed, and dynamic timing alignment instructions are constructed.

9. The method for coordinated control of a building energy system according to claim 1, characterized in that: The way to sense and suppress the ripple effect of the load abrupt change is as follows: After tracking compensation, obtain the load change and coupling contribution of the subsystems within the response system group, and construct the load mutation vector and coupling strength matrix; A ripple effect propagation equation is constructed, and recursive filtering is performed using historical parameters to realize the ripple effect triggering criterion. Once a ripple effect is detected, the instruction timing is adjusted in the subsystem to suppress the ripple effect.

10. A method for coordinated control of a building energy system according to claim 9, characterized in that: The method for implementing the ripple effect triggering criterion is as follows: Establish the system state vector and sudden change influence coefficient vector of the response system group, and construct the ripple effect propagation equation by combining the load sudden change vector and the coupling strength matrix; The ripple effect propagation equation is recursively filtered, and the Euclidean distance deviation of the system state vector is calculated. When the Euclidean distance deviation value is higher than the preset Euclidean distance deviation threshold, it is determined that there is a ripple effect between the subsystems.

Images