CN-121995752-A - Building energy-carbon double control and equipment predictive maintenance linkage energy consumption optimization method and system

CN121995752ACN 121995752 ACN121995752 ACN 121995752ACN-121995752-A

Abstract

The invention discloses a construction energy-carbon double control and equipment predictive maintenance linkage energy consumption optimization method, which aims to solve the problems that the equipment efficiency degradation is 'non-failure but high energy consumption' is difficult to identify in time and difficult to link with energy-carbon control, and the invention generates a control state characteristic sequence and outputs an equipment health risk sequence by acquiring construction operation multi-source data and time alignment cleaning, establishes a health constraint multi-objective optimization model of energy consumption, carbon emission and health risk to generate a control objective parameter set comprising punishment parameters and constraint parameters, adopts a safety reinforcement learning controller to generate and verify the issuing execution of corrective control actions, the method comprises the steps of constructing an action condition expected energy consumption model to obtain expected energy consumption and form a total residual error with actual energy consumption, combining execution feedback to obtain an execution residual error and decomposing the execution residual error to generate an efficiency residual error, screening a health window according to health risks and the execution residual error, automatically updating a baseline model by using only health window data, determining a degradation threshold value to calculate an efficiency degradation index and additional energy consumption and additional carbon emission based on the health window, triggering a maintenance instruction and updating a control target parameter set, and realizing the technical effects of optimizing the energy consumption and the carbon emission, identifying the efficiency degradation and maintaining linkage.

Inventors

CHEN JUNCHENG
YANG YI
WANG YUYANG
CHEN QILIANG
XU SHIJIE
SHEN SHULING

Assignees

江苏尔讯智能科技股份有限公司

Dates

Publication Date: 20260508
Application Date: 20260121

Claims (10)

1. A building energy-carbon double control and equipment predictive maintenance linkage energy consumption optimization method comprises the following steps: S1, acquiring building operation data of a target building, performing time alignment and data cleaning to obtain aligned cleaned building operation data, S2, generating a state characteristic sequence for control decision based on the aligned cleaned building operation data, outputting a device health risk sequence through a health assessment model, S3, establishing a health constraint multi-target optimization model of energy consumption, carbon emission and health risk based on the state characteristic sequence, the device health risk sequence, a carbon emission factor and an energy carbon control index, solving the model to generate a control target parameter set, wherein the control target parameter set comprises reward parameters and constraint parameters, S4, checking and correcting the control action sequence according to the control target parameter set and combining the state characteristic sequence through a safety reinforcement learning controller to obtain an execution control action sequence, executing the execution control action sequence, acquiring actual energy consumption and device execution feedback, S5, outputting expected energy consumption through an expected energy consumption model based on action conditions and generating total residual errors by the actual energy consumption, S6, generating an execution control action sequence and device execution feedback by the execution feedback, generating the total residual errors by the execution control action sequence and the device, and determining the total energy consumption corresponding performance window, and the total residual errors are correspondingly degraded by the total energy consumption window and the health constraint window, S7, and the total performance window is correspondingly degraded by the performance window and the performance residual error is calculated when the total performance window is corresponding to the performance window and the performance is correspondingly degraded, the performance is correspondingly degraded by the performance window and the performance of the performance window and the performance is determined and the performance is 8, the performance is correspondingly degraded by the performance and the performance is determined, and updates the control target parameter set for a subsequent control period.
2. The building energy-carbon double control and equipment predictive maintenance linkage energy consumption optimization method according to claim 1, wherein S1 comprises the following steps: Respectively acquiring equipment energy consumption data, equipment operation condition data, equipment control action data, equipment execution feedback data, outdoor environment data, indoor load data, electric power carbon emission factor data, energy consumption control indexes and carbon emission control indexes of a target building in the preset sampling period, and generating corresponding time stamps for various data records; Performing time alignment processing on various data records according to the time stamp, wherein the time alignment processing comprises resampling the data with different sampling frequencies according to the preset sampling period, and merging the resampled various data records according to the same time stamp; carrying out data cleaning treatment on various data records after the time alignment treatment is completed, wherein the data cleaning treatment comprises missing value treatment, abnormal value treatment and numerical value unification treatment, and the numerical value unification treatment comprises unit conversion and dimension unification; And outputting the aligned and cleaned building operation data after the time alignment treatment and the data cleaning treatment, wherein the aligned and cleaned building operation data comprises equipment energy consumption data, equipment operation condition data, equipment control action data, equipment execution feedback data, outdoor environment data, indoor load data, electric carbon emission factor data, energy consumption control indexes and carbon emission control indexes corresponding to the same time stamp.
3. The building energy-carbon double control and equipment predictive maintenance linkage energy consumption optimization method according to claim 1, wherein S2 comprises the following steps: calling the aligned and cleaned building operation data, combining equipment energy consumption data, equipment operation condition data, equipment control action data, equipment execution feedback data, outdoor environment data and indoor load data according to time stamps, and carrying out numerical normalization processing and time serialization processing on the combined data to generate a state feature sequence for control decision; and calculating health characteristics reflecting energy efficiency deviation of the target equipment by using equipment energy consumption data, equipment operation condition data and equipment execution feedback data corresponding to the target equipment in the aligned and cleaned building operation data, and inputting the health characteristics into a health evaluation model so as to output an equipment health risk sequence representing efficiency degradation risk of the target equipment.
4. The building energy-carbon double control and equipment predictive maintenance linkage energy consumption optimization method according to claim 1, wherein the step S3 comprises the following steps: according to the state characteristic sequence and the equipment health risk sequence, combining electric power carbon emission factor data, energy consumption control indexes and carbon emission control indexes, establishing a health constraint multi-objective optimization model taking equipment control braking as a decision variable, wherein the equipment control actions comprise a set value adjusting action and a start-stop scheduling action; In the health constraint multi-objective optimization model, defining an energy consumption target as an optimization target of an energy consumption evaluation result of the equipment control action under the working condition corresponding to the state characteristic sequence, defining a carbon emission target as an optimization target of a carbon emission evaluation result obtained by converting the energy consumption evaluation result based on the electric power carbon emission factor data, and defining a health risk target as an optimization target of a health risk evaluation result corresponding to the equipment health risk sequence; Meanwhile, setting energy consumption constraint corresponding to the energy consumption control index, carbon emission constraint corresponding to the carbon emission control index and equipment operation constraint in the health constraint multi-objective optimization model, wherein the equipment operation constraint comprises amplitude limit, change rate limit and start-stop interval limit of equipment control action; And solving the health constraint multi-objective optimization model to generate a control objective parameter set, wherein the control objective parameter set comprises reward and punishment parameters used for generating a control action sequence by a safety reinforcement learning controller and constraint parameters used for carrying out feasibility verification and correction on the control action sequence.
5. The building energy-carbon double control and equipment predictive maintenance linkage energy consumption optimization method according to claim 1, wherein S4 comprises: Generating a control action sequence through a safety reinforcement learning controller according to a control target parameter set and in combination with a state characteristic sequence, wherein the safety reinforcement learning controller takes reward and punishment parameters corresponding to an energy consumption target, a carbon emission target and a health risk target as return function parameters in the interactive training process, and limits a control action output range by the constraint parameters; Performing feasibility verification on the control action sequence according to the constraint parameters, and when the control actions in the control action sequence do not meet the equipment operation constraint corresponding to the constraint parameters, performing correction processing on the control actions which do not meet the constraint to generate an execution control action sequence meeting the equipment operation constraint; And issuing the execution control action sequence to target equipment for execution, and acquiring an actual energy consumption sequence and an equipment execution feedback sequence corresponding to the execution control action sequence according to the preset sampling period in the execution process.
6. The building energy-carbon double control and equipment predictive maintenance linkage energy consumption optimization method according to claim 1, wherein S5 comprises the following steps: constructing an action condition feature sequence by using the state feature sequence and the execution control action sequence, wherein the action condition feature sequence is used for representing the running condition of the target equipment under the corresponding state feature and the corresponding execution control action; inputting the action condition characteristic sequence into an action condition expected energy consumption model, and outputting an expected energy consumption sequence corresponding to the execution control action sequence; And according to the difference between the actual energy consumption sequence and the expected energy consumption sequence time by time, generating a total residual sequence.
7. The building energy-carbon double control and equipment predictive maintenance linkage energy consumption optimization method according to claim 1, wherein S6 comprises: For each control action in the execution control action sequence, determining an execution feedback value corresponding to the control action time stamp in an equipment execution feedback sequence, and performing deviation calculation on the control action and the execution feedback value to obtain a control deviation sequence for representing the execution deviation degree of the control action; converting the control deviation into an execution residual sequence according to the control deviation sequence, wherein the execution residual sequence is used for representing energy consumption deviation caused by unexpected execution of the control action; And calculating an efficiency residual sequence according to the total residual sequence and the execution residual sequence, wherein the efficiency residual sequence is used for representing energy consumption deviation caused by equipment efficiency degradation under the condition that the control action is executed according to expectations.
8. The building energy-carbon double control and equipment predictive maintenance linkage energy consumption optimization method according to claim 1, wherein S7 comprises: dividing the equipment health risk sequence and the execution residual sequence into a plurality of time windows according to the preset window length; respectively calculating equipment health risk statistics and execution residual error statistics in each time window, and determining the time window in which the equipment health risk statistics are lower than a first threshold value and the execution residual error statistics are smaller than a second threshold value as a health window, so as to generate a health window set; And carrying out parameter updating on the action condition expected energy consumption model by using a state characteristic sequence, an execution control action sequence and an actual energy consumption sequence corresponding to the health window set to generate an updated action condition expected energy consumption model, wherein the parameter updating is only carried out based on data corresponding to the health window set so as to avoid the deviation of the action condition expected energy consumption model caused by the participation of data corresponding to the equipment efficiency degradation in updating.
9. The building energy-carbon double control and equipment predictive maintenance linkage energy consumption optimization method according to claim 1, wherein S8 comprises: Calculating efficiency residual error reference statistics in an efficiency residual error sequence corresponding to a health window set based on the health window set, and determining a degradation judgment threshold according to the efficiency residual error reference statistics; comparing the efficiency residual error sequence with the degradation judgment threshold value, and generating an efficiency degradation index according to a preset calculation rule, wherein the preset calculation rule comprises accumulating and performing time smoothing on the efficiency residual error exceeding the degradation judgment threshold value; calculating additional energy consumption caused by efficiency degradation according to the efficiency residual error sequence, and converting the additional energy consumption into additional carbon emission according to electric power carbon emission factor data; And when the efficiency degradation index meets a preset degradation trigger condition, generating a maintenance instruction, and improving the health risk target weight or tightening equipment operation constraint in the health constraint multi-target optimization model in the next control period to update the control target parameter set, so that the updated control target parameter set is used for the step S4 of the next control period.
10. A building energy-carbon double control and equipment predictive maintenance linkage energy consumption optimization system for executing the building energy-carbon double control and equipment predictive maintenance linkage energy consumption optimization method according to any one of claims 1 to 9, comprising: The system comprises a data preprocessing module, a state and health assessment module, a control target parameter generation module, a safety reinforcement learning control module, a residual analysis and maintenance linkage module, a control target parameter generation module and a control target parameter set, wherein the data preprocessing module is used for acquiring target building operation data and carrying out time alignment and data cleaning, the state and health assessment module is used for generating control state characteristics and outputting equipment health risks, the control target parameter generation module is used for establishing a health constraint multi-target optimization model based on the control state characteristics, the equipment health risks, carbon emission factors and carbon control indexes and solving the health constraint multi-target optimization model to generate a control target parameter set containing punishment parameters and constraint parameters, the safety reinforcement learning control module is used for generating control actions according to the control target parameter set and carrying out issuing after verification and correction according to constraint parameters, and the residual analysis and maintenance linkage module is used for acquiring actual energy consumption and equipment execution feedback, obtaining total residual error, executing residual error and efficiency residual error, screening a health window according to the equipment health risks and executing residual error and updating an expected energy consumption model of action conditions, determining a degradation judgment threshold based on the health window and generating an efficiency degradation index based on the efficiency residual error, calculating additional energy consumption caused by efficiency degradation and corresponding additional carbon emission, and updating the efficiency degradation index, and generating a maintenance instruction and updating the control target parameter set when the efficiency degradation index meets a trigger condition.

Description

Building energy-carbon double control and equipment predictive maintenance linkage energy consumption optimization method and system Technical Field The invention relates to the field of building energy management and equipment operation and maintenance, in particular to a building energy and carbon double control and equipment predictive maintenance linkage energy consumption optimization method and system. Background Under the promotion of national carbon reaching peak and carbon neutralization targets and building energy saving policies, public buildings and park buildings commonly build building automatic control systems, energy management systems and sub-metering platforms, and perform data acquisition and centralized management control on energy utilization equipment such as air conditioner cold source systems, terminal systems, water pumps, fans and the like. In the prior art, data such as equipment energy consumption, operation conditions, control instructions, execution feedback, indoor and outdoor environments, loads and the like are generally collected, and energy consumption monitoring, energy efficiency evaluation and energy saving control are carried out. In the aspect of control optimization, the existing scheme is developed from traditional rule control to set value adjustment and start-stop scheduling based on model predictive control and data driving optimization, and energy consumption and carbon emission collaborative optimization is carried out by combining time-of-use electricity price and electric power carbon emission factors. In the aspect of equipment operation and maintenance, fault detection and diagnosis based on state monitoring and predictive maintenance are gradually applied to identify sensor abnormality, actuator abnormality and dominant faults, and operation and maintenance efficiency and system reliability are improved. The prior art still has the following defects: 1. The energy consumption and carbon emission optimization mainly comprises energy consumption indexes and comfort constraint, equipment health risks are often fractured from a control strategy, and a unified modeling mechanism for incorporating the health risks into objective functions and operation constraint is lacked, so that risks in the equipment efficiency degradation stage are difficult to reflect to the control strategy in time; 2. the fault detection and diagnosis are more good at identifying fault type abnormality, and for the situation of no fault but high energy consumption caused by equipment efficiency degradation, the fault detection and diagnosis are easily confused with load fluctuation, working condition change and energy consumption deviation caused by control action which is not executed according to expectations, and the degradation influence is difficult to accurately identify and quantify in time; 3. the expected energy consumption baseline model is trained for a long time by depending on history data, is easily subjected to strategy change and degradation data to drift in operation, so that a residual error threshold value is unstable, the reliability of degradation judgment and maintenance triggering is further reduced, and closed loop linkage of maintenance decision and follow-up control optimization is difficult to form. Therefore, the building energy-carbon double control and equipment predictive maintenance linkage energy consumption optimizing method and system capable of solving the defects in the prior art are the problems to be solved by the person skilled in the art. Disclosure of Invention The invention aims to provide a building energy-carbon double-control and equipment predictive maintenance linkage energy consumption optimization method, aiming at the problems that equipment efficiency degradation causes no faults but high energy consumption is difficult to identify in time and control optimization and maintenance decisions are difficult to cooperate in the prior art, a health constraint multi-objective optimization model based on state characteristic construction and equipment health risk assessment of time alignment and cleaning of building operation data is provided, energy consumption, carbon emission and health risks are established to generate reward and punishment parameters and constraint parameters, a safety reinforcement learning controller is adopted to output control actions after constraint verification and correction and execute acquisition feedback, a total residual error is formed based on the action condition expected energy consumption model and combined with execution feedback decomposition to obtain an execution residual error and an efficiency residual error, a baseline model is self-updated through health window screening, a degradation judgment threshold value is further determined, an efficiency degradation index, additional energy consumption and corresponding additional carbon emission are calculated, a maintenance instruction is generated under