Water chilling unit performance evaluation method and system

Abstract

The embodiment of the invention provides a performance evaluation method and system of a water chilling unit, electronic equipment, a computer readable storage medium and a computer program product.The performance evaluation method of the water chilling unit comprises the steps that on the basis of historical operation data of a specific refrigeration system, the typical operation condition of the system is identified, the system comprises a water chilling unit. Selecting a static empirical model for predicting the performance of the water chilling unit, and fitting parameters of the model based on historical operation data; and inputting the data of the typical operation condition into the model after parameter fitting, and generating a predicted value and a predicted interval of the performance of the water chilling unit under the typical operation condition, and under the typical operation condition, the actual observed value of the performance of the water chilling unit is located in the predicted interval with a predetermined confidence coefficient.

Description

Technical Field

[0001] This disclosure relates to the field of refrigeration equipment, and more specifically, to a method for evaluating the performance of a chiller unit, a system for evaluating the performance of a chiller unit, an electronic device, a computer-readable storage medium, and a computer program product. Background Technology

[0002] In the global energy demand and carbon emission landscape, buildings account for approximately 30% of energy consumption, with HVAC systems contributing half of that. Compared to large-scale system retrofits, accurately predicting energy demand based on existing equipment and developing reasonable operating strategies for cooling systems can achieve energy savings at a lower cost. In cooling systems such as large commercial buildings or data centers, chillers typically consume more energy than other components like cooling towers and pumps, accounting for roughly 40% or more of the total energy consumption of the cooling equipment.

[0003] Therefore, optimizing the configuration and operation of chiller units to improve their efficiency has become one of the research hotspots in the field of refrigeration equipment. Summary of the Invention

[0004] Embodiments of this disclosure provide methods and systems for evaluating the performance of chiller units, electronic devices, computer-readable storage media, and computer program products that can at least partially solve the aforementioned technical problems or other problems.

[0005] In a first aspect, this disclosure provides a method for evaluating the performance of a chiller unit. The evaluation method includes: identifying typical operating conditions of a system based on historical operating data of a specific refrigeration system, wherein the system includes a chiller unit; selecting a static empirical model for predicting the performance of the chiller unit, and fitting the parameters of the model based on historical operating data; and inputting the data of the typical operating conditions into the model after fitting the parameters to generate predicted values ​​and prediction intervals of the performance of the chiller unit under the typical operating conditions, wherein the actual observed values ​​of the performance of the chiller unit under the typical operating conditions are within the prediction interval with a predetermined confidence level.

[0006] In some embodiments of this disclosure, generating a prediction interval includes: generating a prediction interval by quantifying the uncertainty factors of the model after fitting the parameters, wherein the uncertainty factors include data noise, parameter fitting errors, and errors caused by the lack or sparsity of historical operating data representing all operating conditions of the system.

[0007] In some embodiments of this disclosure, historical operating data includes historical data at the chiller unit level. The method further includes: determining adaptive chiller unit performance values ​​based on the proportion of predicted values ​​and data under typical operating conditions in the historical data at the chiller unit level; and evaluating the performance of multiple chiller units within a specific refrigeration system based on the adaptive chiller unit performance values.

[0008] In some embodiments of this disclosure, evaluating the performance of multiple chillers within a specific refrigeration system includes: determining adaptive chiller performance values ​​for the multiple chillers; determining the average of the adaptive chiller performance values ​​for the multiple chillers; and identifying the chillers corresponding to adaptive chiller performance values ​​that deviate from the average.

[0009] In some embodiments of this disclosure, evaluating the performance of multiple chillers within a specific refrigeration system includes: generating a performance table of the chillers under typical operating conditions; periodically updating the performance table of the chillers under typical operating conditions; and determining the performance status of the chillers based on the performance table of the chillers under typical operating conditions, and comparing the performance differences of multiple chillers, wherein the performance table of the chillers under typical operating conditions includes predicted values ​​and prediction ranges for the chillers under different typical operating conditions.

[0010] In some embodiments of this disclosure, evaluating the performance of multiple chillers within a specific refrigeration system includes: generating curves based on predicted values ​​of the chillers under different typical operating conditions, and generating interval bands based on predicted intervals of the chillers under different typical operating conditions; generating curves and interval bands for multiple chillers within the specific refrigeration system, thereby generating a part load factor-coefficient of performance curve; and determining the optimal operating range of the multiple chillers based on the part load factor-coefficient of performance curve, and comparing the performance differences of the multiple chillers.

[0011] In some embodiments of this disclosure, identifying typical operating conditions of a system based on historical operating data of a specific refrigeration system includes: preprocessing the historical operating data, wherein the preprocessing includes: supplementing missing values ​​in the historical operating data; retaining outliers of engineering significance in the historical operating data; normalizing the preprocessed historical operating data; and identifying typical operating conditions based on the normalized historical operating data.

[0012] In some embodiments of this disclosure, identifying typical operating conditions of a system based on historical operating data of a specific refrigeration system further includes: clustering the historical operating data to obtain multiple operating conditions of the system; and determining typical operating conditions among the multiple operating conditions based on established clustering indices.

[0013] In some embodiments of this disclosure, selecting a static empirical model for predicting the performance of a chiller unit includes selecting a model based on the type and accuracy of sensors within the system.

[0014] In some embodiments of this disclosure, the parameters of the model fitted based on historical operating data include: fitting the parameters using a regression analysis method based on historical operating data, wherein the regression analysis method includes the least squares method.

[0015] In some embodiments of this disclosure, the process of inputting data from typical operating conditions into a model fitted with parameters to generate a prediction interval includes: determining the covariance matrix of the parameters and the mean square error of the historical operating data during the process of determining the parameters of the model based on historical operating data; inputting data from typical operating conditions into a model fitted with parameters to generate predicted values; and determining the boundaries of the prediction interval based on the data from typical operating conditions, predicted values, mean square error, parameter covariance matrix, and a predetermined confidence level.

[0016] In some embodiments of this disclosure, the evaluation method further includes: comparing the observed values ​​with at least one of the design ratings provided by the manufacturer, historical performance values ​​from the building management system, and operating values ​​of similar chillers to determine the state of the chiller, wherein the performance state of the chiller includes a normal performance state, a degraded performance state, an abnormal performance state, and a degraded performance state.

[0017] In some embodiments of this disclosure, the evaluation method further includes: determining the prediction interval coverage probability and the average prediction interval width, wherein the prediction interval coverage probability is the proportion of observations in historical operating data falling within the corresponding prediction interval, and the average prediction interval width is the arithmetic mean of the prediction interval widths corresponding to the observations in historical operating data; determining a coverage width-based criterion based on the prediction interval coverage probability and the average prediction interval width; and evaluating the quality of the prediction interval according to the coverage width-based criterion, wherein the value of the coverage width-based criterion is negatively correlated with the quality of the prediction interval.

[0018] Secondly, this disclosure provides a performance evaluation system for a chiller unit. The system includes: a typical operating condition identification unit, used to identify typical operating conditions of the system based on historical operating data of a specific refrigeration system, wherein the system includes a chiller unit; a model building unit, used to select a static empirical model for predicting the performance of the chiller unit and fit the parameters of the model based on historical operating data; and a prediction unit, used to input the data of the typical operating conditions into the model after fitting the parameters, and generate the predicted value and prediction range of the performance of the chiller unit under the typical operating conditions, wherein the actual observed value of the performance of the chiller unit under the typical operating conditions is within the prediction range with a predetermined confidence level.

[0019] Thirdly, embodiments of this disclosure provide an electronic device comprising: a processor; and a memory communicatively connected to the processor; wherein the memory stores instructions executable by the processor, the instructions being executed by the processor to enable the electronic device to implement the performance evaluation method for a chiller unit as described in any implementation of the first aspect.

[0020] Fourthly, embodiments of this disclosure provide a non-transitory computer-readable storage medium storing computer instructions that enable a computer, when executed, to implement the performance evaluation method for a chiller unit as described in any implementation of the first aspect.

[0021] Fifthly, embodiments of this disclosure provide a computer program product including a computer program that, when executed by a processor, can implement the performance evaluation method for a chiller unit as described in any implementation of the first aspect.

[0022] The performance evaluation method and system, electronic device, computer-readable storage medium, and computer program product for chillers provided in this disclosure can identify typical operating conditions in a specific refrigeration system based on historical operating data, and fit parameters of a selected static empirical model based on this historical operating data. By inputting the data of the typical operating conditions into the model after fitting the parameters, predicted values ​​and prediction intervals of the chiller's performance under typical operating conditions can be generated, wherein the future actual observed values ​​of the chiller can be selected with a confidence level falling within this prediction interval. This closely links the performance of the chiller to be evaluated with the specific characteristics of the refrigeration system in which the chiller is located, effectively reducing the deviation between the predicted performance values ​​and the true values ​​of the chiller.

[0023] Furthermore, the prediction interval can be understood as the range of uncertainty of the future measured true value based on the predicted value. Therefore, the embodiments of this disclosure take into account the potential impact of prediction uncertainty in the evaluation process. By quantifying the probability range of these uncertainties, the selected confidence level can give the limit of the future measured true value.

[0024] Furthermore, it is impractical to perform shutdown measurements on chillers already in operation to evaluate the performance of different chillers. The evaluation method and system provided in this disclosure do not require shutdown testing of the refrigeration system containing the chiller under test; performance evaluation can be achieved solely through analysis of historical operating data. Attached Figure Description

[0025] Other features, objects, and beneficial effects of this disclosure will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings. In the drawings: Figure 1A flowchart of a chiller unit performance evaluation method provided for embodiments of this disclosure; Figure 2 This is a schematic diagram of a performance evaluation system for a chiller unit provided according to an exemplary embodiment of this disclosure; Figure 3 This is a partial schematic diagram of a specific refrigeration system provided according to an exemplary embodiment of the present disclosure; Figures 4 to 14 The figure is a data analysis flowchart of the performance evaluation process for a chiller unit in a specific refrigeration system provided according to an exemplary embodiment of the present disclosure; and Figure 15 A schematic block diagram of an example electronic device that can be used to implement embodiments of the present disclosure is shown.

[0026] Specific methods

[0027] To better understand this disclosure, various aspects of this disclosure will be described in more detail with reference to the accompanying drawings. It should be understood that these detailed descriptions are merely illustrative of exemplary embodiments of this disclosure and are not intended to limit the scope of this disclosure in any way. Throughout the specification, the same reference numerals refer to the same elements. The expression "and / or" includes any and all combinations of one or more of the associated listed items.

[0028] It should be noted that in this specification, the terms "first," "second," "third," etc., are used only to distinguish one feature from another and do not imply any limitation on the features, especially not any order of precedence. Therefore, without departing from the teachings of this disclosure, the first chiller unit discussed herein may also be referred to as the second chiller unit, and vice versa.

[0029] In the accompanying drawings, the thickness, dimensions, and shapes of the parts have been slightly adjusted for ease of illustration. The drawings are for illustrative purposes only and are not drawn to scale. As used herein, the terms “approximately,” “about,” and similar terms are used as expressions of approximation, not as expressions of degree, and are intended to illustrate inherent deviations in measured or calculated values ​​that will be recognized by one of ordinary skill in the art.

[0030] It should also be understood that expressions such as "comprising," "including," "having," "containing," and / or "comprising" are open-ended rather than closed-ended expressions in this specification, indicating the presence of the stated features, elements, and / or components, but not excluding the presence of one or more other features, elements, components, and / or combinations thereof. Furthermore, when expressions such as "at least one of..." appear after a list of listed features, they modify the entire list of features, not just individual elements in the list. Additionally, when describing embodiments of this disclosure, the word "may" is used to mean "one or more embodiments of this disclosure." And the term "exemplary" is intended to refer to an example or illustration.

[0031] Unless otherwise specified, all terms used herein (including engineering and technical terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains. It should also be understood that, unless expressly stated in this disclosure, terms as defined in common dictionaries shall be interpreted as having the meaning consistent with their meaning in the context of the relevant art, and not as having an idealized or overly formalized meaning.

[0032] It should be noted that, unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other. Furthermore, unless explicitly limited or contradicted by the context, the specific steps included in the methods described in this disclosure are not limited to the order in which they are described, but can be performed in any order or in parallel.

[0033] Furthermore, in this disclosure, the use of "connection" or "linkage" may indicate a direct or indirect connection between the corresponding components, and the use of "contact" may indicate a direct connection between the corresponding components, unless otherwise expressly defined or deduced from the context.

[0034] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0035] Figure 1 A flowchart illustrating a chiller unit performance evaluation method 1000 provided for embodiments of this disclosure. Figure 1 As shown, this disclosure provides a method 1000 for evaluating the performance of a chiller unit, which may include the following steps: S1: Based on historical operating data of a specific refrigeration system, identify typical operating conditions of the system, which includes a chiller unit.

[0036] S2: Select a static empirical model for predicting the performance of the chiller unit and fit the model parameters based on historical operating data.

[0037] S3: Input the data of typical operating conditions into the model with fitted parameters to generate the predicted value and prediction range of the chiller unit's performance under typical operating conditions. Under typical operating conditions, the actual observed value of the chiller unit's performance is located within the prediction range with a predetermined confidence level.

[0038] Chiller performance evaluation is beneficial for improving energy efficiency and the reliability of refrigeration systems. During the design phase, matching chiller capacity to the annual load curve improves the annual energy efficiency ratio. During operation, optimizing control algorithms to minimize energy consumption or maximize the coefficient of performance (COP) ensures the refrigeration system operates at its best while meeting cooling demands. Furthermore, in refrigeration systems with multiple chillers, chiller sequencing aims to select the appropriate chiller combination for a specific system based on performance evaluations of each chiller under different operating conditions. This is particularly important in refrigeration systems with chillers of different types, capacities, or performance levels, regardless of whether these configuration parameters are due to differences in design capacity or performance degradation over time. Additionally, the results of chiller performance evaluation can be used for predictive maintenance, ensuring timely maintenance or replacement of underperforming chillers. Therefore, accurate prediction and comprehensive evaluation of chiller performance can improve the reliability of the aforementioned studies.

[0039] However, uncertainty in the fitted model is unavoidable due to model errors and the fact that the data used for model fitting does not cover the entire state space. Therefore, relying solely on the average performance prediction of the chiller unit to set the control strategy is insufficient. One of the main limitations of existing research is that it neglects the necessity of incorporating prediction intervals into performance prediction and evaluation, leading to deviations or inaccuracies between the predicted values ​​of the evaluation results and the actual measured observations. In the embodiments of this disclosure, by selecting appropriate confidence levels for these prediction intervals, the potential increase in costs due to deviations in evaluation results can be mitigated.

[0040] Specifically, observed values ​​can be understood as the actual measurement results of the performance of the chiller unit under test, while predicted values ​​can be understood as the estimation results of the performance of the chiller unit under test based on the selected static empirical model. The prediction interval can quantify the range of uncertainty of future observed values ​​based on the predicted values. In other words, the prediction interval is the probability that future observed values ​​will fall within this interval at a selected confidence level. It takes into account the impact of uncertainties such as model error, data noise, and input variables on the uncertainty of chiller unit performance prediction. By quantifying the probability range of these uncertainties, the selected confidence level can provide the boundary of the observed values.

[0041] Conventional chiller performance evaluation metrics, such as Integrated Part Load Value (IPLV) and Nonstandard Part Load Value (NPLV), provide a comprehensive assessment by weighting the Coefficient of Performance (COP) under different load conditions. These are single-valued quality factors representing the equipment's part-load efficiency, calculated based on weighted operation at different part-load capacities, where part load can be understood as the chiller operating below its rated capacity. However, the data required to calculate these metrics typically necessitates experimental measurements under stringent operating conditions. For chillers already in operation, conducting shutdown measurements to evaluate the performance of different units becomes impractical. Furthermore, the weighting in IPLV and NPLV emphasizes generality and therefore does not accurately reflect the actual operating conditions of a specific refrigeration system. Considering that chillers in these specific systems do not always operate under design conditions, the predicted values ​​obtained based on conventional chiller performance evaluation metrics and their evaluation methods may not be consistent with the actual observed values ​​of chillers in a specific refrigeration system.

[0042] To at least address the aforementioned technical problems, the performance evaluation method for chiller units provided in this disclosure identifies typical operating conditions of a specific refrigeration system based on historical operating data, and fits parameters of a selected static empirical model based on this historical operating data. By inputting the data from the typical operating conditions into the model with the fitted parameters, predicted values ​​and prediction intervals of the chiller unit's performance under typical operating conditions can be generated, wherein the future actual observed values ​​of the chiller unit fall within this prediction interval with a selected confidence level. This closely links the performance of the chiller unit to be evaluated with the specific characteristics of the refrigeration system in which the chiller unit is located, effectively reducing the deviation between the predicted performance values ​​and the true values ​​of the chiller unit.

[0043] Furthermore, the prediction interval can be understood as the range of uncertainty in the future measured true value based on the predicted value. Therefore, the embodiments of this disclosure consider the potential impact of prediction uncertainty in the evaluation process. By quantifying the probability range of these uncertainties, the selected confidence level can provide the boundary of the future measured true value. Additionally, for chillers already in operation, it is impractical to perform shutdown measurements to evaluate the performance of different chillers. The evaluation method and system provided by the embodiments of this disclosure do not require shutdown testing of the refrigeration system containing the chiller under test; performance evaluation can be achieved solely through historical operating data analysis.

[0044] Therefore, the performance evaluation method for chiller units according to the present disclosure can evaluate the performance of chiller units for a specific refrigeration system, which helps to optimize chiller unit sequencing strategies, participate in refrigeration system demand response, and troubleshooting.

[0045] It should be noted that historical operating data comes from a specific refrigeration system, which may include historical data at the chiller unit level and historical data at the refrigeration system level. A specific refrigeration system may include multiple chiller units or only a single chiller unit. The historical data at the chiller unit level can be the dedicated historical data for each chiller unit extracted from the total historical data of the system. When the refrigeration system includes multiple chiller units, their configuration parameters may differ due to variations in design capacity or performance degradation over time. Therefore, using the dedicated historical data for each chiller unit for individual prediction is essential for accurate prediction and comprehensive evaluation of chiller unit performance, and for improving the reliability of the prediction.

[0046] Furthermore, typical operating conditions identified based on historical operating data can include at least one of system-level typical operating conditions and chiller-level typical operating conditions. System-level typical operating conditions can be generated based on historical data at the refrigeration system level, reflecting the operating patterns of a specific refrigeration system and used to optimize group control strategies. Chiller-level typical operating conditions can be generated based on the specific historical data of a single chiller unit, reflecting the performance characteristics of a specific chiller unit and used for health diagnosis of that chiller unit and horizontal comparison of multiple chillers within a specific refrigeration system.

[0047] Step S1

[0048] In some embodiments of this disclosure, step S1, which identifies typical operating conditions of a system based on historical operating data of a specific refrigeration system, may include: preprocessing the historical operating data, wherein the preprocessing includes: supplementing missing values ​​in the historical operating data; retaining outliers of engineering significance in the historical operating data; normalizing the preprocessed historical operating data; and identifying typical operating conditions based on the normalized historical operating data.

[0049] Specifically, after collecting operational data from sensors, the historical operational data can be preprocessed. This preprocessing may include: supplementing missing values ​​and detecting and processing outliers, where engineering-significant outliers in the historical operational data can be retained. Alternatively, outliers in the historical operational data can be evaluated, retaining values ​​that represent system faults or genuine anomalies, and removing invalid data caused by sensor errors or transient disturbances.

[0050] Alternatively, missing values ​​in historical operating data can be supplemented using at least one of interpolation, physical inference, and context-based exclusion methods. Alternatively, after detecting outliers in historical operating data, these outliers need to be evaluated to determine whether they represent a refrigeration system malfunction or an anomaly, and accordingly, the outlier should be retained or deleted. In other words, the retention or deletion of outliers during the preprocessing of historical operating data should be determined based on their engineering significance. For example, if the outlier represents a refrigeration system malfunction, it can be retained; if the outlier originates from data acquisition noise, such as random interference, it can be deleted; if the outlier is caused by special operating conditions such as extreme weather or emergency load switching, it can be retained.

[0051]

[0052] Table 1

[0053] As an example, Table 1 shows the data labels for the historical operating dataset of the chiller unit and summarizes the potential forms of data anomalies, with key information marked with an asterisk (…). The data is labeled to infer the operating status of the chilled water loop and condensing water loop. Data preprocessing may include imputing missing values ​​in flow meter readings; removing outliers caused by unstable pump operation; removing data points generated during the first twenty minutes of chiller startup due to refrigeration system instability; and removing unrecoverable power data. The processed data will be used for subsequent cluster analysis.

[0054] Alternatively, preprocessing historical operating data may include: supplementing missing values ​​in the historical operating data through at least one of interpolation, physical inference, and context-based exclusion methods; evaluating outliers in the historical operating data to determine whether they characterize a refrigeration system failure or an anomaly, and selecting whether to retain or delete the outlier accordingly. Preprocessing historical operating data for a specific refrigeration system can reduce noise in that historical operating data.

[0055] Optionally, before identifying typical operating conditions of a specific refrigeration system based on its historical operating data, the historical operating data can be normalized. Alternatively, after preprocessing the historical operating data, normalization can be performed to optimize its distribution, improve the efficiency of subsequent clustering machine learning algorithms, and help identify different historical operating conditions.

[0056] In some embodiments of this disclosure, step S1, which identifies typical operating conditions of a system based on historical operating data of a specific refrigeration system, may further include: clustering the historical operating data to obtain multiple operating conditions of the system; and determining typical operating conditions among the multiple operating conditions based on established clustering indices.

[0057] Specifically, after normalizing historical operating data, the data can be categorized to generate multiple operating conditions for a specific refrigeration system. For small datasets with simple operating condition transitions, histogram statistics can be used; however, for large datasets with complex operating conditions, clustering algorithms are more suitable. Clustering is a data analysis process that divides a given set of samples into multiple "clusters" based on their characteristics or distance similarity.

[0058] Alternatively, the k-means clustering method can be used to classify the historical dataset. The k-means method aims to classify... n The observations are divided into k Clusters, of which k , n All are positive integers, and k < n In addition, the square of the Euclidean distance can be used to represent the distance between samples, the centroid or mean of the samples can be used to represent the category, and the sum of the squares of the distance between the sample and the centroid of its cluster can be used as the optimization objective function. The relationship can be expressed by formula (1).

[0059]

[0060] in, Indicates the first i The data point belongs to the th data point l Clusters, It is a cluster l The centroid can also be understood as the average value of all data points assigned to that cluster. Representing data points With the centroid of its cluster The square of the Euclidean distance between them.

[0061] Optionally, after clustering the dataset to obtain multiple operating conditions for a specific refrigeration system, the effectiveness of these clusters can be evaluated using established clustering metrics to identify typical operating conditions. Optionally, organizing the categorized data into intuitive representations such as graphical representations can simplify analysis and interpretation.

[0062] For example, criteria such as the Calinski-Harabasz criterion or the Davies-Bouldin criterion can be used to evaluate the effectiveness of these clusters and determine the optimal number of clusters. The number of clusters can be understood as the total number of operating condition categories of a specific refrigeration system divided from its historical operating data. The optimal number of clusters can be understood as the number of clusters that satisfy both mathematical optimality and engineering interpretability, or the typical operating conditions of the specific refrigeration system.

[0063] Step S2

[0064] Traditional methods, when selecting performance prediction models for chiller units (e.g., multinomial regression, neural networks, etc.), often only focus on the average prediction accuracy, neglecting the uncertainties of the static empirical models used to predict chiller power consumption. These uncertainties can include: input data noise caused by sensor measurement errors or outliers; fitting errors of model parameters; and insufficient data coverage due to historical operating data not encompassing all possible operating conditions.

[0065] For example, because historical operating data does not cover all possible operating conditions, it is impossible to classify all operating conditions of a specific refrigeration system from this historical operating data; or, the historical operating data used for model parameter fitting may not cover the entire state space, resulting in significant prediction uncertainty in sparse regions of historical operating data. These uncertainties can be understood as errors caused by the lack or sparse representation of all operating conditions of a specific refrigeration system by historical operating data.

[0066] In some embodiments of this disclosure, step S2, selecting a static empirical model for predicting the performance of the chiller unit, may include: selecting a static empirical model for predicting the performance of the chiller unit based on the type and accuracy of sensors within a specific refrigeration system.

[0067] In addition, step S2, which involves fitting the model parameters based on historical operating data, may include: fitting the parameters using regression analysis methods based on historical operating data, wherein the regression analysis methods include the least squares method.

[0068] Specifically, a suitable static empirical model for chiller performance prediction can be selected; parameters for fitting this model based on historical operating data can be determined by quantifying parameter uncertainty. Furthermore, the performance of the static empirical model can be validated by comprehensively evaluating its fitting accuracy and the quality of the prediction interval construction. This process can generate prediction intervals simultaneously with chiller performance prediction. Constructing appropriate prediction intervals helps to achieve a balance between prediction accuracy and coverage, ensuring the robustness and applicability of the adaptive chiller performance value-based evaluation method under various operating scenarios and requirements.

[0069] Table 2 shows the static empirical models that can be used to predict chiller power consumption, and the variables that may be involved in each static empirical model are indicated by "√". The variables of each static empirical model may include PLR (Part load ratio) Q actual (Actual cooling capacity) T cws (Condenser inlet water temperature) T cwr (Condenser outlet water temperature) T chws (Evaporator outlet water temperature) At least one of them.

[0070] As shown in Table 2, some embodiments of this disclosure summarize static empirical models that can be used to predict the power consumption of chillers and the variables of each static empirical model. Alternatively, a suitable static empirical model can be selected for prediction based on the type and accuracy of sensors within the refrigeration system. For example, if the accuracy of a certain sensor in the refrigeration system containing the chiller is low, a model that does not rely on the accurate data of that sensor can be selected to improve the accuracy of the performance evaluation prediction of the chiller. However, embodiments of this disclosure do not compare the performance of different static empirical models for the following reasons: the refrigeration system containing the chiller under test typically contains only one type of chiller; the performance differences of the six different static empirical models shown in Table 2 largely stem from their adaptability to the characteristics of different types of chillers; however, for the same type of chiller, the prediction trends of different static empirical models are similar, and the comparison results have no decision-making value; furthermore, the accuracy of sensors within the refrigeration system limits the selection and performance of static empirical models.

[0071] As an alternative, a suitable static empirical model can be selected based on the accuracy of the sensors within the refrigeration system. Considering the accuracy of the sensors within the refrigeration system, the following section will describe an evaluation method based on adaptive chiller unit performance values, using static empirical model 6 as an example. The evaluation index is the adjusted R... 2And the root mean square error (RMSE).

[0072] Optionally, regression analysis methods can be used to fit the parameters of a static empirical model based on historical operating data of a specific refrigeration system, wherein the regression analysis method includes the least squares method.

[0073] As shown in formula (2), the adjusted coefficient of determination R 2 It can be represented as It takes into account the number of predictors in a static empirical model and adjusts the degrees of freedom.

[0074]

[0075] in, Indicates the number of observations. This indicates the number of regression coefficients. Represents the sum of squared errors. It represents the sum of squares.

[0076] As shown in formula (3), the root mean square error (RMSE) can quantify the standard deviation of the residuals and measure the predictive accuracy of the static empirical model:

[0077] in, Indicates the observed or actual value. This represents the predicted value from the static empirical model. This indicates the number of observations.

[0078] In other words, in this embodiment of the present disclosure, the SSE is minimized by the least squares method to determine the parameters of the selected static empirical model, and then the adjusted R² and RMSE are calculated to evaluate the quality of the model.

[0079]

[0080] Table 2

[0081] Step S3

[0082] Refer again Figure 1In some embodiments of this disclosure, step S3, which involves inputting data from typical operating conditions into the model after fitting parameters to generate predicted values ​​and prediction intervals for the performance of the chiller unit under typical operating conditions, may include: determining the covariance matrix of the parameters and the mean square error of the historical operating data during the process of determining the parameters of the model based on historical operating data; inputting data from typical operating conditions into the model after fitting parameters to generate predicted values; and determining the boundaries of the prediction interval based on the data from typical operating conditions, predicted values, mean square error, parameter covariance matrix, and a predetermined confidence level.

[0083] Specifically, as described above, traditional methods, when selecting performance prediction models for chiller units (e.g., multinomial regression, neural networks, etc.), often only focus on the average prediction accuracy, neglecting the uncertainties of the static empirical models used to predict chiller power consumption. The uncertainties of static empirical models can include: input data noise caused by sensor measurement errors or outliers; fitting errors of model parameters; and insufficient data coverage due to historical operating data not covering all possible operating conditions.

[0084] Therefore, embodiments of this disclosure propose constructing prediction intervals (PIs) to quantify the uncertainty of the static empirical model used to predict the power consumption of chiller units, so that the selected static empirical model provides not only the predicted value, but also the confidence range of the predicted value.

[0085] For example, model identification may include: selecting a static empirical model for chiller power prediction and fitting the parameters of the model based on historical operating data; and constructing a prediction range by identifying uncertainties in the static empirical model.

[0086] Optionally, generating the prediction interval may include: generating the prediction interval by quantifying the uncertainty factors of the model after fitting the parameters, wherein the uncertainty factors may include data noise, parameter fitting error, and errors caused by the lack or sparsity of historical operating data in representing all operating conditions of the system.

[0087] Optionally, for the values ​​of the predictor variables New observations at [location] It can be expressed as formula (4).

[0088] =

[0089] in, To predict variable values Fitted values ​​at time, This represents the relevant error.

[0090] It should be noted that, This can be understood as when the predictor variable takes the value of At that time, the actual observed values ​​of the chiller unit, such as the chiller unit's power or COP. This can be understood as when the predictor variable takes the value of At that time, the predicted value is obtained based on the static empirical model. This can be understood as prediction error, which includes the uncertainty factors of the static empirical model and can be used to construct prediction intervals.

[0091] Furthermore, to maintain universality, this disclosure also provides a more general prediction boundary expression, where the prediction boundary can be understood as the upper and lower limits of the prediction interval. The prediction boundary of the new observation can be given by formula (5), in other words, the new observation... It will fall on the predicted boundary with a predetermined probability. Inside.

[0092]

[0093] in, The fitted value can be understood as the predicted value obtained based on a static empirical model. Mean square error, It depends on the confidence level and is calculated as the inverse function of the cumulative distribution function (CDF) of the Student's t distribution. The covariance matrix is ​​the coefficient estimate. A row vector of the design matrix or Jacobian matrix to be evaluated at specified predictor variable values.

[0094] Alternatively, this seemingly complex process can be efficiently accomplished using various tools, such as the Python libraries SciPy and scikit-learn, as well as the MATLAB curve fitting toolbox.

[0095] Furthermore, embodiments of this disclosure also provide a method for evaluating prediction intervals. Specifically, the method for evaluating prediction intervals may include: determining a prediction interval coverage probability and an average prediction interval width, wherein the prediction interval coverage probability is the proportion of observations in historical operational data falling within the corresponding prediction interval, and the average prediction interval width is the arithmetic mean of the prediction interval widths corresponding to observations in historical operational data; determining a coverage width-based criterion based on the prediction interval coverage probability and the average prediction interval width; and evaluating the quality of the prediction interval based on the coverage width-based criterion, wherein the value of the coverage width-based criterion is negatively correlated with the quality of the prediction interval.

[0096] Specifically, the prediction boundaries mentioned above define the lower and upper limits of the prediction interval. The prediction interval considers a wider range of uncertainty sources and can be used to quantify uncertainties in the chiller power prediction process. The coverage probability and width of the prediction interval can represent its reliability and sharpness, respectively. To evaluate the spontaneous metrics related to the quality of the constructed prediction interval, N Each predicted value PICP (Prediction Interval Coverage Probability) can be defined as Equation (6).

[0097] Among them, if the predicted value ,but ;otherwise, . It is the first i The lower limit of a prediction interval; It is the first i The upper limit of the prediction interval. If empirical... If the predicted values ​​are much lower than the nominal confidence level preset for that prediction interval, it indicates that the constructed prediction interval lacks reliability. The nominal confidence level can also be understood as the preset confidence level. If all predicted values ​​fall within the range of all prediction intervals, It is 100%.

[0098] Furthermore, as shown in Equation (7), considering that a wide prediction interval cannot convey any information about changes in the target value, MPIW (Mean Prediction Interval Width) can be used to quantify the width of the prediction interval:

[0099] Alternatively, MPIW can be normalized for objective comparison of prediction intervals. Furthermore, given the target width R, NMPIW (Normalized MPIW) can be given by formula (8). Measure the proportion of actual samples that fall within the predicted interval. The average width between the upper and lower bounds of the quantized prediction interval. Optionally, for a given confidence level, a higher PICP and a lower NMPIW indicate better prediction performance. Because and These are contradictory indicators, therefore they can also be used. (Coverage Width-based Criterion) serves as a more comprehensive evaluation to balance and Of these two metrics, CWC is a negative indicator; the smaller the value, the better the predictive performance.

[0100]

[0101] in, It can be a value between 50 and 100; For depend on and The step function. This can be shown by formula (10).

[0102] in, Set to equal the Prediction Interval Nominal Confidence (PINC).

[0103] The metrics described above will be used below to evaluate the reliability of the prediction intervals generated by the static empirical model used. Furthermore, these metrics can also be used to compare the reliability of prediction intervals from different static empirical models, aiding in the selection of a static empirical model.

[0104] The weights of conventional chiller performance evaluation indicators such as IPLV and NPLV emphasize versatility, are set based on industry standards or general scenarios, and are not targeted at the actual operating characteristics of specific systems.

[0105] To address at least the aforementioned technical problems, embodiments of this disclosure propose a new index, the Adaptive Chiller Performance Value (ACPV), as well as an evaluation method and system based on the ACPV. The ACPV can be weighted based on the proportion of typical operating conditions identified from chiller operating data, ensuring that the evaluated chiller performance closely aligns with the specific characteristics of the refrigeration system.

[0106] Specifically, in some embodiments of this disclosure, adaptive chiller performance values ​​are determined based on the proportion of predicted values ​​and typical operating conditions in historical data at the chiller unit level; and the performance of multiple chillers within a specific refrigeration system is evaluated based on the adaptive chiller performance values.

[0107] Adaptive chiller performance values ​​are weighted based on the proportion of typical operating conditions identified from historical operating data of a specific refrigeration system, ensuring that the evaluated chiller performance closely matches the specific characteristics of the refrigeration system. Furthermore, the evaluation method based on adaptive chiller performance values ​​does not require shutdown testing of the refrigeration system containing the chiller under test; performance evaluation can be achieved solely through historical operating data analysis. In addition, this evaluation method fully considers potential uncertainties in chiller performance prediction, facilitating a realistic assessment of chiller performance in a specific refrigeration system. These uncertainties may stem from input data noise caused by sensor measurement errors or outliers, model parameter fitting errors, and insufficient data coverage due to historical operating data not encompassing all possible operating conditions.

[0108] The adaptive chiller performance evaluation method considers the selection of confidence levels for prediction intervals and the potential impact of prediction uncertainties during the evaluation process. The reliability of the proposed method is verified through a case study of a specific refrigeration system. The complete evaluation process was implemented. The results demonstrate that the evaluation method provided by the embodiments of this disclosure is adaptable to various refrigeration systems and provides intuitive and statistically significant chiller performance evaluation data. This facilitates the development of chiller ranking strategies and promotes predictive maintenance, ultimately achieving energy savings and emission reductions.

[0109] Alternatively, the performance of the chiller unit can be compared across multiple dimensions, including the manufacturer's design data (which can be understood as rated values), historical performance inferred from the building management system (BMS), and operating data from similar equipment. This comparison helps determine whether the chiller unit is currently operating well. For example, in the absence of significant performance differences, the chiller unit can be considered to be currently operating well; in the presence of significant differences, maintenance may be required to determine whether these differences are due to sensor errors within the refrigeration system or degradation of components such as the evaporator and condenser.

[0110] Alternatively, the chiller's condition is determined by comparing observed values ​​with at least one of the following: design ratings provided by the manufacturer, historical performance values ​​from the building management system, and operating values ​​of similar chillers. The chiller's performance condition can include normal performance, deterioration performance, abnormal performance, and degradation performance. For example, in a normal performance condition, the actual observed values ​​of the chiller fluctuate within the predicted range; in a deterioration performance condition, the predicted values ​​under the same operating conditions may continuously decrease compared to historical performance values; in an abnormal performance condition, the predicted values ​​may be significantly lower than the operating values ​​of similar chillers; and in a degradation performance condition, the predicted values ​​may be significantly lower than the design ratings.

[0111] As described above, conducting shutdown measurements to assess the performance of different chillers already in operation becomes impractical. In other words, obtaining the IPLV and NPLV of operating chillers experimentally presents significant challenges. Therefore, to evaluate chiller performance in existing refrigeration systems using operational data, embodiments of this disclosure propose an adaptive chiller performance value. This value incorporates the clustering results of annual operational data described above and weights the chiller's COP based on the frequency of historical operating conditions. By continuously adjusting the weights to reflect the latest refrigeration system performance, the adaptive chiller performance value provides a refined assessment of chiller performance, accurately reflecting the typical operating modes of the refrigeration system.

[0112] Adaptive chiller performance values ​​for specific chiller unit performance evaluation It can be given by equation (11):

[0113] in, This represents the number of typical historical operating conditions after clustering; Indication of operating conditions The corresponding percentage of the annual sample. Indicates operating conditions Average COP of the chiller unit.

[0114] Compared to conventional chiller performance evaluation indicators, such as IPLV and NPLV, the weights of conventional chiller performance evaluation indicators emphasize universality, are based on industry standards or general scenarios, and are not specific to the actual operating characteristics of a particular system. Referring to formula (11), the weights of the adaptive chiller performance values ​​provided in the embodiments of this disclosure are derived from cluster analysis of historical operating data of a specific refrigeration system, reflecting the actual proportion of typical operating conditions of the system. The identification of typical operating conditions is based on the historical operating data of the system. The weights and operating conditions of the adaptive chiller performance values ​​are adapted to the actual operating characteristics of a specific refrigeration system, thus enabling a more accurate match to the operating characteristics of a specific refrigeration system. In addition, the adaptive chiller performance values ​​utilize historical operating data and can achieve online performance evaluation without shutdown testing. This helps to monitor, optimize, and predictively maintain the performance of chillers in operation.

[0115] The adaptive chiller performance value provides an overall performance assessment of the chiller within the current refrigeration system and its operating performance. Subsequent evaluation methods based on the adaptive chiller performance value focus more on the performance of the chiller under specific operating conditions.

[0116] Evaluation methods based on adaptive chiller performance values ​​may include: evaluating the performance of multiple chillers within a specific refrigeration system, including: determining the adaptive chiller performance values ​​of the multiple chillers; determining the average value of the adaptive chiller performance values ​​of the multiple chillers; and identifying the chillers corresponding to adaptive chiller performance values ​​that deviate from the average value.

[0117] In addition, the evaluation method based on adaptive chiller performance values ​​may also include: generating performance tables of chillers under typical operating conditions and / or part load factor-coefficient of performance (PLR-COP) curves of chillers under specific operating conditions.

[0118] Optionally, evaluating the performance of multiple chillers within a specific refrigeration system may include: generating a performance table of the chillers under typical operating conditions; periodically updating the performance table of the chillers under typical operating conditions; and determining the performance status of the chillers based on the performance table of the chillers under typical operating conditions, and comparing the performance differences of multiple chillers, wherein the performance table of the chillers under typical operating conditions includes predicted values ​​and prediction ranges for the chillers under different typical operating conditions.

[0119] For example, clustering can identify multiple typical operating conditions from historical operating data. The performance of chillers under these conditions and their predicted ranges are then summarized in a table, resulting in a performance table for chillers under typical operating conditions. Furthermore, this performance table can be updated periodically based on the latest operating data. This helps maintenance personnel to make horizontal comparisons of chiller performance under common operating conditions, develop group control strategies, and determine whether poorly performing chillers require maintenance.

[0120] Optionally, evaluating the performance of multiple chillers within a specific refrigeration system may include: generating curves based on predicted values ​​of the chillers under different typical operating conditions, and generating interval bands based on predicted intervals of the chillers under different typical operating conditions; generating curves and interval bands for multiple chillers within a specific refrigeration system, thereby generating a part load factor-coefficient of performance curve; and determining the optimal operating range of multiple chillers based on the part load factor-coefficient of performance curve, and comparing the performance differences of multiple chillers.

[0121] For example, performance curves and prediction ranges of chiller units under specific operating conditions can be plotted, thereby generating PLR-COP curves for chiller units under specific operating conditions. This graphical representation helps maintenance personnel intuitively understand the optimal operating range of each chiller unit and make horizontal comparisons of their performance.

[0122] The performance tables of chillers under typical operating conditions and the PLR-COP curves of chillers under specific operating conditions can comprehensively evaluate the operating chillers based on real-time operating data, providing a reliable basis for formulating chiller operation strategies and carrying out equipment maintenance.

[0123] Figure 2 This is a schematic diagram of a performance evaluation system 2000 for a chiller unit provided according to an exemplary embodiment of the present disclosure.

[0124] like Figure 2 As shown, this disclosure provides a chiller unit performance evaluation system 2000, which may include: a typical operating condition identification unit 100, a model building unit 200, and a prediction unit 300. The typical operating condition identification unit 100 is used to identify typical operating conditions of a specific refrigeration system based on historical operating data of the system, wherein the system includes a chiller unit. The model building unit 200 is used to select a static empirical model for predicting the performance of the chiller unit and fit the parameters of the model based on historical operating data. The prediction unit 300 is used to input the data of the typical operating conditions into the model after fitting the parameters to generate the predicted value and prediction range of the performance of the chiller unit under the typical operating conditions, wherein the actual observed value of the performance of the chiller unit under the typical operating conditions is within the prediction range with a predetermined confidence level.

[0125] In the performance evaluation system for chiller units provided in this disclosure, typical operating conditions of the system can be identified based on historical operating data of a specific refrigeration system, and parameters of a selected static empirical model can be fitted based on this historical operating data. By inputting the data from the typical operating conditions into the model after fitting the parameters, predicted values ​​and prediction intervals of the chiller unit's performance under typical operating conditions can be generated, wherein the future actual observed values ​​of the chiller unit fall within this prediction interval with a selected confidence level. This closely links the performance of the chiller unit to be evaluated with the specific characteristics of the refrigeration system in which the chiller unit is located, effectively reducing the deviation between the predicted performance values ​​and the true values ​​of the chiller unit.

[0126] Furthermore, the prediction interval can be understood as the range of uncertainty in the future measured true value based on the predicted value. Therefore, the embodiments of this disclosure consider the potential impact of prediction uncertainty in the evaluation process. By quantifying the probability range of these uncertainties, the selected confidence level can provide the boundary of the future measured true value. Additionally, for chillers already in operation, it is impractical to perform shutdown measurements to evaluate the performance of different chillers. The evaluation method and system provided by the embodiments of this disclosure do not require shutdown testing of the refrigeration system containing the chiller under test; performance evaluation can be achieved solely through historical operating data analysis.

[0127] Therefore, the performance evaluation system for chiller units according to the embodiments of this disclosure can evaluate the performance of chiller units for a specific refrigeration system, which helps to optimize chiller unit sequencing strategies, participate in refrigeration system demand response, and troubleshoot.

[0128] Specifically, the typical operating condition identification unit 100 can also be used to preprocess historical operating data, wherein the preprocessing includes: supplementing missing values ​​in the historical operating data; and retaining outliers of engineering significance in the historical operating data. By preprocessing the historical operating data of a specific refrigeration system, noise in the historical operating data can be reduced.

[0129] Optionally, the typical operating condition identification unit 100 can also be used to normalize the pre-processed historical operating data, and to identify typical operating conditions based on the normalized historical operating data. Normalizing the historical operating data optimizes its distribution, improves the efficiency of subsequent clustering machine learning algorithms, and helps identify different historical operating conditions.

[0130] In addition, the typical operating condition identification unit 100 can also be used to cluster historical operating data to obtain multiple operating conditions of the system; and to determine typical operating conditions among multiple operating conditions based on established clustering indicators.

[0131] For example, after normalizing historical operating data, the data can be categorized to generate multiple operating conditions for a specific refrigeration system. For small datasets with simple operating condition transitions, histogram statistics can be used; however, for large datasets with complex operating conditions, clustering algorithms are more suitable.

[0132] The effectiveness of these clusters is evaluated using criteria such as the Kalinsky-Harabas criterion or the Davis-Bourdin criterion, to determine the optimal number of clusters. The number of clusters can be understood as the total number of operating condition categories of a specific refrigeration system divided from its historical operating data. The optimal number of clusters can be understood as the number of clusters that satisfy both mathematical optimality and engineering interpretability, or the typical operating conditions of the specific refrigeration system.

[0133] The model building unit 200 can also be used to select a model based on the type and accuracy of sensors within the system. Furthermore, based on historical operational data, regression analysis methods are used to fit the parameters, including the least squares method.

[0134] The prediction unit 300 can also be used to generate a prediction interval by quantifying the uncertainty factors of the model after fitting the parameters. The uncertainty factors include data noise, parameter fitting error, and errors caused by the lack or sparsity of historical operating data in representing all operating conditions of the system.

[0135] Specifically, the prediction unit 300 can also be used to determine the covariance matrix of the parameters and the mean square error of the historical operating data in the process of determining the parameters of the model based on historical operating data; input the data of typical operating conditions into the model after fitting the parameters to generate predicted values; and determine the boundary of the prediction interval based on the data of typical operating conditions, predicted values, mean square error, parameter covariance matrix and predetermined confidence level.

[0136] In addition, the prediction unit 300 can also be used to evaluate the quality of the prediction interval. For example, it can determine the prediction interval coverage probability and the average prediction interval width, where the prediction interval coverage probability is the proportion of observations in historical operating data that fall within the corresponding prediction interval, and the average prediction interval width is the arithmetic mean of the prediction interval widths corresponding to observations in historical operating data; based on the prediction interval coverage probability and the average prediction interval width, it can determine a criterion based on the coverage width; and based on the coverage width-based criterion, it can evaluate the quality of the prediction interval, where the value of the coverage width-based criterion is negatively correlated with the quality of the prediction interval.

[0137] For example, after selecting a suitable static empirical model for chiller performance prediction, the model's parameters can be fitted based on historical operating data, and the parameter uncertainty can be quantified to determine the model's parameters. Furthermore, the performance of the static empirical model can be validated by comprehensively evaluating its fitting accuracy and the quality of the prediction interval construction. This process can generate prediction intervals simultaneously with chiller performance prediction. Constructing appropriate prediction intervals helps to achieve a balance between prediction accuracy and coverage, ensuring that the evaluation method based on adaptive chiller performance values ​​is robust and applicable under various operating scenarios and requirements.

[0138] Optionally, the chiller performance evaluation system 2000 may also include an evaluation module (not shown), which can be used to determine adaptive chiller performance values ​​based on the proportion of predicted values ​​and typical operating conditions in historical data at the chiller level; and to evaluate the performance of multiple chillers in a specific refrigeration system based on the adaptive chiller performance values.

[0139] For example, by determining the adaptive chiller performance values ​​of multiple chiller units, the average value of the adaptive chiller performance values ​​of multiple chiller units is determined, and the chiller units corresponding to the adaptive chiller performance values ​​that deviate from the average value are identified.

[0140] In addition, the evaluation module can also be used to generate performance tables and / or partial load factor-performance coefficient curves for chillers under typical operating conditions.

[0141] For example, by generating a performance table of chillers under typical operating conditions, regularly updating the performance table of chillers under typical operating conditions, and determining the performance status of chillers based on the performance table of chillers under typical operating conditions, the performance differences of multiple chillers can be compared. The performance table of chillers under typical operating conditions includes the predicted values ​​and prediction ranges of chillers under different typical operating conditions.

[0142] By generating curves based on predicted values ​​of chiller units under different typical operating conditions, and generating interval bands based on predicted intervals of chiller units under different typical operating conditions, curves and interval bands of multiple chiller units within a specific refrigeration system are generated, thereby generating part load factor-coefficient of performance curves. Based on the part load factor-coefficient of performance curves, the optimal operating range of multiple chiller units is determined, and the performance differences of multiple chiller units are compared.

[0143] Optionally, the evaluation module can also be used to compare the observed values ​​with at least one of the following: design ratings provided by the manufacturer, historical performance values ​​from the building management system, and operating values ​​of similar chillers, to determine the status of the chiller, wherein the performance status of the chiller includes normal performance, degraded performance, abnormal performance, and degraded performance.

[0144] Adaptive chiller performance values ​​are weighted based on the proportion of typical operating conditions identified from historical operating data of a specific refrigeration system, ensuring that the evaluated chiller performance closely matches the specific characteristics of the refrigeration system. Furthermore, the evaluation system built upon adaptive chiller performance values ​​does not require shutdown testing of the refrigeration system containing the chiller under test; performance evaluation can be achieved solely through historical operating data analysis. In addition, this evaluation system fully considers potential uncertainties in chiller performance prediction to ensure a realistic assessment of chiller performance within a specific refrigeration system. These uncertainties may stem from input data noise caused by sensor measurement errors or outliers, model parameter fitting errors, and insufficient data coverage due to historical operating data not encompassing all possible operating conditions.

[0145] The adaptive chiller performance evaluation system considers the selection of confidence levels for prediction intervals and the potential impact of prediction uncertainties during the evaluation process. The reliability of the proposed system is verified through a case study of a specific refrigeration system. The complete evaluation process was implemented. The results demonstrate that the evaluation system provided by the embodiments of this disclosure can adapt to various refrigeration systems and provides intuitive and statistically significant chiller performance evaluation data. This facilitates the development of chiller prioritization strategies and promotes predictive maintenance, ultimately achieving energy savings and emission reductions.

[0146] Example

[0147] The following section will use an example of the operation records of a real data center cooling equipment over a period of one year to describe the performance evaluation method and system of the chiller unit provided in the embodiments of this disclosure.

[0148] Figure 3 This is a partial schematic diagram of a specific refrigeration system provided according to an exemplary embodiment of the present disclosure.

[0149] Specifically, such as Figure 3 As shown, the data used in this embodiment comes from the operating records of a real data center cooling system, spanning from March 2022 to March 2023 (5-minute intervals). This cooling system comprises six identical centrifugal chillers, such as Chiller1, Chiller2, Chiller3, Chiller4, Chiller5, and Chiller6, each with a rated cooling capacity of 4396 kW (nominal COP of 7.61), all equipped with water-side economizers (plate heat exchangers) and flow meters. Because the set chilled water supply temperature needs to be maintained around the clock to ensure an optimal operating environment for the data center servers, these chillers operate in a rotating and uninterrupted manner during transitional seasons and summer to meet cooling demands. In addition, the refrigeration system includes six dual-unit cooling towers (e.g., CT1-1 and CT1-2, CT2-1 and CT2-2, CT3-1 and CT3-2, CT4-1 and CT4-2, CT4-1 and CT4-2 and CT5-1 and CT5-2), six variable-speed condenser water pumps (e.g., CWP1, CWP2, CWP3, CWP4, CWP5 and CWP6), six variable-speed chilled water pumps (e.g., CHWP1, CHWP2, CHWP3, CHWP4, CHWP5 and CHWP6), six heat exchangers (e.g., HEX1, HEX2, HEX3, HEX4, HEX5 and HEX6), and four cold storage tanks (e.g., Tank1, Tank2, Tank3 and Tank4, which are only activated in emergency situations such as power outages).

[0150] Optionally, the above-mentioned refrigeration system further includes: a bypass valve for the cold storage tank water pipe (e.g., V-1, V-2 and V-3), a bypass valve for the chilled water supply and return main pipes (e.g., VT1, VT2), and a water valve for switching the heat exchanger mode (V1, V2, V3 and V4).

[0151] To ensure the validity of the proposed adaptive chiller performance indicators, cluster analysis can be performed on the operating data of each chiller unit separately. Furthermore, considering that although the six chillers in this specific refrigeration system are identical, many other refrigeration systems typically use chillers of different capacities and types. For example, combinations of screw and centrifugal chillers, or pairings of large and small capacity chillers, are used to meet different refrigeration needs. Therefore, according to the specifications of Model 6, cluster analysis was performed on the operating conditions of each chiller unit using three features: condenser inlet water temperature (…). T cws ), Evaporator outlet water temperature ( T chws The data was normalized before clustering to improve the distribution. The goal of each clustering iteration was to determine the optimal number of clusters by minimizing the Weiss-Bourdin criterion. k , k The range is from 4 to 20.

[0152] Detailed clustering results are shown in Tables A1 to A6 of the appendix. It can be seen that the operating conditions of the chiller units are quite complex. Therefore, during data processing, to simplify the complex operating conditions, a PLR interval of 0.1 and a temperature difference interval of 2°C can be selected, where the temperature difference interval is... T cws －T chws After analyzing the preliminary clustering results, these operating conditions can be summarized in Table 3.

[0153]

[0154] Table A1

[0155] Table A2

[0156] Table A3

[0157] Table A4

[0158] Table A5

[0159] Table A6

[0160] As shown in Table 3, the chiller unit operates under 17 different conditions, C1 to C17, during the annual operation period. The eight operating conditions with the highest frequency are indicated by asterisks (…). These operating conditions are marked as such, and each of these operating conditions accounts for at least 5% of the annual operating data. In addition, these operating conditions can account for a total of 86% of the annual operating data. Understanding the distribution of historical operating conditions helps in subsequent evaluation of chiller performance using the Adaptive Chiller Performance Value (ACPV).

[0161]

[0162] Table 3

[0163] Table 4

[0164] Model 6 was used for chiller unit performance prediction. The fitted model coefficients and goodness of fit are shown in Table 4. As described above, higher... and lower RMSE This indicates that the model has higher prediction accuracy. Therefore, based on the results, Model 6 has a high fitting accuracy for the COP of the six chiller units, indicating that Model 6 is suitable for further research.

[0165] Figure 4 and Figure 5 The chiller unit 01 provided according to the exemplary embodiments of this disclosure is respectively in PINC The historical COP values ​​and corresponding predicted values ​​are 80% and 90% of the data.

[0166] like Figure 4 and Figure 5 As shown, chiller unit 01 constructs a prediction interval (PI) based on the COP of 60 test data points, where PINC They are 80% and 90% respectively.

[0167] As shown in Table 5, selecting an appropriate prediction interval requires a balance between reliability and sharpness. An overly wide prediction interval will not improve prediction accuracy, while an overly narrow interval may miss important information. Therefore, it is possible to utilize... , and The quality of the constructed prediction intervals is evaluated at confidence levels of 80% and 90%.

[0168]

[0169] Table 5

[0170] At a 90% confidence level, all six chiller units Both are close to 90%, indicating good coverage at this confidence level; The lower value indicates a moderate prediction interval. At an 80% confidence level, except for chiller unit 3... The percentage was 79.91%, slightly below the target of 80%, for the remaining chiller units. All exceeded 80%, indicating satisfactory coverage of the actual data; The value remains low, indicating that the prediction interval size is moderate. Furthermore, at both confidence levels, The values ​​are all low, indicating that the constructed prediction interval has achieved a good balance between prediction accuracy and coverage.

[0171] Since data centers maintain chilled water supply temperatures between 13°C and 15°C and return water temperatures between 19°C and 21°C year-round, and the cooling water is supplied by cooling towers, the temperature difference ΔT between the condenser inlet and evaporator outlet can serve as an indicator of seasonal variation. Historical operational data analysis shows that during transitional seasons, the chiller condenser inlet temperature (which can be understood as the cooling water outlet temperature from the water-side economizer) is typically between 19°C and 22°C. Therefore, a ΔT value less than 7°C is classified as a transitional season, while a value exceeding 7°C is classified as summer.

[0172] Figures 6 to 11 The average COP of the six chillers under 17 historical operating conditions is shown. It can be seen that, due to the identical structure and similar operating times of the six chillers, the performance distributions of chillers 01 to 06 are similar. Specifically, the COP of the six chillers is generally lower in summer than in the transitional season, attributed to the higher condenser inlet water temperature. Furthermore, the analysis shows that the COP under medium-high PLR can be greater than that under low PLR, confirming that the chillers generally perform best within a PLR range of 0.7 to 0.9.

[0173] In addition, such as Figure 12 As shown, the adaptive chiller performance value (ACPV) of six chiller units can be calculated using two methods. Method 1 can include the ACPV for all 17 operating conditions, while Method 2 can consider only the ACPV for 8 typical operating conditions that account for more than 5% of historical occurrences. main The results show that chillers 02, 05, and 06 outperform the average performance of their respective refrigeration systems, while chillers 01, 03, and 04 underperform the average performance of the refrigeration system. Furthermore, considering ACPV... main The maximum difference between the two is only 0.47%, and when providing performance references for chillers, the typical operating conditions of the chillers should be given priority.

[0174]

[0175] Table 6

[0176] The eight typical operating conditions mentioned above can be divided into five main groups based on seasonal characteristics and PLR. As shown in Table 6, Table 6 compiles the COP of six chiller units and their corresponding 80% prediction intervals. Each column in the table represents a typical operating condition; red cells indicate the chiller unit with the lowest performance under that condition, and green cells indicate the chiller unit with the highest performance under that condition. Using 80% prediction intervals strikes a balance between reliability and sharpness, clearly distinguishing between the best and worst performing chiller units, with no overlap in their prediction intervals. Other chiller units in the same group require further analysis. For example, under operating condition C2, the average COP values ​​of chiller units 03 and 06 differ by 0.36, which is less than the sum of their standard deviations of 0.38. This indicates the existence of prediction uncertainty, leading to overlapping prediction intervals; therefore, it cannot be concluded that chiller unit 06 is superior to chiller unit 03 under this operating condition. This uncertainty can significantly affect the control algorithm for dependency point estimation, because random errors may lead to suboptimal control actions in actual refrigeration systems, deviating from the expected results.

[0177] Figures 13 to 14 The PLR-COP curves of the six chiller units are shown separately.

[0178] like Figure 13 and Figure 14 As shown, the PLR-COP curves effectively illustrate the performance of the chiller unit, and combining them with appropriate prediction ranges can enhance the clarity of these comparisons. Figure 13 and Figure 14 Two examples are provided, depicting the PLR-COP curves and their 80% prediction intervals for chillers 03, 04, and 06 with condenser inlet and evaporator outlet temperature differences of 3.3°C and 10.8°C, respectively. Figure 2 Preliminary inspections indicate that the optimal efficiency of the chillers occurs within a PLR range of 0.7 to 0.9. At a temperature difference of 3.3°C, the performance difference between chiller 06 and chiller 03 is significantly greater than that between chiller 04 and chiller 03. This pattern is similar at a temperature difference of 10.8°C. Furthermore, at 10.8°C, the performance difference between chiller 03 and chiller 04 is less pronounced than at 3.3°C because their prediction ranges largely overlap, especially within a PLR range of 0.3 to 0.4. Within this range, the performance of chiller 03 and chiller 04 is similar, making it difficult to assess which unit performs better.

[0179] Therefore, the performance evaluation process and adaptive chiller performance index proposed in the embodiments of this disclosure have been tested in actual refrigeration equipment. Unlike existing indices such as IPLV and NPLV, the performance evaluation process and adaptive chiller performance index proposed in this disclosure do not require shutdown for measurement; they can be obtained simply by analyzing historical operating data.

[0180] In the performance evaluation method and system for chiller units provided in the embodiments of this disclosure, typical operating conditions in the system can be identified based on historical operating data of a specific refrigeration system, and parameters of a selected static empirical model can be fitted based on this historical operating data. By inputting the data of the typical operating conditions into the model after fitting the parameters, predicted values ​​and prediction intervals of the chiller unit's performance under typical operating conditions can be generated, wherein the future actual observed values ​​of the chiller unit can be selected with a confidence level falling within this prediction interval. This closely links the performance of the chiller unit to be evaluated with the specific characteristics of the refrigeration system in which the chiller unit is located, effectively reducing the deviation between the predicted performance values ​​and their true values. Furthermore, the prediction interval can be understood as the range of uncertainty of the future measured true value based on the predicted value. Therefore, the embodiments of this disclosure consider the potential impact of prediction uncertainty in the evaluation process. By quantifying the probability range of these uncertainties, the selected confidence level can provide the boundary of the future measured true value.

[0181] It should be noted that, in order to at least address the technical issues described in this paper, two widely accepted concepts in statistical analysis are employed: confidence intervals and prediction intervals. A confidence interval describes the uncertainty in estimating an unknown but fixed value, while a prediction interval manages the uncertainty associated with the future realization of a random variable. Therefore, the prediction interval includes additional sources of uncertainty, such as model misspecification and noise variance, making it wider than the corresponding confidence interval.

[0182] The chiller performance evaluation method of the present disclosure can evaluate the performance of chillers for a specific refrigeration system, which helps to optimize chiller sequencing strategies, participate in refrigeration system demand response, and troubleshoot.

[0183] Figure 15 A schematic block diagram of an example electronic device 3000 that can be used to implement embodiments of the present disclosure is shown.

[0184] like Figure 15As shown, the electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0185] Electronic device 3000 includes a computing unit 301, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 302 or a computer program loaded from storage unit 308 into random access memory (RAM) 303. The RAM 303 may also store various programs and data required for the operation of electronic device 3000. The computing unit 301, ROM 302, and RAM 303 are interconnected via bus 304. Input / output (I / O) interface 305 is also connected to bus 304.

[0186] Multiple components in electronic device 3000 are connected to I / O interface 305, including: input unit 306, such as keyboard, mouse, etc.; output unit 307, such as various types of displays, speakers, etc.; storage unit 308, such as disk, optical disk, etc.; and communication unit 309, such as network card, modem, wireless transceiver, etc. Communication unit 309 allows electronic device 3000 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0187] The computing unit 301 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 301 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 301 performs the various methods and processes described above, such as the performance evaluation method for a chiller unit. For example, in some embodiments, the performance evaluation method for a chiller unit may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as storage unit 308. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 3000 via ROM 302 and / or communication unit 309. When the computer program is loaded into RAM 303 and executed by the computing unit 301, one or more steps of the performance evaluation method for a chiller unit described above may be performed. Alternatively, in other embodiments, the computing unit 301 may be configured to perform a performance evaluation method for the chiller unit by any other suitable means (e.g., by means of firmware).

[0188] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0189] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0190] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0191] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0192] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0193] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and Virtual Private Server (VPS) services, such as high management difficulty and weak business scalability.

[0194] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0195] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A method of evaluating performance of a water chiller, characterized by, The method comprises: identifying typical operating conditions of a specific refrigeration system based on historical operating data of the system, wherein the system comprises a chiller; selecting a static empirical model for predicting performance of the chiller, and fitting parameters of the model based on the historical operating data; and inputting data of the typical operating conditions into the model with the fitted parameters to generate a predicted value and a prediction interval of the performance of the chiller under the typical operating conditions, wherein an actual observed value of the performance of the chiller under the typical operating conditions is within the prediction interval with a predetermined confidence level.

2. The evaluation method according to claim 1, wherein Generating the prediction interval comprises: generating the prediction interval by quantifying uncertainty factors of the model with the fitted parameters, wherein the uncertainty factors include data noise, fitting errors of the parameters, and errors caused by missing or sparsity of representation of all operating conditions of the system by the historical operating data.

3. The evaluation method according to claim 1, wherein, The historical operating data comprises historical data at a chiller level, and the method further comprises: determining an adaptive chiller performance value based on the predicted value and a proportion of data of the typical operating conditions in the historical data at the chiller level; and evaluating performance of multiple chillers in the specific refrigeration system based on the adaptive chiller performance value.

4. The evaluation method according to claim 3, wherein Evaluating performance of multiple chillers in the specific refrigeration system comprises: determining the adaptive chiller performance values of multiple chillers; determining an average value of the adaptive chiller performance values of multiple chillers; and identifying a chiller corresponding to an adaptive chiller performance value deviating from the average value.

5. The evaluation method according to claim 3, wherein Evaluating performance of multiple chillers in the specific refrigeration system comprises: generating a performance table of chillers under typical operating conditions; periodically updating the performance table of chillers under typical operating conditions; and determining performance statuses of the chillers based on the performance table of chillers under typical operating conditions, and comparing performance differences of multiple chillers, wherein the performance table of chillers under typical operating conditions comprises the predicted values and the prediction intervals of the chillers under different typical operating conditions.

6. The evaluation method according to claim 3, wherein Evaluating performance of multiple chillers in the specific refrigeration system comprises: generating a curve based on the predicted values of the chillers under different typical operating conditions, and generating an interval band based on the prediction intervals of the chillers under different typical operating conditions; generating the curve and the interval band of multiple chillers in the specific refrigeration system to generate a part load ratio-coefficient of performance curve; and determining optimal operating ranges of multiple chillers based on the part load ratio-coefficient of performance curve, and comparing performance differences of multiple chillers.

7. The evaluation method according to claim 1, wherein Identifying typical operating conditions of a specific refrigeration system based on historical operating data of the system comprises: preprocessing the historical operation data, wherein the preprocessing comprises: supplementing missing values in the historical operation data; and retaining outliers with engineering significance in the historical operation data; normalizing the historical operation data that has completed the preprocessing; and identifying the typical operation condition based on the historical operation data that has completed the normalization.

8. The evaluation method according to claim 1, wherein, identifying the typical operation condition of a specific refrigeration system based on historical operation data of the system further comprises: clustering the historical operation data to obtain a plurality of operation conditions of the system; and determining the typical operation condition among the plurality of operation conditions based on the established clustering index.

9. The evaluation method according to claim 1, wherein, selecting a static empirical model for predicting the performance of the chiller unit comprises: selecting the model based on the types and accuracies of the sensors in the system.

10. The evaluation method according to claim 1, wherein, fitting parameters of the model based on the historical operation data comprises: fitting the parameters using a regression analysis method based on the historical operation data, wherein the regression analysis method comprises a least squares method.

11. The evaluation method according to claim 10, wherein inputting data of the typical operation condition into the model fitted with the parameters to generate the prediction interval comprises: determining a covariance matrix of the parameters and a mean square error of the historical operation data in the process of determining the parameters of the model based on the historical operation data; inputting data of the typical operation condition into the model fitted with the parameters to generate the prediction value; and determining the boundaries of the prediction interval based on the data of the typical operation condition, the prediction value, the mean square error, the parameter covariance matrix, and the predetermined confidence level.

12. The evaluation method according to claim 1, wherein, The method further comprises: comparing the observation value with at least one of a design rating value provided by a manufacturer, a historical performance value from a building management system, and an operation value of a similar chiller unit to determine a state of the chiller unit, wherein the performance state of the chiller unit comprises a normal performance state, a performance degradation state, an abnormal performance state, and a performance deterioration state.

13. The evaluation method according to claim 1, wherein, The method further comprises: determining a prediction interval coverage probability and an average prediction interval width, wherein the prediction interval coverage probability is a proportion of observation values in the historical operation data falling within the corresponding prediction interval, and the average prediction interval width is an arithmetic mean of the prediction interval widths corresponding to the observation values in the historical operation data; determining a coverage width-based criterion based on the prediction interval coverage probability and the average prediction interval width; and evaluating the quality of the prediction interval based on the coverage width-based criterion, wherein a value of the coverage width-based criterion is negatively correlated with the quality of the prediction interval.

14. A performance evaluation system for a water chiller, the system comprising: comprises: an identifying typical operation condition unit configured to identify a typical operation condition of a specific refrigeration system based on historical operation data of the system, wherein the system comprises the chiller unit; a model building unit configured to select a static empirical model for predicting the performance of the chiller unit, and fit parameters of the model based on the historical operation data; and ​ a prediction unit configured to input data of the typical operating condition into the model after fitting the parameters, to generate a predicted value and a prediction interval of performance of the water chiller under the typical operating condition, wherein an actual observed value of the performance of the water chiller under the typical operating condition is within the prediction interval with a predetermined confidence level.

15. An electronic device, comprising: comprise: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the electronic device to implement the performance evaluation method of the water chiller according to any one of claims 1 to 13.

16. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is used to enable a computer to execute the performance evaluation method of the water chiller according to any one of claims 1 to 13.

17. A computer program product comprising a computer program, characterized in that, The computer program is executed by a processor to implement the performance evaluation method of the water chiller according to any one of claims 1 to 13.

Images