Hot well liquid level remote transmission monitoring system

Abstract

The invention discloses a monitoring system, belongs to the technical field of liquid level monitoring, and particularly relates to a hot well liquid level remote transmission monitoring system which comprises a signal acquisition unit, an automatic control unit and a cloud platform unit, and the signal acquisition unit is composed of a liquid level acquisition module and a visual module; the liquid level acquisition module and the visual module are used for double acquisition of liquid level values and data processing, filtering processing, temperature compensation processing and vibration compensation processing are carried out on two signals respectively, then data correction is carried out, stable and accurate reading data are obtained, meanwhile, the cloud platform unit is used for achieving real-time display of identification results, and the identification accuracy is improved. According to the system, unmanned and real-time monitoring of the boiler double-color liquid level meter is achieved, meanwhile, the recognition result is stable, and the safety of factory production is guaranteed.

Description

Technical Field

[0001] The invention discloses a monitoring system, belongs to the technical field of liquid level monitoring, and specifically relates to a hot well liquid level remote transmission monitoring system. Background Art

[0002] Principle of polyester hot well: Polyester hot well is an important component in polyester device, mainly used to store and adjust the temperature and pressure of polyester melt (PET). Hot well is usually located after the polycondensation reactor of polyester device, and is used to temporarily store the melt for subsequent filtering and conveying operations. The temperature and pressure in the hot well need to be strictly controlled to ensure the stability and quality of the melt. The pressure in the hot well also needs to be strictly controlled, usually kept within a certain range to prevent the melt from overflowing or inhaling air. The pressure in the hot well can be maintained stable by adjusting the valve and pressure control system.

[0003] In order to ensure the stability of the continuous production process and prevent product quality fluctuations in the prior art, a pressure transmitter is added to the polyester hot well to observe the liquid level in the polyester hot well; however, this monitoring method cannot automatically alarm or upload data, and the monitoring of the polyester hot well liquid level meter is mainly based on manual inspection, which cannot be carried out uninterruptedly, and errors are prone to occur in the reading and recording links. Summary of the invention

[0004] Purpose of the invention: To provide a hot well liquid level remote monitoring system to solve the above-mentioned problems.

[0005] Technical solution: A hot well liquid level remote monitoring system, the monitoring system comprising:

[0006] The signal acquisition unit is composed of a liquid level acquisition module and a visual module. The liquid level acquisition module detects the liquid level signal of the hot well according to the pressure transmitter, and the visual module collects the liquid level image signal of the hot well according to the two-color liquid level meter.

[0007] The automatic control unit receives the liquid level signal and image signal processed by the signal acquisition unit, calculates the liquid level at this time, and determines the accuracy of the signals collected by the liquid level acquisition module and the visual module;

[0008] The cloud platform unit is used to obtain the liquid level data uploaded by the automatic control unit through the gateway. The cloud platform will automatically generate new connection information and save data information, and realize the connection of sensor devices in each hot well through the newly built API gateway and gateway sub-device of the human cloud platform.

[0009] In a further embodiment, the liquid level acquisition module of the signal acquisition unit inserts an air blowing pipe into the hot well, the upper end of the air blowing pipe is connected to the pressure transmitter through a tee, and nitrogen is connected to the air blowing pipe through a pressure reducing valve. The pressure of the pressure reducing valve is adjusted according to the liquid level, so that a constant flow of gas escapes from the lower port of the air blowing pipe inserted into the liquid, bubbles and is discharged into the atmosphere through the liquid. When a small amount of bubbles escapes from the lower end of the air blowing pipe, the air pressure in the air blowing pipe is almost equal to the static pressure of the liquid level. Therefore, the pressure value indicated by the pressure transmitter can reflect the liquid level.

[0010] In a further embodiment, the visual module of the signal acquisition unit collects the pressure transmitter video through a camera and extracts the reading from the video stream data, which is mainly divided into the following steps:

[0011] Preprocessing: image denoising and image conversion to HSV channel, then threshold filtering to obtain the salient features of the identified object;

[0012] Object segmentation, using manually segmented masks and images to perform AND operations to obtain visual analysis objects;

[0013] Line detection: first filter the visual analysis object, then perform edge extraction, and finally perform line detection on the edge extraction results.

[0014] Indication calculation: convert the indication of the pressure transmitter through relative distance and output a numerical signal.

[0015] In a further embodiment, after the image is converted into an HSV channel in preprocessing, the image is binarized in blue using the inRange function to obtain the salient features of the identified object.

[0016] In a further embodiment, the edge extraction in the line detection stage uses the Canny operator to perform edge extraction, and the specific algorithm is as follows:

[0017] Let f(x, y) represent the input image and G(x, y) represent the Gaussian function, then we get:

[0018]

[0019] Convolve G and f to form a smoothed image:

[0020] f s (x, y) = G(x, y) * f(x, y)

[0021] Compute the gradient magnitude and direction:

[0022]

[0023] The maximum gradient value on each pixel point is retained, and other non-maxima are removed to perform non-maximum suppression. After completing the straight line detection, the image is subjected to Hough line detection to extract the straight lines and their endpoints, that is, the coordinates of the red-green dividing point.

[0024] In a further embodiment, the indication of the pressure transmitter is converted by the relative distance in the indication calculation stage, that is:

[0025] Readings=D-(yH)

[0026] The actual maximum range is D>0, the ordinate of the maximum range in the image position is H, and the pressure transmitter reading is Readings; y is the point with the larger ordinate value among the two end points of the straight line obtained by Hough line detection.

[0027] In a further embodiment, in the automatic control unit, after receiving the liquid level numerical data of the liquid level signal and the image signal, filtering processing, temperature compensation processing and vibration compensation processing are performed on the two signals respectively;

[0028] The filtering process adaptively adjusts the system noise covariance matrix Q and the observation noise covariance matrix R, and its mathematical model is:

[0029] Assume that the mathematical model of the linear discrete system is:

[0030]

[0031] Z(k)=H(k)X(k)+V(k)

[0032] Where X(k) is the state vector, is the state transfer matrix, w(k-1) is the noise vector; Z(k) is the observation vector; H(k) is the observation matrix; V(k) is the measurement noise vector;

[0033] Then adaptive filtering is performed:

[0034] For state quantity And the accompanying covariance P is used for prediction:

[0035]

[0036] P(k|k-1)=Φ(k,k-1)P(k-1)Φ T (k,k-1)+Г(k,k-1)Ω(k)Г T (k,k

[0037] -1)

[0038] Then the derivative of the gain coefficient K in the update part is performed:

[0039]

[0040] Where b is the forgetting factor. The value of the observation error R will gradually stabilize over time and change slightly. The change of the gain coefficient K of the corresponding update part will also gradually stabilize.

[0041] The derivative of the gain coefficient K corresponding to the update part has the main difference that the observation error R is based on d k Adjustment, as the number of frames k changes, R(k)>R(k-1), that is, the observation error R tends to be stable, where R(k) is the observation covariance; R(k-1) is the prediction state error covariance matrix;

[0042] The temperature compensation process first establishes a mathematical model by polynomial fitting. Assuming that the temperature is the same, the output value of the sensor is x, and the actual measured value is y, the polynomial function is obtained as follows:

[0043]

[0044] where a i The value of is calculated using the least squares fitting method, and then the temperature effect is considered. When the temperature of the surrounding environment changes, a i The value of will also change. Similarly, we use the polynomial fitting method to establish the relationship between temperature t and a i The mathematical model is as follows:

[0045]

[0046] The a calculated by selecting different temperatures i The value of and the temperature t at this time are fitted by the least squares method to obtain b ji ; The final true measurement value y can be expressed as the formula:

[0047]

[0048] The vibration compensation process collects the tank vibration data and analyzes it, giving the final measurement result a weight compensation to improve the weighing accuracy and stability. The vibration sensor measures the acceleration as a, and the formula for the force on the support column due to vibration is obtained:

[0049] G f =k 3 a

[0050] where k 3 is the vibration compensation coefficient, which is a constant.

[0051] In a further embodiment, after completing the filtering process, the temperature compensation process and the vibration compensation process, the liquid level value result is corrected, and the average value of the liquid level acquisition module and the visual module is used as the ambient temperature t at a certain moment, and the ambient environment compensation weight is G t , which is calculated as follows:

[0052] G t =k 1 *(t i -t 0 )

[0053] Finally, the weight G is obtained by actual measurement calculation:

[0054]

[0055] Among them, G y1 is the liquid level value of the liquid level signal, G y2 is the liquid level value of the image signal.

[0056] Beneficial effect: The present invention collects liquid level values ​​dually and processes data through the liquid level acquisition module and the visual module, and performs filtering, temperature compensation and vibration compensation on the two signals respectively, and corrects the data to obtain stable and accurate reading data. At the same time, the cloud platform unit is used to realize real-time display of recognition results and automatic alarm. The system realizes unmanned and real-time monitoring of the boiler two-color liquid level meter. At the same time, the recognition result is stable, which ensures the safety of factory production. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a schematic diagram of the present invention.

[0058] Figure 2 It is a workflow diagram of the vision module of the present invention.

[0059] Figure 3 It is a schematic diagram of the filtering process of the automatic control unit of the present invention.

[0060] Figure 4 It is a working flow chart of the automatic control unit of the present invention. DETAILED DESCRIPTION

[0061] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0062] A hot well liquid level remote monitoring system, such as Figure 1As shown, the monitoring system comprises:

[0063] The signal acquisition unit is composed of a liquid level acquisition module and a visual module. The liquid level acquisition module detects the liquid level signal of the hot well according to the pressure transmitter, and the visual module collects the liquid level image signal of the hot well according to the two-color liquid level meter.

[0064] The automatic control unit receives the liquid level signal and image signal processed by the signal acquisition unit, calculates the liquid level at this time, and determines the accuracy of the signals collected by the liquid level acquisition module and the visual module;

[0065] The cloud platform unit is used to obtain the liquid level data uploaded by the automatic control unit through the gateway. The cloud platform will automatically generate new connection information and save data information, and realize the connection of sensor devices in each hot well through the newly built API gateway and gateway sub-device of the human cloud platform.

[0066] In one embodiment, the liquid level acquisition module of the signal acquisition unit inserts an air blowing pipe into the hot well, the upper end of the air blowing pipe is connected to the pressure transmitter through a tee, and nitrogen is connected to the air blowing pipe through a pressure reducing valve. The pressure of the pressure reducing valve is adjusted according to the liquid level, so that a constant flow of gas escapes from the lower port of the air blowing pipe inserted into the liquid, bubbles and is discharged into the atmosphere through the liquid. When a small amount of bubbles escape from the lower end of the air blowing pipe, the air pressure in the air blowing pipe is almost equal to the static pressure of the liquid level, so the pressure value indicated by the pressure transmitter can reflect the liquid level.

[0067] In one embodiment, Figure 2 As shown, the visual module of the signal acquisition unit collects the pressure transmitter video through the camera and extracts the reading from the video stream data, which is mainly divided into the following steps:

[0068] Preprocessing: image denoising and image conversion to HSV channel, then threshold filtering to obtain the salient features of the identified object;

[0069] Object segmentation, using manually segmented masks and images to perform AND operations to obtain visual analysis objects;

[0070] Line detection: first filter the visual analysis object, then perform edge extraction, and finally perform line detection on the edge extraction results.

[0071] Indication calculation: convert the indication of the pressure transmitter through relative distance and output a numerical signal.

[0072] In one embodiment, in preprocessing, after converting the image into HSV channels, the inRange function is used to perform blue binarization on the image to obtain the salient features of the identified object.

[0073] In one embodiment, the edge extraction in the line detection stage uses the Canny operator to extract the edge, and the specific algorithm is as follows:

[0074] Let f(x, y) represent the input image and G(x, y) represent the Gaussian function, then we get:

[0075]

[0076] Convolve G and f to form a smoothed image:

[0077] f s (x, y) = G(x, y) * f(x, y)

[0078] Compute the gradient magnitude and direction:

[0079]

[0080] The maximum gradient value on each pixel point is retained, and other non-maxima are removed to perform non-maximum suppression. After completing the straight line detection, the image is subjected to Hough line detection to extract the straight lines and their endpoints, that is, the coordinates of the red-green dividing point.

[0081] In one embodiment, in the indication calculation stage, the indication of the pressure transmitter is converted by the relative distance, that is:

[0082] Readings=D-(yH)

[0083] The actual maximum range is D>0, the ordinate of the maximum range in the image position is H, and the pressure transmitter reading is Readings; y is the point with the larger ordinate value among the two end points of the straight line obtained by Hough line detection.

[0084] In one embodiment, Figure 3 As shown, in the automatic control unit, after receiving the liquid level numerical data of the liquid level signal and the image signal, the two signals are subjected to filtering processing, temperature compensation processing and vibration compensation processing respectively;

[0085] The filtering process adaptively adjusts the system noise covariance matrix Q and the observation noise covariance matrix R, and its mathematical model is:

[0086] Assume that the mathematical model of the linear discrete system is:

[0087]

[0088] Z(k)=H(k)X(k)+V(k)

[0089] Where X(k) is the state vector, is the state transfer matrix, w(k-1) is the noise vector; Z(k) is the observation vector; H(k) is the observation matrix; V(k) is the measurement noise vector;

[0090] Then adaptive filtering is performed:

[0091] For state quantity And the accompanying covariance P is used for prediction:

[0092]

[0093] P(k|k-1)=Φ(k,k-1)P(k-1)Φ T (k,k-1)+Г(k,k-1)Ω(k)Г T (k,k

[0094] -1)

[0095] Then the derivative of the gain coefficient K in the update part is performed:

[0096]

[0097] Where b is the forgetting factor. The value of the observation error R will gradually stabilize over time and change slightly. The change of the gain coefficient K of the corresponding update part will also gradually stabilize.

[0098] The derivative of the gain coefficient K corresponding to the update part has the main difference that the observation error R is based on d k Adjustment, as the number of frames k changes, R(k)>R(k-1), that is, the observation error R tends to be stable, where R(k) is the observation covariance; R(k-1) is the prediction state error covariance matrix;

[0099] The temperature compensation process first establishes a mathematical model by polynomial fitting. Assuming that the temperature is the same, the output value of the sensor is x, and the actual measured value is y, the polynomial function is obtained as follows:

[0100]

[0101] where a i The value of is calculated using the least squares fitting method, and then the temperature effect is considered. When the temperature of the surrounding environment changes, a i The value of will also change. Similarly, we use the polynomial fitting method to establish the relationship between temperature t and a i The mathematical model is as follows:

[0102]

[0103] The a calculated by selecting different temperatures iThe value of and the temperature t at this time are fitted by the least squares method to obtain b ji ; The final true measurement value y can be expressed as the formula:

[0104]

[0105] The vibration compensation process collects the tank vibration data and analyzes it, giving the final measurement result a weight compensation to improve the weighing accuracy and stability. The vibration sensor measures the acceleration as a, and the formula for the force on the support column due to vibration is obtained:

[0106] G f =k 3 a

[0107] where k 3 is the vibration compensation coefficient, which is a constant.

[0108] In one embodiment, Figure 4 As shown, after completing the filtering, temperature compensation and vibration compensation processes, the liquid level value result is corrected, and the average value of the liquid level acquisition module and the visual module is used as the ambient temperature t at a certain moment, and the ambient environment compensation weight is G t , which is calculated as follows:

[0109] G t =k 1 *(t i -t 0 )

[0110] Finally, the weight G is obtained by actual measurement calculation:

[0111]

[0112] Among them, G y1 is the liquid level value of the liquid level signal, G y2 is the liquid level value of the image signal.

[0113] In one embodiment, the cloud platform unit and the automatic control unit need to communicate through a communication channel between the two. The establishment of the communication channel is inseparable from the cloud node of the control algorithm. When the communication module of the monitoring system is connected to the session module generated by the cloud node, the cloud server will automatically create a data transmission channel for the automatic control unit of the hot well. At this time, all kinds of monitoring data uploaded by the automatic control unit to the cloud platform will be directly forwarded to the monitoring system. In order to ensure the smooth transmission of data in the cloud, a buffer needs to be built in the cloud to save buffered data. The buffer uses a first-in-first-out queue transmission method. When real-time data is quickly acquired by the cloud, if the monitoring system end is partially delayed due to slow data transmission, it can still be transmitted normally with the help of the buffer, so the data flow will not be affected.

[0114] The data is parsed through the cloud platform, and the required data is saved in the cloud platform database; the message management module is then used to receive and manage the hardware system messages. Since the received data format is JSON, the data used needs to be extracted through certain encoding and then saved to the database, and finally transmitted to the client through the communication link and gateway to display the message.

[0115] The system adopts multi-threading for data collection and transmission, uses 2-way Socket in 4G gateway to realize communication connection, and uses recursive loop to realize real-time reception of data packets and real-time update of data. The system assigns the functions of liquid level collection module, visual module data acquisition, display to different threads to realize parallel operation, and can monitor various data of each hot well in a timely and accurate manner.

[0116] Obviously, the above embodiments are merely examples for the purpose of clear explanation, and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the scope of protection of the invention.

Claims

1. A hot well liquid level remote monitoring system, characterized in that: The monitoring system comprises: The signal acquisition unit is composed of a liquid level acquisition module and a visual module. The liquid level acquisition module detects the liquid level signal of the hot well according to the pressure transmitter, and the visual module collects the liquid level image signal of the hot well according to the two-color liquid level meter. The automatic control unit receives the liquid level signal and image signal processed by the signal acquisition unit, calculates the liquid level at this time, and determines the accuracy of the signals collected by the liquid level acquisition module and the visual module; The cloud platform unit is used to obtain the liquid level data uploaded by the automatic control unit through the gateway. The cloud platform will automatically generate new connection information and save data information, and realize the connection of sensor devices in each hot well through the newly built API gateway and gateway sub-device of the human cloud platform.

2. A hot well liquid level remote monitoring system according to claim 1, characterized in that: The liquid level acquisition module of the signal acquisition unit inserts an air blowing pipe into the hot well, the upper end of the air blowing pipe is connected to the pressure transmitter through a tee, nitrogen is connected to the air blowing pipe through a pressure reducing valve, and the pressure of the pressure reducing valve is adjusted according to the liquid level, so that a constant flow of gas escapes from the lower port of the air blowing pipe inserted into the liquid, bubbles and is discharged into the atmosphere through the liquid. When a small amount of bubbles escape from the lower end of the air blowing pipe, the air pressure in the air blowing pipe is almost equal to the static pressure of the liquid level, so the pressure value indicated by the pressure transmitter can reflect the liquid level.

3. A hot well liquid level remote monitoring system according to claim 2, characterized in that: The visual module of the signal acquisition unit collects the pressure transmitter video through a camera and extracts the reading from the video stream data, which is mainly divided into the following steps: Preprocessing: image denoising and image conversion to HSV channel, then threshold filtering to obtain the salient features of the identified object; Object segmentation, using manually segmented masks and images to perform AND operations to obtain visual analysis objects; Line detection: first filter the visual analysis object, then perform edge extraction, and finally perform line detection on the edge extraction results. Indication calculation: convert the indication of the pressure transmitter through relative distance and output a numerical signal.

4. A hot well liquid level remote monitoring system according to claim 3, characterized in that: In the preprocessing, the image is converted into HSV channel and then the inRange function is used to perform blue binarization on the image to obtain the salient features of the identified object.

5. A hot well liquid level remote monitoring system according to claim 3, characterized in that: In the line detection stage, edge extraction uses the Canny operator for edge extraction. The specific algorithm is as follows: Let f(x, y) represent the input image and G(x, y) represent the Gaussian function, then we get: Convolve G and f to form a smoothed image: f s (x,y)=G(x,y)*f(x,y) Compute the gradient magnitude and direction: The maximum gradient value on each pixel point is retained, and other non-maxima are removed to perform non-maximum suppression. After completing the straight line detection, the image is subjected to Hough line detection to extract the straight lines and their endpoints, that is, the coordinates of the red-green dividing point.

6. A hot well liquid level remote monitoring system according to claim 5, characterized in that: In the indication calculation stage, the indication of the pressure transmitter is converted by the relative distance, that is: Readings=D-(yH) The actual maximum range is D>0, the ordinate of the maximum range in the image position is H, and the pressure transmitter reading is Readings; y is the point with the larger ordinate value among the two end points of the straight line obtained by Hough line detection.

7. A hot well liquid level remote monitoring system according to claim 1, characterized in that: In the automatic control unit, after receiving the liquid level numerical data of the liquid level signal and the image signal, the two signals are subjected to filtering processing, temperature compensation processing and vibration compensation processing respectively; The filtering process adaptively adjusts the system noise covariance matrix Q and the observation noise covariance matrix R, and its mathematical model is: Assume that the mathematical model of the linear discrete system is: Z(k)=H(k)X(k)+V(k) Where X(k) is the state vector, is the state transfer matrix, w(k-1) is the noise vector; Z(k) is the observation vector; H(k) is the observation matrix; V(k) is the measurement noise vector; Then adaptive filtering is performed: For state quantity And the accompanying covariance P is used for prediction: P(k|k-1)=Ф(k,k-1)P(k-1)Φ T (k,k-1)+Г(k,k-1)Ω(k)Г T (k,k -1) Then the derivative of the gain coefficient K in the update part is performed: Where b is the forgetting factor. The value of the observation error R will gradually stabilize over time and change slightly. The change of the gain coefficient K of the corresponding update part will also gradually stabilize. The derivative of the gain coefficient K corresponding to the update part has the main difference that the observation error R is based on d k Adjustment, as the number of frames k changes, R(k)>R(k-1), that is, the observation error R tends to be stable, where R(k) is the observation covariance; R(k-1) is the prediction state error covariance matrix; The temperature compensation process first establishes a mathematical model by polynomial fitting. Assuming that the temperature is the same, the output value of the sensor is x, and the actual measured value is y, the polynomial function is obtained as follows: where a i The value of is calculated using the least squares fitting method, and then the temperature effect is considered. When the temperature of the surrounding environment changes, a i The value of will also change. Similarly, we use the polynomial fitting method to establish the relationship between temperature t and a i The mathematical model is as follows: The a calculated by selecting different temperatures i The value of and the temperature t at this time are fitted by the least squares method to obtain b ji ; The final true measurement value y can be expressed as the formula: The vibration compensation process collects the tank vibration data and analyzes it, giving the final measurement result a weight compensation to improve the weighing accuracy and stability. The vibration sensor measures the acceleration as a, and the formula for the force on the support column due to vibration is obtained: <h2 style=";text-align:left;direction:ltr">G<h2 style=";text-align:left;direction:ltr"> f <h2 style=";text-align:left;direction:ltr"> =k3a Where k3 is the vibration compensation coefficient, which is a constant.

8. A hot well liquid level remote monitoring system according to claim 7, characterized in that: After completing the filtering, temperature compensation and vibration compensation processes, the liquid level value result is corrected, and the average value of the liquid level acquisition module and the visual module is used as the ambient temperature t at a certain moment. The ambient environment compensation weight is G t , which is calculated as follows: G t =k1*(t i -t0) Finally, the weight G is obtained by actual measurement calculation: Among them, G y1 is the liquid level value of the liquid level signal, G y2 is the liquid level value of the image signal.

Images