An intelligent control system for material feeding of array rectifying column

Abstract

The application discloses an array type rectifying column material feeding intelligent control system, which is characterized by comprising a plunger pump, a rocker arm, an extension arm, an embedded system, a stepping motor and a syringe needle. A first stepping motor is arranged on the upper end of the side surface of the rocker arm, a second stepping motor is arranged on the lower end of the side surface of the rocker arm, and a third stepping motor is arranged on the front end of the extension arm. The extension arm is arranged above the rocker arm. The syringe needle head is fixed to the end of the extension arm, and an industrial CCD camera is fixed above the syringe needle head, and both are vertically downward. The embedded system sends out pulse control to the stepping motor and the high-precision plunger pump. The syringe pump is connected with the syringe needle through a plastic connecting pipe, and the syringe pump realizes material filling by pulse control of the internal motor to extract and inject the material. The array type rectifying column material filling system based on the embedded system is used to realize rapid and accurate filling of the material.

Description

TECHNICAL FIELD

[0001] The application is based on the application field of automatic control technology of array type intelligent integrated system of rectification unit, and designs an array type rectification tower material feeding intelligent control system, which can realize accurate material filling for the array type rectification tower. BACKGROUND

[0002] In the prior art, the material filling of the array type rectification tower is usually carried out in a manual or semi-automatic manner. The manual filling method is inefficient, prone to errors, and difficult to achieve precise control. Although the semi-automatic method partially realizes automatic filling through mechanical equipment, it still relies on manual operation, especially in the material conveying, positioning and injection process, the precision and real-time performance are poor. In addition, the existing semi-automatic filling system is difficult to adapt to the dynamic change of material demand, and cannot realize accurate regulation and control of the filling amount, position and speed, which is easy to cause material waste or uneven filling problems.

[0003] One of the difficulties in the array type rectification tower material filling scheme is how to control the speed of material pipetting. CN113295469A discloses a method for quickly and accurately positioning a pipetting needle, which gives an acceleration distribution method. Compared with CN113295469A, the present application has better performance in speed control and filling accuracy, especially in the dynamic adjustment process, the real-time detection and feedback system of the present application can ensure more stable material filling effect. SUMMARY

[0004] Based on the microchemical technology in the background art, the application proposes an array type rectification tower material feeding intelligent control system, which includes an automatic rocker, a first stepper motor arranged at the upper end of the side of the automatic rocker, a second stepper motor arranged at the lower end of the automatic rocker, a third stepper motor arranged at the front end of the telescopic arm, a needle fixed at the end of the telescopic arm, and the needle head is vertically downward. The plastic connecting pipe of the plunger pump passes through the automatic rocker and connects the needle. A camera is arranged directly above the array type rectification tower.

[0005] The control system controls the first stepper motor, the second stepper motor and the third stepper motor through pulse control to realize the rotation, up and down movement of the automatic rocker, and the extension and contraction of the telescopic arm. The stepper motor inside the injection pump controls the extraction and filling of the material based on pulse control. The initial position of the automatic rocker is positioned directly above the storage tank. An industrial CCD camera acquires a top view of the array type rectification tower and returns to the control system through TCP communication. The detection algorithm acquires accurate positioning of the rectification tower, and then controls the three stepper motors through pulse control to realize accurate filling of the material.

[0006] The industrial CCD camera shoots an array rectifying tower picture, and the position information of the array rectifying tower is fed back to the control system through RetinaNet detection positioning plate to realize position feedback.

[0007] The position information of the array rectifying tower is obtained through the following steps:

[0008] S1-1, two methods of manual shooting based on unmanned laboratory real scene are used by Python to collect data set images of target objects;

[0009] S1-2, according to the array rectifying tower packing condition in the actual production scene, the area that can be packed on the data set target to be grabbed is labeled.

[0010] S1-3, common experimental equipment in unmanned factory is shot. In order to save efficiency, LabelImg is used to complete the labeling work, and Yolo format labeling is carried out on the data set.

[0011] S1-4, the real-time performance of the mobile manipulator packing system is good, which can meet the characteristics of dynamic packing, and various recognition detection algorithms are tried, and the optimal algorithm RetinaNet suitable for the application of the application is selected through the comparison of accuracy and processing frame rate;

[0012] S1-5, the RGB image preprocessed by the RetinaNet target detection algorithm is analyzed, the array rectifying tower is recognized and positioned, and the inherent objects in the scene that should not be operated are excluded, and the boundary box information is output. The best injection packing position is obtained through the output of the frame information.

[0013] The best injection packing position is obtained through the following formula:

[0014]

[0015] Wherein, x i is the coordinate of each pixel in the image in the x direction, p i is the corresponding pixel value, and the optimal position in the x direction is x. Similarly, the y coordinate can be obtained.

[0016] The mechanical arm control system comprises the following steps:

[0017] (1) When taking materials, the air column is first extracted, and the required pulse number m1 is a given value;

[0018] (2) When taking materials, the pulse number m2 required for taking materials is determined by the following formula:

[0019]

[0020] Wherein VM V is the sample volume, M is the number of pulses required for the full range of the plunger pump, V c V is the volume of the plunger pump;

[0021] (3) The step motor sampling is controlled by using the phased variable speed control method, which is divided into acceleration phase and deceleration phase, and the acceleration a1 and speed v1 of the acceleration phase are determined by the following formula:

[0022]

[0023]

[0024] Wherein, t1 is the cutoff time when the acceleration in the acceleration phase increases to the maximum value, t2 is the cutoff time of the acceleration phase, v z is the speed value when the acceleration reaches the maximum value, t is the time change, t1, t2, v z are all given values;

[0025] (4) The acceleration a2 and speed v2 of the deceleration phase in the sampling process are determined by the following formula:

[0026]

[0027]

[0028] Wherein, t1 is the cutoff time when the acceleration in the acceleration phase increases to the maximum value, t3 is the cutoff time when the acceleration in the deceleration phase increases to the maximum value, t4 is the cutoff time of the deceleration phase, v z is the speed value when the acceleration reaches the maximum value, t is the time change, t1, t3, t4, v z are all given values

[0029] Advantages of the present application

[0030] The technology has significant innovation and universality, and the present application has better performance in speed control and filling precision, especially in the dynamic adjustment process, the real-time detection and feedback system of the present application can ensure more stable material filling effect. By providing comprehensive intelligent equipment and technical services, it has important influence on the resources, environment and health problems of chemical industry, plays an active role in promoting the development and progress of society, and the social benefits brought by it are immeasurable. BRIEF DESCRIPTION OF DRAWINGS

[0031] Figure 1 is a device structure schematic diagram of the system

[0032] Figure 2 is an operation flowchart of the system

[0033] Figure 3The schematic diagram of velocity and acceleration change of the phased variable speed control method of the system DETAILED DESCRIPTION

[0034] The application will be further described in connection with the embodiments, but the scope of protection of the application is not limited thereto:

[0035] In combination Figure 1 An array type rectifying tower material feeding intelligent control system, the system comprises an automatic swing arm 3, a first stepper motor 1 arranged at the upper end of the side of the automatic swing arm, a second stepper motor 2 arranged at the lower end of the automatic swing arm 3, a third stepper motor arranged at the front end of the telescopic arm, a needle 7 fixed at the end of the telescopic arm 5, and the needle head of the needle 7 is vertically downward; a plastic connecting pipe of the injection pump 6 passes through the automatic swing arm 3 and the telescopic arm 5 to connect the needle 7; an industrial CCD camera 8 is arranged directly above the array type rectifying tower 9;

[0036] The control system controls the first stepper motor 1, the second stepper motor 2 and the third stepper motor by pulse to realize the rotation and up-down movement of the automatic swing arm 3 and the telescopic movement of the telescopic arm 5; the initial position of the swing arm is positioned directly above the storage tank; a liquid level sensor is arranged on the surface of the needle to determine whether the needle tube contacts the liquid level; the stepper motor inside the injection pump controls the extraction and filling of the material based on pulse control, the industrial CCD camera 8 obtains the overhead view of the cluster and returns to the control system through TCP communication, and accurate positioning is realized through the material detection algorithm of the camera.

[0037] In combination Figure 2 The working steps of the array type rectifying tower material filling control system comprise:

[0038] S1, the industrial CCD camera 8 collects overhead view data of the cluster and returns to the control system;

[0039] S2, according to the material detection algorithm of the camera, the array type rectifying tower is positioned;

[0040] S3, the control system controls the automatic swing arm 3 to reach the target position;

[0041] S4, the control system controls the injection pump 6 to work to realize the material filling of the needle 7;

[0042] S5, the control system controls the automatic swing arm 3 to return to the initial position.

[0043] In the preferred embodiment, the steps of the material detection algorithm of the camera are as follows:

[0044] Step S1: constructing a target detection data set, the data set contains array type rectifying tower overhead views under multiple light conditions;

[0045] Step S2: analyze the pre-processed RGB image using the RetinaNet object detection algorithm to identify and locate the array distillation column while excluding inherent objects within the scene that should not be operated on, and output bounding box information.

[0046] In a preferred embodiment, in step S1, images of two methods are taken under real scenes of unmanned laboratories using Python to construct a data set of target objects.

[0047] The first method uses a normal RGB image input for direct target detection.

[0048] The second method adds an image enhancement operation in the preprocessing stage to adapt to image detection under complex lighting conditions.

[0049] In a preferred embodiment, according to the packing condition of the array distillation column in the actual production scene, the region that can be packed on the target to be grabbed in the data set is labeled; LabelImg is used to complete the labeling work, and the data set is labeled in Yolo format.

[0050] In a preferred embodiment, in step S2, the RetinaNet object detection algorithm is used to analyze the pre-processed RGB image to identify and locate the array distillation column while excluding inherent objects within the scene that should not be operated on, and output bounding box information; the best injection packing position is obtained through the output bounding box information.

[0051] In a preferred embodiment, the specific steps of introducing a feature pyramid network (FPN) in the backbone network of RetinaNet for multi-scale feature fusion are as follows:

[0052] First, different levels of feature maps are extracted through the backbone network, which correspond to different scale information.

[0053] Then, FPN fuses these feature maps from bottom to top, transmits high-level semantic information to low-level detailed information through cross-layer connection, and gradually constructs a series of feature maps with the same resolution but containing multi-scale information.

[0054] Finally, these fused feature maps are used for target detection to ensure that different size and scale target objects can be detected, thereby improving the accuracy and robustness of detection.

[0055] In a preferred embodiment, the best injection packing position is obtained through the following position:

[0056]

[0057] where x is the value of the best injection packing position in the x direction, x ip is the x-direction coordinate of each pixel on the frame in the predicted image i q is the y-direction coordinate of each pixel on the frame in the predicted image i corresponding pixel value.

[0058]

[0059] wherein y is the value of the optimal filler injection position in the y-direction, y i p is the x-direction coordinate of each pixel on the frame in the predicted image i q is the y-direction coordinate of each pixel on the frame in the predicted image i corresponding pixel value.

[0060] In the preferred embodiment, the array rectifying tower material filling control system realizes material filling control through the following steps:

[0061] (1) When taking the material, first extract the air column, and the required pulse number m1 is a given value;

[0062] (2) After the air column is extracted, the pulse number m2 required for taking the material is determined by the following formula:

[0063]

[0064] wherein V m is the sampling amount, M is the pulse number required for the full range of the plunger pump, V c is the capacity of the plunger pump;

[0065] (3) Combined with Figure 3 , a phased variable speed control method is used to control the stepping motor sampling, which is divided into an acceleration stage and a deceleration stage. The acceleration a1 and the speed v1 of the acceleration stage are determined by the following formula:

[0066]

[0067]

[0068] wherein t1 is the cutoff time for the acceleration to increase to the maximum value in the acceleration stage, t2 is the cutoff time of the acceleration stage, v z is the speed value when the acceleration reaches the maximum value, t is the time change, t1, t2, v z are all given values;

[0069] (4) The acceleration a2 and the speed v2 of the deceleration stage in the sampling process are determined by the following formula:

[0070]

[0071]

[0072] wherein t1 is the cut-off time for the acceleration to increase to the maximum value in the acceleration phase, t3 is the cut-off time for the acceleration to increase to the maximum value in the deceleration phase, t4 is the cut-off time in the deceleration phase, v z is the speed value when the acceleration reaches the maximum value, t is the time variation, t1, t3, t4, v z are all given values.

[0073] (5) When taking the material, the stepper motor controls the injection pump, and the injection needle first moves above the storage tank. The liquid level sensor or related detection equipment confirms that the injection needle has reached the liquid level and ensures that the material can be smoothly sucked in. Then, during the material filling stage, the stepper motor continues to control the movement of the injection needle above the rectifying tower, while the CCD camera monitors the injection process in real time to ensure that the injection needle is correctly aligned with the filling position and to avoid errors or deviations. During the filling process, the control system adjusts the flow rate and filling amount of the material according to the preset pulse number and the real-time state of the material, ensuring uniform and accurate filling. After filling is completed, the system confirms whether the material has reached the expected filling amount through the liquid level detection or feedback mechanism, and performs corresponding reset operations on the mechanical arm to complete the entire filling process.

[0074] The specific embodiments described herein are merely illustrative of the spirit of the present application. Those skilled in the art of the present application can make various modifications or supplements to the described specific embodiments or replace them with similar ways, without departing from the spirit of the present application or exceeding the scope defined by the appended claims.

Claims

1. An intelligent control system for material feed of an array distillation tower, characterized in that The system includes an automated rocker arm (3), a first stepper motor (1) arranged at the upper end of the side of the automated rocker arm, a second stepper motor (2) arranged at the lower end of the automated rocker arm (3), a third stepper motor arranged at the front end of a telescopic arm, an injection needle (7) fixed at the end of the telescopic arm (5), and the needle head of the injection needle (7) pointing vertically downward; a plastic connecting tube of an injection pump (6) passes through the automated rocker arm (3) and the telescopic arm (5) and is connected to the injection needle (7); an industrial CCD camera (8) is arranged directly above an array-type distillation tower (9); The control system controls the first stepper motor (1), the second stepper motor (2), and the third stepper motor through pulses to realize the rotation and up-and-down movement of the automatic rocker arm (3), as well as the extension and retraction of the telescopic arm (5); the initial position of the rocker arm is positioned directly above the material storage tank; a liquid level sensor is provided on the surface of the injection needle to determine whether the needle tube contacts the liquid surface; the stepper motor inside the injection pump controls the extraction and filling of the material based on the pulse control injection pump, and the industrial CCD camera (8) obtains the top view of the stack and transmits it back to the control system through TCP communication, and accurate positioning is achieved through the material detection algorithm of the camera; The camera's material detection algorithm steps are as follows: Step S1: constructing a target detection dataset, which includes top views of array distillation towers under various lighting conditions; Step S2: Analyze the preprocessed RGB image using the RetinaNet object detection algorithm to identify and locate the array distillation tower, while excluding inherent objects in the scene that should not be operated, and output bounding box information. The output bounding box information is used to obtain the optimal injection filling position; The optimal injection position of the filler is obtained by: Where x is the value of the optimal injection position in the x direction, x i is the x-coordinate of each pixel on the border of the predicted image, p i For x i The corresponding pixel value; Where y is the value of the optimal injection filler position in the y direction, i is the y-coordinate of each pixel on the border of the predicted image, q i For y i The corresponding pixel value.

2. The system according to claim 1, characterized in that The working steps of the system include: S1, industrial CCD camera (8) collects the top view data of the cluster and transmits it back to the control system; S2. Positioning the array distillation tower based on the camera's material detection algorithm; S3, the control system controls the automated rocker arm (3) to reach the target position; S4, the control system controls the injection pump (6) to operate to fill the injection needle (7) with material; S5, the control system controls the automatic rocker arm (3) to return to the initial position.

3. The system according to claim 1, characterized in that In step S1, Python is used to capture images of the two methods in a real scene in an unmanned laboratory to construct a dataset of the target object; The first method uses ordinary RGB image input and directly performs target detection; The second method is to add image enhancement operations in the preprocessing stage to adapt to image detection under complex lighting conditions.

4. The system according to claim 3, characterized in that According to the packing conditions of the array distillation tower in the actual production scenario, the packing area is marked on the target to be captured in the dataset; LabelImg is used to complete the labeling work and the dataset is labeled in Yolo format.

5. The system according to claim 1, characterized in that The specific steps of introducing the feature pyramid network FPN into the backbone network of RetinaNet for multi-scale feature fusion are as follows: First, the backbone network is used to extract feature maps at different levels, which correspond to different scale information. Then, FPN fuses these feature maps from bottom to top, passing high-level semantic information to low-level detail information through cross-layer connections, and gradually constructs a series of feature maps with the same resolution but containing multi-scale information; Finally, these fused feature maps are used for target detection to ensure that target objects of different sizes and scales can be detected, thereby improving the accuracy and robustness of detection.

6. The system according to claim 1 or 2, characterized in that The array type distillation tower material filling control system realizes material filling control through the following steps: (1) When taking the material, first extract the air column, and the required pulse number m1 is a given value; (2) After the air column is extracted, the number of pulses m2 required to take the material is determined by the following formula: Where V m is the sampling volume, M is the number of pulses required for the full range of the plunger pump, V c is the capacity of the plunger pump; (3) The stepper motor sampling is controlled by a phased variable speed control method, which is divided into an acceleration phase and a deceleration phase. The acceleration a1 and speed v1 in the acceleration phase are determined by the following formula: Among them, t1 is the deadline for the acceleration to reach the maximum value in the acceleration phase, t2 is the deadline for the acceleration phase, v z is the velocity value when the acceleration reaches the maximum value, t is the time change, t1, t2, v z All are given values; (4) The acceleration a2 and velocity v2 during the deceleration phase of the sampling process are determined by the following formula: Among them, t1 is the deadline for the acceleration to increase to the maximum value in the acceleration phase, t3 is the deadline for the acceleration to increase to the maximum value in the deceleration phase, t4 is the deadline for the deceleration phase, v z is the velocity value when the acceleration reaches the maximum value, t is the time change, t1, t3, t4, v z All are given values; (5) When taking materials, the injection pump is controlled by a stepper motor, and the injection needle first moves to the top of the storage tank. The liquid level sensor is used to confirm that the injection needle has reached the liquid surface and ensure that the material can be smoothly inhaled; then, in the material filling stage, the stepper motor continues to control the injection needle to move to the top of the distillation tower, and the CCD camera monitors the injection process of the material in real time to ensure that the injection needle is correctly aligned with the filling position to avoid errors or offsets; during the filling process, the control system adjusts the flow rate and filling amount of the material according to the preset number of pulses and the real-time status of the material to ensure that the filling is completed evenly and accurately; after the filling is completed, the system confirms whether the material has reached the expected filling amount through liquid level detection or feedback mechanism, and performs corresponding reset operations on the robotic arm to complete the entire filling process.

Images