CN-121977249-A - Smart city-oriented abnormal energy consumption identification and operation and maintenance alarm method for heating and ventilation system

CN121977249ACN 121977249 ACN121977249 ACN 121977249ACN-121977249-A

Abstract

The invention relates to the technical field of heating ventilation abnormal energy consumption identification, in particular to a heating ventilation system abnormal energy consumption identification and operation and maintenance alarming method for a smart city. The method comprises the steps of collecting real-time operation data and external environment disturbance data of a target building heating and ventilation system, obtaining equivalent weather temperature through thermal resistance hysteresis correction, inputting the equivalent weather temperature into a lightweight gradient lifting tree prediction theory energy consumption with thermodynamic physical lower limit penalty, calculating total energy consumption residual errors, carrying out orthogonal decomposition, eliminating colinear disturbance of weather and passenger flow, extracting pure equipment operation decline components, finally matching an operation and maintenance expert rule base, and generating and pushing an alarm work order containing fault space positions and processing suggestions. The invention eliminates the prediction distortion of the algorithm black box, thoroughly peels off false alarm caused by external disturbance, and realizes automatic closed loop from macroscopic anomaly perception to microscopic precise obstacle removal.

Inventors

YU JIANGLONG
WANG HAO

Assignees

中电系统建设工程有限公司

Dates

Publication Date: 20260505
Application Date: 20260409

Claims (9)

1. The heating and ventilation system abnormal energy consumption identification and operation and maintenance alarming method for the smart city is characterized by comprising the following steps of: S1, collecting real-time operation data of a target building heating and ventilation system, and external environment disturbance data comprising microclimate data and building real-time people stream density data, and carrying out time sequence correction on the microclimate data to obtain equivalent outdoor meteorological temperature; Preprocessing the real-time operation data, the equivalent outdoor meteorological temperature and the building real-time people stream density data, respectively extracting equipment state characteristics and external load demand characteristics, carrying out time sequence caching on the equipment state characteristics, and constructing a historical operation characteristic set according to the equipment state characteristics; S2, acquiring external load demand characteristics at the current moment in real time, inputting the external load demand characteristics into a pre-trained lightweight gradient lifting tree model, and calculating a theoretical expected energy consumption value for maintaining heat balance of a heating and ventilation system; S3, acquiring an actual total energy consumption value of the heating ventilation system at the current moment, calculating a total energy consumption residual error between the actual total energy consumption value and the theoretical expected energy consumption value, performing orthogonal decomposition on the total energy consumption residual error, and directly extracting an equipment operation degradation component representing the self performance degradation of equipment; If and only if the continuous set time window of the equipment operation decay component exceeds a preset equipment health tolerance threshold, judging to trigger an abnormal energy consumption event of a real equipment level; S4, after triggering the real equipment-level abnormal energy consumption event, extracting the equipment state characteristics in a trigger event time window; calculating the characteristic contribution degree of each equipment state characteristic to the equipment operation decline component by adopting a preset degradation characteristic classification regression tree model, and identifying a core degradation parameter causing sudden increase of system energy consumption according to the characteristic contribution degree; S5, matching a preset operation and maintenance expert rule base according to the identified core degradation parameters, and determining a corresponding specific fault type; And generating an operation and maintenance alarm work order containing the fault position, the specific fault type and the processing suggestion, and pushing the operation and maintenance alarm work order to the intelligent city operation and maintenance terminal.
2. The smart city-oriented abnormal energy consumption identification and operation and maintenance alarm method for a heating and ventilation system according to claim 1, wherein the real-time operation data at least comprises real-time power of a chiller, chilled water supply/return water temperature, chilled water supply/return water pressure, chilled water supply/return water temperature, cooling pump operation frequency, chilled pump operation frequency and cooling tower fan operation frequency; the microclimate data at least comprises outdoor dry bulb temperature, solar radiation intensity and real-time outdoor wind speed; the building real-time people stream density data includes at least a real-time passenger stream density index.
3. The smart city-oriented abnormal energy consumption identification and operation and maintenance alarm method for a heating and ventilation system of claim 1, wherein the method is characterized by performing time sequence correction on the microclimate data, and comprises the following specific steps: obtaining current time from microclimate data Outdoor dry bulb temperature of (2) And intensity of solar radiation And introducing the solar radiation absorption coefficient of the target building envelope Heat exchange coefficient of outdoor surface Calculating the outdoor comprehensive temperature at the current moment ; Introducing a thermal resistance and hysteresis time constant of the enclosure structure based on the outdoor comprehensive temperature at the current moment and the historical moment Data sampling period Performing dynamic weighting calculation of phase lag and amplitude attenuation, and outputting the equivalent outdoor weather temperature 。
4. The smart city-oriented abnormal energy consumption identification and operation and maintenance alarm method for a heating and ventilation system of claim 1, wherein the steps of constructing and training the lightweight gradient lifting tree model are as follows: Constructing a lightweight gradient lifting tree model based on an addition model and a forward distribution algorithm, wherein the lightweight gradient lifting tree model is formed by Accumulating the classification regression tree containing the tree topology structure and the leaf node weight; sample data in a normal operation time period of the heating ventilation system are extracted from the historical operation characteristic set, the external load demand characteristic is taken as an input variable, and the synchronous actual total energy consumption value is taken as a real label value; Constructing a joint loss function for model iterative training, wherein the joint loss function consists of a data fitting error term and a thermodynamic energy conservation penalty term between a real label value and a predicted energy consumption value of a current model iteration; In the training process of the lightweight gradient lifting tree model, calculating a first derivative and a second derivative of the joint loss function about a predicted energy consumption value, and substituting the first derivative and the second derivative into split gain calculation of leaf nodes of a classification regression tree; Obtaining the pre-trained lightweight gradient-lifted tree model by minimizing the joint loss function; And in an online prediction stage, acquiring external load demand characteristics at the current moment in real time, inputting the external load demand characteristics into the pre-trained lightweight gradient lifting tree model, and outputting a theoretical expected energy consumption value.
5. The smart city-oriented abnormal energy consumption identification and operation and maintenance alarm method for a heating and ventilation system of claim 4, wherein the construction steps of the thermodynamic energy conservation penalty term specifically include: extracting the equivalent outdoor meteorological temperature and the building real-time people stream density data in the external load demand characteristics; Introducing a comprehensive heat transfer coefficient of a target building enclosure structure and single average heat dissipation capacity, and calculating dynamic cold and hot load of the target building under the current working condition by combining the equivalent outdoor meteorological temperature and the real-time people stream density data of the building; Calculating a theoretical minimum energy consumption physical lower limit for maintaining thermodynamic equilibrium of the heating ventilation system based on the theoretical maximum energy efficiency ratio of the dynamic cold and hot load and the heating ventilation system; and when the predicted energy consumption value of the current iteration of the lightweight gradient lifting tree model is lower than the theoretical minimum energy consumption physical lower limit, activating the thermodynamic energy conservation penalty term, and applying penalty calculation to the predicted energy consumption value.
6. The smart city-oriented abnormal energy consumption identification and operation and maintenance alarm method for a heating and ventilation system according to claim 1, wherein in S3, an equipment operation degradation component representing the performance degradation of equipment is extracted, and the specific steps involved are as follows: obtaining the actual total energy consumption value and the theoretical expected energy consumption value at the current moment, and performing difference calculation to obtain a total energy consumption residual error; Based on a sliding analysis time window with a set length, carrying out time sequence caching on the total energy consumption residual error obtained by continuous calculation, and splicing the total energy consumption residual error in the sliding analysis time window into a total energy consumption residual error vector according to time sequence; Carrying out standard orthogonalization processing on the microclimate disturbance base vector and the induced abortion disturbance base vector to construct a multidimensional disturbance feature space; performing projection calculation on the total energy consumption residual vector to the multidimensional disturbance feature space to respectively obtain a microclimate disturbance projection vector mapped to the microclimate disturbance base vector direction and a people stream surge disturbance projection vector mapped to the people stream surge disturbance direction after the collinearity is eliminated; Filtering and stripping the microclimate disturbance projection vector and the induced disturbance projection vector from the total energy consumption residual vector by adopting vector subtraction to obtain an equipment degradation residual vector; and extracting a last element scalar corresponding to the current moment in the equipment fading residual vector to serve as the equipment running fading component at the current moment.
7. The smart city-oriented abnormal energy consumption identification and operation and maintenance alarm method for a heating and ventilation system according to claim 1, wherein the identification of the core degradation parameters causing sudden increase of the energy consumption of the system comprises the following specific steps: After triggering the real equipment-level abnormal energy consumption event, acquiring the equipment state characteristics synchronously cached in the triggering event time window for triggering the event as an independent variable input set; Invoking a pre-trained degradation characteristic classification regression tree model, wherein the degradation characteristic classification regression tree model takes the equipment state characteristics of a historical operation stage as input variables, and takes the corresponding equipment operation decay components as nonlinear mapping networks obtained by fitting label training; Inputting the independent variable input set into the degradation characteristic classification regression tree model, introducing a SHAP interpretability algorithm based on a game theory, calculating marginal contribution values of all the equipment state characteristics in the independent variable input set to the currently output equipment operation degradation components, and aggregating to generate a global characteristic contribution degree; And ordering the equipment state features in a descending order according to the global feature contribution degree, and taking the equipment state features meeting a preset contribution degree threshold as the core degradation parameters, thereby generating an operation and maintenance alarm work order aiming at the target building heating and ventilation system.
8. The smart city-oriented abnormal energy consumption identification and operation and maintenance alarm method for a heating and ventilation system of claim 1, wherein in S5, the specific step of generating an operation and maintenance alarm work order comprises: constructing a preset operation and maintenance expert rule base, wherein the operation and maintenance expert rule base comprises a multidimensional mapping matrix; The mapping key value of the multi-dimensional mapping matrix at least comprises a physical label of a core degradation parameter, a characteristic value deviation direction and an abnormal state of a combined association parameter, wherein the mapping target value of the multi-dimensional mapping matrix is a corresponding specific fault type and processing suggestion; Extracting the physical label of the core degradation parameter identified in the step S4, calculating the characteristic average value of the core degradation parameter in a set time window for triggering the real equipment-level abnormal energy consumption event, comparing the characteristic average value with a reference expected value of the core degradation parameter in a historical absolute healthy operation stage, and determining a corresponding characteristic value deviation direction; Combining the real-time deviation state of the combination association parameters of the physical label and the characteristic value deviation direction, which belong to the same physical subsystem as the physical label, and jointly inputting the combination association parameters into the operation and maintenance expert rule base to carry out multidimensional mapping matching, and determining the specific fault type and the processing suggestion corresponding to the heating and ventilation system of the current target building; Extracting a unique equipment identification code directly bound with the core degradation parameters, and addressing a space address in an equipment topology network to obtain corresponding physical equipment node space coordinates as fault positions; And packaging the fault position, the specific fault type, the processing suggestion and the global feature contribution degree of the core degradation parameter into an operation and maintenance alarm work order in a structured message form, and pushing the operation and maintenance work order to the intelligent city operation and maintenance terminal.
9. The smart city-oriented abnormal energy consumption identification and operation and maintenance alarm method for a heating and ventilation system of claim 8, wherein the specific mapping logic of the multidimensional mapping matrix stored in the preset operation and maintenance expert rule base at least comprises: When the physical label of the extracted core degradation parameter contains cooling water approximation degree and the corresponding characteristic value deviation direction is forward higher, and the characteristic value of the combined related cooling water pump inlet-outlet pressure difference parameter is abnormally increased, the specific fault type output by the multidimensional mapping matrix is cooling side heat exchange abnormality, and the corresponding bottom layer physical fault source is mapped to be positioned as severe scaling of cooling tower heat radiation filler or blockage of a cooling water pipe network; When the physical label of the extracted core degradation parameter contains a chilled water supply and return water temperature difference, the corresponding characteristic value deviation direction is negative and low, and the characteristic value deviation direction of the combined related water chilling unit operating frequency parameter is positive and high, the specific fault type output by the multidimensional mapping matrix is a large-flow small-temperature difference syndrome, and the corresponding bottom layer physical fault source is mapped and positioned as severe dust accumulation of a surface cooler of the tail-end air processing unit or the water side two-way regulating valve is out of order.

Description

Smart city-oriented abnormal energy consumption identification and operation and maintenance alarm method for heating and ventilation system Technical Field The invention relates to the technical field of heating ventilation abnormal energy consumption identification, in particular to a heating ventilation system abnormal energy consumption identification and operation and maintenance alarming method for a smart city. Background In a smart energy management and control system of a smart city building group, a Heating Ventilation Air Conditioning (HVAC) system is used as a maximum flexible energy consumption carrier, and the running state of the HVAC system directly determines the energy saving and emission reduction effects of a large commercial complex and a public building. Under actual physical working conditions, the transient total energy consumption of the heating ventilation system is surge, which is essentially macroscopic manifestation of deep coupling interweaving of external dynamic load disturbance (such as urban microclimate drafts and transient passenger flow surges) and internal physical equipment degradation (such as severe scaling of heat exchangers and abrasion of hydraulic pipe networks), and currently, a main-stream building automatic control system (BMS) and an existing energy consumption monitoring and early warning algorithm mostly depend on a static energy efficiency threshold or a simple surface data driving model. However, these prior art techniques, facing complex and varied urban building microenvironments, commonly suffer from the following serious technical bottlenecks in terms of stripping ambient white noise, locating true equipment-level anomalies: In actual operation, when extremely high temperature is encountered outdoors or large passenger flow is suddenly generated in a market, the energy consumption of the system naturally rises, but the prior art only needs to trigger an alarm by comparing the actual energy consumption with the simple residual error of the predicted energy consumption, the false high energy consumption induced by microclimate sudden change and people flow surge cannot be stripped on the bottom mathematical logic, and the interference of multiple collinearity characteristics leads to massive false alarms which are often received by operation and maintenance terminals of the smart city in the load peak period, so that the trust degree of operation and maintenance personnel on the system is greatly reduced. Disclosure of Invention The invention aims to provide a heating and ventilation system abnormal energy consumption identification and operation and maintenance alarm method for a smart city, which aims to solve the problems that a smart city operation and maintenance terminal always receives massive false alarms in a load peak period due to the interference of multiple collinearity characteristics provided in the background technology, so that the trust degree of operation and maintenance personnel on the system is greatly reduced. In order to achieve the above purpose, the present invention provides a method for identifying abnormal energy consumption and alarming operation and maintenance of a heating and ventilation system for a smart city, comprising the following steps: S1, collecting real-time operation data of a target building heating and ventilation system, and external environment disturbance data comprising microclimate data and building real-time people stream density data, and carrying out time sequence correction on the microclimate data to obtain equivalent outdoor meteorological temperature; Preprocessing the real-time operation data, the equivalent outdoor meteorological temperature and the building real-time people stream density data, respectively extracting equipment state characteristics and external load demand characteristics, carrying out time sequence caching on the equipment state characteristics, and constructing a historical operation characteristic set according to the equipment state characteristics; S2, acquiring external load demand characteristics at the current moment in real time, inputting the external load demand characteristics into a pre-trained lightweight gradient lifting tree model, and calculating a theoretical expected energy consumption value for maintaining heat balance of a heating and ventilation system; S3, acquiring an actual total energy consumption value of the heating ventilation system at the current moment, calculating a total energy consumption residual error between the actual total energy consumption value and the theoretical expected energy consumption value, performing orthogonal decomposition on the total energy consumption residual error, and directly extracting an equipment operation degradation component representing the self performance degradation of equipment; If and only if the continuous set time window of the equipment operation decay component exceeds a preset equipment health toler