Refrigerating unit control method based on parameter prediction and threshold correction and related equipment

Abstract

The invention relates to the technical field of refrigerating unit control, in particular to a refrigerating unit control method based on parameter prediction and threshold correction and related equipment. The method comprises the steps of collecting and preprocessing operation parameters, generating an air return pressure prediction curve through a long short-term memory network, dynamically adjusting a starting pressure threshold value, recognizing potential fault features and correcting the threshold value, and monitoring pressure changes to trigger backup system switching or restrain mistaken switching. Preventive maintenance is realized, the reliability and efficiency of the system are improved, and the cost is reduced.

Description

Technical Field

[0001] This application relates to the field of refrigeration unit control technology, and in particular to refrigeration unit control methods and related equipment based on parameter prediction and threshold correction. Background Technology

[0002] In modern industrial and commercial sectors, refrigeration units, as critical equipment, are widely used in food processing, cold chain logistics, data center cooling, and other scenarios. Currently, the control methods and systems for refrigeration units mainly rely on traditional monitoring and control modes. These typically monitor the operating parameters (such as temperature, pressure, and current) of the refrigeration unit using preset fixed thresholds. When the operating parameters exceed the preset thresholds, the system issues an alarm signal, prompting the operator to take action. For example, in publicly available related technologies, the backup system is only activated when the compressor return gas pressure exceeds the hardware switching threshold; this control method is a typical reactive measure.

[0003] However, this traditional control method has serious shortcomings. It only uses parameter exceeding a threshold as the basis for alarms, and cannot detect potential problems in time when operating parameters show abnormal trends even before exceeding the threshold. This makes it difficult to meet the needs of preventive maintenance of refrigeration units. Due to the lack of prediction of abnormal parameter trends, equipment failures often occur without warning, which not only forces the interruption of production and operation activities, causing significant economic losses, but also shortens the service life of refrigeration units due to frequent failures, and greatly increases equipment maintenance and replacement costs.

[0004] Therefore, based on the above problems, the existing technology still needs to be improved. Summary of the Invention

[0005] The purpose of this application is to provide a control method and related equipment for refrigeration units based on parameter prediction and threshold correction, which aims to solve the problems that traditional refrigeration unit control methods cannot predict abnormal trends in operating parameters and are difficult to implement preventive maintenance, thus leading to sudden equipment failures, production interruptions and increased maintenance costs.

[0006] The purpose of this application is to provide a control method for refrigeration units based on parameter prediction and threshold correction, including: The compressor's operating parameters are collected in real time, and the operating parameters are preprocessed by filtering outliers and standardizing them. The preprocessed operating parameters are modeled using a long short-term memory network to generate a predicted return gas pressure curve for a future preset time period. When the deviation between the return gas pressure prediction curve and the hardware switching threshold of the mechanical backup system is less than the system's preset safety margin, the start-up pressure threshold of the main computer control system is dynamically increased according to a preset proportional coefficient to reduce the compressor load. Construct an operational parameter correlation matrix, analyze the combined characteristics of return gas pressure fluctuation and current deviation in real time, and identify potential compressor fault characteristics; If the potential fault characteristics are detected, a delay time threshold correction value and a pressure switching threshold correction value are generated based on the preset rules of the correlation matrix to obtain the adjusted delay threshold and the pressure threshold of the dynamic safe zone. The switching control system monitors the return air pressure changes in real time based on the start-up pressure threshold, the delay threshold, and the pressure threshold. It can trigger the mechanical backup system switching in advance by a preset time before the main computer control system fails, or suppress erroneous switching caused by load fluctuations.

[0007] By adopting the above technical solution, long short-term memory networks are used to model the trend of compressor operating parameters and generate a return gas pressure prediction curve. This allows for the prediction of parameter change trends before a fault occurs. Combined with dynamic adjustment of the start-up pressure threshold, the compressor load is effectively reduced, and the equipment service life is extended. By constructing an operating parameter correlation matrix to analyze the combined characteristics of return gas pressure fluctuations and current deviations, potential faults are accurately identified. Based on preset rules, delay time thresholds and pressure switching thresholds are corrected, allowing the system to flexibly adjust control strategies according to actual conditions. The switching control system monitors pressure changes in real time based on the adjusted thresholds. This not only triggers the mechanical backup system to switch in time before the main computer control system fails, ensuring the continuous and stable operation of the refrigeration unit, but also suppresses erroneous switching caused by load fluctuations, improving system reliability and stability. Ultimately, this achieves intelligent and refined control of the refrigeration unit, significantly improving equipment operating efficiency and reducing maintenance costs and fault risks.

[0008] In one possible implementation of this application, the method further includes: When applied to scenarios involving multiple refrigeration units in parallel, a unit status correlation matrix is ​​established, and the return gas pressure prediction curve, the start-up pressure threshold, and the upper and lower limits of the dynamic safety range of each unit are exchanged in real time. Based on the unit status correlation matrix, start-up and shutdown sequence optimization instructions for each unit are generated to stagger the mechanical backup switching time by a preset interval. When the predicted pressure of any unit reaches the lower limit of the corresponding dynamic safety range, the load distribution is adjusted by increasing the operating frequency of other units, thereby reducing the rate of return gas pressure rise of that unit and delaying the switching time.

[0009] By adopting the above technical solutions, in scenarios where multiple refrigeration units are connected in parallel, a unit status correlation matrix is ​​established and key parameters are exchanged in real time, enabling each unit to share operating information and achieve data collaboration and linkage control. The generated start-stop sequence optimization instructions can effectively stagger the mechanical backup switching time, avoiding excessive instantaneous system load caused by simultaneous switching of multiple units, and reducing the impact on the power grid and refrigeration system. When the predicted pressure of a certain unit reaches the lower limit of the dynamic safety range, the load is redistributed by adjusting the operating frequency of other units. This not only reduces the rate of return gas pressure rise of that unit and delays the switching time, but also balances the workload of each unit, avoiding over-operation of a single unit and extending the overall equipment lifespan. The above methods significantly improve the stability, reliability, and collaborative efficiency of multiple units operating in parallel, reduce the risk of system failure, optimize energy utilization, and reduce unnecessary energy consumption and maintenance costs.

[0010] In one possible implementation of this application, if the potential fault characteristics are detected, the steps of generating a delay time threshold correction value and a pressure switching threshold correction value based on a preset rule of the correlation matrix, and obtaining the adjusted delay threshold and the pressure threshold of the dynamic safe zone, include: A temperature correction coefficient model is established based on historical data of the external ambient temperature and compressor return gas pressure of the refrigeration unit. Real-time acquisition of external ambient temperature data of the refrigeration unit, calculation of temperature correction coefficient through the temperature correction coefficient model, and superposition onto the dynamic safety range; When the rate of fluctuation of ambient temperature exceeds the preset rate, a fast response mechanism is activated to temporarily increase the preset safety margin of the system by a preset percentage.

[0011] By adopting the above technical solution, a temperature correction coefficient model is established based on historical data of the external ambient temperature and compressor return gas pressure of the refrigeration unit. This model can uncover the potential correlation between ambient temperature and return gas pressure, making the pressure threshold adjustment more closely match actual operating conditions. Real-time acquisition of ambient temperature data and calculation of the correction coefficient, superimposed on the dynamic safety range, allows for dynamic optimization of the pressure threshold based on changes in ambient temperature, preventing system misjudgments or omissions due to temperature fluctuations. When the rate of ambient temperature fluctuation exceeds a preset rate, a rapid response mechanism is activated, temporarily expanding the system's preset safety margin. This quickly enhances the system's fault tolerance when ambient temperature changes drastically, effectively addressing the challenges posed by extreme ambient temperatures, reducing the risk of failures caused by environmental factors, improving the stability and reliability of the refrigeration unit under different environmental conditions, and ensuring the system always operates in a safe and efficient state.

[0012] In one possible implementation of this application, the steps of performing outlier filtering and standardization preprocessing on the operating parameters include: Establish a sensor data mutual verification mechanism and calculate the theoretical return gas pressure value based on the compressor thermodynamic model; When the deviation between the actual collected return gas pressure value and the theoretical return gas pressure value exceeds a preset threshold, a sensor fault warning is triggered. Activate backup sensor data or use a Kalman filter algorithm to reconstruct the faulty sensor data to ensure the accuracy of outlier filtering.

[0013] By adopting the above technical solutions, a sensor data cross-verification mechanism is established, and the theoretical return gas pressure value is calculated in conjunction with the compressor thermodynamic model. This enables cross-verification of the accuracy of sensor-collected data from the data source, overcoming the limitations of traditional reliance on single sensor data. When the deviation between the actual collected value and the theoretical value exceeds a preset threshold, an early warning is triggered, which can promptly detect potential sensor faults and prevent erroneous operating parameters from entering the subsequent control process due to sensor anomalies. Activating backup sensor data or using the Kalman filter algorithm to reconstruct fault data can ensure the continuity and accuracy of data when sensors fail, ensuring the reliability of outlier filtering. This provides a precise data foundation for subsequent core control steps such as predictive modeling and fault identification based on operating parameters, effectively improving the stability and anti-interference capability of the refrigeration unit control system and reducing the risk of equipment failure caused by data errors.

[0014] In one possible implementation of this application, the step of using a long short-term memory network to model the trend of the preprocessed operating parameters includes: The confidence interval of the predicted return gas pressure curve is calculated in real time. When the prediction error exceeds the confidence interval, the relevant model is triggered to adaptively optimize. An online learning algorithm is used to iteratively update the weight matrix of the Long Short-Term Memory network using the latest collected operating parameters; When a sudden change in operating conditions is detected, the system switches to a pre-trained sub-model for specific operating conditions to predict return gas pressure.

[0015] By adopting the above technical solutions, the confidence interval of the return gas pressure prediction curve is calculated in real time, and the prediction error is used as the trigger condition for model optimization. This allows for dynamic evaluation of the reliability of the prediction results, ensuring that the accuracy of the prediction curve remains at a high level. The online learning algorithm, combined with the latest operating parameters, iteratively updates the weight matrix of the long short-term memory network, enabling the model to adapt to changes in operating conditions during the refrigeration unit's operation in real time, continuously optimizing prediction capabilities and avoiding prediction failures due to changes in data distribution. When a sudden change in operating conditions is detected, the model quickly switches to a pre-trained sub-model for specific operating conditions, effectively addressing complex scenarios such as sudden load changes and rapid changes in ambient temperature. This compensates for the prediction deficiencies of general models under extreme conditions, significantly improving the timeliness and accuracy of return gas pressure prediction, providing reliable data support for the intelligent control of the refrigeration unit, thereby more accurately predicting the equipment's operating status and reducing the probability of failure.

[0016] In one possible implementation of this application, the step of identifying potential compressor fault characteristics includes: The combined features of return gas pressure fluctuation and current deviation are mapped to a multi-dimensional feature space to generate a fault feature vector. Time series analysis is performed on the fault feature vector to extract dynamic feature parameters, including pressure fluctuation period, current deviation duration, pressure fluctuation amplitude change rate, and current deviation rise rate. A graded prediction model is trained based on historical fault data. The input of the graded prediction model is the dynamic feature parameters, and the output is the fault severity level. Input the current fault feature vector into the hierarchical prediction model and output the corresponding risk assessment result; Based on the risk assessment results, the corresponding threshold correction strategy type is matched from the preset correction strategy library.

[0017] By adopting the above technical solution, the combined characteristics of return gas pressure fluctuation and current deviation are mapped to a multi-dimensional feature space to generate a fault feature vector. This overcomes the limitations of traditional single-parameter analysis and comprehensively characterizes the compressor's operating status from multiple dimensions. Time series analysis of the fault feature vector and extraction of dynamic feature parameters such as pressure fluctuation period and current deviation duration can deeply explore the temporal patterns in the fault development process and capture early subtle changes in potential faults. Based on historical fault data, a graded prediction model is trained, realizing automated judgment from data to fault severity level, which is more objective and accurate than manual experience judgment. Inputting the current fault feature vector into the model to obtain risk assessment results and matching the corresponding threshold correction strategy type according to the results, the system can implement differentiated processing according to the severity of the fault. This not only improves the sensitivity and accuracy of fault identification but also allows for timely and precise adjustment of control strategies at different stages of potential fault development, effectively preventing fault escalation, reducing equipment downtime, improving the reliability and maintenance efficiency of refrigeration unit operation, and reducing overall operation and maintenance costs.

[0018] In one possible implementation of this application, the step of dynamically increasing the start-up pressure threshold of the main computer control system according to a preset proportional coefficient to reduce the compressor load includes: Establish an energy cost model by inputting five parameters: real-time electricity price, compressor energy efficiency ratio, current load rate, ambient temperature, and cooling capacity reserve. When electricity prices are at their peak, the optimal proportional coefficient is calculated based on the energy cost model. Within the system's preset safety margin, the preset proportional coefficient is lowered to reduce the compressor's start-up frequency. When electricity prices are at their lowest point, the optimal ratio coefficient is calculated based on the energy cost model, and the preset ratio coefficient is increased to reserve cooling capacity in advance, thereby achieving peak shaving and valley filling.

[0019] By adopting the above technical solutions, an energy cost model is established that covers multiple dimensions of parameters, including real-time electricity price, compressor energy efficiency ratio, current load rate, ambient temperature, and cooling capacity reserve. This model comprehensively considers key factors affecting the operating cost of refrigeration units. During peak electricity price periods, the optimal proportional coefficient is calculated based on the energy cost model, and the preset proportional coefficient is lowered to effectively reduce the compressor start-up frequency, reduce equipment operating energy consumption during high electricity price periods, and directly reduce electricity costs. During off-peak electricity price periods, the proportional coefficient is increased to reserve cooling capacity in advance, achieving peak shaving and valley filling. This fully utilizes off-peak electricity prices to reduce energy expenditure while ensuring cooling capacity supply during peak periods, avoiding equipment damage caused by frequent start-ups and shutdowns. This method of dynamically adjusting the start-up pressure threshold based on energy costs combines refrigeration unit operation with grid electricity prices, achieving dual optimization of energy consumption and operating costs. This significantly improves the economy and energy utilization efficiency of the refrigeration system, while reducing the load impact on the power grid and promoting the rational allocation and efficient utilization of energy.

[0020] The second objective of this application is to provide a refrigeration unit control system based on parameter prediction and threshold correction, the system comprising: Operating parameter acquisition and processing module: Acquires compressor operating parameters in real time, and performs outlier filtering and standardization preprocessing on the operating parameters; Return gas pressure prediction curve generation module: Utilizes a long short-term memory network to perform trend modeling on the preprocessed operating parameters and generates a return gas pressure prediction curve for a future preset time period; Start-up pressure threshold adjustment module: When the deviation between the return gas pressure prediction curve and the hardware switching threshold of the mechanical backup system is less than the system's preset safety margin, the start-up pressure threshold of the main computer control system is dynamically increased according to a preset proportional coefficient to reduce the compressor load. Potential fault feature identification module: Constructs an operational parameter correlation matrix, analyzes the combined characteristics of return gas pressure fluctuation and current deviation in real time, and identifies potential fault features of the compressor; Delay threshold and pressure threshold adjustment module: If the potential fault characteristics are detected, a delay time threshold correction value and a pressure switching threshold correction value are generated based on the preset rules of the correlation matrix to obtain the adjusted delay threshold and the pressure threshold of the dynamic safe zone. Mechanical backup system switching module: The switching control system monitors the return air pressure change in real time based on the start-up pressure threshold, the delay threshold and the pressure threshold. It can trigger the mechanical backup system switching in advance after a preset time before the computer main control system fails, or suppress false switching caused by load fluctuations.

[0021] By adopting the above technical solution, long short-term memory networks are used to model the trend of compressor operating parameters and generate a return gas pressure prediction curve. This allows for the prediction of parameter change trends before a fault occurs. Combined with dynamic adjustment of the start-up pressure threshold, the compressor load is effectively reduced, and the equipment service life is extended. By constructing an operating parameter correlation matrix to analyze the combined characteristics of return gas pressure fluctuations and current deviations, potential faults are accurately identified. Based on preset rules, delay time thresholds and pressure switching thresholds are corrected, allowing the system to flexibly adjust control strategies according to actual conditions. The switching control system monitors pressure changes in real time based on the adjusted thresholds. This not only triggers the mechanical backup system to switch in time before the main computer control system fails, ensuring the continuous and stable operation of the refrigeration unit, but also suppresses erroneous switching caused by load fluctuations, improving system reliability and stability. Ultimately, this achieves intelligent and refined control of the refrigeration unit, significantly improving equipment operating efficiency and reducing maintenance costs and fault risks.

[0022] The third objective of this application is to provide a refrigeration unit control device based on parameter prediction and threshold correction, the device comprising: The system includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed by the processor to perform the aforementioned refrigeration unit control method based on parameter prediction and threshold correction.

[0023] The fourth objective of this application is to provide a storage medium.

[0024] The fourth objective of this application is achieved through the following technical solution: A storage medium storing a computer program capable of being loaded by a processor and executed by the aforementioned chiller control method based on parameter prediction and threshold correction.

[0025] In summary, this application includes at least one of the following beneficial technical effects: 1. By utilizing a Long Short-Term Memory (LSTM) network to model the trends of compressor operating parameters and generate a return gas pressure prediction curve, the system can predict parameter change trends in advance before a fault occurs. Combined with dynamic adjustment of the start-up pressure threshold, this effectively reduces compressor load and extends equipment lifespan. Furthermore, by constructing an operating parameter correlation matrix to analyze the combined characteristics of return gas pressure fluctuations and current deviations, the system accurately identifies potential faults. Based on preset rules, it corrects delay time thresholds and pressure switching thresholds, allowing the system to flexibly adjust control strategies according to actual conditions. The switching control system monitors pressure changes in real time based on the adjusted thresholds. This not only triggers the mechanical backup system to switch in time before the main computer control system fails, ensuring the continuous and stable operation of the refrigeration unit, but also suppresses erroneous switching caused by load fluctuations, improving system reliability and stability. Ultimately, this achieves intelligent and refined control of the refrigeration unit, significantly improving equipment operating efficiency and reducing maintenance costs and fault risks.

[0026] 2. In scenarios involving multiple refrigeration units operating in parallel, a unit status correlation matrix is ​​established, and key parameters are exchanged in real time. This allows each unit to share operational information, enabling data collaboration and coordinated control. The generated start-stop sequence optimization commands effectively stagger the mechanical backup switching time, preventing simultaneous switching of multiple units from causing excessive instantaneous system load and reducing the impact on the power grid and refrigeration system. When the predicted pressure of a unit reaches the lower limit of the dynamic safety range, load redistribution is achieved by adjusting the operating frequency of other units. This not only reduces the rate of return gas pressure rise of that unit and delays the switching time but also balances the workload of each unit, preventing excessive operation of a single unit and extending the overall equipment lifespan. These methods significantly improve the stability, reliability, and collaborative efficiency of multiple units operating in parallel, reduce the risk of system failure, optimize energy utilization, and reduce unnecessary energy consumption and maintenance costs. Attached Figure Description

[0027] Figure 1 This is a flowchart illustrating the chiller unit control method based on parameter prediction and threshold correction provided in the embodiments of this application; Figure 2 This is a schematic diagram of the virtual structure of the refrigeration unit control system based on parameter prediction and threshold correction provided in the embodiments of this application. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0029] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article, unless otherwise specified, generally indicates that the preceding and following related objects have an "or" relationship.

[0030] The embodiments of this application will now be described in further detail with reference to the accompanying drawings.

[0031] This application provides a refrigeration unit control method based on parameter prediction and threshold correction, referring to... Figure 1 The main process of the method is described as follows: S1: Real-time acquisition of compressor operating parameters, and outlier filtering and standardization preprocessing of the operating parameters; The system uses temperature, pressure, and current sensors to collect compressor operating parameters such as return gas pressure, exhaust temperature, and operating current once per second. A sensor data cross-verification mechanism is established. The theoretical return gas pressure value is calculated based on the compressor's thermodynamic model. When the actual collected return gas pressure value deviates from the theoretical value by more than ±3%, a sensor fault warning is triggered. If a backup sensor is available, data is automatically switched to the backup sensor; otherwise, a Kalman filter algorithm is used to reconstruct the data from the faulty sensor. After outlier processing, a normalization method is used to map the operating parameters uniformly to the [0, 1] interval, eliminating dimensional influences and facilitating subsequent analysis.

[0032] S2: Use a long short-term memory network to model the trend of the preprocessed operating parameters and generate a return gas pressure prediction curve for a future preset time period; The preprocessed operating parameters were divided into fixed-length sequence samples in chronological order. Each sample contained parameter data from the past 30 minutes, which served as input to a Long Short-Term Memory (LSTM) network. The LSTM network was configured with a two-layer structure, with 64 neurons in each layer. It was trained using the backpropagation algorithm and the Adam optimizer to minimize the mean squared error between the predicted and actual return gas pressure values. After the model was trained and put into operation, the latest collected parameter data was input in real time to predict the return gas pressure trend over the next 15 minutes, outputting a continuous return gas pressure prediction curve.

[0033] S3: When the deviation between the return gas pressure prediction curve and the hardware switching threshold of the mechanical backup system is less than the system's preset safety margin, the start-up pressure threshold of the main computer control system is dynamically increased according to a preset proportional coefficient to reduce the compressor load. The system's preset safety margin is set at 8% of the hardware switching threshold. The initial value of the proportional coefficient is preset to 1.0. When the deviation between the return gas pressure prediction curve and the hardware switching threshold enters the safety margin range, this coefficient is dynamically adjusted based on actual operating conditions. By comprehensively considering factors such as compressor operational stability, within the safety margin range, the preset proportional coefficient is increased, thereby raising the start-up pressure threshold of the main computer control system. For example, if the original start-up pressure threshold is 8 Bar, after the proportional coefficient is increased to 1.1, the start-up pressure threshold is adjusted to 8.8 Bar, reducing unnecessary compressor starts and lowering its load.

[0034] S4: Construct an operational parameter correlation matrix, analyze the combined characteristics of return gas pressure fluctuation and current deviation in real time, and identify potential compressor fault characteristics; The collected return gas pressure fluctuation amplitude and current deviation data are used to construct a two-dimensional operational parameter correlation matrix. By calculating the correlation of data in the matrix, the intrinsic relationship between return gas pressure fluctuation and current deviation is analyzed. A sliding window method is used to analyze the data in the correlation matrix in real time, extracting characteristic parameters such as pressure fluctuation period and current deviation duration. Based on historical normal operation data and fault data, judgment rules for potential fault characteristics are established. When the characteristic parameters analyzed in real time meet the judgment rules, the compressor is determined to have potential fault characteristics.

[0035] S5: If the potential fault characteristics are detected, generate a delay time threshold correction value and a pressure switching threshold correction value based on the preset rules of the correlation matrix to obtain the adjusted delay threshold and the pressure threshold of the dynamic safety zone. Upon detecting potential fault characteristics, the delay time threshold correction value and pressure switching threshold correction value are calculated according to pre-defined rules in the correlation matrix. For example, if a more severe potential fault characteristic is detected, the delay time threshold is shortened by 20%, and the pressure switching threshold safety range is tightened by ±0.1 Bar; if a less severe potential fault characteristic is detected, the correction magnitude is adjusted appropriately. By correcting the delay time threshold and pressure switching threshold, the adjusted delay threshold and the upper and lower limits of the dynamic safety range pressure are redefined, providing more accurate parameter basis for subsequent control.

[0036] S6: The switching control system monitors the return air pressure change in real time based on the start-up pressure threshold, the delay threshold and the pressure threshold, and triggers the mechanical backup system switching in advance for a preset time before the computer main control system fails, or suppresses erroneous switching caused by load fluctuations.

[0037] The switching control system collects return gas pressure data in real time with a period of 100ms and compares it with the adjusted start-up pressure threshold, delay threshold, and dynamic safety zone pressure threshold. When the return gas pressure continues to drop and reaches the lower limit of the dynamic safety zone, and the duration exceeds the delay threshold, if the main computer control system is still running, the system will trigger the mechanical backup system to start in advance, ensuring that the backup system can take over in time in the event of a failure in the main control system. If the return gas pressure fluctuation is detected to be caused by a brief load fluctuation, and the fluctuation amplitude does not actually reach the danger zone, the system will adjust the compressor's operating parameters to suppress unnecessary switching of the mechanical backup system due to misjudgment, thus ensuring the stable operation of the refrigeration unit.

[0038] Specifically, in some possible embodiments, the method further includes: When applied to scenarios involving multiple refrigeration units in parallel, a unit status correlation matrix is ​​established, and the return gas pressure prediction curve, the start-up pressure threshold, and the upper and lower limits of the dynamic safety range of each unit are exchanged in real time. Based on the unit status correlation matrix, start-up and shutdown sequence optimization instructions for each unit are generated to stagger the mechanical backup switching time by a preset interval. When the predicted pressure of any unit reaches the lower limit of the corresponding dynamic safety range, the load distribution is adjusted by increasing the operating frequency of other units, thereby reducing the rate of return gas pressure rise of that unit and delaying the switching time.

[0039] When applied to scenarios involving multiple refrigeration units operating in parallel, a unit status correlation matrix is ​​established via industrial Ethernet. The predicted return gas pressure curves, start-up pressure thresholds, and upper and lower limits of the dynamic safety range for each unit are exchanged in real-time at 100ms intervals. The matrix uses an adjacency matrix structure to store the correlation coefficients of operating parameters between units. The status matrix data is processed using a particle swarm optimization algorithm to generate start-up and shutdown sequence optimization instructions that stagger mechanical backup switching times by at least 5 minutes, preventing grid impact caused by simultaneous switching of multiple units. When the predicted pressure of any unit reaches the lower limit of the dynamic safety range, the operating frequency of other units is increased via a fuzzy PID controller (the adjustment range does not exceed 20% of the rated frequency), and a load balancing algorithm is simultaneously activated to redistribute cooling demand, reducing the rate of return gas pressure rise of that unit by at least 0.1 Bar / min and delaying the switching time by at least 15 seconds. Minutes; In addition, an extended set of unit fault redundancy mechanism is set. When a unit is taken out of operation due to a fault, the state association matrix automatically updates the topology structure, redistributes its load according to the energy efficiency ratio weight of the remaining units, and triggers the prediction model of other units to adaptively switch to the fault emergency sub-model to ensure the stability of the system cooling capacity.

[0040] Specifically, in some possible embodiments, if the potential fault characteristics are detected, the steps of generating a delay time threshold correction value and a pressure switching threshold correction value based on the association matrix preset rules, and obtaining the adjusted delay threshold and the pressure threshold of the dynamic safe zone, include: A temperature correction coefficient model is established based on historical data of the external ambient temperature and compressor return gas pressure of the refrigeration unit. Real-time acquisition of external ambient temperature data of the refrigeration unit, calculation of temperature correction coefficient through the temperature correction coefficient model, and superposition onto the dynamic safety range; When the rate of fluctuation of ambient temperature exceeds the preset rate, a fast response mechanism is activated to temporarily increase the preset safety margin of the system by a preset percentage.

[0041] Specifically, a temperature correction coefficient model is established using a support vector regression algorithm based on historical data of the external ambient temperature and compressor return gas pressure of the refrigeration unit. The external ambient temperature data of the refrigeration unit is collected in real time, and the correction coefficient corresponding to the current temperature is calculated through the temperature correction coefficient model (e.g., the correction coefficient is 1.05 when the ambient temperature is 35℃). This coefficient is linearly superimposed on the upper and lower limits of the dynamic safety range. When the rate of fluctuation of the ambient temperature exceeds the preset rate of 0.3℃ / min, a fast response mechanism is activated to temporarily increase the preset safety margin of the system by 30%. At the same time, the temperature trend prediction model (using the ARIMA algorithm) is triggered to predict the ambient temperature change in the next 10 minutes. If the predicted temperature continues to rise, the safety margin is further increased to 50% to ensure the stability of the system under extreme ambient temperature changes.

[0042] Specifically, in some possible embodiments, the steps of outlier filtering and standardization preprocessing of the operating parameters include: Establish a sensor data mutual verification mechanism and calculate the theoretical return gas pressure value based on the compressor thermodynamic model; When the deviation between the actual collected return gas pressure value and the theoretical return gas pressure value exceeds a preset threshold, a sensor fault warning is triggered. Activate backup sensor data or use a Kalman filter algorithm to reconstruct the faulty sensor data to ensure the accuracy of outlier filtering.

[0043] The system establishes a sensor data mutual verification mechanism. Based on a compressor thermodynamic model (combining the laws of mass and energy conservation), the theoretical return gas pressure is calculated. Model input parameters include real-time data such as compressor power, exhaust temperature, and refrigerant flow rate. When the deviation between the actual collected return gas pressure and the theoretical value exceeds a preset threshold of ±5%, a three-level early warning mechanism is triggered (yellow warning: deviation 5%-8%, orange warning: 8%-12%, red warning: >12%). For yellow warnings, the system automatically starts using backup sensor data for comparison and verification. For orange and red warnings, it immediately switches to backup sensor data and uses a Kalman filter algorithm to reconstruct the faulty sensor data (the state vector contains 8 parameters including pressure, temperature, and flow rate; the process noise covariance matrix is ​​determined through maximum likelihood estimation of historical data). Simultaneously, the sensor fault diagnosis process is initiated, using wavelet transform to analyze the frequency domain characteristics of the sensor signal and locate the fault source. Outlier filtering uses a sliding window midpoint filtering method (window size of 11 sampling points) to ensure that the accuracy of the preprocessed data meets the input requirements of the subsequent LSTM model.

[0044] Specifically, in some possible embodiments, the step of using a long short-term memory network to model the trend of the preprocessed operating parameters includes: The confidence interval of the predicted return gas pressure curve is calculated in real time. When the prediction error exceeds the confidence interval, the relevant model is triggered to adaptively optimize. An online learning algorithm is used to iteratively update the weight matrix of the Long Short-Term Memory network using the latest collected operating parameters; When a sudden change in operating conditions is detected, the system switches to a pre-trained sub-model for specific operating conditions to predict return gas pressure.

[0045] The system employs a real-time Bootstrap resampling method to calculate the 95% confidence interval of the return gas pressure prediction curve based on 100 sets of past prediction data and actual data. When the latest prediction error exceeds the boundary of the confidence interval, the model's adaptive optimization program is triggered. The online learning algorithm uses a stochastic gradient descent combined with momentum optimization strategy. Every 20 sets of newly collected operating parameter data are used to iteratively update the weight matrix of the Long Short-Term Memory (LSTM) network, with a learning rate set to 0.001 and an exponential decay strategy employed. The system detects sudden changes in operating conditions by calculating the rate of change of operating parameters. When the rate of change of return gas pressure exceeds 0.2 Bar / min, the rate of change of current exceeds 10% / min, or the rate of change of ambient temperature exceeds 1℃ / min, it is determined to be a sudden change in operating conditions. The system immediately switches to a pre-trained sub-model for a specific operating condition (such as a high-temperature operating condition sub-model or a high-load operating condition sub-model) to predict return gas pressure. The training data for the sub-model comes from 1000 sets of data samples under the corresponding historical operating conditions, ensuring the accuracy and stability of the prediction results during sudden changes in operating conditions.

[0046] Specifically, in some possible embodiments, the steps for identifying potential compressor fault characteristics include: The combined features of return gas pressure fluctuation and current deviation are mapped to a multi-dimensional feature space to generate a fault feature vector. Time series analysis is performed on the fault feature vector to extract dynamic feature parameters, including pressure fluctuation period, current deviation duration, pressure fluctuation amplitude change rate, and current deviation rise rate. A graded prediction model is trained based on historical fault data. The input of the graded prediction model is the dynamic feature parameters, and the output is the fault severity level. Input the current fault feature vector into the hierarchical prediction model and output the corresponding risk assessment result; Based on the risk assessment results, the corresponding threshold correction strategy type is matched from the preset correction strategy library.

[0047] Specifically, the combined features of return gas pressure fluctuation and current deviation are mapped to a four-dimensional feature space (pressure fluctuation amplitude, fluctuation frequency, mean current deviation, and deviation variance) through principal component analysis to generate a standardized fault feature vector. Wavelet transform is used to perform time-frequency analysis on the fault feature vector to extract dynamic feature parameters such as pressure fluctuation period, current deviation duration, pressure fluctuation amplitude change rate, and current deviation rise rate. Based on historical fault data (including 100 cases of bearing wear, 80 cases of valve plate damage, and 120 cases of refrigerant leakage), a random forest algorithm is used to train a hierarchical prediction model. The current fault feature vector is input into the trained hierarchical prediction model, which outputs three levels of risk assessment results: Level I (emergency), Level II (early warning), and Level III (caution). According to the risk level, the corresponding threshold correction strategy type is matched from the preset correction strategy library (e.g., Level I risk triggers an emergency shutdown plan, while shortening the delay threshold by 50% and narrowing the pressure safety range by 0.2 Bar; Level II risk starts the backup system preheating, and shortens the delay threshold by 30%; Level III risk...). For level-one risks, only control parameters are adjusted and the monitoring cycle is extended to achieve accurate fault identification and differentiated response.

[0048] Specifically, in some possible embodiments, the step of dynamically increasing the start-up pressure threshold of the main computer control system according to a preset proportional coefficient to reduce the compressor load includes: Establish an energy cost model by inputting five parameters: real-time electricity price, compressor energy efficiency ratio, current load rate, ambient temperature, and cooling capacity reserve. When electricity prices are at their peak, the optimal proportional coefficient is calculated based on the energy cost model. Within the system's preset safety margin, the preset proportional coefficient is lowered to reduce the compressor's start-up frequency. When electricity prices are at their lowest point, the optimal ratio coefficient is calculated based on the energy cost model, and the preset ratio coefficient is increased to reserve cooling capacity in advance, thereby achieving peak shaving and valley filling.

[0049] The system establishes an energy cost model, employing a multiple linear regression algorithm to construct the mathematical relationship between five parameters—real-time electricity price, compressor energy efficiency ratio, current load rate, ambient temperature, and cooling capacity reserve—and operating costs. The system obtains real-time electricity price data every 15 minutes through the electricity supplier interface, calculates the energy efficiency ratio in real time based on compressor operating parameters, and simultaneously obtains the cooling capacity reserve through a cooling capacity monitoring device. When electricity prices are detected to be in peak hours (e.g., 10:00-15:00 daily), five parameters are input into the energy cost model. A genetic algorithm is used to calculate the optimal proportional coefficient that minimizes operating costs. Within the system's preset safety margin (10% of the hardware switching threshold), the preset proportional coefficient is lowered, for example, from 1.2 to 0.9, which raises the start-up pressure threshold from 8 Bar to 8.8 Bar, reducing the number of compressor starts. When electricity prices are in off-peak hours (e.g., 23:00-7:00 the next day), the same parameters are input to calculate the optimal proportional coefficient and it is raised, for example, to 1.5, which lowers the start-up pressure threshold to 7.5 Bar, causing the compressor to run at a higher frequency, storing cooling capacity in advance, achieving peak shaving and valley filling, and reducing overall operating costs.

[0050] Another embodiment of this application provides a refrigeration unit control system based on parameter prediction and threshold correction, wherein, see reference Figure 2 A chiller unit control system based on parameter prediction and threshold correction includes: Operating parameter acquisition and processing module 100: Acquires the operating parameters of the compressor in real time, and performs outlier filtering and standardization preprocessing on the operating parameters; Return gas pressure prediction curve generation module 200: uses a long short-term memory network to perform trend modeling on the preprocessed operating parameters and generates a return gas pressure prediction curve for a future preset time period. Start-up pressure threshold adjustment module 300: When the deviation between the return gas pressure prediction curve and the hardware switching threshold of the mechanical backup system is less than the system preset safety margin, the start-up pressure threshold of the main computer control system is dynamically increased according to the preset proportional coefficient to reduce the compressor load. Potential Fault Feature Identification Module 400: Constructs an operating parameter correlation matrix, analyzes the combined characteristics of return gas pressure fluctuation and current deviation in real time, and identifies potential fault features of the compressor; Delay threshold and pressure threshold adjustment module 500: If the potential fault characteristics are detected, a delay time threshold correction value and a pressure switching threshold correction value are generated based on the preset rules of the correlation matrix to obtain the adjusted delay threshold and the pressure threshold of the dynamic safety zone. Mechanical backup system switching module 600: The switching control system monitors the return air pressure change in real time based on the start-up pressure threshold, the delay threshold and the pressure threshold, and triggers mechanical backup system switching in advance for a preset time before the computer main control system fails, or suppresses erroneous switching caused by load fluctuations.

[0051] The refrigeration unit control system based on parameter prediction and threshold correction provided in this embodiment can realize the steps of the aforementioned embodiments due to the functions of each module and the logical connections between them. Therefore, it can achieve the same technical effect as the aforementioned embodiments. For the principle analysis, please refer to the relevant description of the steps of the aforementioned refrigeration unit control method based on parameter prediction and threshold correction, which will not be repeated here.

[0052] This application also provides a refrigeration unit control device based on parameter prediction and threshold correction, including a memory and a processor. The memory stores a computer program that can be loaded by the processor and executed by the above-described refrigeration unit control method based on parameter prediction and threshold correction.

[0053] This application also provides a storage medium storing a computer program that can be loaded by a processor and executed by the above-described refrigeration unit control method based on parameter prediction and threshold correction.

[0054] The storage medium provided in this embodiment can achieve the same technical effect as the aforementioned embodiments because the computer program therein, after being loaded and run on the processor, will implement the various steps of the aforementioned embodiments. For the principle analysis, please refer to the relevant description of the aforementioned method steps, which will not be repeated here.

[0055] The storage medium includes, for example, various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0056] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0057] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0058] Furthermore, features defined by the terms "first" and "second" may explicitly or implicitly include at least one of those features. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., and unless otherwise explicitly specified, is used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features.

[0059] Therefore, any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

[0060] The embodiments described in this specific implementation are preferred embodiments of this application and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A method of refrigeration unit control based on parameter prediction and threshold correction, characterized in that, The method comprises the following steps: Real-time acquisition of compressor operating parameters, and abnormal value filtering and standardization preprocessing of the operating parameters; Using a long short-term memory network to model the trend of the preprocessed operating parameters, and generating a return gas pressure prediction curve for a future preset time period; When the deviation of the return gas pressure prediction curve and the hardware switching threshold of the mechanical backup system is less than the system preset safety margin, the startup pressure threshold of the computer main control system is dynamically adjusted by a preset proportion coefficient to reduce the compressor load; Constructing an operating parameter correlation matrix, real-time analysis of the combined characteristics of return gas pressure fluctuation and current deviation, and identification of potential compressor fault characteristics; If the potential fault characteristics are detected, generate a delay time threshold correction value and a pressure switching threshold correction value based on the correlation matrix preset rules to obtain the adjusted delay threshold and the pressure threshold of the dynamic safety interval; The switching control system monitors the return gas pressure changes based on the startup pressure threshold, the delay threshold, and the pressure threshold, and triggers the mechanical backup system to switch in advance by a preset time before the computer main control system fails, or suppresses the false switching caused by load fluctuation.

2. The refrigerant unit control method based on parameter prediction and threshold correction according to claim 1, characterized by, The method further comprises the following steps: When applied to a parallel refrigeration unit scene, a unit state correlation matrix is established to exchange the return gas pressure prediction curve, the startup pressure threshold, and the upper and lower limit values of the dynamic safety interval of each unit in real time; Based on the unit state correlation matrix, generate each unit's start-stop sequence optimization instruction to stagger the mechanical backup switching time by a preset interval; When the predicted pressure of any unit reaches the lower limit value of the corresponding dynamic safety interval, adjust the load distribution by increasing the operating frequency of other units to reduce the return gas pressure rise rate of the unit and delay the switching time.

3. The refrigerant unit control method based on parameter prediction and threshold correction according to claim 1, characterized by, If the potential fault characteristics are detected, generate a delay time threshold correction value and a pressure switching threshold correction value based on the correlation matrix preset rules to obtain the adjusted delay threshold and the pressure threshold of the dynamic safety interval, which comprises the following steps: Based on the historical data of the external environment temperature of the refrigeration unit and the compressor return gas pressure, a temperature correction coefficient model is established; Real-time acquisition of refrigeration unit external environment temperature data, calculation of temperature correction coefficient through the temperature correction coefficient model, and superposition on the dynamic safety interval; When the environmental temperature fluctuation rate exceeds the preset rate, a fast response mechanism is enabled to temporarily expand the system preset safety margin by a preset percentage.

4. The refrigerant unit control method based on parameter prediction and threshold correction according to claim 1, characterized by, The step of abnormal value filtering and standardization preprocessing of the operating parameters comprises the following steps: Establish a sensor data mutual verification mechanism, and calculate the theoretical return gas pressure value based on the compressor thermodynamic model; When the deviation of the actual collected return gas pressure value and the theoretical return gas pressure value exceeds the preset threshold, trigger the sensor fault warning; Start the backup sensor data or use the Kalman filter algorithm to reconstruct the fault sensor data to ensure the accuracy of abnormal value filtering.

5. The refrigerant unit control method based on parameter prediction and threshold correction of claim 1, wherein, The step of using a long short-term memory network to model the trend of the preprocessed operating parameters comprises the following steps: Real-time calculation of the confidence interval of the return gas pressure prediction curve, and adaptive optimization of the related model when the prediction error exceeds the confidence interval; An online learning algorithm is used to iteratively update the weight matrix of the long short-term memory network using the latest collected operating parameters. When a sudden change in operating condition is detected, switch to the pre-trained specific operating condition sub-model for return gas pressure prediction.

6. The refrigerant unit control method based on parameter prediction and threshold correction of claim 1, wherein, The step of identifying potential compressor fault features includes: Mapping the combined features of return gas pressure fluctuation and current deviation to a multi-dimensional feature space to generate a fault feature vector; Performing time series analysis on the fault feature vector to extract dynamic feature parameters, including pressure fluctuation period, current deviation duration, pressure fluctuation amplitude change rate, and current deviation rise rate; Training a hierarchical prediction model based on historical fault data, with the dynamic feature parameters as input and the fault severity level as output; Input the current fault feature vector into the hierarchical prediction model to output the corresponding risk assessment result; According to the risk assessment result, match the corresponding threshold correction strategy type from the pre-set correction strategy library.

7. The refrigerant unit control method based on parameter prediction and threshold correction of claim 1, wherein, The step of dynamically adjusting the startup pressure threshold of the computer main control system by a preset proportion coefficient to reduce the compressor load includes: Establish an energy cost model by inputting five parameters: real-time electricity price, compressor energy efficiency ratio, current load rate, environmental temperature, and cold reserve capacity; When the electricity price is in the peak period, calculate the optimal proportion coefficient based on the energy cost model, and within the system's pre-set safety margin, lower the pre-set proportion coefficient to reduce the compressor startup frequency; When the electricity price is in the valley period, calculate the optimal proportion coefficient based on the energy cost model, and increase the pre-set proportion coefficient to reserve cold energy in advance, achieving peak load shifting.

8. A chiller unit control system based on parameter prediction with threshold correction, characterized by, Comprise: Operating parameter acquisition and processing module: real-time acquisition of compressor operating parameters, and abnormal value filtering and standardization preprocessing of the operating parameters; Return gas pressure prediction curve generation module: using long short-term memory network to model the trend of preprocessed operating parameters, generating return gas pressure prediction curve for future preset time period; Startup pressure threshold adjustment module: when the deviation between the return gas pressure prediction curve and the hardware switching threshold of the mechanical backup system is less than the system's pre-set safety margin, dynamically adjust the startup pressure threshold of the computer main control system by a preset proportion coefficient to reduce the compressor load; Potential fault feature identification module: construct an operating parameter correlation matrix, and real-time analyze the combined features of return gas pressure fluctuation and current deviation to identify potential compressor fault features; Delay threshold and pressure threshold adjustment module: if the potential fault feature is detected, generate delay time threshold correction value and pressure switching threshold correction value based on the correlation matrix pre-set rules to obtain the adjusted delay threshold and dynamic safety interval pressure threshold; Mechanical backup system switching module: the switching control system monitors the return gas pressure changes based on the startup pressure threshold, the delay threshold, and the pressure threshold, triggers the mechanical backup system switching in advance by a preset time before the computer main control system fails, or suppresses the false switching caused by load fluctuation.

9. Refrigeration unit control apparatus based on parameter prediction and threshold correction, characterized in that Comprise: The memory and processor, wherein the memory stores a computer program that can be loaded by the processor and executed the chiller control method based on parameter prediction and threshold correction according to any one of claims 1-7.

10. A storage medium, characterized by The computer program is stored in which the refrigeration unit control method based on parameter prediction and threshold correction of any one of claims 1-7 can be loaded by a processor and executed.

Images