Efficient variable-frequency low-temperature compressor system and compression method thereof

Abstract

The invention discloses an efficient variable-frequency low-temperature compressor system and a compression method thereof. The method comprises the steps that the actual oil-gas separation efficiency is obtained; inputting the working condition parameters of the compressor into a preset neural network prediction model to obtain an oil-gas separation efficiency prediction value and a theoretical solubility reference value; according to the actual oil-gas separation efficiency, the oil-gas separation efficiency predicted value and the theoretical solubility reference value, working condition deviation data is calculated, the comprehensive performance of the compressor at different frequency points is evaluated and calculated through an optimization algorithm, and a frequency adjustment curve is generated; according to the frequency adjusting curve, the operation frequency of the compressor is divided into a plurality of adjusting intervals, the operation frequency is adjusted in a stepped mode according to the adjusting intervals, the temperature and vibration parameters of the bearing are monitored in the adjusting process, and frequency adjusting feedback data are obtained; and according to the frequency adjustment feedback data, oil-gas separation efficiency deviation is periodically calculated, and compensation adjustment is conducted on the frequency adjustment curve. According to the technical scheme, insufficient lubrication and energy efficiency reduction of the compressor can be avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of adaptive control systems, and in particular to a high-efficiency variable-frequency low-temperature compressor system and a compression method thereof. Background Art

[0002] The variable frequency low temperature compressor system is a key device in the refrigeration field and is widely used in refrigerators, freezers and other refrigeration equipment in our daily lives. This type of system achieves precise control of the refrigeration capacity by adjusting the operating frequency of the compressor, and can operate continuously and stably in a low temperature environment. For example, in a refrigerator, the variable frequency low temperature compressor system can automatically adjust the operating state according to the temperature requirements of the storage room, ensuring the fresh storage of food and achieving energy-saving operation.

[0003] However, under low temperature conditions, traditional variable frequency compressor control systems face severe technical challenges. As the ambient temperature decreases, the solubility of the refrigerant in the compressor lubricant oil will change significantly. Especially when the system load changes dynamically, the refrigerant will quickly precipitate from the lubricant oil, which not only affects the effective refrigerant circulation volume of the system, but also causes the problem of insufficient lubrication of the compressor. Due to the lack of accurate grasp of this dynamic characteristic, the traditional variable frequency control method cannot accurately control the oil-gas separation efficiency, which leads to reduced compressor performance and system energy efficiency. Summary of the invention

[0004] The main purpose of the present invention is to solve the technical problem that in a low-temperature environment, the solubility of the refrigerant in the lubricating oil changes significantly due to the decrease in temperature, and the traditional variable frequency compressor control method cannot accurately predict and control the oil-gas separation efficiency, which easily causes insufficient lubrication of the compressor and reduced energy efficiency.

[0005] A first aspect of the present invention provides a compression method for a high-efficiency variable frequency low-temperature compressor system, the compression method for a high-efficiency variable frequency low-temperature compressor system comprising: Collecting the operating parameters of the preset parts of the compressor and performing data processing on the collected operating parameters to obtain the actual oil-gas separation efficiency, wherein the operating parameters include temperature, pressure, current, and power; The operating parameters of the compressor are input into a preset neural network prediction model to perform a prediction operation on the oil-gas separation efficiency to obtain a predicted value of the oil-gas separation efficiency and a theoretical solubility reference value, wherein the operating parameters include the compressor inlet and outlet temperatures, the system pressure, and the refrigerant characteristic parameters; According to the actual oil-gas separation efficiency, the predicted value of the oil-gas separation efficiency and the theoretical solubility reference value, the operating condition deviation data is calculated, and the comprehensive performance of the compressor at different frequency points is evaluated and calculated using an optimization algorithm to generate a frequency adjustment curve; According to the frequency adjustment curve, the operating frequency of the compressor is divided into a plurality of adjustment intervals, and the operating frequency is adjusted stepwise according to the adjustment intervals, and the bearing temperature and vibration parameters are monitored during the adjustment process to obtain frequency adjustment feedback data; According to the frequency adjustment feedback data, the oil-gas separation efficiency deviation is periodically calculated, and the frequency adjustment curve is compensated and adjusted.

[0006] Optionally, collecting operating parameters of preset parts of the compressor and performing data processing on the collected operating parameters to obtain actual oil-gas separation efficiency includes: The compressor inlet and outlet temperatures, oil tank pressure, bearing temperature, motor current and output power are sampled in sections, and the sampling period is dynamically adjusted according to the temperature change rate to obtain an adaptive sampling sequence; The adaptive sampling sequence is layered according to the temperature gradient, and weighted average and variance analysis are performed on the data of each layer to obtain layered processed data; A temperature-pressure-current three-dimensional feature space is established according to the layered processing data, and an oil-gas separation trajectory is constructed in the three-dimensional feature space to obtain an oil-gas separation state quantity; According to the corresponding relationship between the oil-gas separation state quantity and the mass change rate of the refrigerant-oil mixture in the compressor oil tank, the actual oil-gas separation efficiency is obtained.

[0007] Optionally, establishing a temperature-pressure-current three-dimensional feature space according to the layered processing data, and constructing an oil-gas separation trajectory in the three-dimensional feature space to obtain an oil-gas separation state quantity includes: The three parameters of temperature, pressure and current are used to construct characteristic points in a rectangular coordinate system, and cluster analysis is performed on the characteristic points according to the change direction of the temperature gradient to obtain a characteristic point group; Performing principal component analysis on the characteristic point group, extracting the characteristic vector corresponding to the maximum eigenvalue as the main axis direction of oil and gas separation, and establishing a characteristic space coordinate system; In the characteristic space coordinate system, an isothermal surface is constructed according to the variation trends of temperature and pressure, and the current variation trajectory is mapped on the isothermal surface to obtain an initial separation trajectory; A low temperature compensation coefficient is applied to the initial separation trajectory, and the trajectory curvature is corrected according to the change law of the refrigerant physical properties to obtain the oil-gas separation state quantity.

[0008] Optionally, the preset neural network prediction model includes a first hidden layer, a second hidden layer and a third hidden layer having nine neuron nodes; The operating parameters of the compressor are input into a preset neural network prediction model, and the oil-gas separation efficiency is predicted and calculated to obtain the predicted value of the oil-gas separation efficiency and the theoretical solubility reference value, including: The compressor inlet and outlet temperatures, system pressure and refrigerant characteristic parameters are combined, and the critical temperature, molecular weight and eccentricity factor in the refrigerant characteristic parameters are set with a weight coefficient of 0.2-0.5 according to the solubility characteristics under low temperature conditions to obtain a low temperature compensation characteristic vector; The low temperature compensation feature vector is subjected to tangent function nonlinear mapping through the oil-gas separation feature extraction unit of the first hidden layer, and the oil-gas separation dynamic feature is calculated according to the coupling relationship between temperature and pressure to obtain oil-gas separation feature data; The oil-gas separation characteristic data is subjected to exponential function feature reconstruction through the solubility prediction unit of the second hidden layer, and solubility prediction data is generated based on the corresponding relationship between the system pressure and the refrigerant characteristic parameters; The solubility prediction data is input into the dynamic balance calculation unit of the third hidden layer, and the oil-gas separation efficiency prediction value and the theoretical solubility reference value are output through linear mapping.

[0009] Optionally, the operating condition deviation data is calculated according to the actual oil-gas separation efficiency, the predicted oil-gas separation efficiency value and the theoretical solubility reference value, and the comprehensive performance of the compressor at different frequency points is evaluated and calculated using an optimization algorithm to generate a frequency adjustment curve, including: The actual oil-gas separation efficiency is calculated by difference with the predicted value of oil-gas separation efficiency, and the difference is weighted in combination with the theoretical solubility reference value to obtain the oil-gas separation deviation index; Performing time series analysis on the oil-gas separation deviation index, extracting deviation variation characteristics, and setting correction coefficients according to dynamic characteristics of refrigerant solubility under low temperature conditions to obtain compensated deviation data; According to the compensated deviation data, a genetic algorithm is used to perform traversal optimization on different frequency points of the compressor within a set frequency range, and candidate frequency points are selected based on a multi-objective evaluation function of oil-gas separation efficiency and energy consumption; The candidate frequency points are sorted according to the change gradient of the oil-gas separation efficiency, and the frequency points are dynamically weighted in combination with the refrigerant physical property parameters to obtain an optimized frequency sequence; The optimized frequency sequence is subjected to curve fitting using a cubic spline interpolation algorithm to generate a continuous frequency adjustment curve.

[0010] Optionally, the operating frequency of the compressor is divided into a plurality of adjustment intervals according to the frequency adjustment curve, and the operating frequency is step-adjusted according to the adjustment intervals, and the bearing temperature and vibration parameters are monitored during the adjustment process to obtain frequency adjustment feedback data, including: According to the slope variation characteristics of the frequency regulation curve, the operating frequency is divided into a rapid regulation area, a transition regulation area and a stable regulation area, and the frequency threshold of each area is determined according to the phase change characteristics of the refrigerant under low temperature conditions; In the fast adjustment area, the maximum frequency change rate is used for adjustment, and the bearing temperature change rate and vibration acceleration are monitored at the same time. When the detection parameters exceed the preset threshold, the adjustment rate is automatically reduced; In the transition adjustment area, the frequency change increment is dynamically calculated according to the coupling change law of the bearing temperature and the vibration parameters, and the increment is corrected according to the change trend of the oil-gas separation efficiency; When the compressor enters the stable regulation zone, the frequency is fine-tuned using a fuzzy adaptive control strategy based on the comprehensive evaluation indicators of bearing temperature, vibration parameters and oil-gas separation efficiency to obtain frequency regulation feedback data.

[0011] Optionally, the frequency is fine-tuned using a fuzzy adaptive control strategy based on the comprehensive evaluation index of bearing temperature, vibration parameters and oil-gas separation efficiency, including: A three-dimensional fuzzy rule base is established for the bearing temperature change rate, vibration acceleration and oil-gas separation efficiency, wherein the three-dimensional fuzzy rule base includes membership functions of temperature change rate [-5°C / s, 5°C / s], vibration acceleration [0, 10m / s²], and oil-gas separation efficiency [60%, 95%]; The parameters collected in real time are fuzzified according to the membership function to obtain a fuzzy control value in the interval [0, 1]; The trigger strength is calculated according to the fuzzy control amount and the preset rule base, and the defuzzification operation is performed using the center of gravity method to obtain the frequency adjustment increment within the range of [-2Hz, 2Hz]; Based on the variation trend of the frequency adjustment increment and in combination with the temperature sensitivity coefficient of the refrigerant solubility, the controller parameters are adjusted online to achieve adaptive fine-tuning of the frequency.

[0012] Optionally, the periodically calculating the oil-gas separation efficiency deviation according to the frequency adjustment feedback data and compensating and adjusting the frequency adjustment curve include: The bearing temperature curve and vibration curve in the frequency regulation feedback data are decomposed by 4 layers of wavelet, and the frequency division threshold of 0.5-2kHz is selected according to the solubility characteristics of the refrigerant under low temperature conditions to obtain the characteristic waveform sequence; The characteristic waveform sequence is projected onto the temperature-vibration two-dimensional plane within a 100 ms sampling period, and a support vector machine algorithm based on a radial basis kernel function is used to identify the stable working area of ​​the system to generate a stability evaluation index; A 5×5 dimensional state transfer matrix is ​​established based on the stability evaluation index and the current oil-gas separation efficiency, and the operation trend of the next 10 cycles is predicted through a three-layer long short-term memory network to obtain a trend prediction vector; Performing singular value decomposition on the trend prediction vector, selecting characteristic components with a cumulative contribution rate of 90%, and performing online correction on the control parameters of the frequency regulation curve in combination with the refrigerant physical property parameters to obtain a compensation control sequence; The compensation control sequence is adaptively fused with the original frequency regulation curve, and a dynamic weight of [0.3, 0.7] is allocated according to the change rate of the oil-gas separation efficiency to achieve real-time compensation of the frequency regulation curve.

[0013] A second aspect of the present invention provides a high-efficiency variable frequency low-temperature compressor system, the high-efficiency variable frequency low-temperature compressor system comprising: A data acquisition module is used to collect operating parameters of preset parts of the compressor and perform data processing on the collected operating parameters to obtain actual oil-gas separation efficiency. The operating parameters include temperature, pressure, current, and power. A prediction calculation module is used to input the operating parameters of the compressor into a preset neural network prediction model, perform prediction calculations on the oil-gas separation efficiency, and obtain a predicted value of the oil-gas separation efficiency and a theoretical solubility reference value, wherein the operating parameters include the compressor inlet and outlet temperatures, the system pressure, and the refrigerant characteristic parameters; An optimization evaluation module is used to calculate the operating condition deviation data according to the actual oil-gas separation efficiency, the predicted value of the oil-gas separation efficiency and the theoretical solubility reference value, and to evaluate and calculate the comprehensive performance of the compressor at different frequency points using an optimization algorithm to generate a frequency adjustment curve; A frequency control module, for dividing the operating frequency of the compressor into a plurality of adjustment intervals according to the frequency adjustment curve, and performing step-by-step adjustment on the operating frequency according to the adjustment intervals, monitoring the bearing temperature and vibration parameters during the adjustment process, and obtaining frequency adjustment feedback data; The compensation adjustment module is used to periodically calculate the oil-gas separation efficiency deviation according to the frequency adjustment feedback data, and to perform compensation adjustment on the frequency adjustment curve.

[0014] The compression method of the high-efficiency variable-frequency low-temperature compressor system proposed in the present invention establishes a dynamic correlation mechanism between the oil-gas separation efficiency and the compressor operating frequency under low-temperature conditions through multi-level data processing and intelligent control strategies, effectively solving the problem of oil-gas separation control under low-temperature conditions.

[0015] In low temperature environments, the solubility of refrigerant in lubricating oil changes significantly with temperature. Specifically, the lower the temperature, the greater the solubility of refrigerant in lubricating oil, which causes more refrigerant to dissolve in the lubricating oil. When the operating conditions of the compressor change (such as load changes leading to pressure changes), the refrigerant dissolved in the oil will precipitate quickly. This precipitation behavior will not only dilute the lubricating oil and reduce its lubrication performance, but also cause a reduction in the amount of effective circulating refrigerant. Therefore, precise control of the compressor operating frequency is crucial: on the one hand, a suitable operating frequency can maintain a moderate solubility state of the oil-gas mixture and avoid violent refrigerant precipitation; on the other hand, a stable operating state is conducive to maintaining a dynamic balance of oil-gas separation.

[0016] The present invention collects operating parameters such as temperature, pressure, current, and power at preset locations of the compressor, and combines the neural network prediction model to make real-time predictions on the oil-gas separation efficiency, thereby establishing a mapping relationship between the oil-gas separation efficiency and the compressor operating status. This prediction mechanism enables the system to perceive the changing trend of the refrigerant's dissolution behavior in advance, providing a decision-making basis for frequency regulation. Based on the comparison between the prediction results and the actual operating data, the system can promptly detect deviations in the oil-gas separation efficiency, and generate the optimal frequency regulation curve through an optimization algorithm.

[0017] In particular, during the frequency regulation process, the present invention adopts a step-by-step regulation strategy between zones. In the fast-changing zone, the system quickly responds to changes in operating conditions through a larger frequency regulation step; in the transition zone, a moderate regulation step is used to maintain system stability; in the stable zone, fine control is achieved through small frequency adjustments. This multi-level dynamic frequency regulation strategy ensures that the compressor can maintain a suitable operating state under any operating condition changes, and will not cause violent precipitation of refrigerant due to too fast frequency regulation, nor will it cause insufficient lubrication due to delayed response.

[0018] By real-time monitoring of bearing temperature and vibration parameters, and dynamically compensating the frequency adjustment curve based on these feedback data, the system achieves closed-loop control of oil-gas separation efficiency. When there is a trend of refrigerant precipitation, the system automatically adjusts the operating frequency to maintain the optimal state of the oil-gas mixture, thereby ensuring the lubricating performance of the lubricating oil and the effective circulation volume of the refrigerant. This precise control mechanism not only ensures the reliable operation of the compressor, but also optimizes the system energy efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0020] Figure 1 A schematic diagram of an embodiment of a compression method of a high-efficiency variable frequency low-temperature compressor system in an embodiment of the present invention; Figure 2 It is a schematic diagram of an embodiment of a high-efficiency variable frequency low-temperature compressor system in an embodiment of the present invention.

[0021] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] It should be noted that if the embodiments of the present invention involve directional indications (such as up, down, left, right, front, back...), the directional indications are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0024] In addition, the descriptions of "first", "second", etc. in the present invention are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, "and / or" in the full text includes three solutions. Taking A and / or B as an example, it includes technical solution A, technical solution B, and technical solution that satisfies both A and B. In addition, the technical solutions between the various embodiments can be combined with each other, which must be based on the ability of ordinary technicians in the field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0025] An embodiment of the present application provides a compression method for a high-efficiency variable-frequency low-temperature compressor system. Figure 1 A flow chart of a compression method of a high-efficiency variable frequency low-temperature compressor system provided in one embodiment of the present application. In this embodiment, the method includes: See also Figure 1 , collecting operating parameters of preset parts of the compressor, and performing data processing on the collected operating parameters to obtain actual oil-gas separation efficiency, wherein the operating parameters include temperature, pressure, current, and power; In one embodiment of the present invention, the collecting of operating parameters of a preset part of the compressor and data processing of the collected operating parameters to obtain the actual oil-gas separation efficiency include: The compressor inlet and outlet temperatures, oil tank pressure, bearing temperature, motor current and output power are sampled in sections, and the sampling period is dynamically adjusted according to the temperature change rate to obtain an adaptive sampling sequence; The adaptive sampling sequence is layered according to the temperature gradient, and weighted average and variance analysis are performed on the data of each layer to obtain layered processed data; A temperature-pressure-current three-dimensional feature space is established according to the layered processing data, and an oil-gas separation trajectory is constructed in the three-dimensional feature space to obtain an oil-gas separation state quantity; According to the corresponding relationship between the oil-gas separation state quantity and the mass change rate of the refrigerant-oil mixture in the compressor oil tank, the actual oil-gas separation efficiency is obtained.

[0026] Specifically, this method collects operating parameters in real time by setting sensors such as temperature, pressure, and current at key locations such as the compressor inlet and outlet, oil tank, bearings, and motors. During the sampling process, the system dynamically adjusts the sampling period according to the temperature change rate: when the temperature changes dramatically, such as at the beginning of the compressor startup or when the load suddenly changes, the system shortens the sampling period from the standard 1 second to 0.1 second to capture more state change information; when the temperature changes slowly, the system extends the sampling period to 5 seconds to avoid generating redundant data. Through this dynamic adjustment, the system generates an adaptive sampling sequence containing parameters such as temperature, pressure, current, and power. The data points in this sequence are relatively dense when the system state changes dramatically, and are relatively sparse when the system is running stably.

[0027] After obtaining the adaptive sampling sequence, the system calculates the temperature change rate between adjacent sampling points and divides the temperature change rate into four intervals according to size: abrupt change zone (change rate greater than 5°C / s), rapid change zone (2-5°C / s), slow change zone (0.5-2°C / s) and stable zone (less than 0.5°C / s). The data of each interval is weighted averaged, with the weight of the abrupt change zone being 0.4, the rapid change zone being 0.3, the slow change zone being 0.2, and the stable zone being 0.1. This weighted method highlights the data characteristics of the interval with significant temperature changes. At the same time, the discrete degree of temperature, pressure, and current data is calculated for each interval to generate hierarchical processing data containing the characteristic values ​​of each interval.

[0028] When constructing the temperature-pressure-current three-dimensional feature space, the system maps the temperature value to the X-axis (-30℃ to 30℃), the pressure value to the Y-axis (0-3MPa), and the current value to the Z-axis (0-20A). The position of each sampling point in space is determined by these three parameters. For example, a data point at a certain moment may be located at (-5℃, 1.5MPa, 10A). The system connects the feature points adjacent in time through straight lines to form a spatial trajectory that characterizes the oil-gas separation process. The construction of this three-dimensional space selects temperature, pressure, and current as feature dimensions because temperature directly affects the solubility of the refrigerant, pressure reflects the system load status, and current reflects the working status of the compressor. The three together determine the dynamic process of oil-gas separation.

[0029] Finally, the system determines the actual oil-gas separation efficiency by measuring the mass change of the refrigerant-oil mixture in the compressor tank. Specifically, a high-precision mass sensor is set at the bottom of the tank to record the mass of the mixture every 10 seconds. When the refrigerant precipitates from the lubricating oil, the total mass of the mixture decreases, and the mass change rate is calculated by comparing the mass difference at adjacent moments. By comparing this change rate with the obtained oil-gas separation state quantity, the actual oil-gas separation efficiency can be obtained. For example, when it is detected that the mass of the mixture decreases by 5% within 1 minute, and the operating state at this time is in a stable range, it means that the oil-gas separation efficiency reaches 95%.

[0030] The uniqueness of this parameter collection and processing mechanism lies in: the adaptive sampling mechanism avoids the problem that traditional fixed-cycle sampling may miss key state changes; the layered processing method based on temperature gradient highlights the characteristic differences in different operating stages; the construction of three-dimensional feature space realizes the intuitive characterization of system status; and the efficiency calculation based on mass change ensures the accuracy of the results. These data provide a reliable basis for subsequent frequency adjustment, enabling the system to accurately respond to the challenges brought about by changes in refrigerant solubility under low-temperature conditions.

[0031] In one embodiment of the present invention, the temperature-pressure-current three-dimensional feature space is established according to the layered processing data, and the oil-gas separation trajectory is constructed in the three-dimensional feature space to obtain the oil-gas separation state quantity, including: The three parameters of temperature, pressure and current are used to construct characteristic points in a rectangular coordinate system, and cluster analysis is performed on the characteristic points according to the change direction of the temperature gradient to obtain a characteristic point group; Performing principal component analysis on the characteristic point group, extracting the characteristic vector corresponding to the maximum eigenvalue as the main axis direction of oil and gas separation, and establishing a characteristic space coordinate system; In the characteristic space coordinate system, an isothermal surface is constructed according to the variation trends of temperature and pressure, and the current variation trajectory is mapped on the isothermal surface to obtain an initial separation trajectory; A low temperature compensation coefficient is applied to the initial separation trajectory, and the trajectory curvature is corrected according to the change law of the refrigerant physical properties to obtain the oil-gas separation state quantity.

[0032] Specifically, in the process of processing the three-dimensional feature space, the three parameters of temperature, pressure, and current are first standardized to a unified interval (0-1), and the corresponding feature points are marked in the rectangular coordinate system. For each feature point, the direction vector of its local temperature gradient is calculated to reflect the temperature change trend at that point. For example, the feature point at a certain moment is located at (0.4, 0.6, 0.5), and its temperature gradient direction is (0.2, 0.1, 0). The system uses the K-means clustering algorithm to divide feature points with similar temperature gradient directions into multiple point groups, and the number of clusters is set to 4 categories, corresponding to the rapid temperature rise, slow rise, stability, and decline stages.

[0033] For each feature point group, the system performs principal component analysis to find the main direction of data change. The specific process is to calculate the covariance matrix of the feature point group and solve its eigenvalues ​​and eigenvectors. The eigenvector corresponding to the maximum eigenvalue represents the direction in which the data changes most significantly, and the system determines it as the main axis direction of oil and gas separation. For example, in a certain feature point group, if the maximum eigenvalue is 0.85 and the corresponding eigenvector is (0.6, 0.5, 0.3), it means that the change of system parameters in this direction is most obvious. Based on this main axis direction, combined with the original coordinate axis, a new feature space coordinate system is established.

[0034] In the newly established feature space coordinate system, the system selects multiple fixed temperature values ​​(such as -20℃, -10℃, 0℃, etc.) and constructs isothermal surfaces on the pressure-current plane corresponding to each temperature value. On each isothermal surface, the system records the trajectory of the current changing with pressure. The connection of these trajectory points forms the initial oil and gas separation trajectory. For example, on the isothermal surface of -10℃, when the pressure changes from 1MPa to 2MPa, the current may change from 8A to 12A, and this change is recorded as a trajectory.

[0035] Taking into account the special physical property changes of refrigerants in low-temperature environments, the system corrects the initial separation trajectory. First, the low-temperature compensation coefficient is determined according to the current operating temperature. The coefficient is linearly reduced from 1.5 to 1.0 in the range of -30°C to 0°C. Then, based on the temperature variation of parameters such as viscosity and surface tension recorded in the refrigerant physical property database, the curvature of the trajectory is corrected. For example, at -20°C, due to the increase in refrigerant viscosity, the separation process slows down, and the system will increase the trajectory curvature accordingly. The corrected trajectory more accurately reflects the oil and gas separation state under low-temperature conditions.

[0036] The advantages of this feature space construction and trajectory analysis method are: cluster analysis effectively identifies the characteristic differences of the system at different operating stages; principal component analysis finds out the main rules of parameter changes; the construction of isothermal surfaces intuitively shows the relationship between temperature, pressure and current; and the low temperature compensation mechanism ensures the accuracy of analysis under extreme working conditions. These processing processes provide a mathematical basis for the system to accurately grasp the dynamic characteristics of oil and gas separation, making subsequent control decisions more targeted.

[0037] Please continue reading Figure 1 , inputting the operating parameters of the compressor into a preset neural network prediction model, performing a prediction operation on the oil-gas separation efficiency, and obtaining a predicted value of the oil-gas separation efficiency and a theoretical solubility reference value, wherein the operating parameters include the compressor inlet and outlet temperatures, the system pressure, and the refrigerant characteristic parameters; In one embodiment of the present invention, the preset neural network prediction model includes a first hidden layer, a second hidden layer and a third hidden layer having nine neuron nodes; The operating parameters of the compressor are input into a preset neural network prediction model, and the oil-gas separation efficiency is predicted and calculated to obtain the predicted value of the oil-gas separation efficiency and the theoretical solubility reference value, including: The compressor inlet and outlet temperatures, system pressure and refrigerant characteristic parameters are combined, and the critical temperature, molecular weight and eccentricity factor in the refrigerant characteristic parameters are set with a weight coefficient of 0.2-0.5 according to the solubility characteristics under low temperature conditions to obtain a low temperature compensation characteristic vector; The low temperature compensation feature vector is subjected to tangent function nonlinear mapping through the oil-gas separation feature extraction unit of the first hidden layer, and the oil-gas separation dynamic feature is calculated according to the coupling relationship between temperature and pressure to obtain oil-gas separation feature data; The oil-gas separation characteristic data is subjected to exponential function feature reconstruction through the solubility prediction unit of the second hidden layer, and solubility prediction data is generated based on the corresponding relationship between the system pressure and the refrigerant characteristic parameters; The solubility prediction data is input into the dynamic balance calculation unit of the third hidden layer, and the oil-gas separation efficiency prediction value and the theoretical solubility reference value are output through linear mapping.

[0038] Specifically, the preset neural network prediction model adopts a three-layer hidden layer structure, each layer contains nine neuron nodes. Before data input, the system first preprocesses and combines the operating parameters. The basic operating parameters include: compressor inlet temperature (measuring range -30℃ to 0℃, sampling interval 1℃), outlet temperature (measuring range 40℃ to 80℃, sampling interval 1℃), system pressure (measuring range 0.5-3MPa, sampling interval 0.1MPa). At the same time, the characteristic parameter group of the refrigerant is introduced: critical temperature (e.g. R32 is 351.26K), molecular weight M (e.g. R32 is 52.024 g / mol), eccentricity factor ω (e.g. R32 is 0.2769).

[0039] According to the solubility characteristics of refrigerants under low temperature conditions, the system designs a weight distribution mechanism for characteristic parameters: the critical temperature weight coefficient is set to 0.5, because the critical temperature directly affects the phase change characteristics of the refrigerant; the molecular weight weight coefficient is set to 0.3, which affects the diffusion rate of the refrigerant; the eccentricity factor weight coefficient is set to 0.2, reflecting the asymmetry of the intermolecular force. Through this weighted combination, a low-temperature compensation characteristic vector is formed. Specific calculation example: For R32 refrigerant, its characteristic vector is calculated as: (351.26×0.5+52.024×0.3+0.2769×0.2), and the obtained vector more accurately characterizes the physical properties of the refrigerant at low temperatures.

[0040] The first hidden layer is used as the oil-gas separation feature extraction unit, and its nine nodes correspond to different working condition combinations: Nodes 1-3 process the temperature characteristics in the low pressure range (0.5-1.5MPa); Nodes 4-6 process the temperature characteristics in the medium pressure range (1.5-2.5MPa); Nodes 7-9 process the temperature characteristics in the high pressure range (2.5-3MPa). Each node uses the tangent function for nonlinear mapping and converts the input characteristics to the [-1, 1] range. For example, when the system pressure is 2MPa and the temperature is -20℃, the corresponding medium pressure interval node will output a characteristic value close to 0.8, indicating that the oil and gas separation trend is strong under this condition.

[0041] The solubility prediction unit of the second hidden layer uses an exponential function to reconstruct features, and the division of labor of the nine nodes is as follows: Nodes 1-3 are responsible for the prediction of the 0-30% solubility interval; Nodes 4-6 are responsible for the prediction of the 30-70% solubility interval; Nodes 7-9 are responsible for the prediction of the 70-100% solubility range. Each node receives the characteristic data of the first layer and uses the exponential function Map the eigenvalue to the corresponding solubility range (where e is the base of the natural logarithm, approximately equal to 2.71828, and x is the input eigenvalue). For example, a node receives an eigenvalue of 0.75 and obtains a solubility prediction value of 0.85 after being processed by the exponential function (the specific calculation is ≈2.117, and then normalized and mapped to the [0, 1] interval), indicating that the refrigerant tends to maintain a higher dissolved state under this condition.

[0042] The dynamic balance calculation unit of the third hidden layer is transformed into a linear mapping function Process the data output by the previous layer (where is the proportional coefficient, which is used to adjust the slope of the mapping, b is the bias term, which is used to adjust the offset of the mapping, and x is the input value). The division of labor of the nine nodes is more refined: Node 1-2 is responsible for the balance calculation in the extremely low temperature range (-30℃ to -20℃); Node 3-4 is responsible for the balance calculation in the low temperature range (-20℃ to -10℃); Nodes 5-6 are responsible for the balance calculation in the medium temperature range (-10℃ to -5℃); Nodes 7-9 are responsible for the balance calculation in the critical temperature range (-5°C to 0°C). Through this refined division of labor, the system can output more accurate prediction values ​​of oil and gas separation efficiency and theoretical solubility benchmark values.

[0043] The design advantages of this three-layer neural network structure are: clear hierarchy, clear functions of each layer; reasonable division of labor among nodes, covering all key working conditions; and strong pertinence in the selection of different activation functions, which improves the prediction accuracy. The entire network structure not only ensures the integrity of feature extraction, but also avoids the problem of excessive computational complexity.

[0044] Please continue reading Figure 1 , according to the actual oil-gas separation efficiency, the predicted value of the oil-gas separation efficiency and the theoretical solubility reference value, the operating condition deviation data is calculated, and the comprehensive performance of the compressor at different frequency points is evaluated and calculated using an optimization algorithm to generate a frequency adjustment curve; In one embodiment of the present invention, the operating condition deviation data is calculated according to the actual oil-gas separation efficiency, the predicted oil-gas separation efficiency value and the theoretical solubility reference value, and the comprehensive performance of the compressor at different frequency points is evaluated and calculated using an optimization algorithm to generate a frequency adjustment curve, including: The actual oil-gas separation efficiency is calculated by difference with the predicted value of oil-gas separation efficiency, and the difference is weighted in combination with the theoretical solubility reference value to obtain the oil-gas separation deviation index; Performing time series analysis on the oil-gas separation deviation index, extracting deviation variation characteristics, and setting correction coefficients according to dynamic characteristics of refrigerant solubility under low temperature conditions to obtain compensated deviation data; According to the compensated deviation data, a genetic algorithm is used to perform traversal optimization on different frequency points of the compressor within a set frequency range, and candidate frequency points are selected based on a multi-objective evaluation function of oil-gas separation efficiency and energy consumption; The candidate frequency points are sorted according to the change gradient of the oil-gas separation efficiency, and the frequency points are dynamically weighted in combination with the refrigerant physical property parameters to obtain an optimized frequency sequence; The optimized frequency sequence is subjected to curve fitting using a cubic spline interpolation algorithm to generate a continuous frequency adjustment curve.

[0045] Specifically, the system first calculates the difference between the actual oil-gas separation efficiency and the predicted value. The actual efficiency is measured by the mass sensor, ranging from 0-100%; the predicted value is output by the neural network model, also in the range of 0-100%. The calculated difference is weighted with the theoretical solubility reference value (range 0-1): when the theoretical solubility reference value is high (such as above 0.8), the weight of the difference is set to 0.7, and more emphasis is placed on the actual measured value; when the reference value is low (such as below 0.4), the weight of the difference is reduced to 0.3, and more reliance is placed on the predicted value. For example, at a certain moment, the actual efficiency is 85%, the predicted value is 80%, and the theoretical solubility reference value is 0.9, then the weighted deviation index is 3.5%.

[0046] When performing time series analysis on the oil-gas separation deviation index, the system uses a 60-second sliding time window, sliding every 10 seconds, to extract the variation characteristics of the deviation within the window. Four main features are extracted: mean, standard deviation, rate of change, and trend. At the same time, the dynamic characteristic correction coefficient of the refrigerant solubility is determined according to the current operating temperature: in the range of -30°C to -20°C, the correction coefficient is 1.5 (the solubility changes dramatically); in the range of -20°C to -10°C, the correction coefficient drops linearly to 1.2; in the range above -10°C, the correction coefficient remains at 1.0. Multiply the extracted characteristic value by the correction coefficient to obtain the compensated deviation data.

[0047] The system uses genetic algorithm to optimize the search in the frequency range of 20-80Hz. The population size is set to 100 and the evolution generation is 50 generations. The evaluation function contains two objectives: oil and gas separation efficiency E (E ranges from 0 to 100%) and system energy consumption P (P is normalized and ranges from 0 to 1), expressed as F= ,in is the weight coefficient of oil-gas separation efficiency (value is 0.6), is the weight coefficient of system energy consumption (value is 0.4). For each frequency point, F is the comprehensive evaluation value. For each frequency point, the system calculates its corresponding evaluation value. For example, if the oil-gas separation efficiency E of a certain frequency point is 90%, and the normalized energy consumption P is 0.7, then its evaluation value is 0.54-0.28=0.26. Through the selection, crossover, and mutation operations of the genetic algorithm, 30 candidate frequency points with higher evaluation values ​​are finally screened out.

[0048] For the selected frequency candidate points, the system calculates the change gradient of the oil-gas separation efficiency at each point (the difference between the efficiencies of adjacent points divided by the difference in frequency). The points are sorted in descending order according to the gradient size, while considering the physical properties of the refrigerant at the frequency point (such as viscosity, thermal conductivity, etc.). A dynamic weight allocation mechanism is designed: the weight of the frequency point in the area with a large gradient (such as greater than 2% / Hz) is 0.5; the weight of the medium gradient area (1-2% / Hz) is 0.3; the weight of the area with a small gradient (less than 1% / Hz) is 0.2. Through this weight allocation, an optimized frequency sequence that reflects the response characteristics of the system is obtained.

[0049] Finally, the optimized frequency sequence is smoothed using the cubic spline interpolation algorithm. Between adjacent frequency points, a cubic polynomial is used for interpolation to ensure that the first-order and second-order derivatives of the curve at the nodes are continuous. The interpolation interval is set to 0.1 Hz to generate a smooth frequency adjustment curve. For example, the adjustment curve between 35 Hz and 40 Hz can be expressed as: .

[0050] in, For any time The corresponding frequency value, is a time variable (its value range is [0, 1], corresponding to the normalized value of the actual time interval), , , , are the polynomial coefficients (determined by the locations of the interpolation points and the continuity conditions of the first and second derivatives), , , They represent the cube, square and line of the time variable respectively.

[0051] The advantages of this frequency optimization method are: through difference calculation and weighted processing, it accurately reflects the actual operation deviation of the system; time series analysis and dynamic characteristic correction improve the accuracy of compensation; the multi-objective optimization of genetic algorithm balances efficiency and energy consumption; the dynamic weight allocation mechanism ensures the rationality of frequency regulation; and cubic spline interpolation ensures the smoothness of frequency regulation. Tests show that this method can keep the compressor in the optimal operating state under low temperature conditions.

[0052] Please continue reading Figure 1 , according to the frequency adjustment curve, the operating frequency of the compressor is divided into a plurality of adjustment intervals, and the operating frequency is adjusted stepwise according to the adjustment intervals, and the bearing temperature and vibration parameters are monitored during the adjustment process to obtain frequency adjustment feedback data; In one embodiment of the present invention, the operating frequency of the compressor is divided into a plurality of adjustment intervals according to the frequency adjustment curve, and the operating frequency is step-adjusted according to the adjustment intervals, and the bearing temperature and vibration parameters are monitored during the adjustment process to obtain frequency adjustment feedback data, including: According to the slope variation characteristics of the frequency regulation curve, the operating frequency is divided into a rapid regulation area, a transition regulation area and a stable regulation area, and the frequency threshold of each area is determined according to the phase change characteristics of the refrigerant under low temperature conditions; In the fast adjustment area, the maximum frequency change rate is used for adjustment, and the bearing temperature change rate and vibration acceleration are monitored at the same time. When the detection parameters exceed the preset threshold, the adjustment rate is automatically reduced; In the transition adjustment area, the frequency change increment is dynamically calculated according to the coupling change law of the bearing temperature and the vibration parameters, and the increment is corrected according to the change trend of the oil-gas separation efficiency; When the compressor enters the stable regulation zone, the frequency is fine-tuned using a fuzzy adaptive control strategy based on the comprehensive evaluation indicators of bearing temperature, vibration parameters and oil-gas separation efficiency to obtain frequency regulation feedback data.

[0053] Specifically, the system divides the operating range based on the slope characteristics of the frequency adjustment curve. The slope of the frequency curve at each point is calculated. When the slope is greater than 2Hz / s, it is divided into a rapid adjustment zone. When the slope is between 0.5-2Hz / s, it is divided into a transition adjustment zone. When the slope is less than 0.5Hz / s, it is divided into a stable adjustment zone. At the same time, the frequency threshold is set considering the phase change characteristics of the refrigerant under low temperature conditions: in the range of -30℃ to -20℃, due to the drastic phase change of the refrigerant, the frequency range of the rapid adjustment zone is set to 20-35Hz, the transition zone is 35-50Hz, and the stable zone is 50-80Hz; in the range above -20℃, the frequency ranges of these three intervals are adjusted to 20-40Hz, 40-60Hz and 60-80Hz respectively.

[0054] In the fast adjustment area, the system uses a maximum frequency change rate of 5Hz / s for adjustment, while monitoring two key parameters in real time: the bearing temperature change rate and vibration acceleration. The monitoring threshold for bearing temperature is set to 2℃ / s, and the monitoring threshold for vibration acceleration is set to 5m / s². When any parameter exceeds the threshold, the system automatically reduces the frequency change rate by 50%. For example, when the bearing temperature change rate is detected to reach 2.5℃ / s, the frequency change rate is immediately reduced to 2.5Hz / s. If the parameter continues to exceed the standard, the system will further reduce the adjustment rate until it reaches a safe range.

[0055] In the transition adjustment area, the system establishes a coupling evaluation model for bearing temperature and vibration parameters. The bearing temperature ranges from 40-80°C, and the vibration parameter ranges from 0-10m / s². According to the coupling state of these two parameters, the system dynamically calculates the frequency change increment: when the bearing temperature is lower than 60°C and the vibration acceleration is less than 3m / s², the frequency increment is set to 1Hz / s; when the parameter is close to the warning value, the increment is reduced to 0.5Hz / s. At the same time, the system monitors the changing trend of the oil-gas separation efficiency. When the efficiency shows a downward trend (such as a decrease of more than 5% within 3 minutes), the frequency increment is attenuated.

[0056] After the compressor enters the stable regulation zone, the system adopts a more sophisticated control strategy. A comprehensive evaluation index is designed, which includes three parameters: bearing temperature (weight 0.3), vibration parameter (weight 0.3) and oil-gas separation efficiency (weight 0.4). Based on these three parameters, a fuzzy control rule base is established, which contains 25 rules. The membership function of the input variables is divided into five levels: very low, low, medium, high, and very high. For example, when the bearing temperature is at a "medium" level, the vibration parameter is at a "low" level, and the oil-gas separation efficiency is at a "high" level, the corresponding control rule outputs a small positive frequency adjustment (0.1Hz / s).

[0057] The advantages of this partition adjustment mechanism are: through the differentiated control strategies in different intervals, both the adjustment speed and the operation safety are guaranteed; the protection mechanism in the fast adjustment zone avoids the drastic fluctuation of system parameters; the coupling evaluation model in the transition zone achieves a smooth transition; and the fuzzy control strategy in the stable zone ensures the precise adjustment of the operating parameters. Test results show that this mechanism can enable the compressor to complete the entire process from startup to stable operation within 30 seconds, without severe vibration and temperature fluctuations throughout the process.

[0058] In one embodiment of the present invention, the frequency is fine-tuned using a fuzzy adaptive control strategy based on the comprehensive evaluation index of bearing temperature, vibration parameters and oil-gas separation efficiency, including: A three-dimensional fuzzy rule base is established for the bearing temperature change rate, vibration acceleration and oil-gas separation efficiency, wherein the three-dimensional fuzzy rule base includes membership functions of temperature change rate [-5°C / s, 5°C / s], vibration acceleration [0, 10m / s²], and oil-gas separation efficiency [60%, 95%]; The parameters collected in real time are fuzzified according to the membership function to obtain a fuzzy control value in the interval [0, 1]; The trigger strength is calculated according to the fuzzy control amount and the preset rule base, and the defuzzification operation is performed using the center of gravity method to obtain the frequency adjustment increment within the range of [-2Hz, 2Hz]; Based on the variation trend of the frequency adjustment increment and in combination with the temperature sensitivity coefficient of the refrigerant solubility, the controller parameters are adjusted online to achieve adaptive fine-tuning of the frequency.

[0059] Specifically, the three-dimensional fuzzy rule base established by the system covers the range of changes of three key parameters. The bearing temperature change rate is divided into five levels in the range of [-5℃ / s, 5℃ / s]: negative fast (-5 to -3℃ / s), negative slow (-3 to -1℃ / s), stable (-1 to 1℃ / s), positive slow (1 to 3℃ / s), positive fast (3 to 5℃ / s). The vibration acceleration is divided into four levels in the range of [0, 10m / s²]: normal (0-3m / s²), slight (3-5m / s²), medium (5-7m / s²), severe (7-10m / s²). The oil-gas separation efficiency is divided into three levels in the range of [60%, 95%]: low efficiency (60-75%), medium efficiency (75-85%), and high efficiency (85-95%).

[0060] When fuzzifying the parameters collected in real time, the system uses a triangular membership function. For example, when the bearing temperature change rate is measured to be 2.5℃ / s at a certain moment, its membership in the "positive slow" level is 0.75, and its membership in the "positive fast" level is 0.25; when the vibration acceleration is 4.5m / s², its membership in the "mild" level is 0.75, and its membership in the "normal" level is 0.25; when the oil-gas separation efficiency is 82%, its membership in the "medium efficiency" level is 0.7, and its membership in the "high efficiency" level is 0.3. The fuzzification results of these three parameters together constitute the fuzzy control quantity in the [0, 1] interval.

[0061] The system matches the preset rule base according to the fuzzy control quantity and calculates the rule trigger strength. The rule base contains 60 rules (5×4×3), and the form of each rule is as follows: IF (temperature change rate = positive slow) AND (vibration acceleration = normal) AND (oil and gas separation efficiency = high efficiency) THEN (frequency adjustment = positive small). The minimum operator is used to calculate the rule trigger strength. For example, if the three antecedent memberships of a rule are 0.75, 0.75, and 0.3, respectively, the trigger strength of the rule is 0.3. Then the centroid method is used to perform the defuzzification operation, and the output of each rule is weighted averaged according to the trigger strength to obtain the frequency adjustment increment in the range of [-2Hz, 2Hz].

[0062] Finally, the system adjusts the controller online based on the temperature sensitivity of the refrigerant. The temperature sensitivity coefficient of the refrigerant is calculated based on the current temperature: in the range of -30℃ to -20℃, the temperature sensitivity coefficient is 1.5; in the range of -20℃ to -10℃, the coefficient drops linearly to 1.2; in the range above -10℃, the coefficient remains at 1.0. Multiply the frequency adjustment increment by the temperature sensitivity coefficient to obtain the actual adjustment amount. For example, when the increment obtained by the defuzzification operation is 1.5Hz and the temperature sensitivity coefficient is 1.2, the actual frequency adjustment amount is 1.8Hz.

[0063] The advantages of this fuzzy control strategy are: the three-dimensional rule base fully considers the changing characteristics of the key parameters of the system; the fuzzy processing effectively reduces the impact of parameter fluctuations; the center of gravity method defuzzification ensures the smoothness of the control output; and the introduction of the temperature sensitivity coefficient makes the control more suitable for the characteristics of low-temperature conditions. Tests show that this control strategy can keep the compressor running stably in a low-temperature environment, and the frequency adjustment process is smooth and oscillating.

[0064] Please continue reading Figure 1 According to the frequency adjustment feedback data, the oil-gas separation efficiency deviation is periodically calculated, and the frequency adjustment curve is compensated and adjusted.

[0065] In one embodiment of the present invention, periodically calculating the oil-gas separation efficiency deviation according to the frequency adjustment feedback data and compensating and adjusting the frequency adjustment curve include: The bearing temperature curve and vibration curve in the frequency regulation feedback data are decomposed by 4 layers of wavelet, and the frequency division threshold of 0.5-2kHz is selected according to the solubility characteristics of the refrigerant under low temperature conditions to obtain the characteristic waveform sequence; The characteristic waveform sequence is projected onto the temperature-vibration two-dimensional plane within a 100 ms sampling period, and a support vector machine algorithm based on a radial basis kernel function is used to identify the stable working area of ​​the system to generate a stability evaluation index; A 5×5 dimensional state transfer matrix is ​​established based on the stability evaluation index and the current oil-gas separation efficiency, and the operation trend of the next 10 cycles is predicted through a three-layer long short-term memory network to obtain a trend prediction vector; Performing singular value decomposition on the trend prediction vector, selecting characteristic components with a cumulative contribution rate of 90%, and performing online correction on the control parameters of the frequency regulation curve in combination with the refrigerant physical property parameters to obtain a compensation control sequence; The compensation control sequence is adaptively fused with the original frequency regulation curve, and a dynamic weight of [0.3, 0.7] is allocated according to the change rate of the oil-gas separation efficiency to achieve real-time compensation of the frequency regulation curve.

[0066] Specifically, the system uses four-layer wavelet decomposition to process the bearing temperature curve and vibration curve in the frequency regulation feedback data. Using db4 wavelet as the basis function, the signal is decomposed into four detail coefficient layers: and an approximate coefficient layer According to the solubility characteristics of the refrigerant under low temperature conditions, the frequency division threshold is set: the first layer is 0.5-1kHz (reflecting high-frequency vibration), the second layer is 1-1.5kHz (reflecting medium-frequency vibration), the third layer is 1.5-2kHz (reflecting low-frequency vibration), and the fourth layer is above 2kHz (reflecting basic vibration). For example, after the temperature curve at a certain moment is decomposed, Layers reflect rapid fluctuations in temperature, The layers reflect the slowly changing trend of temperature.

[0067] When the characteristic waveform sequence is projected onto the temperature-vibration two-dimensional plane, a sampling period of 100ms is used for data acquisition. On this plane, the horizontal axis represents the temperature range (-30℃ to 80℃), and the vertical axis represents the vibration range (0-10m / s²). The radial basis kernel function (RBF) support vector machine algorithm is used to identify the stable working area of ​​the system, and the kernel function parameter σ is set to 0.5. The training sample contains 1000 sets of labeled data, of which 70% is used for training and 30% is used for verification. The algorithm divides the working points into three categories: stable area (evaluation index greater than 0.8), metastable area (evaluation index 0.5-0.8), and unstable area (evaluation index less than 0.5).

[0068] Based on the stability evaluation index and the current oil-gas separation efficiency, the system constructs a 5×5 dimensional state transfer matrix. The rows of the matrix represent the current state (divided into five levels), and the columns represent the next state (also divided into five levels). A three-layer long short-term memory network (LSTM) is used to predict the operating trend of the system in the next 10 cycles, with each layer containing 64 neurons. The network input is the state sequence of the most recent 100 sampling points, and the output is the state prediction value for the next 10 cycles. For example, if the current state is at the third level and the oil-gas separation efficiency is 85%, the network may predict that the system will enter the fourth level in the next 3 cycles.

[0069] Perform singular value decomposition on the trend prediction vector to obtain the eigenvalue sequence. Select the eigenvalues ​​with a cumulative contribution rate of 90%. Usually, the requirements can be met by retaining the first 3-4 eigenvalues. At the same time, consider the physical parameters of the refrigerant (such as thermal conductivity and specific heat capacity) to establish a compensation coefficient matrix. For example, when the temperature rises from -20°C to -10°C, the thermal conductivity of the refrigerant increases by 15%, and the corresponding compensation coefficient is 1.15. Multiply the eigenvalue by the compensation coefficient to obtain the compensation control sequence.

[0070] Finally, the system adaptively merges the compensation control sequence and the original frequency adjustment curve according to the change rate of the oil-gas separation efficiency. When the oil-gas separation efficiency rises rapidly (the change rate is greater than 2% / s), the weight of the compensation control sequence is set to 0.7 and the weight of the original curve is 0.3; when the efficiency changes slowly, the weights of the two are equal; when the efficiency decreases, the weight of the compensation control sequence drops to 0.3. Through this dynamic weight distribution, real-time compensation of the frequency adjustment curve is achieved.

[0071] The advantages of this compensation mechanism are: wavelet decomposition effectively extracts the multi-scale features of the signal; the support vector machine algorithm accurately identifies the stable area of ​​the system; the LSTM network achieves accurate prediction of the system trend; the singular value decomposition reduces the data dimension and retains key information; and the dynamic weight allocation ensures the smoothness of the compensation. Tests show that this compensation mechanism can keep the oil-gas separation efficiency fluctuation range of the compressor within ±2% under low temperature conditions.

[0072] The compression method of the high-efficiency variable frequency low-temperature compressor system in the embodiment of the present invention is described above. The high-efficiency variable frequency low-temperature compressor system in the embodiment of the present invention is described below. Figure 2 , an embodiment of the high-efficiency variable frequency low-temperature compressor system in the embodiment of the present invention includes: The data acquisition module 201 is used to collect the operating parameters of the preset parts of the compressor and perform data processing on the collected operating parameters to obtain the actual oil-gas separation efficiency. The operating parameters include temperature, pressure, current and power. The prediction calculation module 202 is used to input the operating parameters of the compressor into a preset neural network prediction model, perform prediction calculation on the oil-gas separation efficiency, and obtain the predicted value of the oil-gas separation efficiency and the theoretical solubility reference value. The operating parameters include the inlet and outlet temperatures of the compressor, the system pressure, and the characteristic parameters of the refrigerant; The optimization evaluation module 203 is used to calculate the operating condition deviation data according to the actual oil-gas separation efficiency, the predicted value of the oil-gas separation efficiency and the theoretical solubility reference value, and evaluate and calculate the comprehensive performance of the compressor at different frequency points using an optimization algorithm to generate a frequency adjustment curve; The frequency control module 204 is used to divide the operating frequency of the compressor into a plurality of adjustment intervals according to the frequency adjustment curve, and to perform step-by-step adjustment on the operating frequency according to the adjustment intervals, and to monitor the bearing temperature and vibration parameters during the adjustment process to obtain frequency adjustment feedback data; The compensation adjustment module 205 is used to periodically calculate the oil-gas separation efficiency deviation according to the frequency adjustment feedback data, and to perform compensation adjustment on the frequency adjustment curve.

[0073] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. All equivalent structural changes made by using the contents of the present invention specification and drawings under the inventive concept of the present invention, or directly / indirectly applied in other related technical fields are included in the patent protection scope of the present invention.

Claims

1. A compression method for a high-efficiency variable frequency low-temperature compressor system, characterized in that: include: Collecting the operating parameters of the preset parts of the compressor and performing data processing on the collected operating parameters to obtain the actual oil-gas separation efficiency, wherein the operating parameters include temperature, pressure, current, and power; The operating parameters of the compressor are input into a preset neural network prediction model to perform a prediction operation on the oil-gas separation efficiency to obtain a predicted value of the oil-gas separation efficiency and a theoretical solubility reference value, wherein the operating parameters include the compressor inlet and outlet temperatures, the system pressure, and the refrigerant characteristic parameters; According to the actual oil-gas separation efficiency, the predicted value of the oil-gas separation efficiency and the theoretical solubility reference value, the operating condition deviation data is calculated, and the comprehensive performance of the compressor at different frequency points is evaluated and calculated using an optimization algorithm to generate a frequency adjustment curve; According to the frequency adjustment curve, the operating frequency of the compressor is divided into a plurality of adjustment intervals, and the operating frequency is adjusted stepwise according to the adjustment intervals, and the bearing temperature and vibration parameters are monitored during the adjustment process to obtain frequency adjustment feedback data; According to the frequency adjustment feedback data, the oil-gas separation efficiency deviation is periodically calculated, and the frequency adjustment curve is compensated and adjusted.

2. The compression method of the high-efficiency variable frequency low-temperature compressor system according to claim 1, characterized in that: The collecting of the operating parameters of the preset parts of the compressor and data processing of the collected operating parameters to obtain the actual oil-gas separation efficiency include: The compressor inlet and outlet temperatures, oil tank pressure, bearing temperature, motor current and output power are sampled in sections, and the sampling period is dynamically adjusted according to the temperature change rate to obtain an adaptive sampling sequence; The adaptive sampling sequence is layered according to the temperature gradient, and weighted average and variance analysis are performed on the data of each layer to obtain layered processed data; A temperature-pressure-current three-dimensional feature space is established according to the layered processing data, and an oil-gas separation trajectory is constructed in the three-dimensional feature space to obtain an oil-gas separation state quantity; According to the corresponding relationship between the oil-gas separation state quantity and the mass change rate of the refrigerant-oil mixture in the compressor oil tank, the actual oil-gas separation efficiency is obtained.

3. The compression method of the high-efficiency variable frequency low-temperature compressor system according to claim 2, characterized in that: The step of establishing a temperature-pressure-current three-dimensional feature space according to the layered processing data, and constructing an oil-gas separation trajectory in the three-dimensional feature space to obtain an oil-gas separation state quantity includes: The three parameters of temperature, pressure and current are used to construct characteristic points in a rectangular coordinate system, and cluster analysis is performed on the characteristic points according to the change direction of the temperature gradient to obtain a characteristic point group; Performing principal component analysis on the characteristic point group, extracting the characteristic vector corresponding to the maximum eigenvalue as the main axis direction of oil and gas separation, and establishing a characteristic space coordinate system; In the characteristic space coordinate system, an isothermal surface is constructed according to the variation trends of temperature and pressure, and the current variation trajectory is mapped on the isothermal surface to obtain an initial separation trajectory; A low temperature compensation coefficient is applied to the initial separation trajectory, and the trajectory curvature is corrected according to the change law of the refrigerant physical properties to obtain the oil-gas separation state quantity.

4. The compression method of the high-efficiency variable frequency low-temperature compressor system according to claim 1, characterized in that: The preset neural network prediction model includes a first hidden layer, a second hidden layer and a third hidden layer having nine neuron nodes; The operating parameters of the compressor are input into a preset neural network prediction model, and the oil-gas separation efficiency is predicted and calculated to obtain the predicted value of the oil-gas separation efficiency and the theoretical solubility reference value, including: The compressor inlet and outlet temperatures, system pressure and refrigerant characteristic parameters are combined, and the critical temperature, molecular weight and eccentricity factor in the refrigerant characteristic parameters are set with a weight coefficient of 0.2-0.5 according to the solubility characteristics under low temperature conditions to obtain a low temperature compensation characteristic vector; The low temperature compensation feature vector is subjected to tangent function nonlinear mapping through the oil-gas separation feature extraction unit of the first hidden layer, and the oil-gas separation dynamic feature is calculated according to the coupling relationship between temperature and pressure to obtain oil-gas separation feature data; The oil-gas separation characteristic data is subjected to exponential function feature reconstruction through the solubility prediction unit of the second hidden layer, and solubility prediction data is generated based on the corresponding relationship between the system pressure and the refrigerant characteristic parameters; The solubility prediction data is input into the dynamic balance calculation unit of the third hidden layer, and the oil-gas separation efficiency prediction value and the theoretical solubility reference value are output through linear mapping.

5. The compression method of the high-efficiency variable frequency low-temperature compressor system according to claim 1, characterized in that: The operating condition deviation data is calculated according to the actual oil-gas separation efficiency, the predicted value of the oil-gas separation efficiency and the theoretical solubility reference value, and the comprehensive performance of the compressor at different frequency points is evaluated and calculated using an optimization algorithm to generate a frequency adjustment curve, including: The actual oil-gas separation efficiency is calculated by difference with the predicted value of oil-gas separation efficiency, and the difference is weighted in combination with the theoretical solubility reference value to obtain the oil-gas separation deviation index; Performing time series analysis on the oil-gas separation deviation index, extracting deviation variation characteristics, and setting correction coefficients according to dynamic characteristics of refrigerant solubility under low temperature conditions to obtain compensated deviation data; According to the compensated deviation data, a genetic algorithm is used to perform traversal optimization on different frequency points of the compressor within a set frequency range, and candidate frequency points are selected based on a multi-objective evaluation function of oil-gas separation efficiency and energy consumption; The candidate frequency points are sorted according to the change gradient of the oil-gas separation efficiency, and the frequency points are dynamically weighted in combination with the refrigerant physical property parameters to obtain an optimized frequency sequence; The optimized frequency sequence is subjected to curve fitting using a cubic spline interpolation algorithm to generate a continuous frequency adjustment curve.

6. The compression method of the high-efficiency variable frequency low-temperature compressor system according to claim 1, characterized in that: According to the frequency adjustment curve, the operating frequency of the compressor is divided into a plurality of adjustment intervals, and the operating frequency is adjusted stepwise according to the adjustment intervals, and the bearing temperature and vibration parameters are monitored during the adjustment process to obtain frequency adjustment feedback data, including: According to the slope variation characteristics of the frequency regulation curve, the operating frequency is divided into a rapid regulation area, a transition regulation area and a stable regulation area, and the frequency threshold of each area is determined according to the phase change characteristics of the refrigerant under low temperature conditions; In the fast adjustment area, the maximum frequency change rate is used for adjustment, and the bearing temperature change rate and vibration acceleration are monitored at the same time. When the detection parameters exceed the preset threshold, the adjustment rate is automatically reduced; In the transition adjustment area, the frequency change increment is dynamically calculated according to the coupling change law of the bearing temperature and the vibration parameters, and the increment is corrected according to the change trend of the oil-gas separation efficiency; When the compressor enters the stable regulation zone, the frequency is fine-tuned using a fuzzy adaptive control strategy based on the comprehensive evaluation indicators of bearing temperature, vibration parameters and oil-gas separation efficiency to obtain frequency regulation feedback data.

7. The compression method of the high-efficiency variable frequency low-temperature compressor system according to claim 6, characterized in that: The frequency is fine-tuned using a fuzzy adaptive control strategy based on the comprehensive evaluation index of bearing temperature, vibration parameters and oil-gas separation efficiency, including: A three-dimensional fuzzy rule base is established for the bearing temperature change rate, vibration acceleration and oil-gas separation efficiency, wherein the three-dimensional fuzzy rule base includes membership functions of temperature change rate [-5°C / s, 5°C / s], vibration acceleration [0, 10m / s²], and oil-gas separation efficiency [60%, 95%]; The parameters collected in real time are fuzzified according to the membership function to obtain a fuzzy control value in the interval [0, 1]; The trigger strength is calculated according to the fuzzy control amount and the preset rule base, and the defuzzification operation is performed using the center of gravity method to obtain the frequency adjustment increment within the range of [-2Hz, 2Hz]; Based on the variation trend of the frequency adjustment increment and in combination with the temperature sensitivity coefficient of the refrigerant solubility, the controller parameters are adjusted online to achieve adaptive fine-tuning of the frequency.

8. The compression method of the high-efficiency variable frequency low-temperature compressor system according to claim 1, characterized in that: The step of periodically calculating the oil-gas separation efficiency deviation according to the frequency adjustment feedback data and compensating and adjusting the frequency adjustment curve includes: The bearing temperature curve and vibration curve in the frequency regulation feedback data are decomposed by 4 layers of wavelet, and the frequency division threshold of 0.5-2kHz is selected according to the solubility characteristics of the refrigerant under low temperature conditions to obtain the characteristic waveform sequence; The characteristic waveform sequence is projected onto the temperature-vibration two-dimensional plane within a 100 ms sampling period, and a support vector machine algorithm based on a radial basis kernel function is used to identify the stable working area of ​​the system to generate a stability evaluation index; A 5×5 dimensional state transfer matrix is ​​established based on the stability evaluation index and the current oil-gas separation efficiency, and the operation trend of the next 10 cycles is predicted through a three-layer long short-term memory network to obtain a trend prediction vector; Performing singular value decomposition on the trend prediction vector, selecting characteristic components with a cumulative contribution rate of 90%, and performing online correction on the control parameters of the frequency regulation curve in combination with the refrigerant physical property parameters to obtain a compensation control sequence; The compensation control sequence is adaptively fused with the original frequency regulation curve, and a dynamic weight of [0.3, 0.7] is allocated according to the change rate of the oil-gas separation efficiency to achieve real-time compensation of the frequency regulation curve.

9. A high-efficiency variable frequency low-temperature compressor system, characterized in that: The compression method used in the high-efficiency variable frequency cryogenic compressor system according to any one of claims 1 to 8 comprises: A data acquisition module is used to collect operating parameters of preset parts of the compressor and perform data processing on the collected operating parameters to obtain actual oil-gas separation efficiency, wherein the operating parameters include temperature, pressure, current and power; A prediction calculation module is used to input the operating parameters of the compressor into a preset neural network prediction model, perform prediction calculations on the oil-gas separation efficiency, and obtain a predicted value of the oil-gas separation efficiency and a theoretical solubility reference value, wherein the operating parameters include the compressor inlet and outlet temperatures, the system pressure, and the refrigerant characteristic parameters; An optimization evaluation module is used to calculate the operating condition deviation data according to the actual oil-gas separation efficiency, the predicted value of the oil-gas separation efficiency and the theoretical solubility reference value, and to evaluate and calculate the comprehensive performance of the compressor at different frequency points using an optimization algorithm to generate a frequency adjustment curve; A frequency control module, used to divide the operating frequency of the compressor into a plurality of adjustment intervals according to the frequency adjustment curve, and to perform step-by-step adjustment on the operating frequency according to the adjustment intervals, and to monitor the bearing temperature and vibration parameters during the adjustment process to obtain frequency adjustment feedback data; The compensation adjustment module is used to periodically calculate the oil-gas separation efficiency deviation according to the frequency adjustment feedback data, and to perform compensation adjustment on the frequency adjustment curve.

Images