Anti-surge control method for compressor unit

Abstract

The invention discloses an anti-surge control method for a compressor unit, and particularly relates to the field of operation control of the compressor unit, a plurality of sensor terminals for pressure, flow, temperature and the like are deployed in the installation and debugging stage of the compressor unit, working parameters such as air inlet pressure, flow, exhaust pressure and the like are collected in real time, and data are preprocessed after collection; smoothing data by adopting a moving average filtering method, cleaning abnormal values according to a 3 sigma principle, processing missing values according to a missing proportion and normalizing the missing values, constructing an operation condition model based on preprocessed data, and calculating anti-surge threshold values in low, medium and high load areas by combining parameters such as air inlet flow and exhaust pressure; and constructing an exhaust pressure and intake flow prediction model by using historical data, and predicting a future state. The surge state is judged by comparing the predicted value with the threshold value, then the opening degree of the air inlet and exhaust adjusting valve is adjusted, stable operation of the compressor unit is guaranteed, the surge phenomenon is effectively prevented, and the operation safety and reliability of the compressor unit are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of compressor unit operation control, and more particularly, to an anti-surge control method for a compressor unit. Background Art

[0002] In modern industrial systems, compressor units are core equipment in many production processes, with a wide range of applications, covering key areas such as petrochemicals, natural gas transportation, refrigeration systems, and power generation. For example, in petrochemical production, compressor units are used to compress, transport, and boost gas pressure, playing a vital role in ensuring the continuity and stability of the production process. In natural gas transportation, they increase gas pressure, enabling efficient long-distance transportation.

[0003] With the continuous expansion of industrial production scale and the increasing complexity of production processes, the requirements for the stability, reliability and efficiency of compressor unit operation are also getting higher and higher. However, the surge problem has always been a key factor that plagues the stable operation of compressor units. Surge is an unstable phenomenon that occurs during the operation of the compressor. Its essence is that when the compressor flow rate decreases to a certain level, the air flow produces strong oscillations and backflows inside the compressor, resulting in drastic fluctuations in parameters such as pressure and flow. At present, in the field of anti-surge control technology, there are many technical solutions. Among them, the more common one is the control method based on fixed thresholds. This method sets a fixed surge control threshold based on experience or simple theoretical calculations. When the compressor operating parameters exceed the threshold, the corresponding control measures are initiated. There are also some technologies that monitor some operating parameters, such as only paying attention to the intake flow and exhaust pressure, to determine whether surge occurs, and perform control adjustments accordingly.

[0004] However, there are still some shortcomings in its actual use. For example, the traditional anti-surge control method has defects in the data acquisition and processing links. The collection of compressor unit operating parameters may not be comprehensive or accurate enough, and it is difficult to obtain multi-dimensional parameters such as cylinder temperature, bearing temperature, intake pipe temperature and exhaust pipe temperature, resulting in an inability to accurately reflect the actual operating status of the unit. At the same time, in terms of setting the anti-surge threshold, previous methods often use fixed thresholds or simple empirical formulas, without fully considering the differences in the operating characteristics of the compressor unit under different operating conditions. This results in the threshold being unable to accurately reflect the actual surge boundary of the compressor, prone to misjudgment, and unable to prevent surge in a timely and effective manner. Moreover, the traditional method for judging the surge state is relatively simple, mainly relying on a few parameters such as intake flow and exhaust pressure, ignoring other important indicators such as vibration amplitude, and the judgment of the surge state is only based on the currently obtained data, which may miss early surge risks, fail to play a predictive role, and delay the start of anti-surge control measures. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides an anti-surge control method for a compressor unit, which solves the problems raised in the above-mentioned background technology through the following scheme.

[0006] To achieve the above-mentioned object, the present invention provides the following technical solutions: an anti-surge control method for a compressor unit, comprising: S1: sensor deployment: during the installation and commissioning phase of the compressor unit, deploying sensor terminals according to the measurement requirements of various parameters;

[0007] The sensor terminal includes: a pressure sensor, a flow sensor, a temperature sensor, a rotation speed sensor and a vibration sensor;

[0008] S2: Compressor unit data collection: Based on the sensor terminals deployed in step S1, the compressor unit operating parameters are collected in real time;

[0009] The operating parameters of the compressor unit include intake pressure, intake flow, exhaust pressure, exhaust flow, operating temperature, compressor speed and vibration amplitude;

[0010] The operating temperature includes cylinder temperature, bearing temperature, intake pipe temperature and exhaust pipe temperature;

[0011] S3: Data preprocessing: Use the moving average filtering method to smooth the operating parameters of the compressor unit, check whether there are abnormal values ​​and missing values ​​in the collected parameters, and perform data cleaning. The cleaned data is normalized to obtain preprocessed data;

[0012] S4: Dynamic anti-surge threshold calculation: Based on the pre-processed data obtained in step S3, a compressor unit operating condition model is constructed. Intake air flow and exhaust pressure are used as main variables. In combination with the operating condition model, anti-surge thresholds for the low load zone, medium load zone, and high load zone of the compressor operation are constructed.

[0013] S5: Future state prediction: Based on the historical operating data, an exhaust pressure prediction model and an intake flow prediction model are constructed. The pre-processed data obtained in step S3 are substituted into the exhaust pressure prediction model and the intake flow prediction model to obtain the exhaust pressure prediction value and the intake flow prediction value at the future moment.

[0014] S6: Surge state determination: comparing the exhaust pressure prediction value and intake flow prediction value obtained in step S5 with the anti-surge threshold value obtained by dynamic calculation to determine the surge state of the compressor unit at a future time;

[0015] S7: Control execution: Based on the judgment result in step S6, the openings of the exhaust control valve and the intake control valve are adjusted until the exhaust pressure prediction value and the intake flow prediction value are compared with the anti-surge threshold value obtained by dynamic calculation and the result is a normal operating state.

[0016] Technical effects and advantages of the present invention:

[0017] 1. This invention deploys multiple sensors such as pressure, flow, temperature, speed, and vibration to comprehensively collect the operating parameters of the compressor unit, including intake pressure, intake flow, exhaust pressure, exhaust flow, operating temperature, compressor speed, and vibration amplitude. This ensures that the acquired data accurately reflects the operating status of the unit. The moving average filtering method is used to smooth the data, and the 3σ principle is combined to identify and clean outliers. Different methods are used to handle missing values ​​based on the missing ratio. Finally, data normalization is performed, effectively improving data quality and providing a reliable basis for subsequent analysis and decision-making.

[0018] 2. The present invention constructs an operating condition model based on collected multi-parameter data, using intake flow and exhaust pressure as the main variables. Combined with historical operating data under different operating conditions, corresponding anti-surge threshold calculation formulas are constructed for low-load, medium-load, and high-load zones. This dynamic threshold calculation method fully considers the operating characteristics of the compressor under different operating conditions, can more accurately reflect the surge boundary, improve the accuracy and timeliness of surge judgment, and effectively prevent surge occurrence;

[0019] 3. The present invention constructs an exhaust pressure prediction model and an intake flow prediction model based on historical operating data. The exhaust pressure and intake flow values ​​at future moments are predicted based on the prediction model. The surge state of the compressor unit can be predicted before surge occurs, effectively preventing surge from occurring. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a schematic diagram of the overall structure of the present invention. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the 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 making creative efforts are within the scope of protection of the present invention.

[0022] As attached Figure 1 An anti-surge control method for a compressor unit is shown, comprising:

[0023] S1: Sensor deployment: During the installation and commissioning phase of the compressor unit, sensor terminals are deployed according to the measurement requirements of various parameters;

[0024] The sensor terminal includes: a pressure sensor, a flow sensor, a temperature sensor, a rotation speed sensor and a vibration sensor;

[0025] It should be further explained that the deployment method of the sensor terminal is as follows:

[0026] Install pressure sensors and flow sensors on the intake pipe near the compressor inlet, and also on the exhaust pipe near the outlet; arrange temperature sensors reasonably on the compressor cylinder and bearings; install speed sensors on the motor shaft; install vibration sensors on the casing and bearing seats. After completing the sensor installation, calibrate all sensors to ensure that their measurement accuracy meets the requirements, and establish a connection between the sensors and the data acquisition module to prepare for data collection.

[0027] S2: Compressor unit data collection: Based on the sensor terminals deployed in step S1, the compressor unit operating parameters are collected in real time;

[0028] Specifically, the compressor unit operating parameters include intake pressure, intake flow, exhaust pressure, exhaust flow, operating temperature, compressor speed, and vibration amplitude;

[0029] The operating temperature includes cylinder temperature, bearing temperature, intake pipe temperature and exhaust pipe temperature;

[0030] S3: Data preprocessing: Use the moving average filtering method to smooth the operating parameters of the compressor unit, check whether there are abnormal values ​​and missing values ​​in the collected parameters, and perform data cleaning. The cleaned data is normalized to obtain preprocessed data;

[0031] It should be specifically noted that the smoothing process refers to setting the window size to 5 for the time series data of each parameter and calculating the average value of the data in the window as the smoothed value of the data in the center of the window;

[0032] The data cleaning refers to the identification of outliers using the statistically based 3σ principle. If a data point deviates from the mean by more than 3 standard deviations, it is considered an outlier and replaced with the mean of the adjacent data. For missing values, different processing methods are selected according to the missing ratio. When the missing ratio is less than 10%, linear interpolation is used to fill it in. When the missing ratio is between 10% and 30%, a regression model is used to predict the missing values. If the missing ratio exceeds 30%, the data segment is discarded.

[0033] S4: Dynamic anti-surge threshold calculation: Based on the pre-processed data obtained in step S3, a compressor unit operating condition model is constructed. Intake air flow and exhaust pressure are used as main variables. In combination with the operating condition model, anti-surge thresholds for the low load zone, medium load zone, and high load zone of the compressor operation are constructed.

[0034] It should be specifically explained that the construction process of the compressor unit operating condition model is as follows:

[0035] Based on the collected intake pressure P in , intake air flow Q in , exhaust pressure P out , exhaust flow Q out , cylinder temperature T cyl , bearing temperature T b , intake duct temperature T in-pipe , exhaust pipe temperature T out-pipe , compressor speed n, vibration amplitude A, and other 10 operating parameters, a multiple linear regression model was used to establish the compressor unit operating condition model, with the critical pressure ratio γ cr As the dependent variable, its expression is:

[0036] γ cr =β0+β1P in +β2Q in +β3P out +β4Q out +β5T cyl +β6T b +β7T in-pipe +β8T out-pipe +β9n+β 10 A+ε

[0037] Among them, γ cr The critical pressure ratio is the ratio of exhaust pressure to intake pressure. It is an important indicator to measure whether the compressor is close to the surge state. It reflects the pressure limit at which the compressor can operate stably under different working conditions. When the actual pressure ratio is close to or exceeds the critical pressure ratio, the compressor is at risk of surge. β0 refers to the constant term in the regression model, which represents the theoretical initial value of the critical pressure ratio when all independent variables are 0. In practical terms, it includes the comprehensive influence of other factors on the critical pressure ratio that are not considered by the model, β1, β2, ..., β 10 The corresponding independent variables P in , Q in 、P out , Q out 、T cyl 、T b 、T in-pipe 、T out-pipe , n, and A regression coefficients, which represent the degree and direction of influence of each independent variable on the critical pressure ratio. For example, β1 represents the intake pressure P in When the other independent variables remain unchanged, the critical pressure ratio γ cr The change of β2 represents the intake air flow rate Q inThe degree of influence on the critical pressure ratio is obtained by collecting at least 1,000 sets of historical operating data under different operating conditions and fitting them using the least squares method to ensure that the model can accurately reflect the relationship between each parameter and the critical pressure ratio. ε refers to the error term, which is used to represent the influence of random factors or measurement errors not considered in the model on the dependent variable. This value is also obtained by fitting the historical operating data using the least squares method. Since there are many factors in actual operation that cannot be accurately modeled, the existence of the error term makes the model more in line with the actual situation.

[0038] The low load region refers to the intake air flow rate Q in <300m 3 / h, low load area exhaust pressure anti-surge threshold P th1 The calculation formula is:

[0039] P th1 =a1Q in +a2P in +a3T in-pipe +a4n+b1;

[0040] Low load area intake flow anti-surge threshold Q th1 The calculation formula is:

[0041] Q th1 =c1P out +c2T out-pipe +c3A+d1;

[0042] Among them, P th1 Refers to the exhaust pressure anti-surge threshold in the low-load area. It is the exhaust pressure critical value to judge whether the compressor will surge when running at low load. When the actual exhaust pressure exceeds this threshold, the compressor is at risk of entering a surge state. th1 Refers to the anti-surge threshold of intake flow in the low-load area, which is used to measure whether the intake flow is in a safe range under low-load conditions. If the actual intake flow is lower than this threshold, surge may occur. a1, a2, a3, a4, b1, c1, c2, c3, and d1 are coefficients determined by analyzing and fitting the historical operating data in the low-load area. These coefficients reflect the weight and direction of the influence of various related parameters on the anti-surge threshold. For example, a1 means that in the low-load area, the intake flow Q in Every time the exhaust pressure anti-surge threshold value P changes by one unit, th1 The change in the exhaust pressure P out Intake air flow anti-surge threshold Q th1 The value is obtained based on the statistical analysis and mathematical fitting of a large amount of actual operating data in the low-load area to ensure the accuracy of the threshold calculation.

[0043] The medium load area refers to 300m 3 / h≤Q in ≤700m 3 / h, medium load exhaust pressure anti-surge threshold P th2 The calculation formula is:

[0044] P th2 =e1Q in +e2P in +e3Q out +e4T cyt +f1;

[0045] Intake flow anti-surge threshold Q in the medium load area th2 The calculation formula is:

[0046] Q th2 =g1P out +g2T b +g3n+h1;

[0047] Among them, P th2 and Q th2 They refer to the exhaust pressure anti-surge threshold in the medium load area and the intake flow anti-surge threshold in the medium load area, which are important bases for judging whether surge occurs when the compressor is operating in medium load conditions. e1, e2, e3, e4, f1, g1, g2, g3, and h1 are coefficients determined by analyzing and fitting the historical operating data in the medium load area. Similar to the low load area coefficients, they quantify the impact of various parameters on the anti-surge threshold in the medium load area. For example, e3 represents the exhaust flow rate Q out The exhaust pressure surge prevention threshold P in the middle load area th2 g2 represents the influence of bearing temperature Tb on the intake flow anti-surge threshold Q in the medium load area. th2 The value of the influence is determined according to the actual operating data characteristics of the medium load area.

[0048] The high load region is Q in >700m 3 / h, high load area exhaust pressure anti-surge threshold P th3 The calculation formula is:

[0049] P th3 =i1Q in +i2P in +i3Q out +i4T out-pipe +j1;

[0050] Intake flow anti-surge threshold Q in high load area th3 The calculation formula is:

[0051] Q th3 =k1P out +k2A+k3n+l1;

[0052] Among them, Pth3 and Qth3 refer to the high-load area exhaust pressure anti-surge threshold and the high-load area intake flow anti-surge threshold, respectively, which are used to define the pressure and flow range of safe operation of the compressor under high-load conditions. i1, i2, i3, i4, j1, k1, k2, k3, l1 are also coefficients determined by analyzing and fitting the historical operating data of the high-load area, reflecting the relationship between the various parameters and the anti-surge threshold under high-load conditions. For example, i4 represents the exhaust pipe temperature T out-pipe The exhaust pressure surge prevention threshold P in the high load area th3 k2 represents the influence of the vibration amplitude A on the anti-surge threshold Q of the intake flow in the high load area. th3 The value is based on the analysis and fitting of the actual operating data in the high-load area to ensure that the threshold can adapt to the characteristics of high-load conditions.

[0053] S5: Future state prediction: Based on the historical operating data, an exhaust pressure prediction model and an intake flow prediction model are constructed. The pre-processed data obtained in step S3 are substituted into the exhaust pressure prediction model and the intake flow prediction model to obtain the exhaust pressure prediction value and the intake flow prediction value at the future moment.

[0054] It should be further explained that the exhaust pressure prediction model is as follows:

[0055]

[0056] Among them, P out (k+1) represents the predicted value of exhaust pressure at time k+1, which is the key result of the model output and provides an important reference for subsequent exhaust valve adjustment. out (k) refers to the actual exhaust pressure at time k, which reflects the continuity of the exhaust pressure in the time series and affects the predicted value at the next moment. Refers to the autoregressive coefficient, which measures the influence of the exhaust pressure at time k on the predicted value of the exhaust pressure at time k+1. The value range is generally between -1 and 1 and is determined through historical data training. θ1 refers to the sliding average coefficient, which describes the influence of the error term ε(k) at time k on the predicted value of the exhaust pressure at time k+1, reflecting the effect of past random interference on the current prediction. ε(k) refers to the error term at time k, which includes random factors or measurement errors not considered by the model, etc. in (k), Q in (k), T in-pipe(k), n(k), and A(k) are the intake pressure, intake flow, intake duct temperature, compressor speed, and vibration amplitude at time k, respectively. These parameters affect the exhaust pressure in different aspects and are important input variables of the model. u1(k), u2(k), u3(k), u4(k), and u5(k) are the coefficients of intake pressure, intake flow, intake duct temperature, compressor speed, and vibration amplitude, respectively, which are used to quantify the influence weight of each parameter on the predicted value of exhaust pressure at time k+1.

[0057] The intake air flow prediction model is as follows:

[0058] Q in (k+1)=λ1Q in (k)+δ1ξ(k)+w1(k)P in (k)+w2(k)P out (k)+w3(k)T in-pipe (k)+w4(k)n(k)+w5(k)A(k)

[0059] Among them, Q in (k+1) represents the intake flow prediction value at time k+1. This value is the key result of the model output and is used to predict the intake flow of the compressor at future times. The control system can plan the adjustment strategy in advance based on this prediction value to ensure that the intake flow is in a reasonable range and effectively prevent surge. in (k) is the actual intake flow at moment k, reflecting the current intake state of the compressor. As the input of the autoregressive part, it reflects the continuity and inertia of the intake flow in the time series, which means that the intake flow at the next moment is closely related to the intake flow at the current moment. λ1 refers to the autoregressive coefficient, which is used to measure the influence of the intake flow at moment k on the intake flow prediction value at moment k+1. Its value range is usually between -1 and 1. It is determined by statistical analysis and model training of a large amount of historical operating data. If λ1 is close to 1, it indicates that the intake flow has a strong autocorrelation in time, that is, the intake flow at the current moment has a greater impact on the next moment. If λ1 is close to 0, it means that the autocorrelation is weak. δ1 refers to the sliding average coefficient, which describes the error term ξ( k) on the predicted value of the intake flow at time k+1, which reflects the degree of influence of random interference or unmodeled factors in the past on the current prediction. It is also obtained through data training. The size of δ1 determines the degree of dependence of the model on historical errors. Reasonable adjustment of this coefficient helps to improve the prediction accuracy of the model. ξ(k) refers to the error term at time k, which represents the influence of random factors or measurement errors not considered in the model on the dependent variable. In actual operation, there are many factors that cannot be accurately modeled, such as sudden changes in gas composition and slight blockages in pipelines. These factors will cause deviations between the model prediction value and the actual value. The introduction of the error term makes the model more in line with the actual situation.in (k) refers to the intake pressure at time k. Changes in intake pressure will directly affect the amount of gas flowing in, and thus affect the intake flow rate. For example, an increase in intake pressure will usually increase the intake flow rate when other conditions remain unchanged. Therefore, it is one of the important influencing factors in the model. w1(k) refers to the intake pressure P in The coefficient corresponding to (k) is used to quantify the influence weight of the intake pressure at time k on the intake flow prediction value at time k+1. Its value is determined by data fitting and optimization algorithm, reflecting the specific correlation strength between intake pressure and intake flow. out (k) refers to the exhaust pressure at time k. The exhaust pressure will have an indirect effect on the intake flow. When the exhaust pressure is too high, it may hinder the flow of gas, resulting in a decrease in the intake flow. Conversely, a decrease in exhaust pressure may help increase the intake flow. w2(k) refers to the exhaust pressure P out The coefficient corresponding to (k) is used to measure the influence of the exhaust pressure at time k on the intake flow prediction value at time k+1, reflecting the quantitative relationship between exhaust pressure and intake flow. in-pipe (k) refers to the intake duct temperature at time k. Temperature affects the density and volume of the gas, and thus affects the intake flow rate. For example, if the intake duct temperature increases, the gas density decreases, which may cause the intake flow rate to change at the same intake pressure. w3(k) refers to the intake duct temperature T in-pipe The coefficient corresponding to (k) is used to determine the weight of the impact of the intake duct temperature at time k on the predicted intake flow at time k+1. n(k) refers to the compressor speed at time k, which directly affects the gas intake capacity. An increase in speed generally increases intake flow and is one of the important operating variables for regulating intake flow. w4(k) is the coefficient corresponding to the compressor speed n(k), which quantifies the degree of influence of the compressor speed at time k on the predicted intake flow at time k+1. A(k) refers to the vibration amplitude at time k. The vibration amplitude reflects the operating stability and mechanical condition of the compressor. Abnormal vibration may indicate an internal fault or unstable operating condition of the compressor, which in turn affects the intake flow. For example, excessive vibration may cause loose pipe connections or reduced sealing performance, indirectly affecting the intake flow. w5(k) is the coefficient corresponding to the vibration amplitude A(k), which is used to weigh the impact of the vibration amplitude at time k on the predicted intake flow at time k+1.

[0060] It should be further explained that the coefficients in the exhaust pressure and flow prediction model are as follows: θ1, u1(k)-u5(k) and λ1, δ1, w1(k)-w5(k) in the intake air flow prediction model are mainly obtained by parameter estimation based on historical operating data, specifically using the least squares method for estimation;

[0061] The least squares method is as follows:

[0062] Taking the exhaust pressure prediction model as an example, a large amount of historical operating data is collected, including P under different working conditions. out (k), P in (k), Q in (k), T in-pipe The actual measured values ​​of parameters such as (k), n(k), and A(k) are substituted into the exhaust pressure prediction model formula to construct the objective function:

[0063]

[0064] Where n is the number of data samples. By minimizing the objective function S, the coefficient value that makes S the smallest is calculated, which is the estimated value of the model coefficient. The coefficient of the exhaust flow prediction model is calculated using the same principle;

[0065] The error terms in the exhaust pressure and flow prediction model, such as ε(k) and ξ(k), include random factors and measurement errors that are not considered by the model. The mean and variance of the error terms are calculated through statistical analysis of historical operating data. For example, over a period of time, the difference between the model prediction value and the actual measurement value is recorded each time, and these differences are used as sample data for the error terms to calculate their mean μ and variance σ. 2 In the subsequent model calculation, it is assumed that the error term has a mean of μ and a variance of σ. 2 Normal distribution, that is, ε(k)~N(μ,σ 2 ),ω(k)~N(μ,σ 2 ), used for model calculation.

[0066] S6: Surge state judgment: The exhaust pressure prediction value and the intake flow prediction value obtained in step S5 are combined with the anti-surge threshold value obtained by dynamic calculation to perform surge judgment, and judge the surge state of the compressor unit at a future time;

[0067] The method for judging surge is as follows:

[0068] When the intake air flow rate Q in Less than the intake flow anti-surge threshold Q under the corresponding working conditions th , and the exhaust pressure P out Greater than the exhaust pressure anti-surge threshold P under the corresponding working conditions th When the vibration amplitude A exceeds 2 times the standard deviation of the average value of the normal operating range, even if the operating parameters have not completely exceeded the anti-surge threshold range, it is determined that there is a surge risk and anti-surge control measures are started in advance.

[0069] S7: Control execution: Based on the judgment result in step S6, the openings of the exhaust control valve and the intake control valve are adjusted until the exhaust pressure prediction value and the intake flow prediction value are compared with the anti-surge threshold value obtained by dynamic calculation and the result is a normal operating state.

[0070] Secondly: The drawings of the embodiments disclosed in the present invention only involve structures related to the embodiments disclosed in the present invention. Other structures may refer to conventional designs. The same embodiment and different embodiments of the present invention may be combined with each other without conflict.

[0071] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for controlling surge prevention in a compressor unit, characterized in that: include: S1: Sensor deployment: During the installation and commissioning phase of the compressor unit, sensor terminals are deployed according to the measurement requirements of various parameters; The sensor terminal includes: a pressure sensor, a flow sensor, a temperature sensor, a rotation speed sensor and a vibration sensor; S2: Compressor unit data collection: Based on the sensor terminals deployed in step S1, the compressor unit operating parameters are collected in real time; The operating parameters of the compressor unit include intake pressure, intake flow, exhaust pressure, exhaust flow, operating temperature, compressor speed and vibration amplitude; The operating temperature includes cylinder temperature, bearing temperature, intake pipe temperature and exhaust pipe temperature; S3: Data preprocessing: Use the moving average filtering method to smooth the operating parameters of the compressor unit, check whether there are abnormal values ​​and missing values ​​in the collected parameters, and perform data cleaning. The cleaned data is normalized to obtain preprocessed data; S4: Dynamic anti-surge threshold calculation: Based on the pre-processed data obtained in step S3, a compressor unit operating condition model is constructed. Intake air flow and exhaust pressure are used as main variables. In combination with the operating condition model, anti-surge thresholds for the low load zone, medium load zone, and high load zone of the compressor operation are constructed. S5: Future state prediction: Based on the historical operating data, an exhaust pressure prediction model and an intake flow prediction model are constructed. The pre-processed data obtained in step S3 are substituted into the exhaust pressure prediction model and the intake flow prediction model to obtain the exhaust pressure prediction value and the intake flow prediction value at the future moment. S6: Surge state determination: comparing the exhaust pressure prediction value and intake flow prediction value obtained in step S5 with the anti-surge threshold value obtained by dynamic calculation to determine the surge state of the compressor unit at a future time; S7: Control execution: Based on the judgment result in step S6, the openings of the exhaust control valve and the intake control valve are adjusted until the exhaust pressure prediction value and the intake flow prediction value are compared with the anti-surge threshold value obtained by dynamic calculation and the result is a normal operating state.

2. The anti-surge control method for a compressor unit according to claim 1, characterized in that: The smoothing process refers to setting the window size to 5 for the time series data of each parameter and calculating the average value of the data in the window as the smoothed value of the data in the center of the window; Data cleaning refers to the identification of outliers using the statistical 3σ principle. If a data point deviates from the mean by more than 3 standard deviations, it is considered an outlier and replaced with the mean of the adjacent data. For missing values, different processing methods are selected according to the missing ratio. When the missing ratio is less than 10%, linear interpolation is used to fill it. When the missing ratio is between 10% and 30%, a regression model is used to predict the missing value. If the missing ratio exceeds 30%, the data segment is discarded.

3. The anti-surge control method for a compressor unit according to claim 1, characterized in that: The construction process of the compressor unit operating condition model is as follows: Based on the collected intake pressure P in , intake air flow Q in , exhaust pressure P out , exhaust flow Q out , cylinder temperature T cyl , bearing temperature T b , intake duct temperature T in-pipe , exhaust pipe temperature T out-pipe , compressor speed n, vibration amplitude A, and other 10 operating parameters, a multiple linear regression model was used to establish the compressor unit operating condition model, with the critical pressure ratio γ cr As the dependent variable, its expression is: c cr =β0+β1P in +β2Q in +β3P out +β4Q out +β5T cyl +β6T b +β7T in-pipe +β8T out-pipe +β9n+β 10 A+e Among them, γ cr Refers to the critical pressure ratio, which is the ratio of exhaust pressure to intake pressure. β0 refers to the constant term in the regression model, which represents the theoretical initial value of the critical pressure ratio when all independent variables are 0. β1, β2,…, β 10 The corresponding independent variables P in , Q in 、P out , Q out 、T cyl 、T b 、T in-pipe 、T out-pipe , n, and the regression coefficient of A. ε refers to the error term, which is used to represent the impact of random factors or measurement errors not considered in the model on the dependent variable.

4. The anti-surge control method for a compressor unit according to claim 1, characterized in that: The low load region refers to the intake air flow rate Q in <300m 3 / h, low load area exhaust pressure anti-surge threshold P th1 The calculation formula is: P th1 =a1Q in +a2P in +a3T in-pipe +a4n+b1; Low load area intake flow anti-surge threshold Q th1 The calculation formula is: Q th1 =c1P out +c2T out-pipe +c3A+d1; Among them, P th1 Refers to the exhaust pressure anti-surge threshold in the low load area, Q th1 Refers to the intake flow anti-surge threshold in the low-load area, and a1, a2, a3, a4, b1, c1, c2, c3, and d1 are coefficients determined by analyzing and fitting the historical operating data in the low-load area; The medium load area refers to 300m 3 / h≤Q in ≤700m 3 / h, medium load area exhaust pressure anti-surge threshold P th2 The calculation formula is: P th2 =e1Q in +e2P in +e3Q out +e4T cyl +f1; Intake flow anti-surge threshold Q in the medium load area th2 The calculation formula is: Q th2 =g1P out +g2T b +g3n+h1; Among them, P th2 and Q th2 They refer to the exhaust pressure anti-surge threshold and the intake flow anti-surge threshold in the medium load area respectively. e1, e2, e3, e4, f1, g1, g2, g3, and h1 are coefficients determined by analyzing and fitting the historical operating data in the medium load area. The high load region is Q in >700m 3 / h, high load area exhaust pressure anti-surge threshold P th3 The calculation formula is: P th3 =i1Q in +i2P in +i3Q out +i4T out-pipe +j1; Intake flow anti-surge threshold Q in high load area th3 The calculation formula is: Q th3 =k1P out +k2A+k3n+l1; Among them, Pth3 and Qth3 refer to the exhaust pressure anti-surge threshold and intake flow anti-surge threshold in the high load area, respectively. i1, i2, i3, i4, j1, k1, k2, k3, and l1 are coefficients determined by analyzing and fitting the historical operating data in the high load area.

5. The anti-surge control method for a compressor unit according to claim 1, characterized in that: The exhaust pressure prediction model is as follows: Among them, P out (k+1) represents the predicted exhaust pressure at time k+1, P out (k) refers to the actual exhaust pressure at time k, refers to the autoregressive coefficient, θ1 refers to the sliding average coefficient, ε(k) refers to the error term at time k, P in (k), Q in (k), T in-pipe (k), n(k), A(k) are the intake pressure, intake flow, intake duct temperature, compressor speed and vibration amplitude at time k, respectively; u1(k), u2(k), u3(k), u4(k), u5(k) are the coefficients of intake pressure, intake flow, intake duct temperature, compressor speed and vibration amplitude, respectively.

6. The anti-surge control method for a compressor unit according to claim 1, characterized in that: The intake air flow prediction model is as follows: Q in (k+1)=λ1Q in (k)+δ1ξ(k)+w1(k)P in (k)+w2(k)P out (k)+w3(k)T in-pipe (k)+w4(k)n(k)+w5(k)A(k) Among them, Q in (k+1) represents the intake air flow prediction value at time k+1, Q in (k) is the actual intake flow rate at time k, λ1 refers to the autoregressive coefficient, δ1 refers to the sliding average coefficient, ξ(k) refers to the error term at time k, P in (k) refers to the intake pressure at time k, w1(k) refers to the intake pressure P in (k) The corresponding coefficient, P out (k) refers to the exhaust pressure at time k, w2(k) refers to the exhaust pressure P out (k) The corresponding coefficient, T in-pipe (k) refers to the intake duct temperature at time k, w3(k) refers to the intake duct temperature T in-pipe (k), n(k) refers to the compressor speed at time k, w4(k) refers to the coefficient corresponding to the compressor speed n(k), A(k) refers to the vibration amplitude at time k, and w5(k) refers to the coefficient corresponding to the vibration amplitude A(k).

7. The anti-surge control method for a compressor unit according to claim 1, characterized in that: The method for judging surge is as follows: When the intake air flow rate Q in Less than the intake flow anti-surge threshold Q under the corresponding working conditions th , and the exhaust pressure P out Greater than the exhaust pressure anti-surge threshold P under the corresponding working conditions th When the vibration amplitude A exceeds 2 times the standard deviation of the average value of the normal operating range, even if the operating parameters have not completely exceeded the anti-surge threshold range, it is determined that there is a surge risk and anti-surge control measures are started in advance.

Images