Sub-item energy consumption reference value calibration method based on heating ventilation air conditioning equipment

Abstract

The invention discloses a subitem energy consumption reference value calibration method based on heating ventilation air conditioning equipment, and relates to the technical field of building energy conservation. The invention discloses a subitem energy consumption reference value calibration method based on heating ventilation air conditioning equipment. The method comprises the steps of energy consumption reference value abnormity judgment, energy consumption reference value preliminary calibration and energy consumption reference value optimization calibration. According to the invention, through serialized analysis of energy consumption of different subitems, fine management of energy consumption of the heating ventilation and air conditioning equipment is realized, and more accurate identification of energy consumption abnormity is facilitated; by analyzing the energy consumption correlation degree, energy consumption can be monitored in real time, abnormal conditions can be fed back quickly, and the subitem energy consumption of the abnormal target heating, ventilation and air conditioning equipment can be recognized accurately; environmental information, climate information and equipment state information are considered, so that the reason of energy consumption change can be more comprehensively understood; by dynamically adjusting the energy consumption reference value according to the real-time data and the influence factor, the adaptability and accuracy of the reference value are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of building energy conservation, and in particular to a method for calibrating reference values of sub-item energy consumption based on heating, ventilation and air-conditioning equipment. Background Art

[0002] With the increase in global energy consumption and growing awareness of environmental protection, improving energy efficiency has become particularly important. As one of the largest energy consumers in a building, the energy efficiency of HVAC equipment directly affects the energy efficiency of the entire building. Traditional HVAC equipment energy consumption benchmarks are typically set based on historical data, but this approach may have limitations. Due to the complexity and diversity of building operating conditions, a simple energy consumption benchmark may not accurately reflect actual energy consumption. Due to the constant changes in the environment and the equipment's own status, fixed energy consumption benchmarks may not be able to adapt to these changes, necessitating calibration of the HVAC equipment's individual energy consumption benchmarks. Summary of the Invention

[0003] The present invention aims to provide a method for calibrating the reference value of sub-item energy consumption based on HVAC equipment, and calibrates the reference value of sub-item energy consumption in combination with multiple targets.

[0004] The calibration method for the benchmark value of energy consumption of HVAC equipment includes: S1. Determine abnormality of energy consumption baseline value; Obtain the actual energy consumption sequence of different items within the preset time range and obtain the energy consumption sequence to be analyzed X n , X n =[X n（1） , X n（2） ,…,X n（I） ]; N is the number of energy consumption categories of target HVAC equipment, X n（i） Represents the energy consumption sequence X to be analyzed n The energy consumption of the i-th target HVAC equipment in the set, n=1, 2, ..., N; I represents the energy consumption sequence X to be analyzed within the preset time range n The actual energy consumption value number in, i=1, 2, ..., I; based on the energy consumption sequence to be analyzed X n and the current energy consumption benchmark value D n Perform correlation analysis to obtain the correlation Z of sub-item energy consumption n ; If the correlation degree of sub-item energy consumption Z n If the energy consumption associated threshold is reached, the energy consumption abnormality result is normal, and the current sub-item energy consumption baseline value D is retained. n Otherwise, the energy consumption abnormality result is abnormal, and the correlation degree Z of the sub-item energy consumption is recorded. n Corresponding target HVAC equipment energy consumption X n（i） For abnormal target HVAC equipment energy consumption, proceed to step S2; S2. Preliminary calibration of energy consumption baseline value; Obtain the environmental information and climate information within a preset time range to obtain the environmental information to be analyzed and the climate information to be analyzed; obtain the current status information of the target HVAC equipment to obtain the status information to be analyzed; input the status information to be analyzed, the environmental information to be analyzed, the climate information to be analyzed, and the abnormal target HVAC equipment sub-item energy consumption into the energy consumption impact factor judgment model for analysis to obtain the energy consumption baseline value impact factor; S3. Optimized calibration of energy consumption baseline value; Based on the energy consumption baseline value impact factor, optimize the calculation of the abnormal target HVAC equipment sub-item energy consumption to obtain the optimized sub-item energy consumption baseline value D n ’; Evaluate the optimized sub-item energy consumption baseline value D n ’. If the baseline value evaluation result is passed, then use the optimized sub-item energy consumption baseline value D n ’ as the calibration result of the current sub-item energy consumption baseline value D n ; Otherwise, continue to optimize the optimized sub-item energy consumption baseline value D n ’ until it meets the calibration result of the current sub-item energy consumption baseline value D n .

[0005] As a preferred technical solution of the present invention, the specific steps of calculating the sub-item energy consumption correlation degree Z n in step S1 include: Perform calculations on the abnormal target HVAC equipment sub-item energy consumption X n（i） : Q1. Calculate the sub-item energy consumption correlation coefficient G n（i） ; Use the formula

[0006] to calculate the sub-item energy consumption correlation coefficient G n（i） . The sub-item energy consumption correlation coefficient G n（i） represents the correlation coefficient of X n in the energy consumption sequence X n（i） to be analyzed; R A（n） represents the two-level minimum difference, R A（n） = minminY n（i） ; R B（n） represents the two-level maximum difference, R B（n） = maxmaxY n（i） ; Y n（i） represents the absolute value of the difference between the current sub-item energy consumption baseline value D n and X n（i） , Y n（i） = |D n - X n（i）|; E is the discrimination coefficient, which is confirmed by the sub-item energy consumption X of the target HVAC equipment n（i） for confirmation; Q2. Determine the weight W of the sub-item energy consumption correlation coefficient n ; Using the nine-level scale method for all sub-item energy consumptions X of the target HVAC equipment n（i） to establish a sub-item energy consumption evaluation matrix; calculate the eigenvalue for the sub-item energy consumption evaluation matrix to obtain the maximum eigenvalue of the sub-item energy consumption evaluation matrix; use the sub-item energy consumption evaluation matrix and the maximum eigenvalue of the sub-item energy consumption evaluation matrix to calculate the eigenvector to obtain the eigenvector of the sub-item energy consumption evaluation matrix; perform normalization processing on the eigenvector of the sub-item energy consumption evaluation matrix to obtain the preprocessed eigenvector of the sub-item energy consumption evaluation matrix; the nth item in the preprocessed eigenvector of the sub-item energy consumption evaluation matrix is the weight W of the sub-item energy consumption correlation coefficient n ; Q3. Calculate the sub-item energy consumption correlation Z n ; Using the formula

[0007] to calculate the sub-item energy consumption correlation Z n .

[0008] As a preferred technical solution of the present invention, the specific steps for determining the discrimination coefficient E based on the sub-item energy consumption X of the target HVAC equipment n（i） include: For the sub-item energy consumption X of the target HVAC equipment n（i） and the sub-item energy consumption X of the target HVAC equipment n’（i） , using the formula

[0009] to calculate the discrimination coefficient E （n，n’） , and the discrimination coefficient E （n，n’） represents the influence between different sub-item energy consumptions X of the target HVAC equipment at the same moment n（i） and the sub-item energy consumption X of the target HVAC equipment n’（i） ; the sub-item energy consumption X of the target HVAC equipment n（i） and the sub-item energy consumption X of the target HVAC equipment n’（i） represent any two combinations of the sub-item energy consumptions X of the target HVAC equipment n（i） for calculating the discrimination coefficient, indicating the energy consumption deviation degree between the sub-item energy consumptions of different sub-item energy consumption categories of the target HVAC equipment within the same moment; R B（n，n’） represents the maximum value between R B（n） and R B（n’） ; a total of n*(n - 1) / 2 groups of discrimination coefficients E are calculated （n，n’） ; Calculate the mean value of all discrimination coefficients E, and determine the discrimination coefficient E based on the interval where the mean value is located. （n，n’）

[0010] As a preferred technical solution of the present invention, the energy consumption impact factor judgment model in step S2 includes a feature extraction layer, an internal feature analysis layer, an external feature analysis layer, and a result output layer; The feature extraction layer is used to extract features from the environmental information and climate information to be analyzed to obtain environmental information features and climate information features; extract features from the status information to be analyzed to obtain status information features; The external feature analysis layer is used to fuse the environmental information features and climate information features to obtain external fusion features; analyze the external fusion features to obtain external energy consumption impact factors; The internal feature analysis layer is used to analyze the status information features to obtain internal energy consumption impact factors; The result output layer is used to match the corresponding impact factor weights in the sub-item energy consumption factor impact library according to the sub-item energy consumption of the abnormal target HVAC equipment to obtain the target impact factor weights; perform weighted calculation on the external energy consumption impact factors and internal energy consumption impact factors according to the target impact factor weights to obtain the energy consumption benchmark value impact factor.

[0011] As a preferred technical solution of the present invention, the specific steps for training the external feature analysis layer include: Collect several groups of external feature analysis training samples. Each group of external feature analysis training samples contains an external fusion feature and the corresponding marked external impact factor value; combine several groups of external feature analysis training samples to obtain an external feature analysis training set; Input the external feature analysis training set into the energy consumption impact factor judgment model to train the external feature analysis layer with the corresponding marked external impact factor value as the target to obtain an initial external feature analysis layer; evaluate the initial external feature analysis layer to obtain the model evaluation result of the initial external feature analysis layer; if the model evaluation result of the initial external feature analysis layer passes, use the initial external feature analysis layer as the external feature analysis layer in the energy consumption impact factor judgment model; otherwise, continue to perform model training using the external feature analysis training set.

[0012] As a preferred technical solution of the present invention, the specific steps for training the internal feature analysis layer include: Collect several groups of internal feature analysis training samples. Each group of internal feature analysis training samples contains an internal status feature and the corresponding marked internal impact factor value; combine several groups of internal feature analysis training samples to obtain an internal feature analysis training set; Input the internal feature analysis training set into the energy consumption impact factor judgment model to train the internal feature analysis layer with the internal impact factor values corresponding to the marks as the target, and obtain the initial internal feature analysis layer; perform model evaluation on the initial internal feature analysis layer to obtain the model evaluation result of the initial internal feature analysis layer; if the model evaluation result of the initial internal feature analysis layer passes, use the initial internal feature analysis layer as the internal feature analysis layer in the energy consumption impact factor judgment model; otherwise, continue model training using the internal feature analysis training set.

[0013] As a preferred technical solution of the present invention, the specific steps for optimizing the calculation of the abnormal target HVAC equipment sub-item energy consumption based on the energy consumption benchmark value impact factor include: Obtain the energy consumption sequence X to be analyzed corresponding to the abnormal target HVAC equipment sub-item energy consumption n , and obtain the abnormal energy consumption sequence; perform abnormal judgment on the abnormal energy consumption sequence to obtain the optimal screened energy consumption sequence; construct K optimized sub-item energy consumption benchmark value individuals H k ; each optimized sub-item energy consumption benchmark value individual H k contains a set of energy consumption benchmark values to be optimized established based on the current sub-item energy consumption benchmark value D n and the energy consumption benchmark value impact factor; combine the K optimized sub-item energy consumption benchmark value individuals H k to obtain the optimized sub-item energy consumption benchmark value iterative population; set the maximum number of iterations T, and set the current number of iterations as t; Perform simulation calculations on the optimized sub-item energy consumption benchmark value individuals H k in the optimized sub-item energy consumption benchmark value iterative population, perform predictive analysis based on the optimal screened energy consumption sequence to obtain the predicted energy consumption sequence; perform regression analysis based on the predicted energy consumption sequence to obtain the optimal predicted energy consumption mean value; calculate the Euclidean distance between the optimal predicted energy consumption mean value and the energy consumption benchmark value to be optimized in the optimized sub-item energy consumption benchmark value individual H k to obtain the energy consumption benchmark difference C k ; take the reciprocal of the energy consumption benchmark difference C k as the fitness S k of the optimized sub-item energy consumption benchmark value individual H k ; During the population iteration process, perform iterative update on the optimized sub-item energy consumption benchmark value iterative population; When the maximum number of iterations is reached, output the optimized sub-item energy consumption benchmark value individual H k corresponding to the maximum fitness, which is the optimal optimized sub-item energy consumption benchmark value individual; use the energy consumption benchmark value to be optimized in the optimal optimized sub-item energy consumption benchmark value individual as the optimized sub-item energy consumption benchmark value D n '.

[0014] As a preferred technical solution of the present invention, the specific steps for iteratively updating the optimized sub-item energy consumption benchmark value iteration population include: Divide the optimized sub-item energy consumption benchmark value iteration population into two parts to obtain the first optimized sub-item energy consumption benchmark value iteration population and the second optimized sub-item energy consumption benchmark value iteration population; Set the iteration parameter U, U = exp(-t / T); exp() is the natural exponential function; set the iteration factor V, V = exp((t - T) / T) / 2; When V < L1, the first optimized sub-item energy consumption benchmark value iteration population and the second optimized sub-item energy consumption benchmark value iteration population are randomly exchanged and updated; When U < L2, the first optimized sub-item energy consumption benchmark value iteration population and the second optimized sub-item energy consumption benchmark value iteration population are updated to the optimized sub-item energy consumption benchmark value individuals with larger fitness; When V > L1 and U > L2, select the optimized sub-item energy consumption benchmark value individuals with larger fitness in the first optimized sub-item energy consumption benchmark value iteration population and the second optimized sub-item energy consumption benchmark value iteration population for mutation to obtain a new optimized sub-item energy consumption benchmark value iteration population.

[0015] The present invention has the following advantages: 1. Through the serialized analysis of different sub-item energy consumptions, the present invention realizes the refined management of the energy consumption of HVAC equipment, which helps to more accurately identify abnormal energy consumption; through the analysis of energy consumption correlation, it helps to monitor energy consumption in real time and quickly feedback abnormal situations, and can accurately identify the abnormal target sub-item energy consumption of HVAC equipment; by considering environmental information, climate information and equipment status information, it helps to more comprehensively understand the reasons for energy consumption changes; by dynamically adjusting the energy consumption benchmark value according to real-time data and influencing factors, the adaptability and accuracy of the benchmark value are improved.

[0016] 2. By calculating the sub-item energy consumption correlation coefficient, the present invention can accurately evaluate the correlation degree between each energy consumption data point and the benchmark value, providing a quantitative basis for energy consumption anomaly judgment; using the nine-level scale method and eigenvector to calculate the weight of the sub-item energy consumption correlation coefficient ensures the reasonable weight distribution of different energy consumption factors in the comprehensive evaluation; by comprehensively considering the sub-item energy consumption correlation coefficient and weight, the calculated sub-item energy consumption correlation can more scientifically reflect the consistency between the energy consumption sequence and the benchmark value; by calculating the influence between different sub-item energy consumptions and dynamically determining the discrimination coefficient, the model can adapt to the characteristics of different energy consumption data, improving the adaptability and accuracy of the correlation analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a structural schematic diagram of the sub-item energy consumption benchmark value calibration method based on HVAC equipment adopted in the embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0019] Based on the calibration method of the energy consumption benchmark value of HVAC equipment, see Figure 1 As shown, including: S1. Determine abnormality of energy consumption baseline value; Obtain the actual energy consumption sequence of different items within the preset time range and obtain the energy consumption sequence to be analyzed X n , X n =[X n（1） , X n（2） ,…,X n（I） ]; N is the number of energy consumption categories of target HVAC equipment, X n（i） Represents the energy consumption sequence X to be analyzed n The energy consumption of the i-th target HVAC equipment in the set, n=1, 2, ..., N; I represents the energy consumption sequence X to be analyzed within the preset time range n The actual energy consumption value number in, i=1, 2, ..., I; based on the energy consumption sequence to be analyzed X n and the current energy consumption benchmark value D n Perform correlation analysis to obtain the correlation Z of sub-item energy consumption n ; If the correlation degree of sub-item energy consumption Z n If the energy consumption associated threshold is reached, the energy consumption abnormality result is normal, and the current sub-item energy consumption baseline value D is retained. n Otherwise, the energy consumption abnormality result is abnormal, and the correlation degree Z of the sub-item energy consumption is recorded. n Corresponding target HVAC equipment energy consumption X n（i） For abnormal target HVAC equipment energy consumption, proceed to step S2; The preset time range is set by professional technicians based on actual conditions and can be the time range during which abnormal signals are continuously received; the energy consumption correlation threshold is set by professional technicians based on actual conditions and reflects the degree of correlation of energy consumption values; In step S1, the correlation degree Z of the energy consumption of each item is calculated. n The specific steps include: Energy consumption of target HVAC equipment X n（i） Perform the calculation: Q1. Calculate the correlation coefficient G of sub-item energy consumption n（i） ; Using the formula

[0020] Calculate the correlation coefficient G of the sub-item energy consumptionn（i） , the sub - item energy consumption correlation coefficient G n（i） represents the energy consumption sequence X to be analyzed n in X n（i） 's correlation coefficient; R A（n） represents the two - level minimum difference, R A（n） = min min Y n（i） ; R B（n） represents the two - level maximum difference, R B（n） = max max Y n（i） ; Y n（i） represents the absolute value of the difference between the current sub - item energy consumption benchmark value D n and X n（i） ; Y n（i） = |D n - X n（i） |; E is the resolution coefficient, which is confirmed by the target HVAC equipment sub - item energy consumption X n（i） ; The resolution coefficient E is an adjustment parameter used to control the sensitivity of the grey correlation degree calculation. It corresponds to the resolution coefficient in the standard grey correlation degree calculation formula; X n（i） represents the energy consumption of the nth sub - item energy consumption category at the ith moment, which is the actually collected energy consumption data, that is, the target HVAC equipment sub - item energy consumption; D n is the benchmark value of the nth sub - item energy consumption category, usually from historical operation data or equipment specifications; Because Y n（i） = |D n - X n（i） |, so R A（n） = min min Y n（i） , represents the point closest to the benchmark value; and R B（n） = max max Y n（i） represents the point most deviated from the benchmark value; The formula of the sub - item energy consumption correlation coefficient G n（i） comes from a form of the grey correlation degree calculation formula in grey correlation analysis method. This method is often used to measure the similarity degree between different data sequences. The essence of this formula is to measure the similarity between the actual energy consumption sequence and the benchmark value. This formula is one of the standard forms of grey correlation degree calculation, where R A（n） and R B（n） represent the two - level minimum difference and the two - level maximum difference respectively, and E is the resolution coefficient used to adjust the sensitivity of the correlation degree. This formula calculates the correlation coefficient of each data point by comparing the relationship between the absolute difference between the actual energy consumption and the benchmark value and the range.

[0021] Q2. Determine the weight W of the sub - item energy consumption correlation coefficient n ; Use the nine - level scaling method for all target HVAC equipment sub - item energy consumption Xn（i） Establish a sub - item energy consumption evaluation matrix; calculate the eigenvalues for the sub - item energy consumption evaluation matrix to obtain the maximum eigenvalue of the sub - item energy consumption evaluation matrix; use the sub - item energy consumption evaluation matrix and the maximum eigenvalue of the sub - item energy consumption evaluation matrix to calculate the eigenvector, obtaining the eigenvector of the sub - item energy consumption evaluation matrix; perform normalization processing on the eigenvector of the sub - item energy consumption evaluation matrix to obtain the pre - processed eigenvector of the sub - item energy consumption evaluation matrix; the nth item in the pre - processed eigenvector of the sub - item energy consumption evaluation matrix is the weight W of the sub - item energy consumption correlation coefficient n ; Q3. Calculate the sub - item energy consumption correlation degree Z n ; Use the formula

[0022] Calculate to obtain the sub - item energy consumption correlation degree Z n ; This formula is a weighted average of the correlation coefficients, where W n is the weight calculated by the nine - layer scaling method and the eigenvector method, which is used to reflect the importance of different energy consumption sub - items

[0023] Grey relational analysis is particularly suitable for systems with small samples and incomplete information. The energy consumption data of HVAC equipment often has uncertainty and dynamics. Therefore, this method is very suitable for calibrating the energy consumption benchmark value. By calculating the correlation degree, the consistency between the actual energy consumption and the benchmark value can be quantified, so as to judge whether calibration is needed; using the nine - layer scaling method and the eigenvector method to calculate the weight ensures a reasonable weight distribution of different energy consumption sub - items in the comprehensive evaluation. This method combines expert experience and mathematical modeling, avoiding subjective arbitrariness and improving the objectivity and scientificity of weight distribution; by introducing the discrimination coefficient E, the formula can dynamically adjust the sensitivity of the correlation degree calculation according to the characteristics of the actual data, so as to meet the energy consumption analysis needs in different scenarios

[0024] The purpose of using the formula is to judge energy consumption anomalies. By calculating the sub - item energy consumption correlation degree, the deviation degree between the actual energy consumption sequence and the benchmark value can be quantified; if the sub - item energy consumption correlation degree Z n is lower than the energy consumption correlation threshold, it is judged as abnormal and triggers the subsequent calibration process; this formula can accurately evaluate the correlation degree of each energy consumption data point, providing a quantitative basis for energy consumption management and realizing the refined management of HVAC equipment energy consumption; through correlation degree analysis, the deviation of the benchmark value caused by factors such as environmental changes and equipment aging can be identified, providing data support for dynamic calibration

[0025] Based on the sub - item energy consumption X of the target HVAC equipment n（i） The specific steps for determining the discrimination coefficient E include: For the sub - item energy consumption X of the target HVAC equipment n（i）and the sub - item energy consumption X of the target HVAC equipment n’（i） , using the formula

[0026] to calculate the discrimination coefficient E （n，n’） , the discrimination coefficient E （n，n’） represents the influence between different sub - item energy consumptions X of the target HVAC equipment at the same moment n（i） and the sub - item energy consumption X of the target HVAC equipment n’（i） ; the sub - item energy consumption X of the target HVAC equipment n（i） and the sub - item energy consumption X of the target HVAC equipment n’（i） represent any two sub - item energy consumptions X of the target HVAC equipment n（i） combined for discrimination coefficient calculation, representing the energy consumption deviation degree between the sub - item energy consumptions of different categories of the target HVAC equipment within the same moment; the core idea of this formula is to calculate an adaptive discrimination coefficient by analyzing the mutual influence between different sub - item energy consumptions; in traditional grey relational analysis, the discrimination coefficient is usually a fixed value (such as 0.5), but in practical applications, the characteristics of different sub - item energy consumptions may vary greatly; by calculating the mutual influence between sub - items, the value of the discrimination coefficient is dynamically determined to adapt to the characteristics of different energy consumption data and improve the accuracy of correlation analysis; constructed through the absolute difference and range of sub - item energy consumption, it ensures the mathematical rigor of the calculation, and the normalization process further eliminates the interference of different sub - items on the discrimination coefficient, making the result more universal.

[0027] In the formula, in the numerator, and are the sum of the absolute differences between two sub - item energy consumptions and the reference value, and the product form is used to quantify the synergistic effect of the overall deviation degree of two sub - item energy consumption sequences; the denominator part is used to eliminate the influence of the data volume on the calculation result. The cumulative sum in the denominator increases as I increases, and dividing by I can ensure that the discrimination coefficient is not directly interfered by the number of data points, standardizing the deviation product in the denominator to the same dimension as the range, avoiding the distortion of the discrimination coefficient caused by the absolute value difference; the coefficient 2 is used to control the value range of the discrimination coefficient, which is set through expert experience.

[0028] The dynamic discrimination coefficient can more accurately reflect the relationship between the actual energy consumption data and the reference value, avoiding problems such as insufficient sensitivity or over - sensitivity caused by a fixed coefficient; by calculating the discrimination coefficients between different sub - items, the mutual influence of multi - sub - item energy consumption can be comprehensively analyzed, providing a basis for the global calibration of the energy consumption reference value; R B（n，n’） represents the maximum value between R B（n） and R B（n’） ; a total of n*(n - 1) / 2 groups of discrimination coefficients E （n，n’） are calculated; Calculate all resolution coefficients E （n，n’） The resolution coefficient E is determined based on the mean of the mean interval. This formula reflects the impact of different sub-items on the resolution coefficient. By calculating the resolution coefficient for every two sub-items and averaging all the resolution coefficients, the impact of different energy consumption sub-items on the correlation calculation can be eliminated. In the formula, Y n（i） =|D n -X n（i） |, Y n’（i） =|D n’ -X n’（i） |, R B（n） =maxmaxY n（i） , R B（n’） =maxmaxY n’（i） , the specific value is determined by X n（i） The energy consumption sequence X to be analyzed corresponding to the nth sub-item energy consumption category n and X n’（i） The energy consumption sequence X to be analyzed corresponding to the n'th sub-item energy consumption category n’ Make a determination; n≠n'; The specific steps for determining the resolution coefficient E based on the interval of the mean are: Normalization is performed based on the calculated result of the mean so that the normalized mean is within the range of (0, 1). The interval is determined based on the normalized mean, and the final resolution coefficient E is selected. For example, when the normalized mean is within the first interval, E=0.3 can be set; when the normalized mean is within the second interval, E=0.4 can be set; when the normalized mean is within the third interval, E=0.5 can be set; when the normalized mean is within the fourth interval, E=0.6 can be set; when the normalized mean is within the fifth interval, E=0.7 can be set. The above values only provide one setting method, and the specific values are manually set by professional and technical personnel.

[0029] By calculating the correlation coefficient of sub-item energy consumption, the correlation degree between each energy consumption data point and the reference value can be accurately evaluated, providing a quantitative basis for the judgment of energy consumption anomalies; using the nine-level scale method and eigenvector to calculate the weight of the correlation coefficient of sub-item energy consumption ensures a reasonable weight distribution of different energy consumption factors in the comprehensive evaluation; by comprehensively considering the correlation coefficient and weight of sub-item energy consumption, the calculated correlation degree of sub-item energy consumption can more scientifically reflect the consistency between the energy consumption sequence and the reference value; by calculating the influence between different sub-item energy consumptions and dynamically determining the discrimination coefficient, the model can adapt to the characteristics of different energy consumption data, improving the adaptability and accuracy of the correlation analysis; through the sub-item correlation analysis, the rationality of the sub-item energy consumption reference value can be directly reflected by the data. If an abnormal sub-item energy consumption correlation degree is judged, it means that there is a problem with the setting of the sub-item energy consumption reference value, such as not being updated with the climate change, or the energy consumption deviating due to equipment aging and other problems, and the sub-item energy consumption reference value needs to be calibrated. S2. Preliminary calibration of the energy consumption reference value; Obtain the environmental information and climate information within the preset time range to obtain the environmental information to be analyzed and the climate information to be analyzed; obtain the current status information of the target HVAC equipment to obtain the status information to be analyzed; input the status information to be analyzed, the environmental information to be analyzed, the climate information to be analyzed, and the abnormal sub-item energy consumption of the target HVAC equipment into the energy consumption impact factor judgment model for analysis to obtain the energy consumption reference value impact factor. The energy consumption impact factor judgment model in step S2 includes a feature extraction layer, an internal feature analysis layer, an external feature analysis layer, and a result output layer. The feature extraction layer is used to extract features from the environmental information to be analyzed and the climate information to be analyzed to obtain environmental information features and climate information features; extract features from the status information to be analyzed to obtain status information features. The external feature analysis layer is used to fuse the environmental information features and the climate information features to obtain external fusion features; analyze the external fusion features to obtain external energy consumption impact factors. The internal feature analysis layer is used to analyze the status information features to obtain internal energy consumption impact factors. The result output layer is used to match the corresponding impact factor weights in the sub-item energy consumption factor impact library according to the abnormal sub-item energy consumption of the target HVAC equipment to obtain the target impact factor weights; perform weighted calculation on the external energy consumption impact factors and the internal energy consumption impact factors according to the target impact factor weights to obtain the energy consumption reference value impact factor. The specific steps for training the external feature analysis layer include: Collect several groups of external feature analysis training samples, where each group of external feature analysis training samples contains an external fusion feature and the corresponding marked value of the external influence factor; combine several groups of external feature analysis training samples to obtain an external feature analysis training set; Input the external feature analysis training set into the energy consumption influence factor judgment model to train the external feature analysis layer with the corresponding marked value of the external influence factor as the target, and obtain the initial external feature analysis layer; conduct model evaluation on the initial external feature analysis layer to obtain the model evaluation result of the initial external feature analysis layer; if the model evaluation result of the initial external feature analysis layer passes, then use the initial external feature analysis layer as the external feature analysis layer in the energy consumption influence factor judgment model; otherwise, continue model training using the external feature analysis training set; The specific steps for training the internal feature analysis layer include: Collect several groups of internal feature analysis training samples, where each group of internal feature analysis training samples contains an internal state feature and the corresponding marked value of the internal influence factor; combine several groups of internal feature analysis training samples to obtain an internal feature analysis training set; Input the internal feature analysis training set into the energy consumption influence factor judgment model to train the internal feature analysis layer with the corresponding marked value of the internal influence factor as the target, and obtain the initial internal feature analysis layer; conduct model evaluation on the initial internal feature analysis layer to obtain the model evaluation result of the initial internal feature analysis layer; if the model evaluation result of the initial internal feature analysis layer passes, then use the initial internal feature analysis layer as the internal feature analysis layer in the energy consumption influence factor judgment model; otherwise, continue model training using the internal feature analysis training set; By simultaneously considering environmental, climatic, and equipment status information, a multi-dimensional analysis of energy consumption influencing factors is achieved, improving the accuracy of energy consumption prediction and analysis; the introduction of the external feature analysis layer and the internal feature analysis layer enables the model to comprehensively consider features from different sources and conduct in-depth analysis on them, enhancing the rationality of energy consumption benchmark value calibration; through automated collection of training samples and construction of the training set, the complexity of manual operations is reduced and the efficiency of model training is improved; the process of model evaluation and iterative training ensures the accuracy and effectiveness of the feature analysis layer, and through continuous evaluation and optimization, the performance of the energy consumption influence factor judgment model is improved; through precise feature analysis and weight matching, a more accurate energy consumption benchmark value influence factor is obtained, providing a reliable basis for the calibration of the energy consumption benchmark value; S3. Optimization and calibration of the energy consumption benchmark value; Based on the energy consumption benchmark value influence factor, optimize the calculation of the abnormal target HVAC equipment sub-item energy consumption to obtain the optimized sub-item energy consumption benchmark value D n ’; for the optimized sub-item energy consumption benchmark value D n’ is evaluated. If the evaluation result of the benchmark value passes, the optimized sub - item energy consumption benchmark value D n ’ is used as the current sub - item energy consumption benchmark value D n ’s calibration result; otherwise, continue to optimize the optimized sub - item energy consumption benchmark value D n ’ until it meets the calibration result of the current sub - item energy consumption benchmark value D n ; The specific steps for optimizing the sub - item energy consumption of abnormal target HVAC equipment based on the energy consumption benchmark value impact factor include: Obtain the energy consumption sequence X to be analyzed corresponding to the sub - item energy consumption of the abnormal target HVAC equipment n , and obtain the abnormal energy consumption sequence; conduct abnormal judgment on the abnormal energy consumption sequence to obtain the optimal screened energy consumption sequence; construct K individual optimized sub - item energy consumption benchmark values H k ; Each individual optimized sub - item energy consumption benchmark value H k contains a group of energy consumption benchmark values to be optimized established based on the current sub - item energy consumption benchmark value D n and the energy consumption benchmark value impact factor; Combine the K individual optimized sub - item energy consumption benchmark values H k to obtain the optimized sub - item energy consumption benchmark value iteration population; Set the maximum number of iterations T, and set the current number of iterations as t; The maximum number of iterations T is set by professional and technical personnel according to the actual situation; Abnormal judgment means manually screening out the energy consumption with real abnormalities and retaining the energy consumption values misjudged due to errors in the benchmark value; The generation of the initial individual uses the impact factor to calculate the current sub - item energy consumption benchmark value D n , delimit the data fluctuation range, and randomly generate the energy consumption benchmark value to be optimized within the data fluctuation range; Conduct simulation calculations on the individual optimized sub - item energy consumption benchmark values H k in the optimized sub - item energy consumption benchmark value iteration population, conduct predictive analysis according to the optimal screened energy consumption sequence to obtain the predicted energy consumption sequence; Conduct regression analysis based on the predicted energy consumption sequence to obtain the optimal predicted energy consumption mean value; Calculate the Euclidean distance between the optimal predicted energy consumption mean value and the energy consumption benchmark value to be optimized in the individual optimized sub - item energy consumption benchmark value H k to obtain the energy consumption benchmark difference C k ; Take the reciprocal of the energy consumption benchmark difference C k as the fitness S k of the individual optimized sub - item energy consumption benchmark value H k ; During the population iteration process, iterate and update the optimized sub - item energy consumption benchmark value iteration population; The specific steps for iteratively updating the optimized sub - item energy consumption benchmark value iteration population include: Divide the optimized sub - item energy consumption benchmark value iterative population into two parts to obtain the first optimized sub - item energy consumption benchmark value iterative population and the second optimized sub - item energy consumption benchmark value iterative population; Set the iteration parameter U, U = exp( - t / T); exp() is the natural exponential function; set the iteration factor V, V = exp((t - T) / T) / 2; When V < L1, the first optimized sub - item energy consumption benchmark value iterative population and the second optimized sub - item energy consumption benchmark value iterative population are randomly exchanged and updated; When U < L2, the first optimized sub - item energy consumption benchmark value iterative population and the second optimized sub - item energy consumption benchmark value iterative population are updated to the optimized sub - item energy consumption benchmark value individuals with larger fitness; When V > L1 and U > L2, select the optimized sub - item energy consumption benchmark value individuals with larger fitness in the first optimized sub - item energy consumption benchmark value iterative population and the second optimized sub - item energy consumption benchmark value iterative population for mutation to obtain a new optimized sub - item energy consumption benchmark value iterative population; The specific values of L1 and L2 are set by professional technicians according to the actual situation. For example, L1 is taken as 0.3 and L2 is taken as 0.7; Using two parameters to iteratively update the population can divide the iterative process into different stages. Referring to the foraging, feeding, and mating stages of the snake population to update the iterative population, approaching the optimal solution in a reasonable update manner, avoiding falling into the local optimal solution in the early stage, and at the same time improving the convergence speed; When the maximum number of iterations is reached, output the optimized sub - item energy consumption benchmark value individual H corresponding to the maximum fitness k which is the optimal optimized sub - item energy consumption benchmark value individual; take the energy consumption benchmark value to be optimized in the optimal optimized sub - item energy consumption benchmark value individual as the optimized sub - item energy consumption benchmark value D n ’; Through optimization calculations, it is ensured that the energy use of HVAC equipment reaches the highest efficiency, reducing unnecessary energy waste, realizing the dynamic management and continuous optimization of the energy consumption benchmark value; by optimizing the abnormal energy consumption sequence, it can more accurately reflect the actual energy consumption status of HVAC equipment. The optimized sub - item energy consumption benchmark value helps to identify energy consumption abnormal points, thereby taking measures to reduce energy consumption; adopting an automated iterative optimization method reduces the need for manual intervention and improves work efficiency.

[0030] It should be understood that for those of ordinary skill in the art, improvements or transformations can be made according to the above description, and all such improvements and transformations should fall within the protection scope of the appended claims of the present invention. The parts not described in detail in this specification belong to the well - known prior art of those skilled in the art.

Claims

1. A calibration method for the sub-item energy consumption benchmark value of heating, ventilation and air conditioning equipment, characterized in that Including: S1. Abnormal judgment of energy consumption baseline value; Obtain the actual energy consumption sequences of different sub-items within a preset time range to get the energy consumption sequence X to be analyzed n , X n = [X n（1） , X n（2） , …, X n（I） ; N is the number of sub-item energy consumption categories of the target HVAC equipment, and X n（i） represents the i-th sub-item energy consumption of the target HVAC equipment in the energy consumption sequence X n . n = 1, 2, …, N; I represents the number of actual energy consumption values in the energy consumption sequence X n to be analyzed within the preset time range, and i = 1, 2, …, I; Based on the energy consumption sequence X to be analyzed n and the current sub-item energy consumption benchmark value D n perform correlation analysis to obtain the sub-item energy consumption correlation Z n ; if the sub-item energy consumption correlation Z n reaches the energy consumption correlation threshold, the energy consumption anomaly result is normal, and the current sub-item energy consumption benchmark value D n is retained; otherwise, the energy consumption anomaly result is abnormal, and the sub-item energy consumption correlation Z n corresponding target HVAC equipment sub-item energy consumption X n（i） is the abnormal target HVAC equipment sub-item energy consumption, and proceed to step S2; S2. Preliminary calibration of energy consumption baseline value; Obtain the environmental information and climate information within a preset time range to obtain the environmental information to be analyzed and the climate information to be analyzed; Obtain the current status information of the target HVAC equipment to obtain the status information to be analyzed; input the status information to be analyzed, the environmental information to be analyzed, the climate information to be analyzed, and the abnormal target HVAC equipment sub-item energy consumption into the energy consumption impact factor judgment model for analysis to obtain the energy consumption baseline value impact factor; S3. Optimized calibration of energy consumption baseline value; Optimize the sub - item energy consumption of abnormal target HVAC equipment based on the influencing factors of the energy consumption baseline value to obtain the optimized sub - item energy consumption baseline value D n ’; Evaluate the optimized sub - item energy consumption baseline value D n ’. If the evaluation result of the baseline value passes, then use the optimized sub - item energy consumption baseline value D n ’ as the calibration result of the current sub - item energy consumption baseline value D n . Otherwise, continue to optimize the optimized sub - item energy consumption baseline value D n ’ until it meets the calibration result of the current sub - item energy consumption baseline value D n .

2. The calibration method for the sub-item energy consumption baseline value based on the HVAC equipment according to claim 1, wherein The specific steps for calculating the correlation degree Z of sub-item energy consumption in step S1 n include: For the sub-item energy consumption X of the target HVAC equipment n（i） perform the calculation: Q1. Calculate the correlation coefficient G of the sub-item energy consumption n（i） ; Using the formula Calculate the correlation coefficient G of sub-item energy consumption n（i） , the correlation coefficient G of sub-item energy consumption n（i） represents the correlation coefficient of the energy consumption sequence X n in X n（i） ; R A（n） represents the two-level minimum difference, R A（n） =minminY n（i） ; R B（n） represents the two-level maximum difference, R B（n） =maxmaxY n（i） ; Y n（i） represents the absolute value of the difference between the current sub-item energy consumption reference value D n and X n（i） , Y n（i） =|D n -X n（i） |; E is the resolution coefficient, which is confirmed by the sub-item energy consumption X of the target HVAC equipment n（i） ; Q2. Determine the weight W of the correlation coefficient of sub-item energy consumption n ; Using the nine-level scale method for all target HVAC equipment sub-item energy consumptions X n（i） Establish a sub-item energy consumption evaluation matrix; calculate the eigenvalues for the sub-item energy consumption evaluation matrix to obtain the maximum eigenvalue of the sub-item energy consumption evaluation matrix; use the sub-item energy consumption evaluation matrix and the maximum eigenvalue of the sub-item energy consumption evaluation matrix to calculate the eigenvector to obtain the eigenvector of the sub-item energy consumption evaluation matrix; perform normalization processing on the eigenvector of the sub-item energy consumption evaluation matrix to obtain the preprocessed eigenvector of the sub-item energy consumption evaluation matrix; the nth item in the preprocessed eigenvector of the sub-item energy consumption evaluation matrix is the sub-item energy consumption correlation coefficient weight W n ; Q3. Calculate the correlation degree Z of sub-item energy consumption n ; Using the formula Calculate the sub - item energy consumption correlation degree Z n .

3. The calibration method for itemized energy consumption baseline values based on heating, ventilation, and air conditioning equipment according to claim 2, wherein Based on the sub-item energy consumption X of the target heating, ventilation, and air conditioning equipment n（i） The specific steps for determining the discrimination coefficient E include: For the sub-item energy consumption X of the target HVAC equipment n（i） and the sub-item energy consumption X of the target HVAC equipment n’（i） , using the formula The discrimination coefficient E is calculated （n，n’） , the discrimination coefficient E （n，n’） represents the influence between the sub - item energy consumptions X of different target HVAC equipment at the same moment n（i） and the sub - item energy consumption X of the target HVAC equipment n’（i） ; the sub - item energy consumption X of the target HVAC equipment n（i） and the sub - item energy consumption X of the target HVAC equipment n’（i） represent any two sub - item energy consumptions X of the target HVAC equipment n（i） for which the discrimination coefficient is calculated, indicating the energy consumption deviation degree between the sub - item energy consumptions of different categories of the target HVAC equipment within the same moment; R B（n，n’） represents R B（n） and R B（n’） the maximum value between them; A total of n*(n - 1) / 2 groups of resolution coefficients E are calculated （n，n’） ; Calculate the mean value of all discrimination coefficients E （n，n’） and determine the discrimination coefficient E based on the interval where the mean value lies.

4. The calibration method for sub-item energy consumption benchmark values based on heating, ventilation, and air conditioning equipment according to claim 3, wherein The energy consumption impact factor judgment model in step S2 includes a feature extraction layer, an internal feature analysis layer, an external feature analysis layer, and a result output layer; The feature extraction layer is used to extract features from the environmental information to be analyzed and the climate information to be analyzed to obtain environmental information features and climate information features; Extract features from the status information to be analyzed to obtain status information features; The external feature analysis layer is used to perform feature fusion on the environmental information features and the climate information features to obtain external fusion features; perform feature analysis on the external fusion features to obtain external energy consumption impact factors; The internal feature analysis layer is used to perform feature analysis on the status information features to obtain internal energy consumption impact factors; The result output layer is used to match the corresponding impact factor weights in the sub-item energy consumption factor impact library according to the abnormal target HVAC equipment sub-item energy consumption to obtain the target impact factor weights; Perform weighted calculation on the external energy consumption impact factor and the internal energy consumption impact factor according to the target impact factor weights to obtain the energy consumption baseline value impact factor.

5. The calibration method for the sub-item energy consumption benchmark value based on HVAC equipment according to claim 4, characterized in that, The specific steps for training the external feature analysis layer include: Collect several groups of external feature analysis training samples, each group of external feature analysis training samples contains an external fusion feature and the corresponding marked external impact factor value; combine several groups of external feature analysis training samples to obtain an external feature analysis training set; Input the external feature analysis training set into the energy consumption impact factor judgment model to train the external feature analysis layer with the corresponding marked external impact factor value as the target to obtain an initial external feature analysis layer; perform model evaluation on the initial external feature analysis layer to obtain the initial external feature analysis layer model evaluation result; if the initial external feature analysis layer model evaluation result is passed, use the initial external feature analysis layer as the external feature analysis layer in the energy consumption impact factor judgment model; otherwise, continue to perform model training using the external feature analysis training set.

6. The method for calibrating the sub-item energy consumption benchmark value based on the heating, ventilation and air conditioning equipment according to claim 5, characterized in that The specific steps for training the internal feature analysis layer include: Collect several groups of internal feature analysis training samples, each group of internal feature analysis training samples contains an internal status feature and the corresponding marked internal impact factor value; combine several groups of internal feature analysis training samples to obtain an internal feature analysis training set; Input the internal feature analysis training set into the energy consumption impact factor judgment model to train the internal feature analysis layer with the internal impact factor values corresponding to the tags as the target, and obtain the initial internal feature analysis layer; perform model evaluation on the initial internal feature analysis layer to obtain the model evaluation result of the initial internal feature analysis layer; if the model evaluation result of the initial internal feature analysis layer is passed, use the initial internal feature analysis layer as the internal feature analysis layer in the energy consumption impact factor judgment model; otherwise, continue model training using the internal feature analysis training set.

7. The calibration method for the sub-item energy consumption benchmark value based on HVAC equipment according to claim 6, wherein, The specific steps for optimizing the sub-item energy consumption of abnormal target HVAC equipment based on the energy consumption benchmark value impact factor include: Obtain the energy consumption sequence X to be analyzed corresponding to the abnormal target HVAC equipment's sub-item energy consumption n , and obtain the abnormal energy consumption sequence; perform abnormal judgment on the abnormal energy consumption sequence to obtain the optimal screened energy consumption sequence; construct K optimized sub-item energy consumption benchmark value individuals H k ; each optimized sub-item energy consumption benchmark value individual H k contains a group of energy consumption benchmark values to be optimized established based on the current sub-item energy consumption benchmark value D n and the energy consumption benchmark value influence factor; combine the K optimized sub-item energy consumption benchmark value individuals H k to obtain the optimized sub-item energy consumption benchmark value iterative population; set the maximum number of iterations T, and set the current number of iterations as t; For the optimized sub-item energy consumption benchmark value individual H in the optimized sub-item energy consumption benchmark value iteration population k Perform simulation calculations, conduct predictive analysis based on the optimal screening energy consumption sequence to obtain the predicted energy consumption sequence; perform regression analysis based on the predicted energy consumption sequence to obtain the optimal predicted energy consumption mean; calculate the Euclidean distance between the optimal predicted energy consumption mean and the energy consumption benchmark value to be optimized in the optimized sub-item energy consumption benchmark value individual H k to obtain the energy consumption benchmark difference C k ; Take the reciprocal of the energy consumption benchmark difference C k as the fitness S of the optimized sub-item energy consumption benchmark value individual H k ; k ​ During the population iteration process, iteratively update the optimized sub-item energy consumption benchmark value iteration population. When the maximum number of iterations is reached, output the individual H of the optimized sub-item energy consumption benchmark value corresponding to the maximum fitness k , which is the optimal individual of the optimized sub-item energy consumption benchmark value; use the energy consumption benchmark value to be optimized in the optimal individual of the optimized sub-item energy consumption benchmark value as the optimized sub-item energy consumption benchmark value D n ’.

8. The calibration method for the sub-item energy consumption benchmark value based on HVAC equipment according to claim 7, characterized in that, The specific steps for iteratively updating the optimized sub-item energy consumption benchmark value iteration population include: Divide the optimized sub-item energy consumption benchmark value iteration population into two parts to obtain the first optimized sub-item energy consumption benchmark value iteration population and the second optimized sub-item energy consumption benchmark value iteration population. Set the iteration parameter U, U = exp(-t / T); exp() is the natural exponential function; set the iteration factor V, V = exp((t - T) / T) / 2. When V < L1, randomly exchange and update the first optimized sub-item energy consumption benchmark value iteration population and the second optimized sub-item energy consumption benchmark value iteration population. When U < L2, update the first optimized sub-item energy consumption benchmark value iteration population and the second optimized sub-item energy consumption benchmark value iteration population to the optimized sub-item energy consumption benchmark value individuals with larger fitness. When V > L1 and U > L2, select the optimized sub-item energy consumption benchmark value individuals with larger fitness in the first optimized sub-item energy consumption benchmark value iteration population and the second optimized sub-item energy consumption benchmark value iteration population for mutation to obtain a new optimized sub-item energy consumption benchmark value iteration population.

Images