Extruder hydraulic automatic control method based on machine learning

Abstract

The invention discloses an extruder hydraulic automatic control method based on machine learning, and relates to the technical field of machine learning. The method comprises the following steps: collecting training data to establish a damping coefficient prediction model and a friction coefficient prediction model; hydraulic pressure data, hydraulic flow data and component movement data are collected; establishing a pressure-load control model, a flow-speed control model and a displacement-speed control model; collecting an initial pressure value and real-time operation characteristic data; acquiring a real-time friction coefficient, a real-time damping coefficient, a real-time output pressure value, a real-time output flow value, a real-time displacement change value and a real-time speed change value; calculating a real-time pressure compensation value, a real-time flow compensation value, a real-time displacement adjustment value and a real-time speed adjustment value; and pressure compensation and displacement and speed regulation and control are carried out. According to the method, multiple machine learning models are established, and hydraulic pressure is compensated and comprehensively regulated, so that accurate, efficient and intelligent control of the hydraulic system is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of machine learning, and in particular relates to an automatic hydraulic control method for an extruder based on machine learning. Background Art

[0002] The extrusion processing of small batches of brazing filler metals of various varieties places complex and precise requirements on the pressurization process. Due to the different properties of different brazing filler materials, the pressure needs to be specifically adjusted during the extrusion process to obtain stable and high-quality products. The hydraulic control of the extruder refers to the use of the hydraulic system to adjust and control the pressure, flow and movement inside the extruder to achieve precise forming and processing of the material. Although the current technology can perform partial automatic control according to the preset parameters, the parameter setting still needs to rely on experienced operators in the face of a variety of brazing fillers, which also limits the improvement of the working efficiency of the extruder and causes many problems in hydraulic control, such as: delayed response time, high energy consumption in the standby state, heat generated during the working process causes the temperature to rise and affects the performance, and different materials have different requirements for hydraulic parameters during the extrusion process, which makes the hydraulic system less adaptable. Therefore, in order to improve the processing effect of materials, improve production efficiency and product quality, and achieve efficient and precise hydraulic control and solve the above problems, it is necessary to make more automated and intelligent improvements to the hydraulic control of the extruder. Summary of the invention

[0003] The object of the present invention is to provide an automatic hydraulic control method for an extruder based on machine learning, which can be achieved by the following technical solutions: The embodiment of the present application provides an extruder hydraulic automatic control method based on machine learning, comprising the following steps: S1, setting a data acquisition device at multiple sensor locations in the hydraulic control system of the extruder to collect a variety of data; S2, collecting damping coefficient training data and hydraulic friction coefficient training data of the hydraulic control system; S3, establishing a damping coefficient prediction model based on the damping coefficient training data; establishing a friction coefficient prediction model based on the hydraulic friction coefficient training data; S4, collecting hydraulic pressure data, hydraulic flow data and component movement data of the hydraulic control system when performing hydraulic operation; S5, establishing a pressure-load control model based on the hydraulic pressure data; establishing a flow-speed control model based on the hydraulic flow data; and establishing a displacement-speed control model based on the component motion data; S6, collecting the initial pressure value and real-time operation characteristic data of the hydraulic control system; S7, inputting the friction coefficient characteristic data in the real-time operation characteristic data into the friction coefficient prediction model to obtain the real-time friction coefficient, and then inputting the real-time friction coefficient and the damping coefficient characteristic data in the real-time operation characteristic data into the damping coefficient prediction model to obtain the real-time damping coefficient; S8, inputting the real-time operation characteristic data into the pressure-load control model, the flow-speed control model and the displacement-speed control model respectively, so as to obtain the real-time output pressure value, the load size carried, the real-time output flow value, the real-time displacement change value and the real-time speed change value respectively; S9, calculating a real-time pressure compensation value based on the initial pressure value, the real-time output pressure value, the load size carried, and the real-time damping coefficient; calculating a real-time flow compensation value based on the initial pressure value, the real-time output pressure value, the load size carried, and the real-time output flow value; calculating a real-time displacement adjustment value and a real-time speed adjustment value based on the real-time flow compensation value, the real-time displacement change value, and the real-time speed change value; S10, performing pressure compensation on the hydraulic control system based on the real-time pressure compensation value; and regulating the displacement and speed of the hydraulic control system based on the real-time displacement adjustment value and the real-time speed adjustment value.

[0004] Preferably, the method further comprises the following steps: S11, collecting system standby data of the hydraulic control system in a standby state and system operating data in an operating state; S12, establishing a standby energy consumption prediction model based on the system standby data; establishing an operation heat prediction model based on the system operation data; S13, inputting the real-time standby data in the real-time operation characteristic data into the standby energy consumption prediction model to obtain the real-time standby energy consumption; inputting the real-time operation temperature data in the real-time operation characteristic data into the operation heat prediction model to obtain the real-time operation heat; S14, making a rationality judgment on the real-time standby energy consumption and the real-time operating heat and performing energy conversion between the two according to the judgment result. Preferably, step S14 is specifically: If the real-time operating heat exceeds the heat range of the hydraulic control system, the excess will be stored as heat; If the real-time standby energy consumption exceeds the energy consumption range of the hydraulic control system, the energy difference is calculated and the stored excess heat is converted into the energy difference required by the hydraulic control system; Among them, the energy consumption range and heat range are the ranges when the hydraulic control system achieves optimal operating efficiency.

[0005] Preferably, the friction coefficient characteristic data includes the pressure value of the hydraulic oil, the hydraulic oil temperature, the hydraulic oil viscosity, the average roughness of the hydraulic cylinder surface, the average roughness of the piston surface and the oil film thickness; the damping coefficient characteristic data includes the hydraulic oil viscosity, the cross-sectional area of ​​the hydraulic cylinder, the length of the hydraulic cylinder, the hydraulic friction coefficient, the ambient temperature and the ambient humidity. Preferably, the hydraulic pressure data includes: pump outlet pressure, cylinder pressure and control valve pressure; the hydraulic flow data includes: liquid flow, pump flow, cylinder inlet and outlet flow and leakage flow; the component motion data includes hydraulic cylinder displacement, movement speed of the hydraulic cylinder piston and acceleration change during the movement of the hydraulic cylinder.

[0006] Preferably, establishing the damping coefficient prediction model specifically includes: Feature extraction: performing feature extraction on the damping coefficient training data to obtain multiple sets of damping coefficient feature data; Data combination: combining multiple sets of damping coefficient characteristic data in the form of the first characteristic vector; Model setting: the set of all first eigenvectors is used as the input of the damping coefficient prediction model; the damping coefficient value predicted for each set of damping coefficient characteristic data is used as the output; the actual damping coefficient value corresponding to each set of damping coefficient characteristic data is used as the prediction target; the sum of the first prediction accuracies of all predicted damping coefficient values ​​is minimized as the training target; Model training: The damping coefficient prediction model is trained based on the model settings until the sum of the first prediction accuracy reaches convergence and the training is stopped; Output the trained damping coefficient prediction model. Preferably, establishing the friction coefficient prediction model specifically includes: Feature extraction: performing feature extraction on the hydraulic friction coefficient training data to obtain multiple sets of friction coefficient feature data; Data combination: combining multiple sets of friction coefficient characteristic data in the form of a second characteristic vector; Model setting: The set of all second eigenvectors is used as the input of the friction coefficient prediction model; the friction coefficient value predicted for each set of friction coefficient feature data is used as the output; the actual friction coefficient value corresponding to each set of friction coefficient feature data is used as the prediction target; the sum of the second prediction accuracies of all predicted friction coefficient values ​​is minimized as the training target; Model training: The friction coefficient prediction model is trained based on the model settings until the sum of the second prediction accuracy reaches convergence and the training is stopped; Output the trained friction coefficient prediction model.

[0007] Preferably, a pressure-load control model is established, including the following contents: Feature extraction: performing feature extraction on the hydraulic pressure data to obtain multiple sets of hydraulic pressure feature data; Data combination: combining multiple sets of hydraulic pressure characteristic data in the form of a third characteristic vector; Model setting: The set of all third eigenvectors is used as the input of the pressure-load control model; the pressure value and load size predicted for each set of hydraulic pressure characteristic data are used as the output; the actual pressure value and the actual load size corresponding to each set of hydraulic pressure characteristic data are used as the prediction target; the sum of the third prediction accuracies of minimizing all predicted pressure values ​​and load sizes is used as the training target; Model training: The pressure-load control model is trained based on the model settings until the sum of the third prediction accuracy reaches convergence; Output the trained pressure-load control model. Preferably, the damping coefficient prediction model is any one of a deep neural network model or a deep belief network model; The friction coefficient prediction model is any one of a deep neural network model or a deep belief network model; The pressure-load control model is any one of a deep neural network model or a deep belief network model; The flow-speed control model is any one of a deep neural network model or a deep belief network model; The displacement-velocity control model is any one of a deep neural network model and a deep belief network model.

[0008] Preferably, the method further comprises the following steps: S15, when the hydraulic control system stops running, various control data and operation data in the hydraulic control process are recorded, and the hydraulic historical data set is updated accordingly; according to the set time interval or data update number, data is again selected from the updated hydraulic historical data set as training data for various models. The beneficial effects of the present invention are: (1) The present invention establishes multiple machine learning models to obtain the hydraulic values ​​that need to be adjusted in the hydraulic control system. At the same time, the pressure, hydraulic oil flow, displacement and speed of the moving parts of the hydraulic control system are compensated and comprehensively regulated to achieve accurate, efficient and intelligent control of the hydraulic system, thereby achieving the best processing effect on the material and improving production efficiency and product quality.

[0009] (2) The present invention first sets data acquisition devices at multiple sensor locations within the hydraulic control system of the extruder to collect multiple data; such as pressure sensors, temperature sensors and displacement sensors; then collects damping coefficient training data and hydraulic friction coefficient training data of the hydraulic control system; then establishes a damping coefficient prediction model based on the damping coefficient training data, and establishes a friction coefficient prediction model based on the hydraulic friction coefficient training data; then collects hydraulic pressure data, hydraulic flow data and component movement data of the hydraulic control system during hydraulic operation; and establishes a pressure-load control model based on the hydraulic pressure data; establishes a flow-speed control model based on the hydraulic flow data; and establishes a displacement-speed control model based on the component movement data; further collects the initial pressure value and real-time operation characteristic data of the hydraulic control system; inputs the friction coefficient characteristic data in the real-time operation characteristic data into the friction coefficient prediction model to obtain the real-time friction coefficient, and then The friction coefficient and the damping coefficient characteristic data in the real-time operation characteristic data are input into the damping coefficient prediction model to obtain the real-time damping coefficient; the real-time operation characteristic data are then respectively input into the pressure-load control model, the flow-speed control model and the displacement-speed control model to respectively obtain the real-time output pressure value, the load size carried, the real-time output flow value, the real-time displacement change value and the real-time speed change value; then the real-time pressure compensation value is calculated based on the initial pressure value, the real-time output pressure value, the load size carried and the real-time damping coefficient; the real-time flow compensation value is calculated based on the initial pressure value, the real-time output pressure value, the load size carried and the real-time output flow value; the real-time displacement adjustment value and the real-time speed adjustment value are calculated based on the real-time flow compensation value, the real-time displacement change value and the real-time speed change value; finally, the hydraulic control system is pressure compensated based on the real-time pressure compensation value; the hydraulic control system is displacement and speed regulated based on the real-time displacement adjustment value and the real-time speed adjustment value. The above content realizes accurate, efficient and intelligent control of the hydraulic system. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] For better understanding and implementation, the technical solution of the present application is described in detail below with reference to the accompanying drawings.

[0011] Figure 1 A flowchart of a method for hydraulic automatic control of an extruder based on machine learning provided in an embodiment of the present application; Figure 2 A flowchart of the steps for establishing a damping coefficient prediction model provided in an embodiment of the present application; Figure 3 A flowchart of the steps for establishing a friction coefficient prediction model provided in an embodiment of the present application; Figure 4A flow chart of the steps for establishing a pressure-load control model provided in an embodiment of the present application. DETAILED DESCRIPTION

[0012] In order to further explain the technical means and effects taken by the present invention to achieve the predetermined invention purpose, exemplary embodiments will be described in detail here, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are only examples of methods and systems consistent with some aspects of the present application as detailed in the attached claims.

[0013] The terms used in this application are only for the purpose of describing specific embodiments and are not intended to limit this application. The singular forms of "a", "said" and "the" used in this application and the appended claims are also intended to include plural forms unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used in this article refers to any or all possible combinations of one or more associated listed items.

[0014] The specific implementation methods, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0015] See also Figure 1 The present application embodiment provides an extruder hydraulic automatic control method based on machine learning, comprising the following steps: S1, a data acquisition device is set at multiple sensor locations in the hydraulic control system of the extruder to collect various data; such as pressure sensors, temperature sensors, displacement sensors, etc.; S2, collecting damping coefficient training data and hydraulic friction coefficient training data of the hydraulic control system; S3, establishing a damping coefficient prediction model based on the damping coefficient training data; establishing a friction coefficient prediction model based on the hydraulic friction coefficient training data; S4, collecting hydraulic pressure data, hydraulic flow data and component movement data of the hydraulic control system when performing hydraulic operation; S5, establishing a pressure-load control model based on the hydraulic pressure data; establishing a flow-speed control model based on the hydraulic flow data; and establishing a displacement-speed control model based on the component motion data; S6, collecting the initial pressure value and real-time operation characteristic data of the hydraulic control system; S7, inputting the friction coefficient characteristic data in the real-time operation characteristic data into the friction coefficient prediction model to obtain the real-time friction coefficient, and then inputting the real-time friction coefficient and the damping coefficient characteristic data in the real-time operation characteristic data into the damping coefficient prediction model to obtain the real-time damping coefficient; S8, inputting the real-time operation characteristic data into the pressure-load control model, the flow-speed control model and the displacement-speed control model respectively, so as to obtain the real-time output pressure value, the load size carried, the real-time output flow value, the real-time displacement change value and the real-time speed change value respectively; S9, calculating a real-time pressure compensation value based on the initial pressure value, the real-time output pressure value, the load size carried, and the real-time damping coefficient; calculating a real-time flow compensation value based on the initial pressure value, the real-time output pressure value, the load size carried, and the real-time output flow value; calculating a real-time displacement adjustment value and a real-time speed adjustment value based on the real-time flow compensation value, the real-time displacement change value, and the real-time speed change value; S10, performing pressure compensation on the hydraulic control system based on the real-time pressure compensation value; and regulating the displacement and speed of the hydraulic control system based on the real-time displacement adjustment value and the real-time speed adjustment value.

[0016] Specifically, since the hydraulic control system of the existing extruder has problems such as delayed response time, high energy consumption in standby state, heat generated during operation causing temperature increase affecting performance, and different materials having different requirements for hydraulic parameters during the extrusion process resulting in insufficient adaptability of the hydraulic system, the present application addresses the above-mentioned problems by establishing a variety of machine learning models to derive the hydraulic values ​​that need to be adjusted in the hydraulic control system; at the same time, the hydraulic oil flow rate and the displacement and speed of the control components in the system are also adjusted to achieve the optimal energy consumption state, thereby achieving optimal control of the hydraulic press. The specific contents of the present application are as follows: first, a data acquisition device is set at multiple sensor locations in the hydraulic control system of the extruder to collect various data; such as pressure sensors, temperature sensors and displacement sensors; then the damping coefficient training data and hydraulic friction coefficient training data of the hydraulic control system are collected; then a damping coefficient prediction model is established based on the damping coefficient training data, and a friction coefficient prediction model is established based on the hydraulic friction coefficient training data; then the hydraulic pressure data, hydraulic flow data and component movement data of the hydraulic control system during hydraulic operation are collected; and a pressure-load control model is established based on the hydraulic pressure data; a flow-speed control model is established based on the hydraulic flow data; a displacement-speed control model is established based on the component movement data; furthermore, the initial pressure value and real-time operation characteristic data of the hydraulic control system are collected; the friction coefficient characteristic data in the real-time operation characteristic data is input into the friction coefficient prediction model to obtain the real-time friction coefficient, and then The real-time friction coefficient and the damping coefficient characteristic data in the real-time operation characteristic data are input into the damping coefficient prediction model to obtain the real-time damping coefficient; the real-time operation characteristic data are then respectively input into the pressure-load control model, the flow-speed control model and the displacement-speed control model to respectively obtain the real-time output pressure value, the load size, the real-time output flow value, the real-time displacement change value and the real-time speed change value; then the real-time pressure compensation value is calculated based on the initial pressure value, the real-time output pressure value, the load size and the real-time damping coefficient; the real-time flow compensation value is calculated based on the initial pressure value, the real-time output pressure value, the load size and the real-time output flow value; the real-time displacement adjustment value and the real-time speed adjustment value are calculated based on the real-time flow compensation value, the real-time displacement change value and the real-time speed change value; finally, the hydraulic control system is pressure compensated based on the real-time pressure compensation value; the displacement and speed of the hydraulic control system are regulated based on the real-time displacement adjustment value and the real-time speed adjustment value.

[0017] Through the above contents, this application compensates and comprehensively regulates the pressure, hydraulic oil flow, displacement and speed of moving parts (such as pistons or clamps) of the hydraulic control system, so as to achieve precise, efficient and intelligent control of the hydraulic system, thereby achieving the best processing effect on the material and improving production efficiency and product quality.

[0018] In general, this application mainly focuses on the pressure control, flow control and motion control of the hydraulic control system, specifically: Pressure control: By adjusting the pressure in the hydraulic control system, ensure that the extruder maintains the required pressure under different process and load conditions to avoid overloading or underloading.

[0019] Flow Control: Control the flow of hydraulic oil to adjust the extrusion speed, ensuring that the material is processed at an appropriate speed during the extrusion process.

[0020] Motion Control: Control the displacement and speed of the extruder’s moving parts, such as pistons or clamps, to achieve precise molding and positioning.

[0021] In an embodiment provided in the present application, the hydraulic automatic control method of the extruder further includes the following steps: S11, collecting system standby data of the hydraulic control system in a standby state and system operating data in an operating state; S12, establishing a standby energy consumption prediction model based on the system standby data; establishing an operation heat prediction model based on the system operation data; S13, inputting the real-time standby data in the real-time operation characteristic data into the standby energy consumption prediction model to obtain the real-time standby energy consumption; inputting the real-time operation temperature data in the real-time operation characteristic data into the operation heat prediction model to obtain the real-time operation heat; S14, judging the rationality of the real-time standby energy consumption and the real-time operating heat and performing energy conversion between the two according to the judgment result, specifically: If the real-time operating heat exceeds the heat range of the hydraulic control system, the excess will be stored as heat; If the real-time standby energy consumption exceeds the energy consumption range of the hydraulic control system, the energy difference is calculated and the stored excess heat is converted into the energy difference required by the hydraulic control system; Among them, the energy consumption range and heat range are the ranges when the hydraulic control system achieves optimal operating efficiency.

[0022] Specifically, because traditional hydraulic systems are usually inefficient, especially in standby mode, they may consume too much energy, resulting in unnecessary energy consumption and increased costs. Hydraulic oil generates heat during operation. If the temperature is not properly controlled, it may cause the viscosity of the hydraulic oil to change, thereby affecting the performance of the system. Therefore, in order to solve the above problems, this embodiment adopts a method of storing energy and converting energy at the appropriate time to eliminate the problems caused by energy consumption and heat, thereby improving the comprehensive performance of the hydraulic control system.

[0023] In an embodiment provided in the present application, the friction coefficient characteristic data includes the pressure value of the hydraulic oil, the hydraulic oil temperature, the hydraulic oil viscosity, the average roughness of the hydraulic cylinder surface, the average roughness of the piston surface and the oil film thickness; the damping coefficient characteristic data includes the hydraulic oil viscosity, the cross-sectional area of ​​the hydraulic cylinder, the length of the hydraulic cylinder, the hydraulic friction coefficient, the ambient temperature and the ambient humidity.

[0024] In an embodiment provided in the present application, the hydraulic pressure data includes: pump outlet pressure, cylinder pressure and control valve pressure; the hydraulic flow data includes: liquid flow, pump flow, cylinder inlet and outlet flow and leakage flow; the component motion data includes hydraulic cylinder displacement, movement speed of the hydraulic cylinder piston and acceleration change during the movement of the hydraulic cylinder.

[0025] Specifically, when the hydraulic control system performs hydraulic operation, hydraulic pressure data, hydraulic flow data and component movement data are important monitoring and control parameters, including the following: Hydraulic pressure data: Pump outlet pressure: the hydraulic pressure at the pump output end, reflecting the performance and working status of the pump; cylinder pressure: the pressure inside the hydraulic cylinder, indicating the working status inside the cylinder, usually used to judge the load conditions; control valve pressure: the pressure on both sides of each control valve (such as proportional valve, directional valve, etc.), helps to understand the working status of the valve and the flow direction of the fluid.

[0026] Hydraulic flow data: Liquid flow: the liquid flow flowing through various parts of the hydraulic system, usually monitored by a flow meter; Pump flow: the output flow of the pump, reflecting the oil supply capacity of the pump; Cylinder inlet and outlet flow: the flow at the oil inlet and return port of the hydraulic cylinder, to determine the movement speed and operating efficiency of the cylinder; Leakage flow: the flow loss caused by leakage in the system, used to evaluate the sealing of the system.

[0027] Component motion data: Hydraulic cylinder displacement: The displacement of the hydraulic cylinder piston, usually measured by a displacement sensor or encoder, reflects the actual working position of the cylinder; Speed: The movement speed of the hydraulic cylinder piston, usually obtained by calculating the displacement change rate, indicating work efficiency and responsiveness; Acceleration: The acceleration change during the movement of the hydraulic cylinder, used to evaluate the dynamic performance of the system.

[0028] By monitoring and analyzing these data, the performance of the hydraulic system can be optimized, work efficiency can be improved, and safe operation can be ensured.

[0029] like Figure 2 As shown, in one embodiment provided in the present application, establishing the damping coefficient prediction model specifically includes: Feature extraction: performing feature extraction on the damping coefficient training data to obtain multiple sets of damping coefficient feature data; Data combination: combining multiple sets of damping coefficient characteristic data in the form of the first characteristic vector; Model setting: the set of all first eigenvectors is used as the input of the damping coefficient prediction model; the damping coefficient value predicted for each set of damping coefficient characteristic data is used as the output; the actual damping coefficient value corresponding to each set of damping coefficient characteristic data is used as the prediction target; the sum of the first prediction accuracies of all predicted damping coefficient values ​​is minimized as the training target; Model training: The damping coefficient prediction model is trained based on the model settings until the sum of the first prediction accuracy reaches convergence and the training is stopped; Output the trained damping coefficient prediction model.

[0030] like Figure 3 As shown, in one embodiment provided in the present application, the friction coefficient prediction model is established, specifically including: Feature extraction: performing feature extraction on the hydraulic friction coefficient training data to obtain multiple sets of friction coefficient feature data; Data combination: combining multiple sets of friction coefficient characteristic data in the form of a second characteristic vector; Model setting: The set of all second eigenvectors is used as the input of the friction coefficient prediction model; the friction coefficient value predicted for each set of friction coefficient feature data is used as the output; the actual friction coefficient value corresponding to each set of friction coefficient feature data is used as the prediction target; the sum of the second prediction accuracies of all predicted friction coefficient values ​​is minimized as the training target; Model training: The friction coefficient prediction model is trained based on the model settings until the sum of the second prediction accuracy reaches convergence and the training is stopped; Output the trained friction coefficient prediction model.

[0031] like Figure 4 As shown, in one embodiment provided in the present application, a pressure-load control model is established, including the following contents: Feature extraction: performing feature extraction on the hydraulic pressure data to obtain multiple sets of hydraulic pressure feature data; Data combination: combining multiple sets of hydraulic pressure characteristic data in the form of a third characteristic vector; Model setting: The set of all third eigenvectors is used as the input of the pressure-load control model; the pressure value and load size predicted for each set of hydraulic pressure characteristic data are used as the output; the actual pressure value and the actual load size corresponding to each set of hydraulic pressure characteristic data are used as the prediction target; the sum of the third prediction accuracies of minimizing all predicted pressure values ​​and load sizes is used as the training target; Model training: The pressure-load control model is trained based on the model settings until the sum of the third prediction accuracy reaches convergence; Output the trained pressure-load control model.

[0032] In one embodiment provided in the present application, the damping coefficient prediction model is any one of a deep neural network model or a deep belief network model; The friction coefficient prediction model is any one of a deep neural network model or a deep belief network model; The pressure-load control model is any one of a deep neural network model or a deep belief network model; The flow-speed control model is any one of a deep neural network model or a deep belief network model; The displacement-velocity control model is any one of a deep neural network model and a deep belief network model.

[0033] In an embodiment provided in the present application, the following steps are also included: S15, when the hydraulic control system stops running, various control data and operation data in the hydraulic control process are recorded, and the hydraulic historical data set is updated accordingly; according to the set time interval or data update number, data is again selected from the updated hydraulic historical data set as training data for various models.

[0034] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0035] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A hydraulic automatic control method for an extruder based on machine learning, characterized in that: The steps include: S1, setting a data acquisition device at multiple sensor locations in the hydraulic control system of the extruder to collect a variety of data; S2, collecting damping coefficient training data and hydraulic friction coefficient training data of the hydraulic control system; S3, establishing a damping coefficient prediction model based on the damping coefficient training data; establishing a friction coefficient prediction model based on the hydraulic friction coefficient training data; S4, collecting hydraulic pressure data, hydraulic flow data and component movement data of the hydraulic control system when performing hydraulic operation; S5, establishing a pressure-load control model based on the hydraulic pressure data; establishing a flow-speed control model based on the hydraulic flow data; and establishing a displacement-speed control model based on the component motion data; S6, collecting the initial pressure value and real-time operation characteristic data of the hydraulic control system; S7, inputting the friction coefficient characteristic data in the real-time operation characteristic data into the friction coefficient prediction model to obtain the real-time friction coefficient, and then inputting the real-time friction coefficient and the damping coefficient characteristic data in the real-time operation characteristic data into the damping coefficient prediction model to obtain the real-time damping coefficient; S8, inputting the real-time operation characteristic data into the pressure-load control model, the flow-speed control model and the displacement-speed control model respectively, so as to obtain the real-time output pressure value, the load size carried, the real-time output flow value, the real-time displacement change value and the real-time speed change value respectively; S9, calculating a real-time pressure compensation value based on the initial pressure value, the real-time output pressure value, the load carried, and the real-time damping coefficient; Calculate the real-time flow compensation value based on the initial pressure value, the real-time output pressure value, the load size carried, and the real-time output flow value; Calculate a real-time displacement adjustment value and a real-time speed adjustment value based on the real-time flow compensation value, the real-time displacement change value, and the real-time speed change value; S10, performing pressure compensation on the hydraulic control system based on the real-time pressure compensation value; and regulating displacement and speed of the hydraulic control system based on the real-time displacement adjustment value and the real-time speed adjustment value.

2. The method for automatic hydraulic control of an extruder based on machine learning according to claim 1, characterized in that: The following steps are also included: S11, collecting system standby data of the hydraulic control system in a standby state and system operating data in an operating state; S12, establishing a standby energy consumption prediction model based on the system standby data; establishing an operation heat prediction model based on the system operation data; S13, inputting the real-time standby data in the real-time operation characteristic data into the standby energy consumption prediction model to obtain the real-time standby energy consumption; inputting the real-time operation temperature data in the real-time operation characteristic data into the operation heat prediction model to obtain the real-time operation heat; S14, making a rationality judgment on the real-time standby energy consumption and the real-time operating heat and performing energy conversion between the two according to the judgment result.

3. The method for automatic hydraulic control of an extruder based on machine learning according to claim 2, characterized in that: Step S14 is specifically as follows: If the real-time operating heat exceeds the heat range of the hydraulic control system, the excess will be stored as heat; If the real-time standby energy consumption exceeds the energy consumption range of the hydraulic control system, the energy difference is calculated and the stored excess heat is converted into the energy difference required by the hydraulic control system; Among them, the energy consumption range and heat range are the ranges when the hydraulic control system achieves optimal operating efficiency.

4. The method for automatic hydraulic control of an extruder based on machine learning according to claim 1, characterized in that: The friction coefficient characteristic data include the pressure value of the hydraulic oil, the hydraulic oil temperature, the hydraulic oil viscosity, the average roughness of the hydraulic cylinder surface, the average roughness of the piston surface and the oil film thickness; the damping coefficient characteristic data include the hydraulic oil viscosity, the cross-sectional area of ​​the hydraulic cylinder, the length of the hydraulic cylinder, the hydraulic friction coefficient, the ambient temperature and the ambient humidity.

5. The method for automatic hydraulic control of an extruder based on machine learning according to claim 1, characterized in that: The hydraulic pressure data includes: pump outlet pressure, cylinder pressure and control valve pressure; the hydraulic flow data includes: liquid flow, pump flow, cylinder inlet and outlet flow and leakage flow; the component movement data includes hydraulic cylinder displacement, movement speed of hydraulic cylinder piston and acceleration change during hydraulic cylinder movement.

6. The method for automatic hydraulic control of an extruder based on machine learning according to claim 1, characterized in that: Establish a damping coefficient prediction model, including: Feature extraction: performing feature extraction on the damping coefficient training data to obtain multiple sets of damping coefficient feature data; Data combination: combining multiple sets of damping coefficient characteristic data in the form of the first characteristic vector; Model setting: the set of all first eigenvectors is used as the input of the damping coefficient prediction model; the damping coefficient value predicted for each set of damping coefficient characteristic data is used as the output; the actual damping coefficient value corresponding to each set of damping coefficient characteristic data is used as the prediction target; the sum of the first prediction accuracies of all predicted damping coefficient values ​​is minimized as the training target; Model training: The damping coefficient prediction model is trained based on the model settings until the sum of the first prediction accuracy reaches convergence and the training is stopped; Output the trained damping coefficient prediction model.

7. The method for automatic hydraulic control of an extruder based on machine learning according to claim 1, characterized in that: Establish a friction coefficient prediction model, including: Feature extraction: performing feature extraction on the hydraulic friction coefficient training data to obtain multiple sets of friction coefficient feature data; Data combination: combining multiple sets of friction coefficient characteristic data in the form of a second characteristic vector; Model setting: The set of all second eigenvectors is used as the input of the friction coefficient prediction model; the friction coefficient value predicted for each set of friction coefficient feature data is used as the output; the actual friction coefficient value corresponding to each set of friction coefficient feature data is used as the prediction target; the sum of the second prediction accuracies of all predicted friction coefficient values ​​is minimized as the training target; Model training: The friction coefficient prediction model is trained based on the model settings until the sum of the second prediction accuracy reaches convergence and the training is stopped; Output the trained friction coefficient prediction model.

8. The method for automatic hydraulic control of an extruder based on machine learning according to claim 1, characterized in that: Establish a pressure-load control model, including the following: Feature extraction: performing feature extraction on the hydraulic pressure data to obtain multiple sets of hydraulic pressure feature data; Data combination: combining multiple sets of hydraulic pressure characteristic data in the form of a third characteristic vector; Model setting: The set of all third eigenvectors is used as the input of the pressure-load control model; the pressure value and load size predicted for each set of hydraulic pressure characteristic data are used as the output; the actual pressure value and the actual load size corresponding to each set of hydraulic pressure characteristic data are used as the prediction target; the sum of the third prediction accuracies of all predicted pressure values ​​and load sizes is minimized as the training target; Model training: The pressure-load control model is trained based on the model settings until the sum of the third prediction accuracy reaches convergence; Output the trained pressure-load control model.

9. The method for automatic hydraulic control of an extruder based on machine learning according to claim 1, characterized in that: The damping coefficient prediction model is any one of a deep neural network model or a deep belief network model; The friction coefficient prediction model is any one of a deep neural network model or a deep belief network model; The pressure-load control model is any one of a deep neural network model or a deep belief network model; The flow-speed control model is any one of a deep neural network model or a deep belief network model; The displacement-velocity control model is any one of a deep neural network model and a deep belief network model.

10. The method for automatic hydraulic control of an extruder based on machine learning according to claim 1, characterized in that: The following steps are also included: S15, when the hydraulic control system stops running, various control data and operation data in the hydraulic control process are recorded, and the hydraulic historical data set is updated accordingly; according to the set time interval or data update number, data is again selected from the updated hydraulic historical data set as training data for various models.

Images