Intelligent liquid injection control system and method for air compressor based on artificial intelligence algorithm

Abstract

The application discloses an air compressor intelligent liquid injection control system and method based on an artificial intelligence algorithm and relates to the technical field of air compressor control.The system comprises the following steps: setting an exhaust temperature target value, collecting the actual exhaust temperature of an air compressor host machine in real time, inputting the deviation of the two into an outer ring temperature control module, calculating a target preset flow rate through a temperature special-purpose PID algorithm, receiving the current opening degree of a liquid injection electric regulating valve by an inner ring flow control module, calculating a predicted opening degree based on historical operation data and the target preset flow rate, acquiring the current flow rate of a liquid injection circuit, calculating and outputting a feedback opening degree for correction according to the deviation between the target preset flow rate and the current flow rate, fusing and calculating the predicted opening degree and the feedback opening degree through a preset weighting algorithm to generate a final target valve opening degree and executing the final target valve opening degree, and performing cyclic execution and state monitoring.The application can intelligently and adaptively regulate the liquid injection amount under different working conditions, accurately control the liquid injection flow rate, reduce energy consumption, and ensure safe and reliable operation of equipment.

Description

Technical Field

[0001] This invention relates to the field of air compressor control technology, and in particular to an intelligent liquid injection control system and method for air compressors based on artificial intelligence algorithms. Background Technology

[0002] Existing liquid-injected positive displacement air compressors are mainly used in industrial manufacturing. Their core principle is to inject a suitable amount of liquid cooling medium into the compression chamber during the compression process to achieve cooling, lubrication, and sealing effects, thereby improving compressor efficiency and reducing exhaust temperature. These air compressors include various structural types such as single-screw and twin-screw compressors, and the liquid cooling media include water and lubricating oil. Liquid-injected cooling is widely used, especially in medium-pressure models. Currently, the mainstream control technology for air compressors is variable frequency speed control, which adjusts the motor speed based on collected exhaust pressure data to match changes in the end-user air load. However, existing liquid-injected positive displacement air compressors mainly use simple on / off control or fixed orifices for liquid injection control, which cannot accurately match the cooling flow required under variable frequency operation.

[0003] When operating at low speed with variable frequency, air compressors are prone to "over-spraying" when using fixed nozzles to spray liquid. When the flow rate of the cooling medium exceeds the actual requirement of the compression process, too much cooling medium will affect the compression efficiency, such as premature increase in internal pressure and increased motor load. The energy consumption required per unit volume of air will increase, resulting in wasted power. Moreover, excessive injection of coolant will cause the original compressed gas to fill the compression chamber with liquid before the exhaust port is connected, i.e., during the last compression of the compression chamber. Since the liquid cannot be compressed, the bearing stress will increase significantly, and the screw or star wheel stress will increase significantly, shortening the machine's lifespan and even causing the machine to be damaged after a short period of use.

[0004] Therefore, it is necessary to propose an intelligent liquid injection control system and method suitable for liquid injection positive displacement air compressors, which can achieve adaptive adjustment of liquid injection flow rate without significantly increasing hardware costs. The liquid injection flow rate can be adaptively adjusted according to changes in air volume (the speed of the motor affects the amount of heat generated by the air compressor, thus affecting the amount of liquid injection), thereby improving the system's energy-saving performance, operational stability, and machine reliability, and effectively preventing compressed liquid. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide an intelligent liquid injection control system and method for air compressors based on artificial intelligence algorithms that can accurately control the amount of liquid injected. This system can achieve adaptive adjustment of the liquid injection flow rate and suppress over-injection while ensuring that there is no liquid shortage. It is especially suitable for positive displacement air compressors that use liquid injection cooling in medium-pressure models.

[0006] This invention is achieved through the following technical solution: An intelligent liquid injection control system for an air compressor based on artificial intelligence algorithms includes a drive motor, an air compressor main unit, an air filter, an exhaust temperature sensor, a liquid-gas separator, a liquid cooler, a liquid filter, and a liquid shut-off valve. The drive motor is connected to the air compressor main unit. The air filter is connected to the air inlet of the air compressor main unit via an intake valve. The air outlet of the air compressor main unit is connected to the upper part of the liquid-gas separator via an exhaust check valve, forming a gas-liquid mixture pipeline. The exhaust temperature sensor is installed on the gas-liquid mixture pipeline between the exhaust check valve and the liquid cooler. Between the gas-liquid separators, the actual exhaust temperature of the air compressor is monitored. The liquid-liquid separator is used for gas-liquid separation. After the gas-liquid mixture enters the liquid-liquid separator, it separates into air at the top and coolant at the bottom. The top of the liquid-liquid separator is connected to a vent pipe and an exhaust pipe. A safety valve is installed on the vent pipe, and a minimum safety valve and an exhaust valve are sequentially installed on the exhaust pipe along the gas discharge direction. The exhaust pipe is used to connect to an external air-consuming unit. The liquid outlet at the bottom of the liquid-liquid separator is connected to the liquid injection port of the air compressor through a liquid injection circuit. The liquid return circuit is used for lubrication and cooling of the air compressor main unit. The liquid cooler, liquid filter, and liquid shut-off valve are sequentially arranged on the liquid injection circuit along the liquid flow direction. An electric liquid injection regulating valve is also installed between the liquid shut-off valve and the liquid injection port of the air compressor main unit on the liquid injection circuit. The electric liquid injection regulating valve is electrically connected to a controller, which is also electrically connected to an exhaust temperature sensor. The controller receives the opening data of the electric liquid injection regulating valve and the exhaust temperature data, and controls and regulates the opening of the electric liquid injection regulating valve in real time. The controller includes an outer loop temperature control module and an inner loop flow control module. The outer loop temperature control module and the inner loop flow control module form a cascaded PID feedback structure. The outer loop temperature control module outputs a target preset flow rate to the inner loop flow control module based on the exhaust temperature transmitted by the exhaust temperature sensor using a temperature-specific PID algorithm. The inner loop flow control module obtains a predicted opening based on feedforward offline data, obtains a feedback opening based on an incremental PID algorithm, and calculates and outputs the target valve opening based on a weighted fusion of the predicted opening and the feedback opening, thereby achieving precise control of the opening of the electric liquid injection regulating valve.

[0007] Furthermore, a bypass passage is added to the injection circuit. The bypass passage is connected in parallel to both ends of the injection electric regulating valve. The bypass passage is provided with a throttling orifice. The throttling orifice is used to provide the basic injection flow rate. Even when the valve opening is small, it can still provide the minimum flow rate to avoid the air compressor main unit being damaged by high temperature due to lack of liquid cooling.

[0008] Furthermore, a flow meter is also installed on the injection circuit. The flow meter is located between the electric injection regulating valve and the air compressor host. The flow meter is electrically connected to the controller and is used to collect the flow data of the injection circuit and transmit it to the controller to realize closed-loop precise control of the opening degree of the electric injection regulating valve.

[0009] A method for intelligent liquid injection control of an air compressor based on artificial intelligence algorithms, employing the aforementioned intelligent liquid injection control system for an air compressor based on artificial intelligence algorithms, includes the following steps: Step 1: Initialization and Signal Acquisition: The controller performs a power-on self-test and initialization, receives the user-set exhaust temperature target value (the exhaust temperature set value cannot be lower than the ambient temperature +10℃ and cannot exceed the ambient temperature +20℃), and receives the actual exhaust temperature of the air compressor host collected in real time by the exhaust temperature sensor as the main feedback signal. Step 2: Outer loop temperature control and target flow generation: The deviation between the target exhaust temperature and the actual exhaust temperature is input into the outer loop temperature control module in the controller; the outer loop temperature control module calculates and outputs the target preset flow rate through a temperature-specific PID algorithm, where the target preset flow rate is the amount of cooling medium injected to achieve the target exhaust temperature under the current operating conditions; Step 3: Feedforward control to obtain predicted opening degree: The inner loop flow control module in the controller receives the current opening degree of the liquid injection electric regulating valve and calculates the predicted opening degree based on the feedforward offline data and the target preset flow rate obtained in Step 2. Step 4: Inner loop flow control, obtaining feedback opening: Obtain the current flow rate of the injection circuit. The inner loop flow control module calculates and outputs a feedback opening for correction based on the deviation between the target preset flow rate obtained in Step 2 and the current flow rate through an incremental PID algorithm. Step 5, Weighted Fusion and Command Output: The controller fuses the predicted opening (feedforward) obtained in Step 3 and the feedback opening (feedback correction) obtained in Step 4 using a preset weighted algorithm to generate a final target valve opening, and sends the control command corresponding to this target valve opening to the liquid injection electric regulating valve for execution via a communication protocol. Step Six: Cyclic Execution and Status Monitoring: Steps Two through Five are continuously executed in a cyclical manner. The amount of liquid injected into the injection circuit is dynamically adjusted through the electric regulating valve until the actual exhaust temperature of the gas-liquid mixture pipeline stabilizes within the preset exhaust temperature target range. At the same time, the actual exhaust temperature is monitored throughout the process. If the actual exhaust temperature is detected to exceed the preset safety threshold, the preset alarm or shutdown protection program is triggered.

[0010] Furthermore, the method for calculating the predicted opening based on the feedforward offline data and the target preset flow rate obtained in step two, as described in step three, includes the following steps: Feedforward offline data creation: Collect historical operating data of similar electric control valves for liquid injection. Based on the correspondence between valve forward / return, valve opening degree and flow rate in the historical operating data, calibrate the forward / return opening degree-flow rate characteristic curve model. Then, based on the forward / return opening degree-flow rate characteristic curve model, create an "opening degree-flow rate" correspondence database. Based on the changing trend of the data in the database, construct a forward data array or a return data array, with the forward stroke corresponding to the valve opening process and the return stroke corresponding to the valve closing process. Forward / Return Status Judgment: Compare the current opening of the liquid injection electric regulating valve with the set opening; if the set opening > the current opening, it is determined to be forward; otherwise, it is determined to be return; when the system is started, the set opening is a set value, and after the system is running, the set opening is the target valve opening in step five. Select data array: Based on the forward / return status judgment result, select the corresponding forward data array or return data array from the pre-calibrated "opening degree-flow rate" correspondence database; Find the optimal value by traversing the table: The target preset flow generated in step two is traversed and compared in the selected forward data array or return data array, and the absolute difference between each calculated flow in the selected array and the target preset flow is calculated one by one. Determine the predicted opening: Find the array point that minimizes the absolute difference; the valve opening corresponding to this point is the predicted opening. This opening is the estimated value directly output by the feedforward loop.

[0011] Furthermore, the method for creating an "opening-flow" correspondence database based on the forward / return opening-flow characteristic curve model, and constructing a forward data array or a return data array based on the changing trend of the data in the database, is as follows: Determine the opening degree of a certain valve Based on the changing trend of the valve opening, the forward / return stroke is determined. If the valve opening gradually increases, it is determined to be the forward stroke; if the valve opening gradually decreases, it is determined to be the return stroke. If the electric control valve for liquid injection is a DN25 diameter O-type ball valve with an exhaust pressure of 1MPa, then: For the data set corresponding to the positive stroke, such as valve opening... If the flow rate is ≤37%, the following formula is used to calculate the flow rate. : , Such as valve opening If the flow rate is greater than 37%, the following formula is used to calculate the flow rate. : , Valve opening With the corresponding Form a forward data array; For the data set corresponding to the return stroke, such as valve opening... If the flow rate is ≤30%, the following formula is used to calculate the flow rate. : , Such as valve opening If the flow rate is >30%, the following formula is used to calculate the flow rate. : ; Valve opening With the corresponding This forms the return data array.

[0012] Furthermore, the method for obtaining the current flow rate of the injection circuit in step four is as follows: Hardware measurement mode: If the control system is equipped with a flow meter, the real-time flow rate is directly read from the flow meter as the current flow rate; Soft measurement estimation mode: If the control system is not equipped with a flow meter, the corresponding valve flow characteristic curve model is called according to the type of the liquid injection electric regulating valve. The current opening degree of the liquid injection electric regulating valve is substituted into the valve flow characteristic curve model to calculate the estimated flow rate, which is then used as the current flow rate.

[0013] Furthermore, in the soft measurement estimation mode, when a bypass passage with a throttling orifice is added to the injection circuit, and this bypass passage is connected in parallel with the injection electric regulating valve to ensure that lubricating fluid always enters the air compressor host, the estimated flow rate should be superimposed with the fixed base flow rate corresponding to the throttling orifice.

[0014] Furthermore, the calculation formulas corresponding to the temperature-specific PID algorithm in step two and the incremental PID algorithm in step four are as follows: , Where: t is time, e(t) is the deviation at time t, e(t) = SP - PV, SP is the set value, and PV is the actual value; These represent the deviation at the previous time step and the deviation at the time step before that, respectively; Kp, Ki, and Kd are the proportional coefficient, integral coefficient, and differential coefficient, respectively. Output the increment for the current moment.

[0015] Furthermore, the weighted fusion formula of the weighted algorithm described in step five is as follows: Target valve opening = preset opening α+ Feedback Opening (1-α), Here, α is the weighting coefficient, ranging from 0 to 1. The weighting ratio needs to be determined through actual operation to obtain the optimal weighting ratio.

[0016] The beneficial technical effects of this invention are as follows: 1. At the hardware level, to address the over-spraying problem in existing liquid-injected positive displacement air compressors under variable frequency low-speed conditions caused by switch control or fixed orifice control, a controllable electric liquid-injection regulating valve is added to the liquid injection circuit. Combined with a sensor module and a PLC programmable controller, this enables the system to intelligently and adaptively adjust the liquid injection volume under different operating conditions. This achieves intelligent and precise control of the liquid injection flow rate, dynamically matching the liquid injection flow rate with the compressor status. This reduces energy consumption per unit of air output, improves energy efficiency, and ensures that the compressor does not experience pure liquid compression. Adding an orifice and flow meter to the liquid injection circuit, with the orifice and electric control valve connected in parallel and bypassed, further ensures that the air compressor will not be damaged by high temperatures due to insufficient liquid. Flow feedback from the flow meter, in conjunction with the controller, further improves the accuracy of the electric regulating valve control.

[0017] 2. At the software PLC programming level, a cascade control architecture consisting of an outer temperature loop and an inner flow loop is constructed as the main structure. The outer loop is responsible for precise control, while the inner loop is responsible for rapid response, realizing dynamic matching between the liquid injection flow and the compressor status, and achieving intelligent adaptive adjustment of the liquid injection volume.

[0018] 3. The hardware and software integration introduces a data-driven predictive compensation mechanism. By establishing a model relationship between valve opening and injection flow rate through historical operating data, the system fits the valve opening corresponding to the predicted flow rate by looking up a table and traversing the array, thus achieving feedforward compensation. At the same time, a soft measurement algorithm based on valve characteristic curves is proposed. Based on measured data, a forward and backward flow model of valve flow is constructed. By selecting the corresponding flow characteristic curve for flow estimation based on the current opening and valve direction change, the system replaces hardware sensors, enabling the system to achieve injection flow estimation and closed-loop system under the condition of no flow meter, thereby reducing the hardware cost of closed-loop control. 4. By combining with an industrial controller to achieve online adaptive adjustment, the system can maintain stable operation under different load and speed conditions. This can effectively solve the over-spraying problem of existing technology under low speed conditions of frequency conversion, and at the same time avoid backflow and liquid shortage accidents caused by valve failure, thereby improving the safety and reliability of system operation. Attached Figure Description

[0019] Figure 1 This is a system flowchart of Embodiment 1 of the present invention.

[0020] Figure 2 This is a system flowchart of Embodiment 1 of the present invention.

[0021] Figure 3 This is a system flowchart of Embodiment 1 of the present invention.

[0022] Figure 4This is the control logic diagram in Embodiment 4 of the present invention.

[0023] Figure 5 This is a flowchart of the feedforward offline data creation process in Embodiment 4 of the present invention.

[0024] Figure 6 This is a flowchart of the process for determining the predicted opening degree in Embodiment 4 of the present invention.

[0025] Reference numerals in the attached diagram: 1-Air compressor main unit; 2-Drive motor; 3-Air filter; 4-Intake valve; 5-Liquid-gas separator; 6-Exhaust temperature sensor; 7-Check valve; 8-Safety valve; 9-Exhaust valve; 10-Minimum safety valve; 11-Liquid cooler; 12-Liquid filter; 13-Liquid shut-off valve; 14-Gas-liquid mixture pipeline; 15-Exhaust pipeline; 16-Injection circuit; 17-Injection electric regulating valve; 18-Throttle orifice; 19-Flow meter; 20-Vent pipeline. Detailed Implementation

[0026] Example 1 An intelligent liquid injection control system for air compressors based on artificial intelligence algorithms, such as Figure 1 As shown, the system includes a drive motor 2, an air compressor main unit 1, an air filter 3, an exhaust temperature sensor 6, a liquid-gas separator 5, a liquid cooler 11, a liquid filter 12, a liquid shut-off valve 13, an electric liquid injection regulating valve 17, and a controller. The drive motor 2 is connected to the air compressor main unit 1, which is a screw air compressor main unit 1. The air filter 3 is connected to the air inlet of the air compressor main unit 1 through an intake valve 4. The air outlet of the air compressor main unit 1 is connected to the upper part of the liquid-gas separator 5 through an exhaust one-way valve 7, forming a gas-liquid mixture pipeline 14. The exhaust temperature sensor is installed between the exhaust one-way valve 7 and the liquid-gas separator 5 on the gas-liquid mixture pipeline 14 to monitor the actual exhaust of the air compressor main unit 1. Temperature, the liquid-gas separator 5 is used for gas-liquid separation. After the gas-liquid mixture enters the liquid-gas separator 5, it is separated into air at the top and coolant at the bottom. The top of the liquid-gas separator 5 is connected to a vent pipe 20 and an exhaust pipe 15. A safety valve 8 is installed on the vent pipe 20. A minimum safety valve 10 and an exhaust valve 9 are sequentially installed on the exhaust pipe 15 along the gas discharge direction. The exhaust pipe 15 is used to connect to an external gas supply unit. The liquid outlet at the bottom of the liquid-gas separator 5 is connected to the liquid injection port of the air compressor main unit 1 through a liquid injection circuit 16. The liquid cooler 11, liquid filter 12, liquid shut-off valve 13, and liquid injection electric regulating valve 17 are sequentially arranged on the liquid injection circuit 16 along the liquid flow direction. The liquid injection circuit 16 is used to spray cooling medium into the compression chamber of the air compressor main unit 1.

[0027] The electric regulating valve 17 for liquid injection can be an O-type threaded ball valve, such as Born BN005P, or a V-type ball valve. The V-type ball valve is preferred because its flow characteristics are closer to the ideal flow characteristic curve, which facilitates precise control and improves regulation accuracy.

[0028] The controller is communicatively connected to the electric injection regulating valve 17 and the exhaust temperature sensor 6. Based on the received current opening degree and exhaust temperature, the controller precisely controls the opening degree of the electric injection regulating valve 17 using an intelligent composite control algorithm. The controller can communicate with the electric injection regulating valve 17 and the exhaust temperature sensor 6 via an RS485 communication interface using the Modbus RTU communication protocol.

[0029] The controller includes an outer-loop temperature control module and an inner-loop flow control module, which form a cascaded PID feedback structure. The outer-loop temperature control module outputs a target preset flow rate to the inner-loop flow control module based on the exhaust temperature transmitted by the exhaust temperature sensor 6 using a dedicated temperature PID algorithm. The inner-loop flow control module obtains a predicted valve opening based on feedforward offline data, a feedback valve opening based on an incremental PID algorithm, and calculates the target valve opening based on a weighted fusion of the predicted and feedback valve openings. The output of the outer-loop temperature PID serves as the setting input for the inner-loop incremental PID, forming a cascaded nested structure. The outer loop is responsible for stabilizing the temperature, while the inner loop is responsible for tracking the flow rate.

[0030] This embodiment adds an electric regulating valve 17 to the existing lubrication system to control the amount of lubricating fluid injected. By precisely controlling the opening of the electric regulating valve, the variable frequency liquid-injecting air compressor can intelligently and adaptively adjust the required amount of liquid injected at different speeds. This avoids both excessively low injection volume leading to excessively high temperatures and excessive injection volume causing compressed liquid, thus achieving energy saving and ensuring equipment reliability.

[0031] The intelligent composite control algorithm mainly adopts the technical solution of cascaded PID feedback + internal feedforward compensation + soft measurement. For specific control methods, please refer to the control method section in Example 4.

[0032] Example 2 An intelligent liquid injection control system for an air compressor based on artificial intelligence algorithms, otherwise the same as in Example 1, such as... Figure 2 As shown, the difference lies in that a bypass passage is added to the injection circuit 16. The bypass passage is connected in parallel to both ends of the injection electric regulating valve 17. The bypass passage is equipped with a throttling orifice 18, which is used to provide the basic injection flow rate. Even when the valve opening is small, it can still provide the minimum flow rate, preventing the air compressor main unit 1 from being damaged by high temperature due to lack of liquid cooling. The throttling orifice 18 can be set by means of an orifice plate or a venturi tube.

[0033] Because electric regulating valves have certain malfunctions, such as the electric valve failing to open or close completely, the setting of the throttle orifice 18 ensures that lubricating fluid always enters the air compressor, ensuring that the air compressor will not be damaged by dry compression or dry friction due to lack of liquid cooling under any operating conditions.

[0034] Example 3 An intelligent liquid injection control system for an air compressor based on artificial intelligence algorithms, otherwise the same as in Embodiment 1 or Embodiment 2, such as... Figure 3 As shown, the difference is that a flow meter 19 is also provided on the injection circuit 16. The flow meter 19 is located between the injection electric regulating valve 17 and the air compressor host 1. The flow meter 19 is electrically connected to the controller and is used to collect the flow data of the injection circuit 16 and transmit it to the controller to realize closed-loop precise control of the opening degree of the injection electric regulating valve 17.

[0035] Based on Embodiment 1 or Embodiment 2, a flow meter 19 is added, and flow data is collected through PLC artificial intelligence algorithm to realize closed-loop control of the opening of electric valve, which can further improve the control accuracy and intelligence level of the equipment.

[0036] Example 4 A method for intelligent liquid injection control of an air compressor based on artificial intelligence algorithms, employing the aforementioned intelligent liquid injection control system for an air compressor based on artificial intelligence algorithms, such as... Figure 4 As shown, it includes the following steps: Step 1: Initialization and Signal Acquisition The controller performs a self-test and initialization upon power-on, receives the user-set target exhaust temperature value, and receives the actual exhaust temperature of the air compressor unit 1 collected in real time by the exhaust temperature sensor 6 as the main feedback signal. The target exhaust temperature value can be the temperature most suitable for operating efficiency under the current environment. Generally speaking, the target exhaust temperature setting should not be lower than the ambient temperature +10℃ and should not exceed the ambient temperature +20℃.

[0037] Step 2: Control the outer loop temperature to generate the target flow rate. The deviation between the target exhaust temperature and the actual exhaust temperature is input into the outer loop temperature control module in the controller; the outer loop temperature control module calculates and outputs the target preset flow rate Q using a temperature-specific PID algorithm. set The target preset flow rate Q set This is the amount of cooling medium injected to achieve the target exhaust temperature under the current operating conditions.

[0038] Temperature-dedicated PID (outer loop temperature control): Input: +exhaust temperature setpoint -feedback temperature signal (amplified 10 times to improve calculation resolution); Output = Target set spray flow rate (magnified 10 times); Function: Based on temperature changes and temperature differences, it slowly and steadily provides the ideal flow rate under the current operating conditions.

[0039] The calculation formula for the temperature-specific PID algorithm is as follows: , Where: t is time, e(t) is the deviation at time t, e(t) = SP - PV, SP is the set value, which corresponds to the target value of exhaust temperature here, and PV is the actual value, which is the actual exhaust temperature here; These represent the deviation at the previous time step and the deviation at the time step before that, respectively; Kp, Ki, and Kd are the proportional coefficient, integral coefficient, and differential coefficient, respectively. It outputs the increment for the current moment. However, in practical applications, based on the upper and lower limits and related parameter settings, the final output result is the set spray flow rate that can be directly output to the flow PID.

[0040] Other existing control algorithms, such as fuzzy PID adaptive control, can also be used for outer loop temperature control.

[0041] Step 3: Feedforward control to obtain the predicted opening degree The inner loop flow control module in the controller receives the current opening degree of the electric regulating valve 17 and, based on the feedforward offline data and the target preset flow rate Q obtained in step two,... set The predicted aperture f is calculated. pre .

[0042] When the system is powered on, the initial valve opening can be set to 50%. This is because, during the construction of the forward / reverse mathematical model, it was found that the difference between the fully open flow rate and the 50% opening flow rate is no more than 1 m³ / s. 3 The overall change is small, but the flow rate changes significantly from fully closed to 50% open, and is basically linear. Therefore, 50% is chosen as the initial opening.

[0043] This step is driven by historical data and quickly approximates the target opening to compensate for system lag, such as... Figure 6 As shown, specifically: Feedforward offline data creation: Collect historical operating data of similar electric control valves for liquid injection. Based on the correspondence between valve forward / return, valve opening degree and flow rate in the historical operating data, calibrate the forward / return opening degree-flow rate characteristic curve model. Then, based on the forward / return opening degree-flow rate characteristic curve model, create an "opening degree-flow rate" correspondence database. Based on the changing trend of the data in the database, construct a forward data array or a return data array, with the forward stroke corresponding to the valve opening process and the return stroke corresponding to the valve closing process. Forward / Return State Judgment: Compare the current opening degree of the electric spray regulating valve 17 with the set opening degree; if the set opening degree > the current opening degree, it is determined to be the forward stroke (valve opening process); otherwise, it is determined to be the return stroke (valve closing process); when the system is started, the set opening degree is a set value, which can be 12% of the minimum opening degree, corresponding to a minimum flow rate of 1m³. 3 / h, after the system is running, the set opening degree is the target valve opening degree in step five; Select data array: Based on the forward / return status judgment result, select the corresponding forward data array or return data array from the pre-calibrated "opening degree-flow rate" correspondence database; Traverse and look up the table to find the optimal value: Calculate the target preset flow Q generated in step two. set The selected forward or return data array is traversed and compared, and the flow rate Q of each element in the selected array is calculated one by one. pre With the target preset flow rate Q set The absolute difference between them; Determine the predicted aperture f pre Find the array point that minimizes the absolute difference; the valve opening corresponding to that point is the predicted opening f. pre This opening is the estimated value directly output by the feedforward circuit.

[0044] Because of the hysteresis characteristics of the forward and return strokes, the flow rates corresponding to the forward and return strokes are different at the same opening degree. Therefore, in order to achieve precise control, it is necessary to distinguish between the forward and return strokes.

[0045] Among them, such as Figure 5 As shown, the method for constructing the forward data array or the return data array is as follows: Determine the opening degree of a certain valve Based on the changing trend of the valve opening, the forward / return stroke is determined. If the valve opening gradually increases, it is determined to be the forward stroke; if the valve opening gradually decreases, it is determined to be the return stroke. If the electric control valve for liquid injection is a DN25 diameter O-type ball valve with an exhaust pressure of 1MPa, then: For the data set corresponding to the positive stroke, such as valve opening... If the flow rate is ≤37%, the following formula is used to calculate the flow rate. : , Such as valve opening If the flow rate is greater than 37%, the following formula is used to calculate the flow rate. : , Valve opening With the corresponding Form a forward data array; For the data set corresponding to the return stroke, such as valve opening... If the flow rate is ≤30%, the following formula is used to calculate the flow rate. : , Such as valve opening If the flow rate is >30%, the following formula is used to calculate the flow rate. : ; Valve opening With the corresponding This forms the return data array.

[0046] Step 4: Inner loop flow control, obtaining feedback opening degree This step precisely corrects the flow rate to form a closed-loop or semi-closed-loop control, specifically including the following steps: (1) Obtain the current flow rate of the injection circuit 16. The current traffic can be obtained in one of the following two modes, depending on the actual situation of the system: Hardware measurement mode: If the control system is equipped with flow meter 19, the real-time flow rate is directly read through flow meter 19 as the current flow rate; Soft measurement estimation mode: If the control system is not equipped with a flow meter 19, the corresponding valve flow characteristic curve model is called according to the type of the electric injection regulating valve 17. These valve flow characteristic curve models are pre-stored in the system. The current opening degree of the electric injection regulating valve 17 is substituted into the valve flow characteristic curve model to calculate the estimated flow rate, which is then used as the current flow rate. When a bypass passage with a throttling orifice 18 is added to the injection circuit 16, and this bypass passage is connected in parallel with the electric injection regulating valve 17 to ensure that lubricating fluid always enters the air compressor host 1, the estimated flow rate should be superimposed with the fixed base flow rate corresponding to the throttling orifice 18.

[0047] (2) Inner loop feedback operation The inner loop flow control module calculates and outputs a feedback opening for correction based on the deviation between the target preset flow rate obtained in step two and the current flow rate obtained, using an incremental PID algorithm.

[0048] Incremental PID (Inner Loop Flow Control): Input = + Outer loop output set spray flow rate - Current actual flow rate / Soft measurement flow rate (magnification 10 times); Output = Target valve opening; Function: To quickly and accurately drive an electric regulating valve to track a given target flow rate. The use of incremental PID control avoids abrupt valve movements, resulting in smoother control.

[0049] The calculation formula corresponding to the above incremental PID algorithm is as follows: , Where: t is time, e(t) is the deviation at time t, e(t) = SP - PV, SP is the set value, which is the target preset flow rate here, and PV is the actual value, which is the current flow rate here; These represent the deviation at the previous time step and the deviation at the time step before that, respectively; Kp, Ki, and Kd are the proportional coefficient, integral coefficient, and differential coefficient, respectively. It outputs an increment for the current moment. Also based on the upper and lower limits of the output and parameter settings, the output result can be directly applied to the final electric regulating valve to control the opening degree.

[0050] Step 5: Weighted Fusion and Instruction Output The controller combines the predicted opening (feedforward) obtained in step three with the feedback opening (feedback correction) obtained in step four using a preset weighting algorithm to generate a final target valve opening. The controller then sends the control command corresponding to this target valve opening to the liquid injection electric regulating valve 17 for execution via a communication protocol.

[0051] The weighted fusion formula for the weighted algorithm is: Target valve opening = preset opening α+ Feedback Opening (1-α), Here, α is the weighting coefficient, ranging from 0 to 1, generally not exceeding 0.7; currently, α = 0.25. The weighting ratio needs to be determined through actual operation to obtain the optimal weighting ratio.

[0052] The purpose of controlling the electric control valve is to regulate the injection flow rate. Offline feedback data can provide historical operating data. However, since the feedback model is based on the fitting of offline measured data, it cannot fully adapt to every operating condition. Changes in factors such as temperature and pressure will affect the accuracy of the model. Its response is basically the same as the PID response under low opening conditions. However, when the opening exceeds 50%, the impact of the opening on the flow rate becomes smaller. The mathematical model output opening may result in a situation where a large flow rate is required, but the valve does not operate due to the small difference. Therefore, it is necessary to use PID to calculate the flow rate between the feedback opening and the set flow rate to reduce the situation where changes in the opening affect the flow rate and cause the valve to not operate.

[0053] Step Six: Loop Execution and Status Monitoring Steps two through five are continuously executed in a loop. The amount of liquid injected into the injection circuit 16 is dynamically adjusted via the electric regulating valve 17 until the actual exhaust temperature of the gas-liquid mixture pipeline 14 stabilizes within the preset exhaust temperature target range. The exhaust temperature target range is the temperature range within which the system can operate stably. Simultaneously, the actual exhaust temperature is monitored throughout the process. If the actual exhaust temperature exceeds a preset safety threshold, a preset alarm or shutdown protection procedure is triggered. For safety reasons, the preset safety threshold range is generally larger than the exhaust temperature target range. When the exhaust temperature is within the target range, the exhaust temperature is adjusted by the valve opening to bring it close to the set exhaust temperature target value. When the exhaust temperature reaches the alarm value, a safety warning is issued, with the alarm light flashing and / or an alarm sounding. When the exhaust temperature reaches the safety value, an automatic emergency shutdown is initiated.

[0054] Compared with existing on / off control or fixed orifice control, this invention achieves dynamic matching of injection flow rate and compressor status by adding hardware devices (electric regulating valve and bypass throttling path) and cascade feedback control architecture through an artificial intelligence composite control method; it reduces system lag and improves system response speed through a data-driven predictive compensation mechanism; it realizes injection flow rate estimation control under sensorless conditions by constructing a valve-flow soft measurement algorithm, meeting the basic conditions for closed-loop control and reducing hardware costs; and it improves system stability under low opening conditions by adding a bypass throttling path.

[0055] Through testing and verification of the prototype, the control system maintains a relatively constant exhaust temperature under the pre-set condition of maintaining the same temperature as when the valve is fully open, and reduces the operating power under the same exhaust pressure, effectively reducing the problem of excessive liquid injection and improving the energy-saving performance of the equipment.

[0056] The above detailed description is a specific description of feasible embodiments of the present invention. These embodiments are not intended to limit the patent scope of the present invention. All equivalent implementations or modifications that do not depart from the present invention should be included in the patent scope of this case.

Claims

1. An intelligent liquid injection control system for an air compressor based on an artificial intelligence algorithm, comprising a drive motor, an air compressor main unit, an air filter, an exhaust temperature sensor, a liquid-gas separator, a liquid cooler, a liquid filter, and a liquid shut-off valve. The drive motor is connected to the air compressor main unit. The air filter is connected to the air inlet of the air compressor main unit via an intake valve. The air outlet of the air compressor main unit is connected to the upper part of the liquid-gas separator via an exhaust check valve, forming a gas-liquid mixture pipeline. The exhaust temperature sensor is installed on the gas-liquid mixture pipeline between the exhaust check valve and the liquid-gas separator. The system is used to monitor the actual exhaust temperature of the air compressor main unit. The liquid-gas separation tank is used for gas-liquid separation. A vent pipe and an exhaust pipe are connected to the top of the liquid-gas separation tank. A safety valve is installed on the vent pipe. A minimum safety valve and an exhaust valve are sequentially installed on the exhaust pipe along the gas discharge direction. The exhaust pipe is used to connect to an external gas-using unit. The liquid spray nozzle at the bottom of the liquid-gas separation tank is connected to the liquid spray nozzle of the air compressor main unit through a liquid spray circuit. The liquid cooler, liquid filter, and liquid shut-off valve are sequentially installed on the liquid spray circuit along the liquid flow direction. The system is characterized in that... An electric regulating valve is also installed between the liquid cut-off valve and the liquid injection port of the air compressor main unit in the liquid injection circuit. The electric regulating valve is electrically connected to a controller, which is also electrically connected to an exhaust temperature sensor. The controller includes an outer loop temperature control module and an inner loop flow control module. The outer loop temperature control module and the inner loop flow control module form a cascade PID feedback structure. The outer loop temperature control module is used to output a target preset flow rate to the inner loop flow control module based on the exhaust temperature transmitted by the exhaust temperature sensor using a temperature-specific PID algorithm. The inner loop flow control module is used to obtain a predicted opening degree based on feedforward offline data, obtain a feedback opening degree based on an incremental PID algorithm, and calculate and output the target valve opening degree based on the weighted fusion of the predicted opening degree and the feedback opening degree, so as to achieve precise control of the opening degree of the electric regulating valve.

2. The intelligent liquid injection control system for an air compressor based on an artificial intelligence algorithm according to claim 1, characterized in that, A bypass passage is also added to the liquid injection circuit. The bypass passage is connected in parallel to both ends of the liquid injection electric regulating valve. The bypass passage is provided with a throttling orifice. The throttling orifice is used to provide the basic liquid injection flow rate to prevent the air compressor main unit from being damaged by high temperature due to lack of liquid cooling.

3. The intelligent liquid injection control system for an air compressor based on an artificial intelligence algorithm according to claim 1 or 2, characterized in that, A flow meter is also installed on the injection circuit. The flow meter is located between the electric injection regulating valve and the air compressor host. The flow meter is electrically connected to the controller and is used to collect the flow data of the injection circuit and transmit it to the controller to realize closed-loop precise control of the opening of the electric injection regulating valve.

4. A method for intelligent liquid injection control of an air compressor based on an artificial intelligence algorithm, employing the intelligent liquid injection control system for an air compressor based on an artificial intelligence algorithm as described in claim 1, characterized in that, Includes the following steps: Step 1: Initialization and Signal Acquisition: The controller performs a power-on self-test and initialization, receives the user-set target value for exhaust temperature, and receives the actual exhaust temperature of the air compressor host collected in real time by the exhaust temperature sensor as the main feedback signal; Step 2: Outer loop temperature control to generate target flow: Input the deviation between the target exhaust temperature and the actual exhaust temperature into the outer loop temperature control module in the controller; The outer ring temperature control module calculates and outputs the target preset flow rate through a temperature-specific PID algorithm. The target preset flow rate is the amount of cooling medium injected to achieve the target exhaust temperature under the current operating conditions. Step 3: Feedforward control to obtain predicted opening degree: The inner loop flow control module in the controller receives the current opening degree of the liquid injection electric regulating valve and calculates the predicted opening degree based on the feedforward offline data and the target preset flow rate obtained in Step 2. Step 4: Inner loop flow control, obtaining feedback opening: Obtain the current flow rate of the injection circuit. The inner loop flow control module calculates and outputs a feedback opening for correction based on the deviation between the target preset flow rate obtained in Step 2 and the current flow rate through an incremental PID algorithm. Step 5, Weighted Fusion and Command Output: The controller fuses the predicted opening obtained in Step 3 and the feedback opening obtained in Step 4 using a preset weighted algorithm to generate a final target valve opening, and sends the control command corresponding to this target valve opening to the liquid injection electric regulating valve for execution via a communication protocol. Step Six: Cyclic Execution and Status Monitoring: Steps Two through Five are continuously executed in a cyclical manner. The amount of liquid injected into the injection circuit is dynamically adjusted through the electric regulating valve until the actual exhaust temperature of the gas-liquid mixture pipeline stabilizes within the preset exhaust temperature target range. At the same time, the actual exhaust temperature is monitored throughout the process. If the actual exhaust temperature is detected to exceed the preset safety threshold, the preset alarm or shutdown protection program is triggered.

5. The intelligent liquid injection control method for an air compressor based on an artificial intelligence algorithm according to claim 4, characterized in that, The method for calculating the predicted opening based on the feedforward offline data and the target preset flow rate obtained in step two in the controller described in step three includes the following steps: Feedforward offline data creation: Collect historical operating data of similar electric control valves for liquid injection. Based on the correspondence between valve forward / return, valve opening degree and flow rate in the historical operating data, calibrate the forward / return opening degree-flow rate characteristic curve model. Then, based on the forward / return opening degree-flow rate characteristic curve model, create an "opening degree-flow rate" correspondence database. Based on the changing trend of the data in the database, construct a forward data array or a return data array, with the forward stroke corresponding to the valve opening process and the return stroke corresponding to the valve closing process. Forward / Return Status Judgment: Compare the current opening of the liquid injection electric regulating valve with the set opening; if the set opening > the current opening, it is determined to be forward; otherwise, it is determined to be return; when the system is started, the set opening is a set value, and after the system is running, the set opening is the target valve opening in step five. Select data array: Based on the forward / return status judgment result, select the corresponding forward data array or return data array from the pre-calibrated "opening degree-flow rate" correspondence database; Find the optimal value by traversing the table: The target preset flow generated in step two is traversed and compared in the selected forward data array or return data array, and the absolute difference between each calculated flow in the selected array and the target preset flow is calculated one by one. Determine the predicted opening: Find the array point that minimizes the absolute difference; the valve opening corresponding to that point is the predicted opening.

6. The intelligent liquid injection control method for an air compressor based on an artificial intelligence algorithm according to claim 5, characterized in that, The method for creating an "opening-flow" correspondence database based on the forward / return opening-flow characteristic curve model, and constructing a forward data array or a return data array based on the changing trend of the data in the database, is as follows: Determine the opening degree of a certain valve The forward / return stroke is determined based on the trend of the valve opening. If the valve opening gradually increases, it is judged as the forward stroke; if the valve opening gradually decreases, it is judged as the return stroke. If the electric control valve for liquid injection is a DN25 diameter O-type ball valve with an exhaust pressure of 1MPa, then: For the data set corresponding to the positive stroke, such as valve opening... If the flow rate is ≤37%, the following formula is used to calculate the flow rate. : ， Such as valve opening If the flow rate is greater than 37%, the following formula is used to calculate the flow rate. : ， Valve opening With the corresponding Form a forward data array; For the data set corresponding to the return stroke, such as valve opening... If the flow rate is ≤30%, the following formula is used to calculate the flow rate. : ， Such as valve opening If the flow rate is >30%, the following formula is used to calculate the flow rate. : ； Valve opening With the corresponding This forms the return data array.

7. The intelligent liquid injection control method for an air compressor based on an artificial intelligence algorithm according to claim 4, characterized in that, The method for obtaining the current flow rate of the injection circuit in step four is as follows: Hardware measurement mode: If the control system is equipped with a flow meter, the real-time flow rate is directly read from the flow meter as the current flow rate; Soft measurement estimation mode: If the control system is not equipped with a flow meter, the corresponding valve flow characteristic curve model is called according to the type of the liquid injection electric regulating valve. The current opening degree of the liquid injection electric regulating valve is substituted into the valve flow characteristic curve model to calculate the estimated flow rate, which is then used as the current flow rate.

8. The intelligent liquid injection control method for an air compressor based on an artificial intelligence algorithm according to claim 7, characterized in that, In the soft measurement estimation mode, when a bypass passage with a throttling orifice is added to the injection circuit, and this bypass passage is connected in parallel with the injection electric regulating valve to ensure that lubricating fluid always enters the air compressor host, the estimated flow rate should be superimposed with the fixed base flow rate corresponding to the throttling orifice.

9. The intelligent liquid injection control method for an air compressor based on an artificial intelligence algorithm according to claim 4, characterized in that, The calculation formulas for the temperature-specific PID algorithm described in step two and the incremental PID algorithm described in step four are as follows: ， Where: t is time, e(t) is the deviation at time t, e(t) = SP - PV, SP is the set value, and PV is the actual value; These represent the deviation at the previous time step and the deviation at the time step before that, respectively; Kp, Ki, and Kd are the proportional coefficient, integral coefficient, and differential coefficient, respectively. Output the increment for the current moment.

10. The intelligent liquid injection control method for an air compressor based on an artificial intelligence algorithm according to claim 4, characterized in that, The weighted fusion formula of the weighted algorithm described in step five is: Target valve opening = preset opening α+ Feedback Opening (1-α), Where α is the weighting coefficient, ranging from 0 to 1.

Images