Intelligent detection method for energy-saving efficiency of heating ventilation air conditioner

Abstract

The invention discloses an intelligent detection method for the energy-saving efficiency of a heating ventilation air conditioner, and relates to the technical field of intelligent detection, and the method comprises the steps: obtaining first comprehensive data of heating ventilation air conditioner equipment, and carrying out the preprocessing of the first comprehensive data; a state transition model is constructed based on the preprocessed first comprehensive data, and the current energy efficiency state of the heating ventilation air conditioning equipment is judged in real time through the state transition model; triggering maintenance instruction generation based on a judgment result, performing clustering adjustment on the state boundary again, and generating a preventive maintenance instruction; and optimizing state transition model parameters based on the maintained first comprehensive data. By avoiding static threshold misjudgment, the full-working-condition energy efficiency characteristics of the equipment are truly reflected; a time decay factor is introduced to dynamically update the state transition matrix, so that the state transition model fits the transition rule change caused by equipment aging and maintenance effects; a grading strategy of efficient recording, medium-efficiency automatic adjustment and low-efficiency preventive maintenance is designed, and excessive maintenance and energy waste are reduced.

Description

Technical Field

[0001] This invention relates to the field of intelligent detection technology, and in particular to an intelligent detection method for the energy efficiency of HVAC systems. Background Technology

[0002] As a core energy-consuming device in buildings, the energy efficiency monitoring of HVAC systems is crucial for building energy conservation. However, existing technologies have significant shortcomings: First, there is a lack of comprehensive collection and correlation analysis of "equipment operating parameters, environmental parameters, and energy efficiency indicators," and the data is often affected by noise, inconsistent dimensions, and asynchronous time, resulting in insufficient reliability. Second, energy efficiency status determination relies on static fixed thresholds, failing to consider equipment aging, environmental fluctuations, and load changes, which easily leads to misjudgments. Third, the initial migration probability matrix of the state transition model is fixed, lacking a dynamic update mechanism, and cannot reflect changes in migration patterns caused by equipment aging, maintenance effectiveness, and changes in operating conditions. Fourth, maintenance strategies are mostly post-fault repair or periodic maintenance, lacking preventive maintenance based on real-time energy efficiency status, and there is no automatic parameter adjustment under medium efficiency status, resulting in lagging energy efficiency optimization. Fifth, the failure to utilize the first comprehensive data to optimize model parameters after maintenance leads to a break in the "judgment-maintenance" closed loop. There is an urgent need for an intelligent detection method for HVAC energy efficiency that can achieve accurate preprocessing of multi-dimensional data, dynamic energy efficiency status determination, adaptive updating of the migration model, and a closed loop of preventive maintenance and parameter optimization to solve existing technical problems and improve the accuracy and energy-saving potential of energy efficiency monitoring.

[0003] However, current common solutions have many drawbacks, including: existing HVAC energy efficiency testing technologies have significant shortcomings: energy efficiency status determination relies on fixed static thresholds, does not consider equipment aging, environmental fluctuations, and load changes, is prone to misjudgment, and cannot reflect the true energy efficiency of the equipment; the state transition model lacks a dynamic update mechanism, and the initial transition probability matrix is ​​calculated based on fixed historical data, which is out of sync with the actual transition patterns brought about by equipment aging and maintenance effects, resulting in large prediction deviations; after maintenance, the first comprehensive data is not used to optimize the model parameters, resulting in a broken closed loop and subsequent judgment deviations. Summary of the Invention

[0004] The purpose of this section is to outline some aspects of embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.

[0005] In view of the problems existing in the current intelligent detection method for HVAC energy efficiency, the present invention is proposed.

[0006] Therefore, the purpose of this invention is to provide an intelligent detection method for HVAC energy efficiency, which is applicable to solving the problems of existing HVAC energy efficiency detection technologies that rely on fixed static thresholds for energy efficiency status determination, do not consider equipment aging, environmental fluctuations and load changes, are prone to misjudgment, and cannot reflect the true energy efficiency of the equipment; the state transition model has no dynamic update mechanism, and the initial transition probability matrix is ​​calculated based on fixed historical data, which is out of touch with the actual transition patterns brought about by equipment aging and maintenance effects, resulting in large prediction deviations.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, embodiments of the present invention provide an intelligent detection method for HVAC energy efficiency, comprising: acquiring first comprehensive data of HVAC equipment and preprocessing the first comprehensive data; constructing a state transition model based on the preprocessed first comprehensive data, and judging the current energy efficiency status of the HVAC equipment in real time through the state transition model; triggering maintenance command generation based on the judgment result, re-clustering and adjusting the state boundaries and generating preventive maintenance commands; optimizing the parameters of the state transition model based on the maintained first comprehensive data to complete the iterative update of the state transition model, thereby realizing continuous intelligent detection of HVAC energy efficiency.

[0008] As a preferred embodiment of the intelligent detection method for HVAC energy efficiency described in this invention, the first comprehensive data includes equipment operating parameters, environmental parameters, and energy efficiency indicators; the equipment operating parameters include compressor parameters, refrigerant circulation parameters, water system parameters, and air system parameters; the environmental parameters include indoor environmental parameters and outdoor environmental parameters; the energy efficiency indicators include real-time coefficient of performance (COP), load fluctuation coefficient, and ambient temperature fluctuation; the real-time COP is acquired in real time by a power transmitter and flow sensor installed in the main circuit of the equipment; the load fluctuation coefficient is obtained by acquiring the actual load of the equipment in real time through the load monitoring module built into the air conditioning controller; the equipment operating parameters are acquired by a pressure sensor, a current sensor, and an electromagnetic flow meter; the environmental parameters are acquired by a temperature and humidity sensor, a CO2 sensor, and a pressure sensor.

[0009] As a preferred embodiment of the intelligent detection method for HVAC energy efficiency described in this invention, the specific content of constructing the state transition model is as follows: Cluster analysis is performed based on the preprocessed first comprehensive data to divide it into three basic energy efficiency states: high efficiency, medium efficiency, and low efficiency; the frequency of occurrence of transition paths between the basic energy efficiency states is statistically analyzed to calculate the initial transition probability, forming an initial transition probability matrix and generating an initial state transition model; the initial transition probability matrix is ​​dynamically updated by introducing a time decay factor to form a final state transition model with self-updating capability; the preprocessed first comprehensive data is input into the state transition model, and the Euclidean distance between it and the cluster center vectors of the high efficiency, medium efficiency, and low efficiency states is calculated to obtain the distance values. , and This allows us to determine the current energy efficiency status of the HVAC equipment.

[0010] As a preferred embodiment of the intelligent detection method for energy efficiency of HVAC systems according to the present invention, the specific content of determining the current energy efficiency status of the HVAC equipment is as follows: If < and < If it is, then it is determined to be in a highly efficient state. < and < If the current state point is closest to the cluster center of the efficient state cluster in the high-dimensional feature space, it is determined to be an efficient state; if < and < If the current state point is closest to the cluster center of the intermediate-efficiency state cluster, it is determined to be an intermediate-efficiency state; if... < and < If the current state point is closest to the cluster center of the inefficient state cluster, it is determined to be an inefficient state.

[0011] As a preferred embodiment of the intelligent detection method for HVAC energy efficiency described in this invention, the following is a specific implementation of the method: When the system is determined to be in a high-efficiency state, no maintenance command is generated; instead, the current state data is stored in a historical database for subsequent re-clustering adjustments of state boundaries, and the duration of the high-efficiency state is recorded. If the duration exceeds a set duration, a stable equipment operation report is generated. When the system is determined to be in a medium-efficiency state, a parameter adjustment command is generated to automatically optimize and adjust the equipment operating parameters, including step-wise adjustment of compressor frequency, water pump speed, and fan operating parameters, and energy efficiency indicators are monitored after each adjustment. When the system is determined to be in an inefficient state, a preventative maintenance command is immediately generated to perform corresponding maintenance operations targeting the main causes of inefficiency. If critical parameters exceed limits, an emergency shutdown warning is added.

[0012] As a preferred embodiment of the intelligent detection method for HVAC energy efficiency described in this invention, the following steps are taken: The state boundaries are re-clustered and adjusted to generate preventative maintenance instructions. Specifically, this involves: collecting pre-processed first comprehensive data and corresponding energy efficiency status determination results; removing invalid data and retaining valid data from the stable operation phase; re-clustering the valid data using a density clustering algorithm to update the cluster center vectors for high-efficiency, medium-efficiency, and low-efficiency states; recalculating the first to sixth thresholds based on the updated cluster center vectors; and synchronizing the updated thresholds and cluster center vectors to the state transition model as a new benchmark for subsequent energy efficiency status determination.

[0013] As a preferred embodiment of the intelligent detection method for HVAC energy efficiency described in this invention, the specific content of optimizing model parameters based on the first comprehensive data after maintenance is as follows: collecting the first comprehensive data after maintenance and the energy efficiency status determination results, and performing preprocessing; merging the data after maintenance with data of the same type of status, and re-clustering to correct the cluster center vectors of high-efficiency, medium-efficiency, and low-efficiency statuses; updating the number of status transitions based on the data after maintenance and recalculating the transition probability; adjusting the time decay factor according to the change in the duration of the status after maintenance; verifying the optimized parameters, and if the status determination matching rate meets the requirements, then using it as a new benchmark for the model.

[0014] Secondly, to further solve the above-mentioned technical problems, the present invention provides an intelligent detection system for energy efficiency of HVAC systems, comprising: a data acquisition module for acquiring first comprehensive data of HVAC equipment and preprocessing the first comprehensive data; a model building module for constructing a state transition model based on the preprocessed first comprehensive data and judging the current energy efficiency status of the HVAC equipment in real time through the state transition model; an instruction generation module for triggering maintenance instruction generation based on the judgment result, re-clustering and adjusting the state boundaries and generating preventive maintenance instructions; and a parameter optimization module for optimizing the parameters of the state transition model based on the first comprehensive data after maintenance.

[0015] Thirdly, embodiments of the present invention provide a computer device, including a memory and a processor, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements any step of the intelligent detection method for HVAC energy efficiency as described in the first aspect of the present invention.

[0016] Fourthly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the intelligent detection method for HVAC energy efficiency as described in the first aspect of the present invention.

[0017] The beneficial effects of this invention are as follows: This invention determines energy efficiency status through "cluster analysis + Euclidean distance calculation + dynamic threshold," avoiding misjudgment by static thresholds and truly reflecting the energy efficiency characteristics of equipment under all operating conditions; it introduces a time decay factor to dynamically update the state transition matrix, making the state transition model conform to the changes in migration patterns brought about by equipment aging and maintenance effects; it designs a hierarchical strategy of "high-efficiency recording, medium-efficiency automatic adjustment, and low-efficiency preventive maintenance" to reduce over-maintenance and energy waste, and adds emergency warnings when key parameters exceed limits; it adapts to the changes in characteristics throughout the entire life cycle of equipment by periodically re-clustering and updating state boundaries and thresholds; it uses post-maintenance data to correct cluster center vectors, update migration probabilities, and adjust time decay factors to form a "maintenance-optimization" closed loop, ensuring long-term judgment accuracy; and the supporting modular system can be implemented through conventional computer equipment, without the need for customized hardware, resulting in low implementation and promotion costs. Overall, it achieves a breakthrough in HVAC energy-saving detection from static and extensive to dynamic and precise, and from passive maintenance to proactive prevention, meeting the needs of building energy conservation and possessing broad industrial application value. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a flowchart illustrating the implementation of the present invention in Example 1.

[0019] Figure 2 This is a dynamic adjustment diagram of the state transition model in Example 1. Detailed Implementation

[0020] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0021] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0023] Example 1 Reference Figure 1 and Figure 2 This is the first embodiment of the present invention, which provides an intelligent detection method for HVAC energy efficiency, including the following steps: S1: Obtain the first comprehensive data of the HVAC equipment and preprocess the first comprehensive data.

[0024] Furthermore, the first comprehensive data includes equipment operating parameters, environmental parameters, and energy efficiency indicators.

[0025] Furthermore, the equipment operating parameters include compressor parameters, refrigerant circulation parameters, water system parameters, and air system parameters; environmental parameters include indoor and outdoor environments; and energy efficiency indicators include real-time coefficient of performance (COP), load fluctuation coefficient, and ambient temperature fluctuation.

[0026] Specifically, the real-time energy efficiency score of the equipment in the first comprehensive data is obtained by real-time data collection from the power transmitter and flow sensor installed in the main circuit of the equipment; the load fluctuation coefficient in the first comprehensive data is obtained by real-time reading of the actual load of the equipment through the load monitoring module built into the air conditioner controller.

[0027] Specifically, the equipment operating parameters in the first comprehensive data are obtained through pressure sensors, current sensors, and electromagnetic flow meters; the environmental parameters are obtained through temperature and humidity sensors, CO2 sensors, and air pressure sensors.

[0028] It should be noted that the preprocessing of the first aggregated data includes noise filtering, outlier correction, dimensional normalization, data alignment, and time synchronization.

[0029] Specifically, the method for obtaining the real-time performance coefficient COP is as follows: ; In the formula, For performance coefficients; The average chilled water flow rate within the time window; The density of water; This is the specific heat capacity of water; The average supply and return water temperature difference within the time window; Real-time pressure of the refrigerant; Rated refrigerant pressure; This is the refrigerant pressure correction factor; , Energy consumption weights for water pumps and fans; , , This refers to the real-time power of the compressor, water pump, and fan. Real-time indoor and outdoor temperature difference; For design operating temperature difference; This is the correction factor for ambient temperature difference.

[0030] The specific formula for the load fluctuation coefficient is as follows: ; In the formula, This is the load fluctuation coefficient; The calculation time window is used to control the time range of gradient calculation, balancing real-time performance and anti-interference capability; Using the current moment as the time reference for integration and differentiation, we ensure the time alignment of the data. for The actual load of the equipment at any given time is collected in real time by the load monitoring module built into the HVAC controller, reflecting the current actual output of the equipment. This is the core basic data for calculating the load fluctuation coefficient. The rated load is the parameter listed on the equipment nameplate. for The standard deviation of the load data noise at any given time is used to preprocess the collected load data, characterize the noise level of the load data, and is used for subsequent noise filtering. The rated noise level refers to the standard noise parameters at the time the equipment leaves the factory. for The compressor operating frequency at any given time is collected in real time by compressor parameter sensors, reflecting the compressor's operating intensity and indirectly related to changes in the load fluctuation coefficient; The compressor's rated frequency is used to correct for the compressor's impact on load fluctuation coefficients. for The fluctuation of indoor and outdoor temperature difference at any time is obtained by collecting the indoor and outdoor temperature difference values ​​through temperature and humidity sensors, reflecting the disturbance of environmental changes on equipment load; This serves as the indoor-outdoor temperature difference benchmark under design conditions, and is used as the equipment design parameter to correct for the influence of the environment on the load fluctuation coefficient.

[0031] The specific formula for ambient temperature fluctuation is as follows: ; In the formula, The intensity of ambient temperature fluctuation reflects the cumulative degree of fluctuation in the indoor and outdoor temperature difference from the design state over a certain period of time. The calculation time window is the time interval for statistically analyzing temperature fluctuations. The actual indoor and outdoor temperature difference at any given time is collected in real time by indoor and outdoor temperature and humidity sensors. It reflects the real-time environmental temperature difference and is the basic raw data for temperature fluctuation calculation, directly reflecting the dynamic changes in environmental temperature. The indoor and outdoor temperature difference under design conditions is set through the equipment technical manual or design parameters to quantify the degree to which the actual condition deviates from the design condition; The standard deviation of the temperature data noise at a given moment reflects random interference in the temperature measurement at that moment, quantifies the data noise, and is used for subsequent noise filtering and correction to reduce the interference of noise on fluctuation calculations. The rated temperature noise level, i.e., the maximum allowable noise standard deviation when the equipment is operating normally, is set through equipment parameters or industry standards; To measure outdoor wind speed and reflect the intensity of outdoor air flow, data is collected in real time by an outdoor wind speed sensor. The rated wind speed is the standard wind speed referenced during equipment design, set according to the historical average wind speed of the installation environment. The wind speed influence coefficient is used to adjust the correction strength of wind speed for temperature fluctuations. It is determined by calibration using historical environmental data and equipment energy efficiency data, and controls the weight of the wind speed correction term.

[0032] For example, in a centralized air conditioning system of a large shopping mall, pressure sensors collect compressor discharge pressure and refrigerant pipeline pressure (refrigerant circulation parameters); current sensors collect real-time current of the compressor, water pump, and fan (core equipment operating parameters); electromagnetic flow meters collect chilled water flow (water system parameters); indoor and outdoor temperature and humidity sensors collect real-time temperature and humidity; and CO2 sensors collect CO2 concentration in the mall's public areas (environmental parameters). The air conditioning controller's built-in load monitoring module reads the actual load of the equipment (used to calculate the load fluctuation coefficient). Simultaneously, the power transmitter in the main circuit of the equipment collects the total power, and combined with the flow data, calculates the real-time coefficient of performance (COP) (energy efficiency index) according to a given formula. For the collected first comprehensive data, noise data generated by the temperature and humidity sensors due to instantaneous electromagnetic interference is filtered out, and occasional out-of-range abnormal values ​​from the refrigerant pressure sensor are corrected. The COP (dimensionless) and chilled water flow (m³) are then calculated. 3 The dimensions of the load fluctuation coefficient and other quantities are uniformly normalized to the [0,1] interval, and the data acquisition time of all sensors is aligned to the same second-level timestamp to complete the preprocessing to ensure data reliability.

[0033] S2: Construct a state transition model based on the preprocessed first comprehensive data, and use the state transition model to determine the current energy efficiency status of the HVAC equipment in real time.

[0034] Preferably, the specific content of constructing the state transition model based on the preprocessed first comprehensive data is as follows: Cluster analysis was performed on the preprocessed first comprehensive data to classify the basic energy efficiency status.

[0035] Specifically, the basic energy efficiency status is determined based on preprocessed historical first comprehensive data.

[0036] It should be noted that the basic energy efficiency status includes: High-efficiency state: The real-time performance coefficient COP is greater than or equal to the first threshold, the load fluctuation coefficient is less than or equal to the second threshold, the ambient temperature fluctuation is less than or equal to the third threshold, the equipment operating parameters and environmental parameters are close to the design conditions, and the state transition model determines it to be a high-efficiency state by calculating that the Euclidean distance between the current data and the cluster center vector of the high-efficiency state is the smallest.

[0037] Medium-efficiency state: When the real-time performance coefficient COP is greater than the fourth threshold and less than the first threshold, the load fluctuation coefficient is greater than the second threshold and less than the fifth threshold, and the ambient temperature fluctuation is greater than the third threshold and less than the sixth threshold, it indicates that some operating parameters or environmental parameters deviate from the design conditions, but do not exceed the controllable range. The state transition model determines that the Euclidean distance between the medium-efficiency state cluster center vector and the medium-efficiency state is the smallest, which is the medium-efficiency state.

[0038] Inefficient state: The real-time performance coefficient COP is less than or equal to the fourth threshold, the load fluctuation coefficient is greater than or equal to the fifth threshold, the ambient temperature fluctuation is greater than the sixth threshold, multiple operating parameters and environmental parameters deviate from the design conditions, and the state transition model calculation shows that the current data has the smallest Euclidean distance to the cluster center vector of the inefficient state, so it is judged as an inefficient state.

[0039] It should be noted that the core purpose of classifying basic energy efficiency states is to learn patterns, summarize characteristics, and establish criteria from historical data, and to classify the states into three categories: "high efficiency," "medium efficiency," and "low efficiency" in an unsupervised manner. The output of this process is the standard for state determination.

[0040] Furthermore, the specific method for obtaining the first threshold is to extract the boundary features of the high-efficiency state cluster by clustering analysis of the COP values ​​in the historical operating data of the equipment. The specific steps are as follows: collect historical COP data of the equipment during normal operation and optimal energy efficiency (such as the initial operation of new equipment or the stable period after maintenance); use density clustering algorithm to divide the high-efficiency state data clusters; take the minimum boundary value of all COP values ​​in the cluster as the first threshold, which represents the minimum COP level required to enter the high-efficiency state.

[0041] The second threshold is obtained by defining the stability of load changes under high-efficiency conditions: extract the load fluctuation coefficient data corresponding to historical high-efficiency conditions; calculate the upper limit statistical value of the dataset (such as the 95th percentile) to reflect the maximum load fluctuation intensity allowed under high-efficiency conditions; use this statistical value as the second threshold, and if it is exceeded, it is determined that the system has left the high-efficiency state.

[0042] The specific method for obtaining the third threshold is to reflect the environmental disturbance limit that can be tolerated during efficient operation: filter environmental temperature fluctuation data under historical efficient conditions; identify outlier boundaries of efficient clusters through box plot analysis; use this boundary value as the third threshold, exceeding which indicates that environmental disturbances have affected energy efficiency.

[0043] The specific method for obtaining the fourth threshold is to distinguish the critical point between medium-efficiency and inefficient states: cluster historical data to generate medium-efficiency state clusters and inefficient state clusters; calculate the clustering interval between the two clusters in the COP dimension (e.g., determine the optimal separation point through the silhouette coefficient); take the COP value corresponding to the interval point as the fourth threshold, and if it is lower than this value, it tends to be inefficient.

[0044] The fifth threshold is obtained by identifying the critical condition of drastic load fluctuation: extracting load fluctuation coefficient data under historical inefficient conditions; analyzing its distribution characteristics, and taking the minimum load fluctuation coefficient of the inefficient cluster as the fifth threshold; when the real-time load fluctuation coefficient is continuously higher than this value, an inefficiency warning is triggered.

[0045] The sixth threshold is obtained by defining the starting point of energy efficiency degradation caused by environmental disturbances: statistically analyzing the environmental temperature fluctuation data of inefficient state clusters; fitting the distribution of inefficient clusters through a Gaussian mixture model; and taking the mean of the distribution model minus twice the standard deviation as the sixth threshold, which reflects the typical disturbance intensity of inefficient states.

[0046] Based on the frequency of occurrence of migration paths between basic energy efficiency states, the initial migration probability is calculated to form an initial migration probability matrix, as shown in the following formula: ; In the formula, To start from basic energy efficiency status Migrate to Basic Energy Efficiency State The initial migration probability, where the superscript "(0)" represents "initial" (the original probability before time decay factor update), and the subscript " "Represents the start-target state pair of the migration path; To be within the statistical period, "from the basic energy efficiency status" Migrate to "The actual number of times this specific migration path occurred; To be within the statistical period, "from the basic energy efficiency status" Migrate to any base energy efficiency state The actual number of times a single migration path occurs; Within the statistical period, "based on basic energy efficiency status" The total number of occurrences of all three migration paths that start in the "starting state".

[0047] Specifically, the formula for the initial migration probability matrix is ​​as follows: ; In the formula, This is the initial migration probability matrix.

[0048] By introducing a time decay factor to dynamically update the matrix, the probability value is adapted to the recent operating characteristics and aging trend of the equipment. The specific formula is as follows: ; In the formula, for From time to time Migrate to The updated migration probability dynamically adapts to the recent operating characteristics of the equipment and reflects the migration probability after equipment aging or changes in operating conditions. The time decay factor is used to weight the migration data at different times; For historical moments, the time points of historical data are determined and used to calculate time decay weights; The current update time is the reference time for time decay calculation; For indicator functions, statistics Whether the target migration path occurs at any given time provides discrete data for weighted summation; For indicator functions, statistics time This provides data for determining whether the transition from the initial state has occurred, and for weighted summation of the denominator.

[0049] Furthermore, the specific formula for the updated migration probability matrix is ​​as follows: ; In the formula, for The migration probability matrix updated at each time step.

[0050] The preprocessed first comprehensive data is input into the model. By calculating the Euclidean distance between the current preprocessed first comprehensive data and the cluster center vectors of the three states, the current energy efficiency status of the device is determined.

[0051] Specifically, the Euclidean distance between the first comprehensive data after preprocessing and the efficient state cluster center vector is calculated using the following formula: ; In the formula, This is the Euclidean distance between the first comprehensive data vector after preprocessing and the efficient state cluster center vector; This is the first component of the first synthesized data vector after preprocessing, i.e., the real-time performance coefficient COP after preprocessing; This is the second component of the first comprehensive data vector after preprocessing, i.e., the real-time load fluctuation coefficient after preprocessing. This is the third component of the first integrated data vector after preprocessing, which represents the real-time ambient temperature fluctuation after preprocessing. The first component of the cluster center vector in the efficient state is the typical value of the real-time performance coefficient COP obtained through cluster analysis, representing the efficient state. The second component of the cluster center vector for efficient states, i.e., the load fluctuation coefficient obtained through cluster analysis, representing the efficient state. Typical values; The third component of the cluster center vector for efficient states, obtained through cluster analysis, represents the environmental temperature fluctuation under efficient states. Typical values.

[0052] The Euclidean distance between the preprocessed first composite data and the cluster center vector of the intermediate state is calculated using the following formula: ; In the formula, The Euclidean distance between the first integrated data vector after preprocessing and the cluster center vector of the intermediate state is given. It is the first component of the cluster center vector of the intermediate state, that is, the typical value of the real-time performance coefficient COP obtained by cluster analysis, which represents the intermediate state. The second component of the cluster center vector for the intermediate-efficiency state is the load fluctuation coefficient obtained through cluster analysis, representing the intermediate-efficiency state. Typical values; The third component of the cluster center vector for the intermediate-efficiency state is the environmental temperature fluctuation under the intermediate-efficiency state, obtained through cluster analysis. Typical values.

[0053] The Euclidean distance between the preprocessed first aggregated data and the cluster center vector of the inefficient state is calculated using the following formula: ; In the formula, The Euclidean distance between the first integrated data vector after preprocessing and the cluster center vector of the inefficient state; The first component of the cluster center vector for inefficient states is the typical value of the real-time performance coefficient COP obtained through cluster analysis, representing the inefficient state. The second component of the cluster center vector for inefficient states is the load fluctuation coefficient obtained through cluster analysis, representing the inefficient state. Typical values; The third component of the cluster center vector for inefficient states, obtained through cluster analysis, represents the environmental temperature fluctuations under inefficient states. Typical values.

[0054] Specifically, if < and < If the current state point is closest to the cluster center of the efficient state cluster in the high-dimensional feature space, it is determined to be an efficient state.

[0055] like < and < If the current state point is closest to the cluster center of the intermediate state cluster, it is determined to be an intermediate state.

[0056] like < and < If the current state point is closest to the cluster center of the inefficient state cluster, it is determined to be an inefficient state.

[0057] It should be noted that the core purpose of determining the current energy efficiency status is to classify the unknown state of the current equipment and determine which energy efficiency status the equipment currently belongs to. This process is carried out by applying existing standards for measurement.

[0058] For example, the central air conditioning system on the 10th floor of an office building is first subjected to density clustering using preprocessed comprehensive data from the past month (including COP, load fluctuation coefficient, ambient temperature fluctuation, and related parameters). This results in the following performance categories: high efficiency (COP ≥ 4.2, load fluctuation coefficient ≤ 0.15, ambient temperature fluctuation ≤ 0.2℃), medium efficiency (COP between 3.8 and 4.2, load fluctuation coefficient between 0.15 and 0.2, ambient temperature fluctuation between 0.2 and 0.3℃), and low efficiency (COP ≤ 3.8, load fluctuation coefficient ≥ 0.2, ambient temperature fluctuation ≥ 0.2℃). Three basic energy efficiency states (COP = 4.3℃, gradient = 0.16, and fluctuation = 0.18℃) are used. The first threshold (4.2) is taken from the minimum boundary value of COP of the high-efficiency cluster, the second threshold (0.15) is the 95th quantile of the gradient of the high-efficiency cluster, and the other thresholds are calculated according to the corresponding logic. The number of migrations between the three states in the past three months is counted (e.g., 18 times for high-efficiency → medium-efficiency and 82 times for high-efficiency → high-efficiency). The initial migration probability matrix is ​​calculated according to the formula (18% probability for high-efficiency → medium-efficiency). A time decay factor λ = 0.8 is introduced to dynamically update the matrix. During the working hours of a certain day, the first comprehensive data after real-time preprocessing (COP = 4.3 is in the high-efficiency state, gradient = 0.16 is in the medium-efficiency state, and fluctuation = 0.18℃ is in the high-efficiency state) is input into the model to calculate the distance to the high-efficiency cluster center. =0.25, medium efficiency =0.48, inefficient =0.62, because If the minimum value meets the high-efficiency threshold, the device is determined to be in a high-efficiency state.

[0059] S3: Based on the judgment result, maintenance instructions are generated, the state boundaries are re-clustered, and preventive maintenance instructions are generated.

[0060] Preferably, maintenance instructions are generated based on the judgment result, as follows: When the system is determined to be in an efficient state, no maintenance instructions are generated. Instead, the current efficient state data is stored in the historical database as the benchmark data for subsequent "re-clustering and boundary adjustment". The duration of the efficient state is recorded synchronously. If the duration exceeds the set threshold, a "stable equipment operation" status report can be generated for maintenance personnel to refer to.

[0061] When the system is determined to be in a medium-efficiency state, a "parameter adjustment command" is generated, which automatically adjusts the optimizable parameters. The compressor frequency is increased in a stepwise manner, and after each increase, the COP and load fluctuation coefficient are monitored for a short time to avoid a sudden increase in frequency. The chilled water pump speed is reduced according to the degree of insufficient temperature difference, and after each adjustment, the temperature difference and flow rate are monitored for a short time. The air volume distribution is improved by adjusting the fan guide vane angle and then gradually increasing the fan speed. The proportion of fresh air is reduced in a stepwise manner, and after each adjustment, the temperature difference and CO2 concentration are monitored for a short time. No emergency maintenance command is generated.

[0062] When an inefficient state is detected, a "preventive maintenance instruction" will be triggered immediately: If triggered due to a low real-time performance coefficient (COP): check the compressor power / refrigerant pressure and clean the heat exchanger; if triggered due to excessive fluctuations in the load fluctuation coefficient: calibrate the load monitoring module and check indoor load changes; if triggered due to excessive fluctuations in ambient temperature: check the temperature and humidity sensor calibration status and clear obstacles around the outdoor unit; if the inefficient state is accompanied by critical parameters exceeding limits (such as compressor discharge temperature ≥120℃, refrigerant pressure ≤50% of rated value), an emergency shutdown warning will be added to prevent the equipment failure from escalating.

[0063] Specifically, the state boundaries are re-clustered, and preventative maintenance instructions are generated, as detailed below: Collect the first comprehensive data after preprocessing and the corresponding energy efficiency status judgment results, and remove invalid data generated during equipment shutdown, sensor failure, and maintenance, retaining only the valid data during the stable operation phase.

[0064] The selected effective data are re-clustered according to three labels: "high efficiency", "medium efficiency" and "low efficiency" using density clustering algorithm. The cluster center vectors of the three basic energy efficiency states are then calculated and updated.

[0065] Based on the updated cluster center vectors, the threshold acquisition logic recalculates the first to sixth thresholds: New first threshold: The minimum boundary value of COP within the new efficient cluster is taken, replacing the original first threshold, as the lower limit of COP for the efficient state; New second threshold: Calculates the minimum boundary value of COP within the new efficient cluster... The 95th percentile of the data replaces the original second threshold as the allowable efficient state. Upper limit; New third threshold: Box plot analysis of the new high-efficiency cluster The outlier boundary is used to replace the original third threshold as the allowable threshold for efficient states. Upper limit; New fourth threshold: Calculate the optimal separation point of the silhouette coefficients of the new intermediate-efficiency cluster and the new inefficient cluster in the COP dimension, and replace the original fourth threshold as the COP critical point between intermediate and inefficient clusters; New fifth threshold: Take the value within the new inefficient cluster. The minimum value is used to replace the original fifth threshold, serving as the criterion for medium efficiency and low efficiency. Critical point; New sixth threshold: for new inefficient clusters Perform Gaussian mixture model fitting, and replace the original sixth threshold with "mean - 2 × standard deviation" as the benchmark for medium efficiency and low efficiency. Critical point.

[0066] The updated first to sixth thresholds and the new cluster center vectors are synchronized to the state transition model as a new benchmark for subsequent energy efficiency status determination, ensuring that the model is adapted to the current operating characteristics of the equipment.

[0067] For example, during the operation of a hotel's central air conditioning system, on a certain day, the system determines that it is in a high-efficiency state via S2. The system only stores the COP, load fluctuation coefficient, environmental fluctuation, and equipment operating parameters for that period into the historical database, and simultaneously records that the high-efficiency state has lasted for 48 hours, exceeding the set 36-hour threshold, and automatically generates a "stable equipment operation" report for maintenance personnel to review. One week later, the system determines that it is in a medium-efficiency state and immediately generates parameter adjustment instructions: the compressor frequency is increased by 2Hz in each step (after the increase, the COP and load fluctuation coefficient are monitored for 3 minutes); the chilled water pump speed is reduced by 60r / min because the supply and return water temperature difference is lower than the design value of 0.3℃ (after adjustment, the temperature difference and flow rate are monitored); if the indoor local temperature difference still exceeds 1℃ after increasing the fan guide vane angle by 5°, the fan speed is increased again; and the fresh air ratio is reduced by 3% each time (monitoring). (Measured temperature difference and CO2 concentration), no emergency maintenance command was generated; half a month later, the system determined that it was in an inefficient state (because the real-time performance coefficient COP was lower than the fourth threshold), and generated a preventive maintenance command: check the real-time power and refrigerant pressure of the compressor, clean the heat exchanger, and it was found that the refrigerant pressure had dropped to 45% of the rated value (key parameters exceeded the limit), and added an emergency shutdown warning to remind maintenance personnel to handle it within 2 hours; at the end of the month, the system collected the first comprehensive data and status judgment results after nearly 1 month of preprocessing, removed invalid data of equipment shutdown at night and sensor failure, and only retained the stable daytime operation data. The cluster centers of the three types of states were updated by density clustering, and the first to sixth thresholds were recalculated based on the new centers (such as adjusting the lower limit of high-efficiency COP from 4.2 to 4.1), and the new thresholds and new centers were synchronized to the state transition model.

[0068] S4: Optimize the parameters of the state transition model based on the maintained first comprehensive data to complete the iterative update of the state transition model, thereby realizing continuous intelligent detection of HVAC energy efficiency.

[0069] The preferred method for optimizing model parameters based on the maintained first comprehensive data is as follows: The normal operating data collected after maintenance, where energy efficiency has been restored, is merged into the original historical dataset. Cluster analysis is then re-executed to update the cluster center vectors for high-efficiency, medium-efficiency, and low-efficiency states, making them more consistent with the current actual performance of the equipment.

[0070] The new state transition paths and frequencies observed after maintenance are merged into the original transition frequency statistics, and the initial transition probability matrix is ​​recalculated.

[0071] Statistical analysis is performed on the average duration of the equipment in each energy efficiency state after maintenance. If the average duration increases (e.g., the duration of high efficiency state is extended by more than 30%), the λ value is increased (e.g., from 0.8 to 0.85-0.9) to increase the model's focus on recent data and to more quickly forget the old patterns before maintenance.

[0072] The optimized parameters (new center, new matrix, new λ) are loaded into the model, and the data from a period of time after maintenance (e.g., 1 day) is used for verification testing. The matching rate of the state judgment results (the proportion of samples whose judgment results are consistent with expectations) is calculated. If the matching rate is higher than the preset target (e.g., 90%), the optimization is confirmed to be effective and the model is updated; if it is lower than the target, the data quality needs to be checked or the optimization process needs to be re-executed, forming a complete closed loop of "maintenance-data collection-parameter optimization-model iteration".

[0073] For example, after preventative maintenance such as heat exchanger cleaning and refrigerant replenishment was performed on the central air conditioning system of a factory workshop, the first comprehensive data (including improved COP, reduced load fluctuation coefficient, and related parameters) of stable operation within one week after maintenance was collected. After preprocessing, this data was merged into the original one-year historical dataset, and density clustering analysis was re-performed. The typical COP value of the high-efficiency state cluster centers was updated from 4.2 before maintenance to 4.3, and the typical load fluctuation coefficient value was updated from 0.15 to 0.12. The number of state transitions within two weeks after maintenance (e.g., the number of inefficient → medium efficiency transitions increased from 20 before maintenance to 45 times) was counted and merged into the original transition frequency statistics. The initial transition probability matrix was recalculated according to the formula (the probability of inefficient → medium efficiency increased from 20% to 45%). The duration of equipment status after maintenance was analyzed, and it was found that the average duration of high-efficiency status was extended by 40% compared with before maintenance. The time decay factor was then adjusted accordingly. The value was adjusted from 0.8 to 0.88 to increase the weight of recent data; the optimized cluster centers, initial migration probability matrix, and The values ​​were loaded into the state transition model and verified using real-time data from the 8th day after maintenance. The calculated state determination results matched the actual energy efficiency characteristics by 93%, which was higher than the preset target of 90%. The optimization was confirmed to be effective and the model parameters were updated, forming a complete closed loop of "maintenance-data acquisition-parameter optimization-model iteration".

[0074] In summary, this invention determines energy efficiency status through "cluster analysis + Euclidean distance calculation + dynamic threshold," avoiding misjudgments by static thresholds and accurately reflecting the energy efficiency characteristics of equipment under all operating conditions. It introduces a time decay factor to dynamically update the state transition matrix, ensuring the state transition model aligns with the changes in migration patterns caused by equipment aging and maintenance effects. A tiered strategy of "high-efficiency recording, medium-efficiency automatic adjustment, and low-efficiency preventative maintenance" is designed to reduce over-maintenance and energy waste, with emergency warnings added when key parameters exceed limits. Regular re-clustering and updating of state boundaries and thresholds adapt to changes in equipment characteristics throughout its entire lifecycle. Post-maintenance data is used to correct cluster center vectors, update migration probabilities, and adjust time decay factors, forming a "maintenance-optimization" closed loop to ensure long-term accuracy. Furthermore, the modular system can be implemented using conventional computer equipment, eliminating the need for customized hardware and reducing deployment costs. Overall, this invention represents a breakthrough in HVAC energy efficiency testing, moving from static and extensive methods to dynamic and precise ones, and from passive maintenance to proactive prevention, aligning with building energy conservation needs and possessing broad industrial application value.

[0075] Example 2, an embodiment of the present invention, provides an intelligent detection system for HVAC energy efficiency, comprising: a data acquisition module for acquiring first comprehensive data of HVAC equipment and preprocessing the first comprehensive data; a model building module for constructing a state transition model based on the preprocessed first comprehensive data and determining the current energy efficiency status of the HVAC equipment in real time through the state transition model; an instruction generation module for triggering maintenance instruction generation based on the judgment result, re-clustering and adjusting the state boundaries and generating preventive maintenance instructions; and a parameter optimization module for optimizing the parameters of the state transition model based on the maintained first comprehensive data.

[0076] Example 3 is an embodiment of the present invention, which differs from the previous embodiment in that: If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0077] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0078] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0079] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0080] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for intelligent detection of energy efficiency in HVAC systems, characterized in that: include: Acquire the first comprehensive data of the HVAC equipment and preprocess the first comprehensive data; A state transition model is constructed based on the preprocessed first comprehensive data, and the current energy efficiency status of the HVAC equipment is determined in real time through the state transition model. Based on the judgment result, maintenance instructions are generated, the state boundaries are re-clustered, and preventive maintenance instructions are generated. The parameters of the state transition model are optimized based on the first comprehensive data after maintenance to complete the iterative update of the state transition model, thereby realizing continuous intelligent detection of HVAC energy efficiency.

2. The intelligent detection method for HVAC energy efficiency as described in claim 1, characterized in that: The first comprehensive data includes equipment operating parameters, environmental parameters, and energy efficiency indicators; The equipment operating parameters include compressor parameters, refrigerant circulation parameters, water system parameters, and air system parameters; the environmental parameters include indoor environmental parameters and outdoor environmental parameters; the energy efficiency indicators include real-time coefficient of performance (COP), load fluctuation coefficient, and ambient temperature fluctuation. The real-time performance coefficient COP is acquired in real time by a power transmitter and flow sensor installed in the main circuit of the equipment; the load fluctuation coefficient is obtained by acquiring the actual load of the equipment in real time through the load monitoring module built into the air conditioner controller. The equipment operating parameters are obtained through pressure sensors, current sensors, and electromagnetic flow meters; the environmental parameters are obtained through temperature and humidity sensors, CO2 sensors, and air pressure sensors.

3. The intelligent detection method for HVAC energy efficiency as described in claim 2, characterized in that: The specific details of constructing the state transition model are as follows: Cluster analysis was performed on the preprocessed first comprehensive data, and three basic energy efficiency states were divided into high efficiency, medium efficiency and low efficiency. Based on the frequency of occurrence of migration paths between basic energy efficiency states, the initial migration probability is calculated to form an initial migration probability matrix, and an initial state migration model is generated. By introducing a time decay factor to dynamically update the initial migration probability matrix, a final state migration model with self-updating capability is formed. The preprocessed first comprehensive data is input into the state transition model, and the Euclidean distance between it and the cluster center vectors of efficient, medium-efficiency, and inefficient states is calculated respectively to obtain the distance values. , and This allows us to determine the current energy efficiency status of the HVAC equipment.

4. The intelligent detection method for HVAC energy efficiency as described in claim 3, characterized in that: The specific steps for determining the current energy efficiency status of the HVAC equipment are as follows: like < and < If the current state point is closest to the cluster center of the efficient state cluster in the high-dimensional feature space, it is determined to be an efficient state. like < and < If the current state point is closest to the cluster center of the intermediate state cluster, it is determined to be an intermediate state. like < and < If the current state point is closest to the cluster center of the inefficient state cluster, it is determined to be an inefficient state.

5. The intelligent detection method for HVAC energy efficiency as described in claim 4, characterized in that: Based on the judgment result, a maintenance instruction is generated, the details of which are as follows: When the system is determined to be in an efficient state, no maintenance instructions are generated. Instead, the current state data is stored in the historical database for subsequent re-clustering and adjustment of state boundaries. The duration of the efficient state is also recorded. If the duration exceeds the set duration, a device operation stability report is generated. When the system is determined to be in a medium-efficiency state, a parameter adjustment command is generated to automatically optimize and adjust the equipment operating parameters, including step adjustment of compressor frequency, adjustment of water pump speed, adjustment of fan operating parameters, and monitoring of energy efficiency indicators after each adjustment. When an inefficient state is identified, a preventative maintenance instruction is immediately generated to perform corresponding maintenance operations on the main causes of inefficiency. If critical parameters exceed limits, an emergency shutdown warning is added.

6. The intelligent detection method for HVAC energy efficiency as described in claim 5, characterized in that: The state boundaries are re-clustered and adjusted to generate preventative maintenance instructions, the details of which are as follows: Collect the first comprehensive data after preprocessing and the corresponding energy efficiency status judgment results, remove invalid data, and retain the valid data in the stable operation phase; The effective data is re-clustered using density clustering algorithm, and the cluster center vectors of the efficient, medium-efficiency and inefficient states are updated. Based on the updated cluster center vectors, the first to sixth thresholds are recalculated. The updated threshold and cluster center vector are synchronized to the state transition model as a new benchmark for subsequent energy efficiency state determination.

7. The intelligent detection method for HVAC energy efficiency as described in claim 6, characterized in that: The specific details of the optimized model parameters based on the maintained first comprehensive data are as follows: Collect the first comprehensive data and energy efficiency status assessment results after maintenance, and perform preprocessing. The maintained data is merged with data of the same state, and re-clustered to correct the cluster center vectors of efficient, medium-efficiency, and inefficient states; Based on the number of data update status migrations after maintenance, the migration probability is recalculated. Adjust the time decay factor according to the change in the duration of the post-maintenance state; The optimized parameters are verified. If the state determination matching rate meets the requirements, it is used as a new benchmark for the model.

8. A smart detection system for HVAC energy efficiency, based on the smart detection method for HVAC energy efficiency according to any one of claims 1 to 7, characterized in that: include, The data acquisition module is used to acquire the first comprehensive data of the HVAC equipment and to preprocess the first comprehensive data; The model building module is used to build a state transition model based on the preprocessed first comprehensive data, and to determine the current energy efficiency status of the HVAC equipment in real time through the state transition model. The instruction generation module is used to trigger the generation of maintenance instructions based on the judgment results, re-cluster and adjust the state boundaries, and generate preventive maintenance instructions. The parameter optimization module is used to optimize the parameters of the state transition model based on the first comprehensive data after maintenance.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the intelligent detection method for HVAC energy efficiency as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the intelligent detection method for energy efficiency of HVAC as described in any one of claims 1 to 7.

Images