Intelligent control method and system for compressor based on dynamic pressure relief and storage medium

Abstract

The application provides a compressor intelligent control method and system based on dynamic pressure relief, and a storage medium. Firstly, a four-dimensional sensing network of pressure, temperature, vibration and current is deployed to capture the multi-physical field coupling characteristics of the compressor in a shutdown state in real time. Secondly, based on a preset pressure entropy model, the pressure fluctuation intensity, temperature anomaly index, high-frequency vibration factor and current attenuation coefficient are fused to quantitatively evaluate the pressure entropy value, and a three-level pressure relief mode is triggered based on the pressure entropy value. Then, based on the deviation of the cavity pressure and the target pressure, the pressure relief parameters are dynamically adjusted to realize smooth transition of the pressure. Finally, the full-link data of the pressure relief operation is collected, and the pressure entropy model parameters are dynamically optimized based on reinforcement learning to enable the system to continuously adapt to equipment aging and working condition changes. Through dynamic closed-loop adjustment, the application reduces the waste of pressure relief energy consumption and equipment loss, and improves the stability of the pressure relief process.

Description

Technical Field

[0001] This invention relates to the field of refrigeration equipment, and more specifically, to a method, system, and storage medium for intelligent control of compressors based on dynamic pressure relief. Background Technology

[0002] In the field of compressor control, pressure relief is a crucial step in ensuring the safe start-up of equipment. Traditional compressor pressure relief control generally employs a timed pressure relief mechanism, which executes the pressure relief operation according to a preset fixed time. Even after the compressor has stopped and the cavity pressure has naturally dropped to a safe range, the fixed-duration pressure relief operation will still be forcibly initiated, resulting in wasted energy. Furthermore, there are issues of insufficient and excessive pressure relief. When the fixed time is insufficient to release high pressure, the compressor starts under load, causing current surges and accelerating the aging of the motor windings; when pressure relief is prolonged under low pressure conditions, lubricating oil backflow occurs, causing dry friction in the bearings and reducing the compressor's lifespan.

[0003] Currently, existing technologies exist that control the pressure relief time based on pressure feedback. However, these rely on a single pressure sensor and cannot capture the coupling effects of multiple physical fields such as temperature, vibration, and current. Furthermore, they typically use fixed pressure relief parameters, which cannot be adapted to the actual operating conditions of the compressor. Therefore, there is an urgent need for a dynamic pressure relief intelligent control technology. Summary of the Invention

[0004] In view of the above problems, the purpose of this invention is to provide a compressor intelligent control method, system and storage medium based on dynamic pressure relief. By deploying a four-dimensional sensor network of pressure, temperature, vibration and current, a complete state profile of the compressor is constructed; by integrating pressure dynamic characteristics, temperature characteristics and mechanical state, a quantitative risk assessment is conducted; graded pressure relief is implemented based on risk level, and pressure relief parameters are continuously adjusted in combination with pressure drop rate to achieve smooth pressure transition; and model parameters are continuously optimized based on operating data to adapt to changes throughout the entire life cycle of the equipment.

[0005] The first aspect of this invention provides a compressor intelligent control method based on dynamic pressure relief, the method comprising:

[0006] Multi-source heterogeneous data is collected through a pre-set sensor array;

[0007] Based on the preset cleaning and feature extraction algorithm, feature operation parameters are obtained according to the multi-source heterogeneous data;

[0008] Based on the preset pressure entropy model, the pressure entropy value is obtained according to the characteristic operating parameters;

[0009] Based on a preset graded threshold pressure relief strategy, the pressure relief mode is determined according to the pressure entropy value;

[0010] Based on the preset pressure relief valve parameter table, the pressure relief parameters are set according to the pressure relief mode;

[0011] After performing the pressure relief operation, monitor the chamber pressure in real time;

[0012] The pressure relief parameters are dynamically adjusted based on the deviation between the cavity pressure and the preset target pressure.

[0013] Based on the cavity pressure, when it is determined that the cavity pressure is in a preset stable state, the compressor is started and the pressure relief operation data is recorded;

[0014] The pressure relief operation data is used as a training set, and the weights of the pressure entropy model are updated according to a preset self-training period.

[0015] In this solution, the acquisition of multi-source heterogeneous data through a preset sensor array specifically includes:

[0016] Pressure data is collected synchronously by pressure sensors deployed in the compressor's intake chamber, exhaust chamber, and oil separator to obtain pressure change curves.

[0017] The temperature distribution map is obtained by real-time monitoring of the compressor cylinder surface temperature using an infrared temperature sensor array.

[0018] Vibration curves of the compressor in a stopped state are captured using a triaxial vibration sensor;

[0019] The residual current curve of the compressor is detected using a Hall effect current sensor.

[0020] The pressure change curve, temperature distribution map, vibration curve, and residual current curve are combined with timestamps to form multi-source heterogeneous data.

[0021] In this scheme, the feature operation parameters obtained based on the preset cleaning and feature extraction algorithm and the multi-source heterogeneous data are as follows:

[0022] Based on the pressure change curve, filtering is performed using a preset sliding window, and the pressure change rate and standard deviation within each window are calculated to obtain a pressure feature vector.

[0023] Based on the temperature distribution map, the number, area, and temperature value of local overheated areas are extracted to obtain a temperature feature vector;

[0024] The vibration curve is processed by a preset wavelet packet decomposition to obtain high-frequency vibration characteristics;

[0025] Based on the slope analysis of the residual current curve, the current attenuation coefficient is obtained;

[0026] By integrating pressure characteristic vector, temperature characteristic vector, high-frequency vibration characteristics, and current attenuation coefficient, characteristic operating parameters are obtained.

[0027] In this solution, obtaining the pressure entropy value based on the preset pressure entropy model and the characteristic operating parameters specifically includes:

[0028] Obtain the first pressure weight and the second pressure weight;

[0029] Based on the temperature feature vector, a pressure compensation coefficient is determined to adjust the first pressure weight and the second pressure weight.

[0030] The pressure change factor is obtained by multiplying the adjusted first pressure weight and the pressure change rate of the pressure eigenvector.

[0031] The product of the adjusted second pressure weight and the pressure standard deviation of the pressure eigenvector is calculated to obtain the pressure standard deviation factor.

[0032] Based on the preset correction coefficient mapping relationship, the correction coefficients are obtained according to the high-frequency vibration characteristics;

[0033] The pressure entropy value is obtained based on the pressure change factor, pressure standard deviation factor, and correction coefficient.

[0034] If the current attenuation coefficient exceeds the preset attenuation threshold, the pressure entropy value will be positively compensated.

[0035] In this scheme, the pressure relief strategy based on a preset graded threshold, which determines the pressure relief mode according to the pressure entropy value, specifically includes:

[0036] If the pressure entropy value is lower than the preset first pressure entropy threshold, it is determined to be a zero pressure relief mode, and the pressure relief time and pressure relief opening are both set to the minimum value;

[0037] If the pressure entropy value is between the preset first pressure entropy threshold and the preset second pressure entropy threshold, it is determined to be a short-term pressure relief mode. The pressure relief opening is adjusted according to the pressure relief entropy value, and the pressure relief operation is performed according to the preset first pressure relief time.

[0038] If the pressure entropy value is higher than the preset second pressure entropy threshold, it is determined to be a forced pressure relief mode. Based on the preset three-level pressure relief strategy, pressure relief parameters are set and pressure relief operation is performed.

[0039] In this solution, dynamically adjusting the pressure relief parameters based on the deviation between the cavity pressure and the preset target pressure specifically includes:

[0040] When the forced pressure relief mode is determined, the pressure drop rate is calculated based on the cavity pressure.

[0041] If the pressure drop rate is greater than the preset upper limit of the drop rate, the pressure relief opening is reduced stepwise according to the preset step size.

[0042] If the pressure drop rate is less than the preset lower limit of the drop rate, the pressure relief opening is increased stepwise according to the preset step size.

[0043] The predicted pressure relief time is obtained based on the deviation between the cavity pressure and the preset target pressure and the pressure drop rate.

[0044] If the predicted depressurization time exceeds the set depressurization time threshold, the depressurization opening is increased and the upper limit of the descent rate is raised.

[0045] A second aspect of the present invention provides a compressor intelligent control system based on dynamic pressure relief, including a compressor intelligent control method program based on dynamic pressure relief, wherein the compressor intelligent control method program based on dynamic pressure relief, when executed by the processor, performs the following steps:

[0046] Multi-source heterogeneous data is collected through a pre-set sensor array;

[0047] Based on the preset cleaning and feature extraction algorithm, feature operation parameters are obtained according to the multi-source heterogeneous data;

[0048] Based on the preset pressure entropy model, the pressure entropy value is obtained according to the characteristic operating parameters;

[0049] Based on a preset graded threshold pressure relief strategy, the pressure relief mode is determined according to the pressure entropy value;

[0050] Based on the preset pressure relief valve parameter table, the pressure relief parameters are set according to the pressure relief mode;

[0051] After performing the pressure relief operation, monitor the chamber pressure in real time;

[0052] The pressure relief parameters are dynamically adjusted based on the deviation between the cavity pressure and the preset target pressure.

[0053] Based on the cavity pressure, when it is determined that the cavity pressure is in a preset stable state, the compressor is started and the pressure relief operation data is recorded;

[0054] The pressure relief operation data is used as a training set, and the weights of the pressure entropy model are updated according to a preset self-training period.

[0055] In this solution, the acquisition of multi-source heterogeneous data through a preset sensor array specifically includes:

[0056] Pressure data is collected synchronously by pressure sensors deployed in the compressor's intake chamber, exhaust chamber, and oil separator to obtain pressure change curves.

[0057] The temperature distribution map is obtained by real-time monitoring of the compressor cylinder surface temperature using an infrared temperature sensor array.

[0058] Vibration curves of the compressor in a stopped state are captured using a triaxial vibration sensor;

[0059] The residual current curve of the compressor is detected using a Hall effect current sensor.

[0060] The pressure change curve, temperature distribution map, vibration curve, and residual current curve are combined with timestamps to form multi-source heterogeneous data.

[0061] In this scheme, the feature operation parameters obtained based on the preset cleaning and feature extraction algorithm and the multi-source heterogeneous data are as follows:

[0062] Based on the pressure change curve, filtering is performed using a preset sliding window, and the pressure change rate and standard deviation within each window are calculated to obtain a pressure feature vector.

[0063] Based on the temperature distribution map, the number, area, and temperature value of local overheated areas are extracted to obtain a temperature feature vector;

[0064] The vibration curve is processed by a preset wavelet packet decomposition to obtain high-frequency vibration characteristics;

[0065] Based on the slope analysis of the residual current curve, the current attenuation coefficient is obtained;

[0066] By integrating pressure characteristic vector, temperature characteristic vector, high-frequency vibration characteristics, and current attenuation coefficient, characteristic operating parameters are obtained.

[0067] A third aspect of the present invention provides a computer-readable storage medium comprising a program for a compressor intelligent control method based on dynamic pressure relief, wherein when the program is executed by a processor, it implements the steps of the compressor intelligent control method based on dynamic pressure relief as described in any of the preceding claims.

[0068] This invention provides a compressor intelligent control method, system, and storage medium based on dynamic pressure relief. First, a four-dimensional sensor network (pressure, temperature, vibration, and current) is deployed to capture the multi-physics coupling characteristics of the compressor in real time during shutdown. Second, based on a preset pressure entropy model, pressure fluctuation intensity, temperature anomaly index, high-frequency vibration factor, and current attenuation coefficient are integrated to quantitatively evaluate the pressure entropy value, and a three-stage pressure relief mode is triggered based on the pressure entropy value. Then, based on the deviation between the cavity pressure and the target pressure, the pressure relief parameters are dynamically adjusted to achieve a smooth pressure transition. Finally, data from the entire pressure relief operation chain is collected, and the pressure entropy model parameters are dynamically optimized based on reinforcement learning, enabling the system to continuously adapt to equipment aging and changes in operating conditions. This invention reduces energy waste and equipment wear during pressure relief through dynamic closed-loop regulation, and improves the stability of the pressure relief process. Attached Figure Description

[0069] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope.

[0070] Figure 1 A flowchart of a compressor intelligent control method based on dynamic pressure relief according to the present invention is shown;

[0071] Figure 2 This invention provides a flowchart for integrating multi-source heterogeneous data according to an embodiment of the present invention.

[0072] Figure 3 A flowchart illustrating the extraction of feature operating parameters provided by an embodiment of the present invention is shown;

[0073] Figure 4 A block diagram of a compressor intelligent control system based on dynamic pressure relief according to the present invention is shown. Detailed Implementation

[0074] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0075] Unless otherwise defined, all terms (including technical and scientific terms) used in embodiments of this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should also be understood that terms such as those defined in a common dictionary should be interpreted as having a meaning consistent with their meaning in the context of the relevant art, and not as being interpreted in an idealized or highly formalized sense, unless expressly defined in this embodiment of the invention.

[0076] The terms "first," "second," and similar terms used in the embodiments of this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, terms such as "an," "one," or "the" do not indicate a quantity limitation, but rather indicate the presence of at least one. Likewise, terms such as "including" or "comprising" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects.

[0077] The terms "connection" or "linking" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. The steps preceding or following each other in the method of this invention are not necessarily performed precisely in sequence. Instead, the steps can be processed in reverse order or simultaneously. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0078] In addition, the functional modules in the various embodiments of the present invention can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0079] Figure 1 A flowchart of a compressor intelligent control method based on dynamic pressure relief according to the present invention is shown.

[0080] like Figure 1 As shown, the first aspect of this invention discloses an intelligent control method for a compressor based on dynamic pressure relief, the method comprising:

[0081] S102 collects multi-source heterogeneous data through a preset sensor array;

[0082] S104, Based on the preset cleaning and feature extraction algorithm, the feature operation parameters are obtained according to the multi-source heterogeneous data;

[0083] S106, Based on the preset pressure entropy model, the pressure entropy value is obtained according to the characteristic operating parameters;

[0084] S108, Based on the preset graded threshold pressure relief strategy, determine the pressure relief mode according to the pressure entropy value;

[0085] S110, Based on the preset pressure relief valve parameter table, set the pressure relief parameters according to the pressure relief mode;

[0086] S112, after performing the pressure relief operation, monitor the chamber pressure in real time;

[0087] S114, Based on the deviation between the cavity pressure and the preset target pressure, dynamically adjust the pressure relief parameter;

[0088] S116, Based on the cavity pressure, when it is determined that the cavity pressure is in a preset stable state, start the compressor and record the pressure relief operation data;

[0089] S118, using the pressure relief operation data as a training set, update the weights of the pressure entropy model according to a preset self-training period.

[0090] It should be noted that, in this embodiment, after the system starts, it first synchronously collects heterogeneous data streams such as compressor cavity pressure fluctuation signals, cylinder surface temperature field distribution, mechanical vibration spectrum, and residual current of the motor system through a distributed sensor array. Then, it uses sliding window filtering and wavelet packet decomposition algorithms to denoise the raw data, extracting key characteristic parameters such as extreme values ​​of pressure change rate, local overheating region characteristics, high-frequency vibration energy proportion, and current decay time constant. These characteristic parameters are input into a preset pressure entropy model for fusion calculation, outputting a pressure entropy value characterizing the system's risk level. Based on the threshold range of this entropy value, the system automatically selects zero pressure relief, short-time pressure relief, or forced pressure relief mode and calls the corresponding pressure relief parameter combination. During the pressure relief execution phase, the electronic pressure relief valve initiates pressure relief according to a preset opening degree, while simultaneously monitoring the cavity pressure change curve in real time. When the actual pressure drop rate deviates from the target range, the system dynamically optimizes the pressure relief process through a step-by-step opening adjustment mechanism. Once the pressure value remains stable within the safe range and the mechanical vibration energy drops to within the permissible range, the compressor enters the safe start-up procedure. All process data is encrypted and transmitted to a cloud-based knowledge base. The system periodically updates the feature weight coefficients of the pressure entropy model based on reinforcement learning algorithms, enabling continuous evolution of the control strategy. This embodiment achieves a closed-loop process, significantly reducing the frequency of ineffective pressure relief operations through multi-dimensional state perception and adaptive decision-making mechanisms, thereby improving energy efficiency while ensuring equipment safety.

[0091] Figure 2 A flowchart illustrating the integration process of multi-source heterogeneous data provided by an embodiment of the present invention is shown.

[0092] According to embodiments of the present invention, such as Figure 2 As shown, the process of acquiring multi-source heterogeneous data through a preset sensor array specifically involves:

[0093] S202 uses pressure sensors deployed in the compressor's suction chamber, exhaust chamber, and oil separator to simultaneously collect pressure data and obtain pressure change curves.

[0094] S204 uses an infrared temperature sensor array to monitor the surface temperature of the compressor cylinder in real time and obtain a temperature distribution map;

[0095] S206 uses a triaxial vibration sensor to capture the vibration curve of the compressor when it is stopped;

[0096] S208 uses a Hall effect sensor to detect the compressor's residual current curve;

[0097] S210, the pressure change curve, temperature distribution map, vibration curve and residual current curve are combined with timestamps to form multi-source heterogeneous data.

[0098] It should be noted that this embodiment employs four types of sensors to cover the mechanical, thermal, and electromagnetic core system states. A high-density sensor network is deployed at key nodes of the compressor, including three sets of high-precision pressure sensors embedded in the intake chamber, exhaust chamber, and oil separator inner wall, respectively, capturing transient pressure fluctuations at millisecond-level sampling frequencies and generating pressure-time change curves; an infrared temperature sensor array arranged around the cylinder block uses scanning temperature measurement, collecting and refreshing the surface temperature distribution thermal map according to a preset period; a triaxial vibration sensor installed on the compressor base continuously collects broadband mechanical vibration signals, generating a vibration energy spectrum through fast Fourier transform; and a current Hall sensor connected in series in the power supply circuit records the residual current decay trajectory of the electromagnetic system after power failure. All sensor data are timestamped and spatiotemporally aligned and encapsulated by the edge computing unit to construct a state matrix containing a four-dimensional physical field of pressure, temperature, vibration, and current. This embodiment uses a multi-source heterogeneous sensing architecture to eliminate single-point monitoring blind spots, providing comprehensive and accurate underlying data support for subsequent feature extraction.

[0099] Figure 3 A flowchart illustrating the extraction of feature operating parameters provided by an embodiment of the present invention is shown.

[0100] According to embodiments of the present invention, such as Figure 3 As shown, the preset cleaning and feature extraction algorithm, based on the multi-source heterogeneous data, obtains feature operation parameters, specifically as follows:

[0101] S302, Based on the pressure change curve, filter processing is performed based on a preset sliding window, and the pressure change rate and standard deviation within each window are calculated to obtain a pressure feature vector;

[0102] S304, Based on the temperature distribution map, extract the number, area, and temperature value of the local overheated areas to obtain a temperature feature vector;

[0103] S306, The vibration curve is processed based on a preset wavelet packet decomposition to obtain high-frequency vibration characteristics;

[0104] S308, based on the slope analysis of the residual current curve, the current attenuation coefficient is obtained;

[0105] S310 integrates pressure characteristic vector, temperature characteristic vector, high-frequency vibration characteristics, and current attenuation coefficient to obtain characteristic operating parameters.

[0106] It should be noted that this embodiment extracts features from sensor measurement data to obtain feature data. First, the pressure change curve is processed using a sliding window method. Within each time window, the absolute value of the pressure change rate and the statistical standard deviation are calculated, outputting a feature vector characterizing the dynamic properties of the pressure. Second, the temperature distribution map is used to locate local overheated areas using image recognition algorithms, and its numerical coordinates, projected area, and highest temperature value are extracted to form a temperature feature set. Then, the vibration signal is separated into different frequency band components using wavelet packet decomposition technology, and energy integration is performed on the high-frequency components of the set frequency band to quantify the mechanical impact intensity. Finally, the residual current curve is fitted with the attenuation equation using the least squares method, and the time constant is extracted as an electromagnetic system state index. The pressure fluctuation intensity, temperature anomaly index, high-frequency vibration factor, and current attenuation coefficient are combined into a structured feature vector. This process, through signal processing and pattern recognition technology, condenses massive amounts of raw data into feature parameters with clear physical meaning, significantly improving the efficiency of subsequent decision-making.

[0107] According to an embodiment of the present invention, obtaining the pressure entropy value based on the preset pressure entropy model and the characteristic operating parameters specifically includes:

[0108] Obtain the first pressure weight and the second pressure weight;

[0109] Based on the temperature feature vector, a pressure compensation coefficient is determined to adjust the first pressure weight and the second pressure weight.

[0110] The pressure change factor is obtained by multiplying the adjusted first pressure weight and the pressure change rate of the pressure eigenvector.

[0111] The product of the adjusted second pressure weight and the pressure standard deviation of the pressure eigenvector is calculated to obtain the pressure standard deviation factor.

[0112] Based on the preset correction coefficient mapping relationship, the correction coefficients are obtained according to the high-frequency vibration characteristics;

[0113] The pressure entropy value is obtained based on the pressure change factor, pressure standard deviation factor, and correction coefficient.

[0114] If the current attenuation coefficient exceeds the preset attenuation threshold, the pressure entropy value will be positively compensated.

[0115] It should be noted that in this embodiment, after the feature vector is input into the pressure entropy model, the system first dynamically adjusts the weighting coefficients of the pressure change rate and pressure standard deviation based on the distribution of overheated regions in the temperature feature vector. The compensated weighting coefficients are then multiplied by the corresponding pressure feature values ​​to generate the basic pressure entropy components. Simultaneously, based on the intensity of the high-frequency vibration factor, the entropy value is mechanically corrected according to a preset mapping relationship. For every specific percentage increase in vibration energy, the entropy value is correspondingly increased by the correction coefficient. If the current attenuation coefficient exceeds the safety threshold, an additional compensation mechanism is activated to positively gain the entropy value. After all correction terms are superimposed, a dimensionless pressure entropy value within the range [0,1] is output through normalization. This model, by integrating thermodynamic, mechanical, and electromagnetic multi-physics field characteristics, constructs a comprehensive quantitative indicator reflecting the system's risk level, providing a scientific basis for pressure relief decisions.

[0116] According to an embodiment of the present invention, the pressure relief strategy based on a preset graded threshold, which determines the pressure relief mode according to the pressure entropy value, specifically includes:

[0117] If the pressure entropy value is lower than the preset first pressure entropy threshold, it is determined to be a zero pressure relief mode, and the pressure relief time and pressure relief opening are both set to the minimum value;

[0118] If the pressure entropy value is between the preset first pressure entropy threshold and the preset second pressure entropy threshold, it is determined to be a short-term pressure relief mode. The pressure relief opening is adjusted according to the pressure relief entropy value, and the pressure relief operation is performed according to the preset first pressure relief time.

[0119] If the pressure entropy value is higher than the preset second pressure entropy threshold, it is determined to be a forced pressure relief mode. Based on the preset three-level pressure relief strategy, pressure relief parameters are set and pressure relief operation is performed.

[0120] It should be noted that this embodiment provides a three-stage pressure relief strategy based on pressure entropy values. When the entropy value remains below the first pressure entropy threshold, the system is determined to be in a low-pressure stable state, and the pressure relief valve maintains its minimum opening while directly starting the compressor. When the entropy value is between the first and second pressure entropy thresholds, a short-time pressure relief mode is activated, with the pressure relief valve opening increasing linearly with the entropy value and completing the pressure relief operation within a fixed duration. If the entropy value exceeds the second pressure entropy threshold, a forced pressure relief mode is activated. The forced pressure relief mode employs a three-stage pressure relief process: in the first stage, the pressure relief valve is fully opened to achieve a rapid pressure drop; when the cavity pressure drops to 150% of the critical value, a buffer pressure relief stage is entered, with the pressure relief valve opening decaying exponentially; when the pressure approaches 120% of the safety threshold, a fine-tuning pressure relief stage is entered, using pulse width modulation technology to precisely control the opening. Each stage transition must meet preset pressure gradient conditions to ensure a smooth and controllable pressure relief process. The on-demand graded control strategy provided in this embodiment significantly reduces energy waste under low-risk operating conditions.

[0121] According to an embodiment of the present invention, dynamically adjusting the pressure relief parameter based on the deviation between the cavity pressure and the preset target pressure specifically includes:

[0122] When the forced pressure relief mode is determined, the pressure drop rate is calculated based on the cavity pressure.

[0123] If the pressure drop rate is greater than the preset upper limit of the drop rate, the pressure relief opening is reduced stepwise according to the preset step size.

[0124] If the pressure drop rate is less than the preset lower limit of the drop rate, the pressure relief opening is increased stepwise according to the preset step size.

[0125] The predicted pressure relief time is obtained based on the deviation between the cavity pressure and the preset target pressure and the pressure drop rate.

[0126] If the predicted depressurization time exceeds the set depressurization time threshold, the depressurization opening is increased and the upper limit of the descent rate is raised.

[0127] It should be noted that this embodiment provides a dynamically adjusted pressure relief process. After pressure relief is initiated, the system calculates the deviation between the current pressure drop rate and the preset target rate in real time. When the actual rate is lower than the target lower limit, the pressure relief valve opening is increased in a step-wise manner with a fixed step size; conversely, when the rate exceeds the upper limit, the opening is gradually decreased. Simultaneously, based on the current drop rate and the remaining pressure difference, the system predicts the pressure relief time required to reach the safe pressure. If the predicted time exceeds a preset threshold, the system synchronously increases the upper limit of the pressure relief valve opening and widens the permissible range of the pressure drop rate. After each opening adjustment, at least three control cycles must be maintained to observe the response effect, avoiding frequent actions that could cause pressure oscillations. This dynamic parameter adjustment mechanism, through a combination of real-time feedback and forward prediction, effectively prevents pressure runaway while ensuring pressure relief efficiency.

[0128] It is worth mentioning that the update mechanism of the pressure entropy model is also included, specifically:

[0129] Record the key parameters for each pressure relief operation, including pressure entropy value, total pressure relief duration, average opening degree, pressure drop curve, and compressor starting current peak value;

[0130] Based on the preset pressure relief efficiency evaluation model, the pressure relief operation parameters are used to obtain the pressure relief efficiency;

[0131] If the deviation between the pressure relief efficiency and the theoretical pressure relief efficiency is greater than the preset efficiency threshold for three consecutive times, the model parameter optimization program will be triggered.

[0132] The weight coefficients of each feature in the pressure entropy model are adjusted by using reinforcement learning algorithms.

[0133] It should be noted that this embodiment provides an update mechanism for the pressure entropy model. After each pressure relief operation, the system automatically records key parameters such as the pressure entropy value, total pressure relief duration, average opening degree, pressure drop curve shape, and compressor starting current peak value. Based on this data, a pressure relief efficiency evaluation index is constructed. By comparing the deviation between the actual pressure relief time and the theoretical optimal duration, the efficiency of this operation is quantified. When the efficiency evaluation value of three consecutive operations is lower than a certain percentage of the historical best level, a parameter optimization program is triggered. A reinforcement learning algorithm is used to analyze the operation data and adjust the weight coefficients of each feature parameter in the pressure entropy model. The updated model needs to be verified for its control effect in a simulation environment. Only after confirming the performance improvement can it be deployed to the actual system. The data-driven self-evolution mechanism provided in this embodiment enables the control system to continuously adapt to the characteristic changes caused by equipment aging.

[0134] It is worth mentioning that it also includes a mechanism for handling abnormal pressure relief, specifically:

[0135] If the pressure relief time exceeds the preset safety time limit, immediately close the pressure relief valve and cut off the compressor power supply;

[0136] The sensor data consistency self-test, control command transmission status self-test, and mechanical structure evaluation are initiated sequentially.

[0137] Based on the self-test results, switch to a backup sensor, activate the emergency pressure relief channel, or generate a maintenance report.

[0138] It should be noted that this embodiment provides an anomaly handling implementation logic. When the pressure relief duration exceeds the preset safety time limit, the system immediately closes the pressure relief valve and cuts off the compressor's power supply. Subsequently, a three-level diagnostic process is initiated: Level 1 diagnostics verifies the consistency of multi-sensor data and identifies abnormal reading nodes; Level 2 diagnostics traces the control command transmission path and verifies the actuator's response status; Level 3 diagnostics analyzes vibration spectrum characteristics and assesses the mechanical structure's integrity. Based on the diagnostic results, tiered handling is performed: if it is a sensor failure, the system switches to a backup sensor module and reinitializes the pressure relief process; if it is an actuator failure, the emergency pressure relief channel is activated; when mechanical damage is detected, a three-dimensional damage report is generated and the equipment is locked. The entire process generates encrypted audit logs and stores them in a tamper-proof memory, providing a complete chain of evidence for subsequent fault analysis. This embodiment significantly improves the system's fault tolerance through multi-level diagnostics and redundancy design.

[0139] Figure 4 A block diagram of a compressor intelligent control system based on dynamic pressure relief according to the present invention is shown.

[0140] like Figure 4 As shown, the second aspect of the present invention discloses a compressor intelligent control system 4 based on dynamic pressure relief, including a memory 41 and a processor 42. The memory includes a compressor intelligent control method program based on dynamic pressure relief. When the compressor intelligent control method program based on dynamic pressure relief is executed by the processor, it performs the following steps:

[0141] Multi-source heterogeneous data is collected through a pre-set sensor array;

[0142] Based on the preset cleaning and feature extraction algorithm, feature operation parameters are obtained according to the multi-source heterogeneous data;

[0143] Based on the preset pressure entropy model, the pressure entropy value is obtained according to the characteristic operating parameters;

[0144] Based on a preset graded threshold pressure relief strategy, the pressure relief mode is determined according to the pressure entropy value;

[0145] Based on the preset pressure relief valve parameter table, the pressure relief parameters are set according to the pressure relief mode;

[0146] After performing the pressure relief operation, monitor the chamber pressure in real time;

[0147] The pressure relief parameters are dynamically adjusted based on the deviation between the cavity pressure and the preset target pressure.

[0148] Based on the cavity pressure, when it is determined that the cavity pressure is in a preset stable state, the compressor is started and the pressure relief operation data is recorded;

[0149] The pressure relief operation data is used as a training set, and the weights of the pressure entropy model are updated according to a preset self-training period.

[0150] It should be noted that, in this embodiment, after the system starts, it first synchronously collects heterogeneous data streams such as compressor cavity pressure fluctuation signals, cylinder surface temperature field distribution, mechanical vibration spectrum, and residual current of the motor system through a distributed sensor array. Then, it uses sliding window filtering and wavelet packet decomposition algorithms to denoise the raw data, extracting key characteristic parameters such as extreme values ​​of pressure change rate, local overheating region characteristics, high-frequency vibration energy proportion, and current decay time constant. These characteristic parameters are input into a preset pressure entropy model for fusion calculation, outputting a pressure entropy value characterizing the system's risk level. Based on the threshold range of this entropy value, the system automatically selects zero pressure relief, short-time pressure relief, or forced pressure relief mode and calls the corresponding pressure relief parameter combination. During the pressure relief execution phase, the electronic pressure relief valve initiates pressure relief according to a preset opening degree, while simultaneously monitoring the cavity pressure change curve in real time. When the actual pressure drop rate deviates from the target range, the system dynamically optimizes the pressure relief process through a step-by-step opening adjustment mechanism. Once the pressure value remains stable within the safe range and the mechanical vibration energy drops to within the permissible range, the compressor enters the safe start-up procedure. All process data is encrypted and transmitted to a cloud-based knowledge base. The system periodically updates the feature weight coefficients of the pressure entropy model based on reinforcement learning algorithms, enabling continuous evolution of the control strategy. This embodiment achieves a closed-loop process, significantly reducing the frequency of ineffective pressure relief operations through multi-dimensional state perception and adaptive decision-making mechanisms, thereby improving energy efficiency while ensuring equipment safety.

[0151] Figure 2 A flowchart illustrating the integration process of multi-source heterogeneous data provided by an embodiment of the present invention is shown.

[0152] According to embodiments of the present invention, such as Figure 2 As shown, the process of acquiring multi-source heterogeneous data through a preset sensor array specifically involves:

[0153] Pressure data is collected synchronously by pressure sensors deployed in the compressor's intake chamber, exhaust chamber, and oil separator to obtain pressure change curves.

[0154] The temperature distribution map is obtained by real-time monitoring of the compressor cylinder surface temperature using an infrared temperature sensor array.

[0155] Vibration curves of the compressor in a stopped state are captured using a triaxial vibration sensor;

[0156] The residual current curve of the compressor is detected using a Hall effect current sensor.

[0157] The pressure change curve, temperature distribution map, vibration curve, and residual current curve are combined with timestamps to form multi-source heterogeneous data.

[0158] It should be noted that this embodiment employs four types of sensors to cover the mechanical, thermal, and electromagnetic core system states. A high-density sensor network is deployed at key nodes of the compressor, including three sets of high-precision pressure sensors embedded in the intake chamber, exhaust chamber, and oil separator inner wall, respectively, capturing transient pressure fluctuations at millisecond-level sampling frequencies and generating pressure-time change curves; an infrared temperature sensor array arranged around the cylinder block uses scanning temperature measurement, collecting and refreshing the surface temperature distribution thermal map according to a preset period; a triaxial vibration sensor installed on the compressor base continuously collects broadband mechanical vibration signals, generating a vibration energy spectrum through fast Fourier transform; and a current Hall sensor connected in series in the power supply circuit records the residual current decay trajectory of the electromagnetic system after power failure. All sensor data are timestamped and spatiotemporally aligned and encapsulated by the edge computing unit to construct a state matrix containing a four-dimensional physical field of pressure, temperature, vibration, and current. This embodiment uses a multi-source heterogeneous sensing architecture to eliminate single-point monitoring blind spots, providing comprehensive and accurate underlying data support for subsequent feature extraction.

[0159] Figure 3 A flowchart illustrating the extraction of feature operating parameters provided by an embodiment of the present invention is shown.

[0160] According to embodiments of the present invention, such as Figure 3 As shown, the preset cleaning and feature extraction algorithm, based on the multi-source heterogeneous data, obtains feature operation parameters, specifically as follows:

[0161] Based on the pressure change curve, filtering is performed using a preset sliding window, and the pressure change rate and standard deviation within each window are calculated to obtain a pressure feature vector.

[0162] Based on the temperature distribution map, the number, area, and temperature value of local overheated areas are extracted to obtain a temperature feature vector;

[0163] The vibration curve is processed by a preset wavelet packet decomposition to obtain high-frequency vibration characteristics;

[0164] Based on the slope analysis of the residual current curve, the current attenuation coefficient is obtained;

[0165] By integrating pressure characteristic vector, temperature characteristic vector, high-frequency vibration characteristics, and current attenuation coefficient, characteristic operating parameters are obtained.

[0166] It should be noted that this embodiment extracts features from sensor measurement data to obtain feature data. First, the pressure change curve is processed using a sliding window method. Within each time window, the absolute value of the pressure change rate and the statistical standard deviation are calculated, outputting a feature vector characterizing the dynamic properties of the pressure. Second, the temperature distribution map is used to locate local overheated areas using image recognition algorithms, and its numerical coordinates, projected area, and highest temperature value are extracted to form a temperature feature set. Then, the vibration signal is separated into different frequency band components using wavelet packet decomposition technology, and energy integration is performed on the high-frequency components of the set frequency band to quantify the mechanical impact intensity. Finally, the residual current curve is fitted with the attenuation equation using the least squares method, and the time constant is extracted as an electromagnetic system state index. The pressure fluctuation intensity, temperature anomaly index, high-frequency vibration factor, and current attenuation coefficient are combined into a structured feature vector. This process, through signal processing and pattern recognition technology, condenses massive amounts of raw data into feature parameters with clear physical meaning, significantly improving the efficiency of subsequent decision-making.

[0167] According to an embodiment of the present invention, obtaining the pressure entropy value based on the preset pressure entropy model and the characteristic operating parameters specifically includes:

[0168] Obtain the first pressure weight and the second pressure weight;

[0169] Based on the temperature feature vector, a pressure compensation coefficient is determined to adjust the first pressure weight and the second pressure weight.

[0170] The pressure change factor is obtained by multiplying the adjusted first pressure weight and the pressure change rate of the pressure eigenvector.

[0171] The product of the adjusted second pressure weight and the pressure standard deviation of the pressure eigenvector is calculated to obtain the pressure standard deviation factor.

[0172] Based on the preset correction coefficient mapping relationship, the correction coefficients are obtained according to the high-frequency vibration characteristics;

[0173] The pressure entropy value is obtained based on the pressure change factor, pressure standard deviation factor, and correction coefficient.

[0174] If the current attenuation coefficient exceeds the preset attenuation threshold, the pressure entropy value will be positively compensated.

[0175] It should be noted that in this embodiment, after the feature vector is input into the pressure entropy model, the system first dynamically adjusts the weighting coefficients of the pressure change rate and pressure standard deviation based on the distribution of overheated regions in the temperature feature vector. The compensated weighting coefficients are then multiplied by the corresponding pressure feature values ​​to generate the basic pressure entropy components. Simultaneously, based on the intensity of the high-frequency vibration factor, the entropy value is mechanically corrected according to a preset mapping relationship. For every specific percentage increase in vibration energy, the entropy value is correspondingly increased by the correction coefficient. If the current attenuation coefficient exceeds the safety threshold, an additional compensation mechanism is activated to positively gain the entropy value. After all correction terms are superimposed, a dimensionless pressure entropy value within the range [0,1] is output through normalization. This model, by integrating thermodynamic, mechanical, and electromagnetic multi-physics field characteristics, constructs a comprehensive quantitative indicator reflecting the system's risk level, providing a scientific basis for pressure relief decisions.

[0176] According to an embodiment of the present invention, the pressure relief strategy based on a preset graded threshold, which determines the pressure relief mode according to the pressure entropy value, specifically includes:

[0177] If the pressure entropy value is lower than the preset first pressure entropy threshold, it is determined to be a zero pressure relief mode, and the pressure relief time and pressure relief opening are both set to the minimum value;

[0178] If the pressure entropy value is between the preset first pressure entropy threshold and the preset second pressure entropy threshold, it is determined to be a short-term pressure relief mode. The pressure relief opening is adjusted according to the pressure relief entropy value, and the pressure relief operation is performed according to the preset first pressure relief time.

[0179] If the pressure entropy value is higher than the preset second pressure entropy threshold, it is determined to be a forced pressure relief mode. Based on the preset three-level pressure relief strategy, pressure relief parameters are set and pressure relief operation is performed.

[0180] It should be noted that this embodiment provides a three-stage pressure relief strategy based on pressure entropy values. When the entropy value remains below the first pressure entropy threshold, the system is determined to be in a low-pressure stable state, and the pressure relief valve maintains its minimum opening while directly starting the compressor. When the entropy value is between the first and second pressure entropy thresholds, a short-time pressure relief mode is activated, with the pressure relief valve opening increasing linearly with the entropy value and completing the pressure relief operation within a fixed duration. If the entropy value exceeds the second pressure entropy threshold, a forced pressure relief mode is activated. The forced pressure relief mode employs a three-stage pressure relief process: in the first stage, the pressure relief valve is fully opened to achieve a rapid pressure drop; when the cavity pressure drops to 150% of the critical value, a buffer pressure relief stage is entered, with the pressure relief valve opening decaying exponentially; when the pressure approaches 120% of the safety threshold, a fine-tuning pressure relief stage is entered, using pulse width modulation technology to precisely control the opening. Each stage transition must meet preset pressure gradient conditions to ensure a smooth and controllable pressure relief process. The on-demand graded control strategy provided in this embodiment significantly reduces energy waste under low-risk operating conditions.

[0181] According to an embodiment of the present invention, dynamically adjusting the pressure relief parameter based on the deviation between the cavity pressure and the preset target pressure specifically includes:

[0182] When the forced pressure relief mode is determined, the pressure drop rate is calculated based on the cavity pressure.

[0183] If the pressure drop rate is greater than the preset upper limit of the drop rate, the pressure relief opening is reduced stepwise according to the preset step size.

[0184] If the pressure drop rate is less than the preset lower limit of the drop rate, the pressure relief opening is increased stepwise according to the preset step size.

[0185] The predicted pressure relief time is obtained based on the deviation between the cavity pressure and the preset target pressure and the pressure drop rate.

[0186] If the predicted depressurization time exceeds the set depressurization time threshold, the depressurization opening is increased and the upper limit of the descent rate is raised.

[0187] It should be noted that this embodiment provides a dynamically adjusted pressure relief process. After pressure relief is initiated, the system calculates the deviation between the current pressure drop rate and the preset target rate in real time. When the actual rate is lower than the target lower limit, the pressure relief valve opening is increased in a step-wise manner with a fixed step size; conversely, when the rate exceeds the upper limit, the opening is gradually decreased. Simultaneously, based on the current drop rate and the remaining pressure difference, the system predicts the pressure relief time required to reach the safe pressure. If the predicted time exceeds a preset threshold, the system synchronously increases the upper limit of the pressure relief valve opening and widens the permissible range of the pressure drop rate. After each opening adjustment, at least three control cycles must be maintained to observe the response effect, avoiding frequent actions that could cause pressure oscillations. This dynamic parameter adjustment mechanism, through a combination of real-time feedback and forward prediction, effectively prevents pressure runaway while ensuring pressure relief efficiency.

[0188] It is worth mentioning that the update mechanism of the pressure entropy model is also included, specifically:

[0189] Record the key parameters for each pressure relief operation, including pressure entropy value, total pressure relief duration, average opening degree, pressure drop curve, and compressor starting current peak value;

[0190] Based on the preset pressure relief efficiency evaluation model, the pressure relief operation parameters are used to obtain the pressure relief efficiency;

[0191] If the deviation between the pressure relief efficiency and the theoretical pressure relief efficiency is greater than the preset efficiency threshold for three consecutive times, the model parameter optimization program will be triggered.

[0192] The weight coefficients of each feature in the pressure entropy model are adjusted by using reinforcement learning algorithms.

[0193] It should be noted that this embodiment provides an update mechanism for the pressure entropy model. After each pressure relief operation, the system automatically records key parameters such as the pressure entropy value, total pressure relief duration, average opening degree, pressure drop curve shape, and compressor starting current peak value. Based on this data, a pressure relief efficiency evaluation index is constructed. By comparing the deviation between the actual pressure relief time and the theoretical optimal duration, the efficiency of this operation is quantified. When the efficiency evaluation value of three consecutive operations is lower than a certain percentage of the historical best level, a parameter optimization program is triggered. A reinforcement learning algorithm is used to analyze the operation data and adjust the weight coefficients of each feature parameter in the pressure entropy model. The updated model needs to be verified for its control effect in a simulation environment. Only after confirming the performance improvement can it be deployed to the actual system. The data-driven self-evolution mechanism provided in this embodiment enables the control system to continuously adapt to the characteristic changes caused by equipment aging.

[0194] It is worth mentioning that it also includes a mechanism for handling abnormal pressure relief, specifically:

[0195] If the pressure relief time exceeds the preset safety time limit, immediately close the pressure relief valve and cut off the compressor power supply;

[0196] The sensor data consistency self-test, control command transmission status self-test, and mechanical structure evaluation are initiated sequentially.

[0197] Based on the self-test results, switch to a backup sensor, activate the emergency pressure relief channel, or generate a maintenance report.

[0198] It should be noted that this embodiment provides an anomaly handling implementation logic. When the pressure relief duration exceeds the preset safety time limit, the system immediately closes the pressure relief valve and cuts off the compressor's power supply. Subsequently, a three-level diagnostic process is initiated: Level 1 diagnostics verifies the consistency of multi-sensor data and identifies abnormal reading nodes; Level 2 diagnostics traces the control command transmission path and verifies the actuator's response status; Level 3 diagnostics analyzes vibration spectrum characteristics and assesses the mechanical structure's integrity. Based on the diagnostic results, tiered handling is performed: if it is a sensor failure, the system switches to a backup sensor module and reinitializes the pressure relief process; if it is an actuator failure, the emergency pressure relief channel is activated; when mechanical damage is detected, a three-dimensional damage report is generated and the equipment is locked. The entire process generates encrypted audit logs and stores them in a tamper-proof memory, providing a complete chain of evidence for subsequent fault analysis. This embodiment significantly improves the system's fault tolerance through multi-level diagnostics and redundancy design.

[0199] A third aspect of the present invention provides a computer-readable storage medium comprising a program for a compressor intelligent control method based on dynamic pressure relief, wherein when the program is executed by a processor, it implements the steps of the compressor intelligent control method based on dynamic pressure relief as described in any of the preceding claims.

[0200] In summary, this invention provides a compressor intelligent control method, system, and storage medium based on dynamic pressure relief. First, by deploying a four-dimensional sensor network of pressure, temperature, vibration, and current, the multi-physics coupling characteristics of the compressor under shutdown conditions are captured in real time. Second, based on a preset pressure entropy model, pressure fluctuation intensity, temperature anomaly index, high-frequency vibration factor, and current attenuation coefficient are fused to quantitatively evaluate the pressure entropy value, and a three-stage pressure relief mode is triggered based on the pressure entropy value. Then, based on the deviation between the cavity pressure and the target pressure, the pressure relief parameters are dynamically adjusted to achieve a smooth pressure transition. Finally, data from the entire pressure relief operation chain is collected, and the pressure entropy model parameters are dynamically optimized based on reinforcement learning, enabling the system to continuously adapt to equipment aging and changes in operating conditions. This invention reduces energy waste and equipment wear during pressure relief through dynamic closed-loop regulation, and improves the stability of the pressure relief process.

[0201] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0202] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A compressor intelligent control method based on dynamic pressure relief, characterized in that, The method includes: Multi-source heterogeneous data is collected through a pre-set sensor array, specifically including: pressure data is collected synchronously by pressure sensors deployed in the compressor's suction chamber, exhaust chamber, and oil separator to obtain pressure change curves; the compressor cylinder surface temperature is monitored in real time by an infrared temperature sensor array to obtain a temperature distribution map; the vibration curve of the compressor in a stopped state is captured by a triaxial vibration sensor; and the residual current curve of the compressor is detected by a Hall effect current sensor. The pressure change curves, temperature distribution map, vibration curves, and residual current curves are combined with timestamps to form multi-source heterogeneous data. Based on the preset cleaning and feature extraction algorithm, feature operation parameters are obtained according to the multi-source heterogeneous data; Based on a preset pressure entropy model, the pressure entropy value is obtained according to the characteristic operating parameters, specifically including: obtaining a first pressure weight and a second pressure weight; determining a pressure compensation coefficient based on a temperature feature vector to adjust the first pressure weight and the second pressure weight; calculating the product of the adjusted first pressure weight and the pressure change rate of the pressure feature vector to obtain a pressure change factor; calculating the product of the adjusted second pressure weight and the pressure standard deviation of the pressure feature vector to obtain a pressure standard deviation factor; obtaining a correction coefficient based on a preset correction coefficient mapping relationship and high-frequency vibration characteristics; obtaining the pressure entropy value based on the pressure change factor, the pressure standard deviation factor, and the correction coefficient; if the current attenuation coefficient exceeds a preset attenuation threshold, positive compensation is performed on the pressure entropy value. Based on a preset graded threshold pressure relief strategy, the pressure relief mode is determined according to the pressure entropy value. Specifically, if the pressure entropy value is lower than a preset first pressure entropy threshold, it is determined to be a zero pressure relief mode, and the pressure relief time and pressure relief opening are both set to the minimum value; if the pressure entropy value is between the preset first pressure entropy threshold and the preset second pressure entropy threshold, it is determined to be a short-term pressure relief mode, the pressure relief opening is adjusted according to the pressure relief entropy value, and the pressure relief operation is performed according to the preset first pressure relief time; if the pressure entropy value is higher than the preset second pressure entropy threshold, it is determined to be a forced pressure relief mode, the pressure relief parameters are set based on the preset three-level pressure relief strategy, and the pressure relief operation is performed. Based on the preset pressure relief valve parameter table, the pressure relief parameters are set according to the pressure relief mode; After performing the pressure relief operation, monitor the chamber pressure in real time; The pressure relief parameters are dynamically adjusted based on the deviation between the cavity pressure and the preset target pressure. Based on the cavity pressure, when it is determined that the cavity pressure is in a preset stable state, the compressor is started and the pressure relief operation data is recorded; The pressure relief operation data is used as a training set, and the weights of the pressure entropy model are updated according to a preset self-training period.

2. The intelligent control method for a compressor based on dynamic pressure relief according to claim 1, characterized in that, The preset cleaning and feature extraction algorithm, based on the multi-source heterogeneous data, obtains feature operation parameters, specifically as follows: Based on the pressure change curve, filtering is performed using a preset sliding window, and the pressure change rate and standard deviation within each window are calculated to obtain a pressure feature vector. Based on the temperature distribution map, the number, area, and temperature value of local overheated areas are extracted to obtain a temperature feature vector; The vibration curve is processed by a preset wavelet packet decomposition to obtain high-frequency vibration characteristics; Based on the slope analysis of the residual current curve, the current attenuation coefficient is obtained; By integrating pressure characteristic vector, temperature characteristic vector, high-frequency vibration characteristics, and current attenuation coefficient, characteristic operating parameters are obtained.

3. The intelligent control method for a compressor based on dynamic pressure relief according to claim 1, characterized in that, The dynamic adjustment of the pressure relief parameter based on the deviation between the cavity pressure and the preset target pressure specifically includes: When the forced pressure relief mode is determined, the pressure drop rate is calculated based on the cavity pressure. If the pressure drop rate is greater than the preset upper limit of the drop rate, the pressure relief opening is reduced stepwise according to the preset step size. If the pressure drop rate is less than the preset lower limit of the drop rate, the pressure relief opening is increased stepwise according to the preset step size. The predicted pressure relief time is obtained based on the deviation between the cavity pressure and the preset target pressure and the pressure drop rate. If the predicted depressurization time exceeds the set depressurization time threshold, the depressurization opening is increased and the upper limit of the descent rate is raised.

4. A compressor intelligent control system based on dynamic pressure relief, characterized in that, The system includes a memory and a processor. The memory includes a program for a compressor intelligent control method based on dynamic pressure relief. When the processor executes the program for the compressor intelligent control method based on dynamic pressure relief, it implements the steps of the compressor intelligent control method based on dynamic pressure relief as described in any one of claims 1 to 3.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that, The computer-readable storage medium includes a compressor intelligent control method program based on dynamic pressure relief. When the compressor intelligent control method program based on dynamic pressure relief is executed by a processor, it implements the steps of the compressor intelligent control method based on dynamic pressure relief as described in any one of claims 1 to 3.

Images