A visual interface instrument for chemical industry interface detection and detection method thereof

Abstract

The invention discloses a visual interface instrument for chemical interface detection, comprising a detection host at the front end, a trapezoidal mirror cover shell connected to the rear end of the detection host, a transfer cover shell connected to the rear end of the trapezoidal mirror cover, a mirror surrounding sheet metal, a mirror body at the rear end of the transfer cover shell, an explosion-proof light source at the rear end of the mirror body, and mirror surrounding sheet metals between the mirror body, the transfer cover shell and the explosion-proof light source. The advantages of the invention are: a mirror interface image is collected by a visual sensor, visual feature extraction and calculation are performed at an embedded edge end, the visual signal of the interface is converted into a standard industrial control signal and output in real time, and a real-time video stream is output at the same time, the whole device has stable performance, high reliability, low delay, and the core components are not easily damaged.

Description

Technical Field

[0001] The invention belongs to the technical field of chemical detection, and in particular relates to a visual interface instrument for chemical interface detection and a detection method thereof. Background Art

[0002] In the material separation process of chemical production, chemical raw materials continuously undergo chemical reactions in the production equipment, and then after a specific process (extraction, distillation, etc.), materials in three states of light, middle layer and heavy phase are formed. Finally, the three states of materials are placed in different containers. The chemical interface refers to the interface formed between the light phase and the middle layer, between the light phase and the heavy phase, and between the middle layer and the heavy phase. In today's highly automated chemical production, the chemical interface detection link is still a pain point. Without the accurate interface position, it is impossible to achieve automated control of the separation of materials in the process.

[0003] The existing processing methods mainly include manual detection and instrument detection:

[0004] Install a glass pipe sight glass at the discharge end, observe the position of the interface manually and take corresponding actions. This method has high requirements on the operator, and it is necessary to observe the position of the interface in the glass pipe sight glass for a long time. Visual fatigue often leads to missed interfaces, especially when the colors of the two materials above and below the interface are similar, it is impossible to judge whether the current interface is above or below, resulting in material waste and economic losses. Some chemical plants have a high-risk production environment, and the raw materials contain toxins, which are not suitable for operators to observe on site for a long time.

[0005] Traditional detection instruments mainly include ultrasonic detectors, conductivity meters and mass flow meters. All three instruments need to be in direct contact with the material, which can easily cause corrosion to the contact part of the instrument, resulting in a decrease in measurement accuracy or even instrument failure. Since it is a contact detection, there are certain requirements for the substances in the material, because large particles or thick materials will damage the instrument detection probe part. Traditional detection instruments generally have a certain detection delay and are not suitable for scenarios with high real-time requirements. Another important thing is that traditional instruments can only output industrial control signals and cannot output video streams that can express the material status inside the pipeline. Traditional detection instruments have the disadvantages of limited use scenarios, high latency, non-intuitiveness, and easy wear. Summary of the invention

[0006] The purpose of the present invention is to overcome the above limitations and provide a visual interface instrument for chemical interface detection that collects mirror interface images through visual sensors, performs visual feature extraction calculations at the embedded edge end, converts the visual signals of the interface into standard industrial control signals and outputs them in real time, while outputting real-time video streams.

[0007] The objective of the present invention is achieved through the following technical solutions: a visual interface instrument for chemical interface detection, comprising a detection host at the front end, a trapezoidal mirror cover connected to the rear end of the detection host, a transfer cover connected to the rear end of the trapezoidal mirror cover, a mirror surrounding sheet metal, a mirror body at the rear end of the transfer cover, an explosion-proof light source at the rear end of the mirror body, and a mirror surrounding sheet metal between the mirror body, the transfer cover and the explosion-proof light source.

[0008] Preferably, the detection host includes a visual sensor and an embedded edge device, the visual sensor is used to obtain the original image required by the interface detection algorithm module, and the embedded edge device is provided with an interface detection algorithm module, a video encoding module and a signal processing module.

[0009] A detection method of a visual interface instrument for chemical interface detection, the steps are as follows:

[0010] A. Align the front end of the detection host with the interface. The visual sensor in the detection host obtains the original YUV image of the liquid phase of the pipeline sight glass formed by the fill light of the explosion-proof light source, and transmits the image data in YUV format to the embedded computer;

[0011] B. The interface detection algorithm module in the embedded edge device converts the image data in YUV format into interface position data and transmits it to the external control system through RS485, network port or current loop;

[0012] C. The video encoding module in the embedded edge device encodes the image data and interface position data, and transmits the video stream to the display terminal through the network port.

[0013] Preferably, in step A, the fill light uses a light source with a specific spectrum for fill light.

[0014] Preferably, the specific steps of converting the image data in YUV format into interface position data in step B are:

[0015] a. Detection area calibration: select the rectangular area to be measured on the pipe mirror image obtained by the visual sensor;

[0016] b. Obtain the brightness response value array. Calculate the brightness mean or median of each row from bottom to top in the image of the rectangular detection area and store them in a one-dimensional array in order.

[0017] c. De-noising and smoothing the brightness response value array: median filtering and smoothing the brightness response value one-dimensional array to complete the noise reduction;

[0018] d. Luminance response value gradient calculation: calculate the direction and rate of change of brightness of each row in the interface according to the set step size, and store it in the response value gradient array;

[0019] e. Search for the row position of the interface in the detection box, search for the maximum negative value in the response value gradient array, and use the index of the maximum negative gradient as the row position of the interface in the detection box.

[0020] Preferably, in step C, the state and position information of the current interface are obtained after algorithm processing of the collected image, and are superimposed on the original data of the video stream, so that the state and position information of the interface can be intuitively seen at the monitoring end, and at the same time, the industrial detection signal is output through the signal processing module, and the video stream is transmitted to the display terminal through the communication methods of Modbus RTU and Modbus TCP.

[0021] In summary, the present invention has the following advantages: 1. It adopts a light source with a specific spectrum for fill-in lighting, and utilizes the difference in spectral responses of different materials to effectively solve the problem of identifying materials that the human eye cannot distinguish, such as two colorless and transparent materials; 2. It adopts a closed heat-insulating shell as the optical path design to effectively solve the external interference and the fogging of the mirror window surface; 3. It collects real-time images inside the mirror through a visual sensor, and encapsulates them into a video stream after encoding processing, provides real-time video stream push service, and solves the problem of remote video monitoring; 4. The visual interface instrument adopts a high-performance embedded edge device, and obtains the status and position information of the current interface after algorithm processing of the collected image, and superimposes it on the original data of the video stream, so that the status and position information of the interface can be intuitively seen at the monitoring end, and the industrial detection signal (4-20mA current loop) is output through the signal processing module, and the communication mode supports both Modbus RTU and Modbus TCP mode has stable performance, high reliability and low delay; 5. The visual interface instrument is connected to the pipeline through the viewing mirror part. The detection host and light source part do not directly contact the material. It has the characteristics of non-contact, so it can be used for a long time and the core components are not easily damaged. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is the main structure diagram of the present invention;

[0023] Figure 2 It is a flow chart of the detection method steps of the present invention;

[0024] Numbers in the figure: 1- detection host, 2- trapezoidal mirror cover, 3- adapter cover, 4- mirror surrounding sheet metal, 5- mirror body, 6- explosion-proof light source. DETAILED DESCRIPTION

[0025] In order to deepen the understanding of the present invention, the present invention will be further described in detail below in conjunction with embodiments and drawings. The embodiments are only used to explain the present invention and do not constitute a limitation on the protection scope of the present invention.

[0026] like Figure 1As shown, the present invention provides a visual interface instrument for chemical interface detection, including a detection host 1 at the front end, a trapezoidal mirror cover 2 connected to the rear end of the detection host 1, a transfer cover 3 connected to the rear end of the trapezoidal mirror cover 2, a mirror surrounding sheet metal 4, a mirror body 5 is provided at the rear end of the transfer cover 3, an explosion-proof light source 6 is provided at the rear end of the mirror body 5, and a mirror surrounding sheet metal 4 is provided between the mirror body 5 and the transfer cover 3 and the explosion-proof light source 6.

[0027] The detection host 1 includes a visual sensor and an embedded edge device. The visual sensor is used to obtain the original image required by the interface detection algorithm module. The embedded edge device is provided with an interface detection algorithm module, a video encoding module and a signal processing module.

[0028] The detection method steps of the visual interface instrument of this application are as follows:

[0029] First, align the front end of the detection host 1 with the interface, and the visual sensor in the detection host 1 obtains the original YUV image of the liquid phase of the pipeline mirror formed by the fill light of the explosion-proof light source. The fill light uses a light source with a specific spectrum for fill light, and then transmits the image data in YUV format to the embedded edge device. The mirror part in the mirror body 5 is connected to the pipeline, and the detection host 1 and the light source part do not directly contact the material, which has the characteristics of non-contact;

[0030] The interface detection algorithm module in the embedded edge device converts the image data in YUV format into interface position data, and transmits it to the external control system through RS485, network port or current loop; the video encoding module in the embedded edge device encodes the image data and interface position data, and transmits the video stream to the display terminal through the network port.

[0031] The specific steps of converting the image data in YUV format into interface position data are as follows:

[0032] a. Detection area calibration: that is, select the rectangular area to be measured on the pipe mirror image obtained by the visual sensor;

[0033] The specific steps are: set the width to w and the height to h to get a w*h two-dimensional array I:

[0034]

[0035] b. Obtaining the brightness response value array; that is, calculating the brightness mean or median of each row from bottom to top in the image of the rectangular detection area, and storing them in a one-dimensional array in order;

[0036] The specific steps are: starting from the 0th row of the two-dimensional array I in step a, take out one row of data (one-dimensional array r) in turn, r = (i (r, 0), i (r, 1) ... i (r, w-1)), and find its average or median;

[0037] The average value is:

[0038] Sort the elements in array r from small to large to get a new array r_sort.

[0039] Take the middle value: Median(r) = r_sort(w / 2)

[0040] Store the mean or median of each row in a one-dimensional array in turn to obtain the response value array V = (v0, v1…v h-2 , v h-1 );

[0041] c. De-noise and smooth the brightness response value array. Perform median filtering and smoothing on the brightness response value one-dimensional array to complete the noise reduction, that is, obtain the smoothed response value array V smooth =(vs0, vs1…vs h-2 , vs h-1 );

[0042] d. Luminance response value gradient calculation: calculate the direction and rate of change of brightness of each row in the interface according to the set step size, and store it in the response value gradient array G;

[0043] The specific method is:

[0044] Expanded smoothed response value array V smooth_exp =[vs0, vs1...vs h-1 , vs h-L , vs h-L+1 ...vs h-1 ] Traverse the expanded response value array and calculate its gradient according to the step size. For the i-th row, the gradient calculation formula is:

[0045]

[0046] After the traversal is completed, the gradient array G of the response value is obtained = [g0, g1...g h-1 ];

[0047] e. Search for the row position of the interface in the detection frame, search for the maximum negative value in the response value gradient array G, and use the index of the maximum negative gradient as the row position of the interface in the detection frame.

[0048] After algorithmic processing of the captured images, the status and position information of the current interface is obtained and superimposed on the original data of the video stream. It is provided to the monitoring end in the form of real-time video stream push, so that the status and position information of the interface can be intuitively seen at the monitoring end. At the same time, a 4-20mA standard industrial detection signal is output through the signal processing module, and it can be used as a Modbus slave to provide the status and position information of the interface to the main control device in the control room through Modbus RTU and Modbus TCP in a communication manner.

Claims

1. A detection method of a visual interface instrument for chemical interface detection, characterized in that: The invention comprises a detection host (1) at the front end, the rear end of the detection host (1) is connected to a trapezoidal mirror housing (2), the rear end of the trapezoidal mirror housing (2) is connected to a transfer housing (3), a mirror surrounding sheet metal (4), a mirror body (5) is provided at the rear end of the transfer housing (3), an explosion-proof light source (6) is provided at the rear end of the mirror body (5), and a mirror surrounding sheet metal (4) is provided between the mirror body (5), the transfer housing (3) and the explosion-proof light source (6); The detection host (1) comprises a visual sensor and an embedded edge device, wherein the embedded edge device is provided with an interface detection algorithm module, a video encoding module and a signal processing module; Here are the steps: A. Align the front end of the detection host (1) with the interface, and the visual sensor in the detection host (1) obtains the original YUV image of the liquid phase of the pipeline mirror after being supplemented by the supplementary light of the explosion-proof light source (6), and transmits the image data in YUV format to the embedded edge device; B. The interface detection algorithm module in the embedded edge device converts the image data in YUV format into interface position data and transmits it to the external control system through RS485, network port or current loop; C. The video encoding module in the embedded edge device encodes the image data and interface position data, and transmits the video stream to the display terminal through the network port; The specific steps of converting the image data in YUV format into interface position data in step B are: a. Detection area calibration: select the rectangular area to be measured on the pipe mirror image obtained by the visual sensor; b. Obtain the brightness response value array. Calculate the brightness mean or median of each row from bottom to top in the image of the rectangular detection area and store them in a one-dimensional array in order. c. De-noising and smoothing the brightness response value array, performing median filtering and smoothing on the one-dimensional array of brightness response values ​​to complete noise reduction; d. Luminance response value gradient calculation: calculate the direction and rate of change of brightness of each row in the interface according to the set step size, and store it in the response value gradient array; e. Search for the row position of the interface in the detection box, search for the maximum negative value in the response value gradient array, and use the index of the maximum negative gradient as the row position of the interface in the detection box.

2. The detection method of a visual interface instrument for chemical industry interface detection according to claim 1, characterized in that: In step A, the fill light uses a light source with a specific spectrum for fill light.

3. The detection method of a visual interface instrument for chemical interface detection according to claim 1, characterized in that: In the step C, the state and position information of the current interface are obtained after algorithm processing of the collected image, and are superimposed on the original data of the video stream, so that the state and position information of the interface can be intuitively seen at the monitoring end. At the same time, the industrial detection signal is output through the signal processing module, and the video stream is transmitted to the display terminal through the communication mode of Modbus RTU and Modbus TCP.

Images