Dynamic evaluation method and system for health condition of refrigerating unit

Abstract

The invention relates to the technical field of equipment management, in particular to a dynamic evaluation method and system for the health condition of a refrigerating unit, and the method comprises the steps: constructing a refrigerating capacity and energy efficiency ratio prediction model through a gradient lifting regression tree algorithm, and training a high-precision reference through the full utilization of historical data; a stable and reliable prediction reference and a data analysis basis are provided for subsequent dynamic evaluation; a static model is converted into a dynamic model responding to a real-time working condition, real-time evaluation on the performance of the refrigerating unit is achieved, and meanwhile, the sensitivity and accuracy of state monitoring are effectively improved by combining degradation degree cooperative calculation; the real-time score of the refrigerating unit is compared with the preset judgment threshold value, so that the health state of the refrigerating unit can be quickly and accurately judged; when it is detected that the refrigerating unit is in the abnormal state, the model updating period and the sensor sampling frequency can be dynamically adjusted, and it is ensured that more accurate and timely data are obtained at the critical moment.

Description

Technical Field

[0001] This invention relates to the field of equipment management technology, and in particular to a method and system for dynamic assessment of the health status of refrigeration units. Background Technology

[0002] Against the backdrop of the rapid development and widespread application of refrigeration technology, the field of refrigeration unit management still faces multiple challenges: traditional methods may fail to accurately assess the real-time health status of refrigeration units, leading to misjudgments of unit performance and the inability to promptly detect potential faults or performance degradation; traditional assessment methods may not fully consider the performance changes of refrigeration units under different operating conditions, resulting in assessment results that do not match actual operating conditions; traditional methods may lack real-time monitoring of refrigeration unit performance degradation, leading to repairs only after faults occur, increasing maintenance costs and downtime; traditional maintenance strategies may adopt fixed maintenance cycles and methods, failing to develop targeted maintenance plans based on the actual operating status and performance degradation of the unit. Therefore, this invention proposes a dynamic assessment method and system for the health status of refrigeration units. Summary of the Invention

[0003] The purpose of this invention is to solve the problems in the background art by proposing a method and system for dynamic evaluation of the health status of refrigeration units.

[0004] To achieve the above objectives, the present invention adopts the following technical solution: A method for dynamic assessment of the health status of a refrigeration unit includes: S1. Obtain the electronic expansion valve opening data stream, and use the gradient boosting regression tree algorithm to establish the benchmark model M1 for predicting cooling capacity and the benchmark model M2 for energy efficiency ratio. S2. Transform the trained baseline model M1 and baseline model M2 into a dynamic baseline model that responds to real-time operating conditions; input the real-time electronic expansion valve opening degree into the dynamic baseline model, and perform joint calculation of real-time performance indicators and degradation degree to obtain the real-time score of the refrigeration unit; among which, the real-time performance indicators are standard cooling capacity and standard energy efficiency ratio. S3. Based on the real-time score of the refrigeration unit, compare it with the configuration judgment threshold to determine the status of the refrigeration unit, including healthy, sub-healthy, faulty, serious faulty or completely failed status. S4. Based on the evaluation results, establish a dynamic adjustment strategy: If the refrigeration unit is detected to be in a fault, serious fault, or complete failure state, the update cycle of the dynamic benchmark model and the sensor sampling frequency will be dynamically adjusted.

[0005] Furthermore, the process of acquiring the electronic expansion valve opening data stream and using the gradient boosting regression tree algorithm to establish a baseline model M1 for predicting cooling capacity and a baseline model M2 for energy efficiency ratio includes: S11. Collect historical operation data from the historical operation database of the chiller unit or refrigeration system: electronic expansion valve opening. The cooling capacity Q and energy efficiency ratio EER were calculated; the historical operation dataset was cleaned, and missing and outlier values ​​were removed; the cleaned dataset was randomly shuffled and divided into training and testing sets according to a preset ratio. S12, the opening degree of the electronic expansion valve The main input features are normalized; at the same time, derived features closely related to the physical relationship between cooling capacity Q and energy efficiency ratio EER are constructed. S13, based on the pre-processed electronic expansion valve opening... The derived features share a common input feature set. By replacing the output target, models M1 and M2 are trained separately within the same gradient boosting regression tree training framework. The process is as follows: select the gradient boosting regression tree regressor, preset the total number of iterations, and initialize the initial prediction values ​​of the two models; train iteratively round by round, for a total of... In each round, model M1 and model M2 are updated synchronously; the iterative training process includes calculating the negative gradient, training a new regression tree, calculating the output values ​​of the leaf nodes, and updating the overall prediction function. S14. Use the partitioned test set to evaluate the performance of benchmark model M1 and benchmark model M2 respectively.

[0006] Furthermore, the iterative training process of the model includes: A1. For each sample i in the training set, calculate the negative gradient between the current model's prediction and the true value. ; A2. Using the entire current training set, with the electronic expansion valve opening... As input, the calculated negative gradient To achieve the output target, train a new regression tree and use this new regression tree to divide the data into J leaf nodes. Where J represents the number of leaf nodes in the m-th tree; R is the region that divides the input space into multiple non-overlapping regions. This represents the j-th leaf node region in the m-th tree; A3. Calculate an output value for each leaf node j of this new tree. ,in, This represents the output value of the j-th leaf node in the m-th tree; A4. Add the new tree to the ensemble model and update the overall prediction function; fine-tune the hyperparameters shared by models M1 and M2 on the training set through cross-validation, and find the optimal hyperparameter combination that minimizes the validation set error of models M1 and M2; use the optimal hyperparameter combination to retrain on the entire training set to obtain the final baseline model M1 and baseline model M2.

[0007] Furthermore, the trained baseline models M1 and M2 are transformed into dynamic baseline models that respond to real-time operating conditions; the real-time electronic expansion valve opening is input into the dynamic baseline model, and real-time performance indicators and degradation are calculated collaboratively to obtain the real-time score of the refrigeration unit. The process includes: The trained baseline models M1 and M2 are packaged into a callable model file package and deployed to the real-time monitoring device of the chiller unit; when the device starts up, all parameters of the baseline models M1 and M2 are automatically loaded. Real-time acquisition of current operating condition data of the unit; alignment of real-time operating condition data with the common input feature set of benchmark model M1 and benchmark model M2; If there is a difference between the real-time operating conditions and the operating conditions range of the training data for the baseline models M1 and M2, a dynamic calibration mechanism is initiated; the model obtained after calibration becomes the dynamic baseline model. With dynamic benchmark model ; Input real-time electronic expansion valve opening and aligned real-time operating condition features to the dynamic baseline model Dynamic benchmark model , by dynamic benchmark model Output the standard cooling capacity under the current operating conditions. , by dynamic benchmark model Output the standard energy efficiency ratio under the current operating conditions. ; Real-time cooling capacity is collected synchronously through the unit's sensors. Real-time energy efficiency ratio Based on the calculated standard cooling capacity and standard energy efficiency ratio as benchmarks, the real-time degradation rate of the cooling capacity of the refrigeration unit is calculated respectively. Real-time degradation rate of energy efficiency ratio ; The real-time score of the refrigeration unit is calculated by combining the real-time degradation rate of its cooling capacity and the real-time degradation rate of its energy efficiency ratio. .

[0008] Furthermore, the process of determining the status of the refrigeration unit based on its real-time score and the configuration threshold includes: If the real-time score of the refrigeration unit If so, the refrigeration unit is in a healthy state; if If so, the refrigeration unit is in a sub-healthy state; if If so, the refrigeration unit is in a fault state; if If so, the refrigeration unit is in a state of serious malfunction; if If this occurs, the refrigeration unit is in a state of complete failure; among which, All of these are judgment thresholds.

[0009] Furthermore, based on the evaluation results, a dynamic adjustment strategy is established: if the refrigeration unit is detected to be in a faulty, severely faulty, or completely failed state, the process of dynamically adjusting the update cycle of the dynamic baseline model and the sensor sampling frequency includes: Get the real-time score of the refrigeration unit Real-time degradation rate of cooling capacity Real-time degradation rate of energy efficiency ratio Construct multi-dimensional state quantification indicators: Define performance degradation acceleration It reflects the rate of performance degradation and is used to determine whether degradation is accelerating. If performance degradation accelerates This indicates accelerated performance degradation; if the performance degradation acceleration is... This indicates that the degradation rate has slowed down or stabilized; Introducing fault impact factors Quantify the impact of the current fault on the function of the refrigeration unit; Real-time score of integrated refrigeration unit Accelerated performance degradation Fault influencing factors This quantifies the urgency of maintenance tasks, thereby obtaining a maintenance priority index. ; Real-time score based on refrigeration unit With performance degradation acceleration For dynamic benchmark models Dynamic benchmark model The update cycle is adaptively adjusted: In the formula, For update cycle; The baseline update cycle; The adjustment coefficient for the real-time score of the refrigeration unit; This is the degradation acceleration adjustment coefficient; Based on fault impact factors With maintenance priority index The sensor sampling frequency is dynamically adjusted. In the formula, The adjusted sampling frequency; The reference sampling frequency; The adjustment coefficient is the factor for the impact of faults. This is the priority adjustment coefficient.

[0010] A dynamic health assessment system for refrigeration units, comprising: Benchmark model construction module: Acquire the electronic expansion valve opening data stream, and use the gradient boosting regression tree algorithm to establish benchmark model M1 for predicting cooling capacity and benchmark model M2 for energy efficiency ratio, respectively; Dynamic benchmark conversion module: Converts the trained benchmark model M1 and benchmark model M2 into a dynamic benchmark model that responds to real-time operating conditions; inputs the real-time electronic expansion valve opening degree into the dynamic benchmark model, and performs joint calculation of real-time performance indicators and degradation degree to obtain the real-time score of the refrigeration unit; among which, the real-time performance indicators are standard cooling capacity and standard energy efficiency ratio. Health status determination module: Based on the real-time score of the refrigeration unit and the configuration determination threshold, the status of the refrigeration unit is determined, including healthy, sub-healthy, faulty, serious faulty or completely failed status. Dynamic adjustment strategy module: Based on the evaluation results, a dynamic adjustment strategy is established: if the refrigeration unit is detected to be in a fault, serious fault or complete failure state, the update cycle of the dynamic benchmark model and the sensor sampling frequency are dynamically adjusted.

[0011] Compared with existing technologies, the advantages of this invention, which provides a method and system for dynamically assessing the health status of refrigeration units, are as follows: This invention constructs benchmark models M1 and M2 for cooling capacity and energy efficiency ratio using a gradient boosting regression tree algorithm, accurately capturing the relationship between electronic expansion valve opening and cooling performance. Training the model with historical data effectively improves prediction accuracy and reliability, providing data support for subsequent dynamic evaluation. By transforming the static benchmark model into a dynamic benchmark model responding to real-time operating conditions, real-time evaluation of refrigeration unit performance is achieved. Model parameters can be adjusted according to actual operating conditions, ensuring the accuracy and timeliness of evaluation results and helping to promptly identify potential problems. By comparing the refrigeration unit's real-time score with a judgment threshold, the system accurately determines the refrigeration unit's status, distinguishing between healthy, sub-healthy, and faulty states, providing maintenance guidance for operation and maintenance personnel and reducing unplanned downtime. When a fault or other abnormal state is detected in the refrigeration unit, the model update cycle and sensor sampling frequency are dynamically adjusted to optimize data acquisition and processing efficiency, ensuring more accurate data is obtained at critical moments. Attached Figure Description

[0012] Figure 1 This is a flowchart of a dynamic assessment method for the health status of a refrigeration unit proposed in this invention.

[0013] Figure 2 This is a block diagram of a dynamic health status assessment system for refrigeration units proposed in this invention. Detailed Implementation

[0014] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the implementation regulations described are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0015] Please see Figure 1 This invention provides a method for dynamically assessing the health status of a refrigeration unit, comprising: S1. Obtain the electronic expansion valve opening data stream, and use the gradient boosting regression tree algorithm to establish the benchmark model M1 for predicting cooling capacity and the benchmark model M2 for energy efficiency ratio. S2. Transform the trained baseline model M1 and baseline model M2 into a dynamic baseline model that responds to real-time operating conditions; input the real-time electronic expansion valve opening degree into the dynamic baseline model, and perform joint calculation of real-time performance indicators and degradation degree to obtain the real-time score of the refrigeration unit; among which, the real-time performance indicators are standard cooling capacity and standard energy efficiency ratio. S3. Based on the real-time score of the refrigeration unit, compare it with the configuration judgment threshold to determine the status of the refrigeration unit, including healthy, sub-healthy, faulty, serious faulty or completely failed status. S4. Based on the evaluation results, establish a dynamic adjustment strategy: If the refrigeration unit is detected to be in a faulty, severely faulty, or completely failed state, the update cycle of the dynamic benchmark model and the sensor sampling frequency will be dynamically adjusted; the dynamic adjustment strategy can adjust the operating parameters of the model and sensors in a timely manner according to the changes in the health status of the unit, thereby improving the response speed and accuracy.

[0016] It should be further explained that, in the specific implementation process, the process of acquiring the electronic expansion valve opening data stream and using the gradient boosting regression tree algorithm to establish the benchmark model M1 for predicting cooling capacity and the benchmark model M2 for energy efficiency ratio includes: S11. Collect historical operation data from the historical operation database of the chiller unit or refrigeration system: electronic expansion valve opening. Cooling capacity Q and energy efficiency ratio EER; clean the collected historical operating dataset, handling missing and outlier values ​​(checking for missing electronic expansion valve openings in the dataset). For records of cooling capacity (Q) and energy efficiency ratio (EER), missing samples should be deleted directly or filled using interpolation. Box plots or the 3σ criterion should be used to identify and remove obviously unreasonable data caused by sensor malfunctions or operational errors, such as electronic expansion valve opening. (For negative values, extremely high or extremely low energy efficiency ratio EER); the cleaned dataset is randomly shuffled and divided into a training set and a test set according to a preset ratio (e.g., 8:2). The training set is used to train the model parameters, and the test set is used to finally evaluate the generalization performance of the model. S12, the opening degree of the electronic expansion valve The main input features are normalized to optimize the gradient and improve the convergence speed and performance of the regression tree. At the same time, derived features closely related to the cooling capacity Q and energy efficiency ratio EER are constructed to form an input feature set that can be used by models M1 and M2. The introduction of derived features enhances the model's ability to capture complex physical processes and improves the accuracy of prediction. Specifically, in a refrigeration system, the heat exchange on the evaporator side is considered as the refrigeration capacity Q; by measuring the inlet and outlet water temperatures and flow rates of the evaporator, the formula for calculating the refrigeration capacity Q is as follows: In the formula, The specific heat capacity of water at constant pressure. The mass flow rate of water on side e of the evaporator. The inlet water temperature of evaporator e. Let the outlet water temperature of evaporator e be used; then, the first derived characteristic is constructed, namely, the cooling contribution per unit mass flow rate. : This characteristic reflects the cooling effect produced when a unit mass of water flows through the evaporator. The larger the value, the stronger the cooling capacity at the same flow rate. Understandably, in the process of feature construction, in addition to the cooling contribution per unit mass flow rate, there are other derived features, such as the heat exchange efficiency on the evaporator side (calculated based on the temperature difference and flow rate of the evaporator inlet and outlet water) and the rate of change of the opening of the electronic expansion valve (calculated based on the opening difference between the two time points), so as to more comprehensively reflect the operating status of the refrigeration system. S13, based on the pre-processed electronic expansion valve opening... The derived features share the same input feature set. By replacing the output target, models M1 (predicting cooling capacity Q) and M2 (predicting energy efficiency ratio EER) are trained separately within the same gradient boosting regression tree training framework. The process is as follows: select the gradient boosting regression tree regressor, preset the total number of iterations ( The algorithm iteratively adds new regression trees to gradually reduce prediction errors, thus improving the model's fitting ability. This process is repeated round by round. In each round, models M1 and M2 are updated synchronously. The iterative training process includes calculating the negative gradient (residual), training a new regression tree, calculating the output values ​​of the leaf nodes, and updating the overall prediction function. A1. For each sample i in the training set, calculate the negative gradient between the predicted value and the true value of the current model (the model composed of the (m-1)th tree): In the formula, G is the negative gradient, i is the sample index, and m is the tree or iteration index; The loss function; Let represent the negative gradient of the m-th iteration or the m-th tree. For example, if the current iteration is the 5th tree constructed by the algorithm, then... This represents the negative gradient (residual) calculated for the i-th training sample when constructing the m-th tree. This represents the true value of the i-th sample in model M1. The actual cooling capacity of the i-th sample In model M2, The true energy efficiency ratio of the i-th sample ; This represents the current ensemble model's prediction for the i-th sample. It is a quantity that changes continuously with iteration. When constructing the m-th tree, The predicted value given by the ensemble model consisting of the first m-1 trees is... ; This represents the feature input of a sample i; A2. Using the entire current training set, with the electronic expansion valve opening... As input, the calculated negative gradient To output the target, train a new regression tree. The data is divided into J leaf nodes using a new regression tree. Where J represents the number of leaf nodes that the m-th tree has been divided into, and J is an integer whose size is controlled by hyperparameters such as the maximum depth of the tree; R is a region that divides the input space into multiple non-overlapping regions. This represents the j-th leaf node region in the m-th tree (where j is the leaf node index, and j = 1, 2, ..., J). For example, This represents the 3rd leaf node on the 5th tree; A3. Calculate an output value for each leaf node j of this new tree. ,in, The output value of the j-th leaf node in the m-th tree is obtained by minimizing the loss function through linear search, i.e., it is assigned to that leaf node. negative gradient of all samples The average value; A4. Add the new tree to the ensemble model and update the overall prediction function: In the formula, It is the learning rate, used to control the contribution of each tree, keeping the two models consistent; For indicator functions, when Belonging to the leaf node region Output 1 if the condition is met, otherwise output 0, used for positioning. The partitioning position in the tree; when updating the overall prediction function, determine the learning rate adjustment strategy, such as dynamically adjusting the learning rate based on the performance on the validation set to avoid overfitting or underfitting; fine-tune the hyperparameters shared by models M1 and M2 on the training set through cross-validation to find the optimal hyperparameter combination that minimizes the validation set error of models M1 and M2; where hyperparameters include the number of trees (or number of iterations), learning rate (to reduce the contribution of each tree), maximum depth of each decision tree (to control model complexity), subsampling ratio (to introduce randomness and prevent overfitting), and minimum number of samples required for leaf nodes; retrain iteratively on the entire training set using the optimal hyperparameter combination to obtain the final baseline model M1 (output cooling capacity Q) and baseline model M2 (output energy efficiency ratio EER). S14. Use the partitioned test set to evaluate the performance of the baseline model M1 and the baseline model M2 respectively, and record the mean square error, mean absolute error and coefficient of determination as the benchmark for subsequent model comparison.

[0017] It should be further explained that, in the specific implementation process, the process of transforming the trained baseline model M1 and baseline model M2 into a dynamic baseline model that responds to real-time operating conditions; inputting the real-time electronic expansion valve opening degree into the dynamic baseline model, and performing joint calculation of real-time performance indicators and degradation degree to obtain the real-time score of the refrigeration unit includes: The trained baseline models M1 and M2 (including core parameters such as optimal hyperparameters, tree structure, and leaf node output values) are packaged into a callable model file package and deployed to the real-time monitoring device of the chiller unit. When the device starts up, it automatically loads all parameters of the baseline models M1 and M2 to ensure that the initial state of the model is consistent with the trained baseline state, providing a basic framework for dynamic calculation. Real-time acquisition of current operating condition data of the unit (including real-time electronic expansion valve opening, real-time ambient temperature, chilled water inlet temperature, cooling water inlet temperature, and other key operating condition variables); alignment of real-time operating condition data with the common input feature set of benchmark models M1 and M2 (e.g., calculating the real-time value of the opening change rate - energy efficiency response coefficient based on the real-time electronic expansion valve opening and the average opening within the previous 10 seconds), ensuring that the feature dimensions and format of the input benchmark models M1 and M2 are consistent with those during training, avoiding prediction bias due to feature mismatch; If the real-time operating conditions differ from the operating conditions range of the training data for baseline models M1 and M2 (e.g., the ambient temperature exceeds ±10% of the training data, indicating a significant difference), a dynamic calibration mechanism is activated: the real-time electronic expansion valve opening value over the past hour is used. Valid real-time cooling capacity Q and real-time energy efficiency ratio EER data (≥50 data sets, no outliers) are used to incrementally train the loaded baseline models M1 and M2 (keeping the original hyperparameters fixed and only updating the latest 3-5 regression trees) to correct the model's prediction bias for extreme operating conditions; the calibrated model is the dynamic baseline model. (Real-time cooling capacity Q prediction model) and dynamic benchmark model (Real-time Energy Efficiency Ratio (EER) Prediction Model) Its parameters are dynamically adjusted according to real-time operating conditions to ensure that the standard output value matches the actual operating status of the current unit. During incremental training, the selection strategy for adding new regression trees is determined, such as selecting the optimal number of regression trees based on the performance of the validation set, and balancing the contribution between the new trees and the original trees. Input real-time electronic expansion valve opening (and aligned real-time operating condition characteristics) to the dynamic baseline model Dynamic benchmark model , by dynamic benchmark model Output the standard cooling capacity under the current operating conditions. , by dynamic benchmark model Output the standard energy efficiency ratio under the current operating conditions. ; Real-time cooling capacity is collected synchronously through the unit's sensors. (For example, calculated using chilled water flow rate and temperature difference), real-time energy efficiency ratio (For example, calculated using cooling capacity and input power), where , Based on the calculated standard cooling capacity and standard energy efficiency ratio as benchmarks, the real-time degradation rate of the cooling capacity of the refrigeration unit is calculated respectively. Real-time degradation rate of energy efficiency ratio : , ; The real-time score of the refrigeration unit is calculated by combining the real-time degradation rate of its cooling capacity and the real-time degradation rate of its energy efficiency ratio. : .

[0018] It should be further explained that, in the specific implementation process, the process of determining the status of the refrigeration unit based on its real-time score and the configuration judgment threshold includes: If the real-time score of the refrigeration unit If the refrigeration unit is in a healthy state, operating normally and with excellent performance; if If the refrigeration unit is in a sub-healthy state, it is operating well and performing normally; if If the refrigeration unit is in a fault state, it will exhibit abnormal behavior but can still operate, although its performance will be significantly reduced; if If the refrigeration unit is in a state of serious malfunction, it will be difficult to operate and its performance will be severely degraded; if If this occurs, the refrigeration unit is in a state of complete failure and is essentially unable to operate; among them, All are judgment thresholds, and .

[0019] It should be further explained that, in the specific implementation process, a dynamic adjustment strategy is established based on the evaluation results: if the refrigeration unit is detected to be in a faulty, severely faulty, or completely failed state, the process of dynamically adjusting the update cycle of the dynamic benchmark model and the sensor sampling frequency includes: Get the real-time score of the refrigeration unit Real-time degradation rate of cooling capacity Real-time degradation rate of energy efficiency ratio Construct multi-dimensional state quantification indicators: Define performance degradation acceleration It reflects the rate of performance degradation and is used to determine whether degradation is accelerating. In the formula, This represents the change in the average degradation rate. , Let be the average degradation rate at time t. (Cooling capacity and energy efficiency ratio are weighted equally, and adjusted according to the unit's application, such as in scenarios where cooling is prioritized.) ); For time intervals; If performance degradation accelerates This indicates accelerated performance degradation; if the performance degradation acceleration is... This indicates that the degradation rate has slowed down or stabilized; Introducing fault impact factors Quantify the impact of the current fault on the core functions of the unit: In the formula, The number of currently detected fault types (e.g., both condenser blockage and abnormal fan speed exist simultaneously). ); Index of detected fault types; For the first Weighting coefficients for different types of faults (preset based on the degree of fault impact); For the first State coefficient of class-specific faults (when fault is confirmed) When a suspected malfunction occurs When there is no fault ); For the first The degradation rate contribution value corresponding to the type of fault; where the fault impact factor The value range is [0, 1], and the larger the value, the more severe the impact of the fault. Real-time score of integrated refrigeration unit Accelerated performance degradation Fault influencing factors Quantify the urgency of maintenance tasks: In the formula, To maintain the priority index, its value range is [0, 1]; a preset priority threshold is set, i.e. As a high priority, Medium priority Low priority; All are preset weighting coefficients; This is a reverse indicator of the real-time score of the refrigeration unit; the larger the value of this indicator, the higher its priority. It is a positive degradation acceleration, and its priority is increased only when degradation is accelerated; Real-time score based on refrigeration unit With performance degradation acceleration For dynamic benchmark models Dynamic benchmark model The update cycle is adaptively adjusted: In the formula, For update cycle; The baseline update cycle is 24 hours by default, which is the initial cycle in a healthy state. The adjustment coefficient for the real-time score of the refrigeration unit; This is the degradation acceleration adjustment coefficient; Based on fault impact factors With maintenance priority index The sensor sampling frequency is dynamically adjusted. In the formula, The adjusted sampling frequency; The baseline sampling frequency (default setting is 1 time / minute, the sampling frequency under healthy conditions); The adjustment coefficient is the factor for the impact of faults. This is the priority adjustment coefficient.

[0020] Please see Figure 2 This invention provides a dynamic health status assessment system for refrigeration units, comprising: Benchmark model construction module: Acquire the electronic expansion valve opening data stream, and use the gradient boosting regression tree algorithm to establish benchmark model M1 for predicting cooling capacity and benchmark model M2 for energy efficiency ratio, respectively; Dynamic benchmark conversion module: Converts the trained benchmark model M1 and benchmark model M2 into a dynamic benchmark model that responds to real-time operating conditions; inputs the real-time electronic expansion valve opening degree into the dynamic benchmark model, and performs joint calculation of real-time performance indicators and degradation degree to obtain the real-time score of the refrigeration unit; among which, the real-time performance indicators are standard cooling capacity and standard energy efficiency ratio. Health status determination module: Based on the real-time score of the refrigeration unit and the configuration determination threshold, the status of the refrigeration unit is determined, including healthy, sub-healthy, faulty, serious faulty or completely failed status. Dynamic adjustment strategy module: Based on the evaluation results, a dynamic adjustment strategy is established: if the refrigeration unit is detected to be in a fault, severe fault, or complete failure state, the update cycle of the dynamic benchmark model and the sensor sampling frequency are dynamically adjusted; in particular, the dynamic adjustment strategy can adjust the operating parameters of the model and sensors in a timely manner according to changes in the health status of the unit, thereby improving response speed and accuracy.

[0021] Furthermore, the formulas mentioned above are all dimensionless calculations, derived from software simulation using a large amount of collected data to approximate the real situation. The proportionality coefficients in the formulas and the preset thresholds in the analysis process are set by those skilled in the art based on the actual situation or obtained through large-scale data simulation. The magnitude of the proportionality coefficient is a specific value obtained by quantifying each parameter to facilitate subsequent comparison. The magnitude of the proportionality coefficient depends on the amount of sample data and the processing coefficients initially set by those skilled in the art for each set of sample data. As long as it does not affect the proportional relationship between the parameter and the quantified value, it is acceptable.

[0022] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. The focus of each embodiment is on its differences from other embodiments. In particular, the apparatus embodiments are described simply because they are fundamentally based on the method embodiments; relevant details can be found in the descriptions of the method embodiments.

[0023] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0024] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0025] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0026] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0027] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0028] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other. In conclusion, the above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for dynamically assessing the health status of a refrigeration unit, characterized in that: S1. Obtain the electronic expansion valve opening data stream, and use the gradient boosting regression tree algorithm to establish the benchmark model M1 for predicting cooling capacity and the benchmark model M2 for energy efficiency ratio. S2. Transform the trained baseline model M1 and baseline model M2 into a dynamic baseline model that responds to real-time operating conditions; input the real-time electronic expansion valve opening degree into the dynamic baseline model, and perform joint calculation of real-time performance indicators and degradation degree to obtain the real-time score of the refrigeration unit; among which, the real-time performance indicators are standard cooling capacity and standard energy efficiency ratio. S3. Based on the real-time score of the refrigeration unit, compare it with the configuration judgment threshold to determine the status of the refrigeration unit, including healthy, sub-healthy, faulty, serious faulty or completely failed status. S4. Based on the evaluation results, establish a dynamic adjustment strategy: If the refrigeration unit is detected to be in a fault, serious fault, or complete failure state, the update cycle of the dynamic benchmark model and the sensor sampling frequency will be dynamically adjusted.

2. The method for dynamic assessment of the health status of a refrigeration unit according to claim 1, characterized in that: The process of acquiring the electronic expansion valve opening data stream and establishing a baseline model M1 for predicting cooling capacity and a baseline model M2 for energy efficiency ratio using the gradient boosting regression tree algorithm includes: S11. Collect historical operation data from the historical operation database of the chiller unit or refrigeration system: electronic expansion valve opening. The cooling capacity Q and energy efficiency ratio EER were calculated; the historical operation dataset was cleaned, and missing and outlier values ​​were removed; the cleaned dataset was randomly shuffled and divided into training and testing sets according to a preset ratio. S12, the opening degree of the electronic expansion valve The main input features are normalized; at the same time, derived features closely related to the physical relationship between cooling capacity Q and energy efficiency ratio EER are constructed. S13, based on the pre-processed electronic expansion valve opening... The derived features share a common input feature set. By replacing the output target, models M1 and M2 are trained separately within the same gradient boosting regression tree training framework. The process is as follows: select the gradient boosting regression tree regressor, preset the total number of iterations, and initialize the initial prediction values ​​of the two models; train iteratively round by round, for a total of... In each round, model M1 and model M2 are updated synchronously; the iterative training process includes calculating the negative gradient, training a new regression tree, calculating the output values ​​of the leaf nodes, and updating the overall prediction function. S14. Use the partitioned test set to evaluate the performance of benchmark model M1 and benchmark model M2 respectively.

3. The method for dynamic assessment of the health status of a refrigeration unit according to claim 2, characterized in that: The process of iterative model training includes: A1. For each sample i in the training set, calculate the negative gradient between the current model's prediction and the true value. ; A2. Using the entire current training set, with the electronic expansion valve opening... As input, the calculated negative gradient To achieve the output target, train a new regression tree and use this new regression tree to divide the data into J leaf nodes. Where J represents the number of leaf nodes in the m-th tree; R is the region that divides the input space into multiple non-overlapping regions. This represents the j-th leaf node region in the m-th tree; A3. Calculate an output value for each leaf node j of this new tree. ,in, This represents the output value of the j-th leaf node in the m-th tree; A4. Add the new tree to the ensemble model and update the overall prediction function; fine-tune the hyperparameters shared by models M1 and M2 on the training set through cross-validation, and find the optimal hyperparameter combination that minimizes the validation set error of models M1 and M2; use the optimal hyperparameter combination to retrain on the entire training set to obtain the final baseline model M1 and baseline model M2.

4. The method for dynamic assessment of the health status of a refrigeration unit according to claim 1, characterized in that: The trained baseline model M1 and baseline model M2 are transformed into dynamic baseline models that respond to real-time operating conditions. The process of inputting the real-time electronic expansion valve opening into the dynamic benchmark model and performing joint calculations of real-time performance indicators and degradation degree to obtain the real-time score of the refrigeration unit includes: The trained baseline models M1 and M2 are packaged into a callable model file package and deployed to the real-time monitoring device of the chiller unit; when the device starts up, all parameters of the baseline models M1 and M2 are automatically loaded. Real-time acquisition of current operating condition data of the unit; alignment of real-time operating condition data with the common input feature set of benchmark model M1 and benchmark model M2; If there is a difference between the real-time operating conditions and the operating conditions range of the training data for the baseline models M1 and M2, a dynamic calibration mechanism is initiated; the model obtained after calibration becomes the dynamic baseline model. With dynamic benchmark model ; Input real-time electronic expansion valve opening and aligned real-time operating condition features to the dynamic baseline model Dynamic benchmark model , by dynamic benchmark model Output the standard cooling capacity under the current operating conditions. , by dynamic benchmark model Output the standard energy efficiency ratio under the current operating conditions. ; Real-time cooling capacity is collected synchronously through the unit's sensors. Real-time energy efficiency ratio Based on the calculated standard cooling capacity and standard energy efficiency ratio as benchmarks, the real-time degradation rate of the cooling capacity of the refrigeration unit is calculated respectively. Real-time degradation rate of energy efficiency ratio ; The real-time score of the refrigeration unit is calculated by combining the real-time degradation rate of its cooling capacity and the real-time degradation rate of its energy efficiency ratio. .

5. The method for dynamic assessment of the health status of a refrigeration unit according to claim 1, characterized in that: The process of determining the status of a refrigeration unit based on its real-time score and configuration threshold includes: If the real-time score of the refrigeration unit If so, the refrigeration unit is in a healthy state; if If so, the refrigeration unit is in a sub-healthy state; if If so, the refrigeration unit is in a fault state; if If so, the refrigeration unit is in a state of serious malfunction; if If this occurs, the refrigeration unit is in a state of complete failure; among which, All of these are judgment thresholds.

6. The method for dynamic assessment of the health status of a refrigeration unit according to claim 1, characterized in that: Based on the evaluation results, a dynamic adjustment strategy is established: if the refrigeration unit is detected to be in a faulty, severely faulty, or completely failed state, the process of dynamically adjusting the update cycle of the dynamic baseline model and the sensor sampling frequency includes: Get the real-time score of the refrigeration unit Real-time degradation rate of cooling capacity Real-time degradation rate of energy efficiency ratio Construct multi-dimensional state quantification indicators: Define performance degradation acceleration It reflects the rate of performance degradation and is used to determine whether degradation is accelerating. If performance degradation accelerates This indicates accelerated performance degradation; if the performance degradation acceleration is... This indicates that the degradation rate has slowed down or stabilized; Introducing fault impact factors Quantify the impact of the current fault on the function of the refrigeration unit; Real-time score of integrated refrigeration unit Accelerated performance degradation Fault influencing factors This quantifies the urgency of maintenance tasks, thereby obtaining a maintenance priority index. ; Real-time score based on refrigeration unit With performance degradation acceleration For dynamic benchmark models Dynamic benchmark model The update cycle is adaptively adjusted: In the formula, For update cycle; The baseline update cycle; The adjustment coefficient for the real-time score of the refrigeration unit; This is the degradation acceleration adjustment coefficient; Based on fault impact factors With maintenance priority index The sensor sampling frequency is dynamically adjusted. In the formula, The adjusted sampling frequency; The reference sampling frequency; The adjustment coefficient is the factor for the impact of faults. This is the priority adjustment coefficient.

7. A dynamic health status assessment system for refrigeration units, characterized in that, The system, which is applied to a dynamic health assessment method for a refrigeration unit as described in any one of claims 1-6, comprises: Benchmark model construction module: Acquire the electronic expansion valve opening data stream, and use the gradient boosting regression tree algorithm to establish benchmark model M1 for predicting cooling capacity and benchmark model M2 for energy efficiency ratio, respectively; Dynamic benchmark conversion module: Converts the trained benchmark model M1 and benchmark model M2 into a dynamic benchmark model that responds to real-time operating conditions; inputs the real-time electronic expansion valve opening degree into the dynamic benchmark model, and performs joint calculation of real-time performance indicators and degradation degree to obtain the real-time score of the refrigeration unit; among which, the real-time performance indicators are standard cooling capacity and standard energy efficiency ratio. Health status determination module: Based on the real-time score of the refrigeration unit and the configuration determination threshold, the status of the refrigeration unit is determined, including healthy, sub-healthy, faulty, serious faulty or completely failed status. Dynamic adjustment strategy module: Based on the evaluation results, a dynamic adjustment strategy is established: if the refrigeration unit is detected to be in a fault, serious fault or complete failure state, the update cycle of the dynamic benchmark model and the sensor sampling frequency are dynamically adjusted.

Images