Chemical mechanical system with material level monitoring function

Abstract

The invention discloses a chemical machinery system with a material level monitoring function, and belongs to the technical field of chemical machinery. A monitoring module is connected with a reaction kettle and a tail gas treatment module; the processing and analyzing module is used for processing and analyzing the data acquired by the monitoring module, determining the material state during reaction and timely adjusting feeding and discharging and temperature data; the control module is used for timely adjusting system parameters according to an analysis result of the processing and analyzing module; the central processing unit is used for data transmission among the modules and processing, calculation and storage support; the user interface is used for displaying system parameters and analysis results, monitoring the material level, temperature and material reaction state data in the reaction kettle in real time through the monitoring module, obtaining reaction state-temperature associated feature vectors through processing and analysis, and comparing the reaction state-temperature associated feature vectors with preset reaction parameters; therefore, temperature increase or decrease and material inlet and outlet increase and decrease control are determined, material reaction is monitored in time, and stable and intact reaction is guaranteed.

Description

Technical Field

[0001] The present application relates to the technical field of chemical machinery, and more specifically, to a chemical machinery system with a material level monitoring function. Background Art

[0002] In chemical reactions, a reactor system is often used for heating and reacting chemical materials, and the control of temperature and time changes is often involved during the reaction.

[0003] The prior art document with the publication number CN107670599B provides a chemical machinery system with a material level monitoring function. N vibration sensors and N noise sensors are evenly and cross - arranged in the circumferential direction of a horizontal tank. The vibration sensors mechanically collect vibration signals, and the noise sensors collect noise signals. The material level in the horizontal tank is judged based on the mechanical vibration signals and noise signals.

[0004] Although the above - mentioned prior art solution can achieve relevant beneficial effects through the existing structure, there are still the following defects:

[0005] In the system, only some sensors are used to control the material level of the material. However, the change in the material level will cause the reaction to change dynamically. With different material levels, at the same temperature and time, the reaction rate is also different. Therefore, it is also necessary to monitor the dynamic changes of the material reaction in real - time according to different material levels, so as to further control the adjustment of the material level and temperature, and ensure the stability and safety of the reaction.

[0006] In view of the above - mentioned related technology, the inventor believes that in the system, only some sensors are used to control the material level of the material. However, the change in the material level will cause the reaction to change dynamically. With different material levels, at the same temperature and time, the reaction rate is also different. Therefore, it is also necessary to monitor the dynamic changes of the material reaction in real - time according to different material levels, so as to further control the adjustment of the material level and temperature, and ensure the stability and safety of the reaction.

[0007] In view of this, we propose a chemical machinery system with a material level monitoring function. Summary of the Invention

[0008] 1. Technical Problems to be Solved

[0009] The purpose of the present application is to provide a chemical machinery system with a material level monitoring function, which solves the technical problem that in the system of the above - mentioned background art, only some sensors are used to control the material level of the material. However, the change in the material level will cause the reaction to change dynamically. With different material levels, at the same temperature and time, the reaction rate is also different. Therefore, it is also necessary to monitor the dynamic changes of the material reaction in real - time according to different material levels, so as to further control the adjustment of the material level and temperature, and ensure the stability and safety of the reaction, and achieve the technical effect.

[0010] 2. Technical Solution

[0011] The technical solution of this application provides a chemical machinery system with a material level monitoring function, including:

[0012] An exhaust gas treatment module, which is connected to the oil and gas outlet of the reaction kettle and is used for the treatment of reaction oil and gas;

[0013] A monitoring module, which is connected to the reaction kettle and the exhaust gas treatment module and is used for detecting the material level, reaction state, temperature change and exhaust gas data during the reaction;

[0014] A processing and analysis module, which is used for processing and analyzing the data collected by the monitoring module, determining the material state during the reaction and timely adjusting the feeding and discharging and temperature data;

[0015] A control module, which is used for timely adjusting the system parameters according to the analysis results of the processing and analysis module;

[0016] A central processing unit, which is used for data transmission between modules and providing support for processing, calculation and storage;

[0017] A user interface, which is used for displaying system parameters and analysis results.

[0018] Through the above solution, the material level, temperature and material reaction state data in the reaction kettle are monitored in real time by the monitoring module. Through processing and analysis, a reaction state-temperature correlation feature vector is obtained and compared with the preset reaction parameters, so as to determine whether the temperature should be increased or decreased at this time and the control of the increase and decrease of the material feeding and discharging, so as to timely monitor the material reaction and ensure the stable and good progress of the reaction.

[0019] Optionally, the exhaust gas treatment module includes at least two coolers, and the multiple coolers are connected in sequence. The inlet of the frontmost cooler is connected to the oil and gas outlet of the reaction kettle through a pipeline, and the outlet of the last cooler is connected to the inlet of the filter through a pipeline. The liquid outlets of the coolers are all connected to the liquid collector through pipelines, the outlet of the filter is connected to the discharge port, and a valve is installed at the oil and gas outlet of the reaction kettle.

[0020] Through the above solution, the condensable components in the exhaust gas are cooled and collected by multiple coolers, and the non-condensable gas is discharged after being filtered by the filter, ensuring the cleanliness of the gas discharged into the high altitude and avoiding environmental pollution.

[0021] Optionally, the monitoring module includes: a camera and a plurality of laser sources, both of which are installed at the top of the inner cavity of the reactor for collecting material data inside the reactor; a thermometer installed inside the reactor for collecting temperature data inside the reactor; a pressure sensor installed at the top of the reactor for monitoring the pressure inside the reactor; and a gas flow meter installed on the pipeline between the filter and the discharge port for monitoring the flow rate of the discharged gas.

[0022] Through the above solution, through the combined monitoring of the laser source and the camera, the material level and the corresponding reaction process can be monitored simultaneously according to the image and brightness change of the material. The pressure sensor and the gas flow meter can adjust the discharge of the tail gas and control the pressure in the gas flow channel, so as to keep the pressure inside the reactor stable.

[0023] Optionally, a pressure relief valve is further installed at the top of the reactor for pressure relief and pressure adjustment. An electric heating wire is installed inside the reactor. A material pump for feeding and discharging materials is also installed on the side and bottom of the reactor. The pressure relief valve, the electric heating wire and the material pump are all electrically connected to the control module.

[0024] Through the above solution, the pressure safety is ensured by the pressure relief valve. The electric heating wire is used to control the temperature inside the reactor, and the material pump is used to control the height of the material level.

[0025] Optionally, the processing and analysis module includes: a material level calculation unit for calculating the height of the material to obtain the material volume data; a temperature processing unit for processing the temperature data to obtain the eigenvector of the temperature time series; a state confirmation unit for confirming the process of the material reaction state and extracting the eigenvector of the dynamic change of the material state time series; and a fusion analysis unit for performing fusion analysis on the eigenvector of the temperature time series and the eigenvector of the dynamic change of the material state time series to confirm whether the temperature should increase or decrease at the current time point and the feeding and discharging control of the material.

[0026] Through the above solution, by processing the temperature time series, calculating the material level of the material, and confirming the material reaction state, the characteristic relationship between the dynamic reaction of the material and the dynamic change of the temperature is established, so as to control the material and the temperature and ensure the reaction process.

[0027] Optionally, the material calculation unit includes highlight confirmation, width calculation, and height calculation. For highlight confirmation, a laser source irradiates the material, and a camera captures an image to confirm the position of the laser highlight on the image. The position and range of the highlight are confirmed by the change in grayscale in the captured image. That is, if the difference between the grayscale value and the surrounding grayscale is greater than a preset difference value, it is the range of the laser highlight area, and the grayscale difference value is 20 - 40. The width calculation confirms the highlight width based on the position and range of the laser highlight on the image. The height calculation is based on the fact that when the position of the laser source is constant, there is a one-to-one correspondence between the distance between the top of the material height and the laser source and the highlight width. Thus, the material height is obtained based on the highlight width, and the height and volume data of the material in the reaction kettle are confirmed.

[0028] Through the above solution, through the dual monitoring of the camera and the laser source, not only can the material level be confirmed by the change and range of the laser source highlight, but also, based on the change in grayscale irradiated by the laser source and the image processing of the camera, the image feature comparison corresponding to the reaction process of the material can be confirmed synchronously, thereby assisting in confirming the reaction state of the material.

[0029] Optionally, the temperature processing unit performs temporal encoding on the collected temperature data through a temperature temporal feature encoder to obtain a temperature time series, and then performs cleaning and processing, dealing with missing values and outliers, normalizing the temperature time series vector based on the maximum value, and then applying the normalized data to the temporal feature vector. The normalized feature vector is used for model training and optimization to obtain the feature vector of the temperature time series.

[0030] Through the above solution, the reaction temperature is adjusted in a timely manner through the temperature time change sequence, facilitating the control of the reaction temperature at different material levels.

[0031] Optionally, the state confirmation unit extracts key frames from the video data set collected by the camera at fixed time intervals, preprocesses the extracted key frames by resizing the image, normalizing, and enhancing the contrast, uses the global average pooling layer of a pre-trained CNN model to extract features, and for the extracted sequence of feature vectors, uses a recurrent neural network (RNN) model for sequence modeling to capture the dynamic information in the video, capture the temporal dependence between frames, fuse the static and dynamic features, and obtain the dynamic feature vector of the final video key frame, thereby obtaining the dynamic change feature vector of the material state time series.

[0032] Optionally, the fusion analysis fuses the dynamic change feature vector of the material state time series with the feature vector of the temperature time series to obtain a reaction state-temperature correlation feature vector, and compares it with the preset reaction parameters to determine whether the temperature should be increased or decreased at this time and the control of the increase and decrease of the material in and out.

[0033] Through the above solution, through dual monitoring by a camera and a laser source, a set of monitoring videos is obtained. Through the processing of the monitoring videos, the dynamic reaction characteristics of the material are obtained, the reaction process is confirmed by comparison, the characteristic vectors of the dynamic change of the material state time series and the characteristic vectors of the temperature time series are fused to obtain the reaction state-temperature correlation characteristic vector, and it is compared with the preset reaction parameters, so as to determine whether the temperature should be increased or decreased at this time and the control of the increase and decrease of the material in and out, so as to timely monitor the material reaction and ensure the stable and proper progress of the reaction.

[0034] A chemical machinery system with material level monitoring includes the following steps:

[0035] S1. After the material is added to the reaction kettle, the electric heating wire is heated and raised in temperature according to the preset process. The laser source is irradiated on the material, and the position of the laser bright spot on the image is confirmed by camera shooting. The position and range of the bright spot are confirmed by the gray-scale change in the shooting image, that is, if the difference between the gray-scale value and the surrounding gray-scale is greater than the preset difference value, it is the range of the laser bright spot area. The width of the bright spot is confirmed according to the position and range of the laser bright spot on the image. Based on the constant position of the laser source, there is a one-to-one correspondence between the distance between the top of the material height and the laser source and the width of the bright spot. Therefore, the material height is obtained according to the width of the bright spot, so as to confirm the material height and volume data in the reaction kettle, so as to facilitate timely feedback of the material level when controlling the increase and decrease of the material.

[0036] S2. The collected temperature data is encoded in time series through a temperature time series feature encoder to obtain a temperature time series, and then it is cleaned and processed to handle missing values and outliers. The temperature time series vector is normalized based on the maximum value, and then the normalized data is applied to the time series feature vector. The normalized feature vector is used for model training and optimization to obtain the feature vector of the temperature time series.

[0037] S3. The video data set collected by the camera is used to extract key frames at fixed time intervals. The extracted key frames are preprocessed by adjusting the image size, normalizing, and enhancing the contrast. The global average pooling layer of the pre-trained CNN model is used to extract features. For the extracted sequence of feature vectors, a recurrent neural network (RNN) model is used for sequence modeling to capture the dynamic information in the video and capture the temporal dependence relationship between frames. The static features and dynamic features are fused to obtain the dynamic feature vector of the final video key frame, so as to obtain the characteristic vector of the dynamic change of the material state time series.

[0038] S4. The characteristic vector of the dynamic change of the material state time series and the characteristic vector of the temperature time series are fused to obtain the reaction state-temperature correlation characteristic vector, and it is compared with the preset reaction parameters, so as to determine whether the temperature should be increased or decreased at this time and the control of the increase and decrease of the material in and out, so as to timely monitor the material reaction and ensure the stable and proper progress of the reaction.

[0039] S5. The tail gas of the reaction is discharged after treatment, and the tail gas and the reaction pressure are monitored synchronously to ensure the stability and environmental protection of the reaction.

[0040] 3. Beneficial effects

[0041] One or more technical solutions provided in the technical solution of the present application have at least the following technical effects or advantages:

[0042] 1. Through the dual monitoring of the camera and the laser source in the present application, the material level can be confirmed both by the change of the laser source bright spot and the range, and at the same time, according to the gray change of the laser source irradiation and the image processing of the camera, the image feature contrast corresponding to the material reaction process can be confirmed synchronously, so as to assist in confirming the material reaction state, and the monitoring data is more timely and accurate;

[0043] 2. The collected temperature data is encoded in time series by the temperature time series feature encoder to obtain the temperature time series, and then it is cleaned and processed to handle missing values and outliers. The temperature time series vector is normalized based on the maximum value, and then the normalized data is applied to the time series feature vector. The normalized feature vector is used for model training and optimization to obtain the feature vector of the temperature time series. The video data set collected by the camera is used to extract key frames at fixed time intervals, and the extracted key frames are preprocessed by resizing the image, normalizing, and enhancing the contrast. The global average pooling layer of the pre-trained CNN model is used to extract features. For the extracted feature vector sequence, the recurrent neural network RNN model is used for sequence modeling to capture the dynamic information in the video and capture the time-dependent relationship between frames. The static features and dynamic features are fused to obtain the dynamic feature vector of the final video key frame, so as to obtain the dynamic change feature vector of the material state time series. The dynamic change feature vector of the material state time series is fused with the feature vector of the temperature time series to obtain the reaction state-temperature correlation feature vector, which is compared with the preset reaction parameters to determine whether the temperature should be increased or decreased at this time and the control of the increase and decrease of the material in and out, so as to monitor the material reaction in time and ensure the stable and good progress of the reaction. Description of the drawings

[0044] Figure 1 It is a schematic diagram of a chemical machinery system with material level monitoring disclosed in a preferred embodiment of the present application;

[0045] Figure 2 It is a schematic diagram of the monitoring module disclosed in a preferred embodiment of the present application;

[0046] Figure 3 It is a schematic diagram of the tail gas treatment module disclosed in a preferred embodiment of the present application;

[0047] Figure 4Schematic diagram of the processing and analysis module disclosed in a preferred embodiment of the present application;

[0048] Figure 5 Schematic diagram of the material level calculation unit disclosed in a preferred embodiment of the present application;

[0049] Figure 6 Schematic diagram of the status confirmation unit disclosed in a preferred embodiment of the present application; Detailed implementation manners

[0050] The present application will be further described in detail below with reference to the accompanying drawings of the specification.

[0051] Referring to Figure 1 , an embodiment of the present application provides a chemical machinery system with material level monitoring, including: an exhaust gas treatment module, the exhaust gas treatment module is connected to the oil and gas outlet of the reaction kettle and is used for the treatment of reaction oil and gas; a monitoring module, the monitoring module is connected to the reaction kettle and the exhaust gas treatment module and is used for detecting the material level, reaction state, temperature change and exhaust gas data in the reaction; a processing and analysis module, the processing and analysis module is used for processing and analyzing the data collected by the monitoring module, determining the material state during the reaction and timely adjusting the feeding and discharging and temperature data; a control module, the control module is used for timely adjusting the system parameters according to the analysis result of the processing and analysis module; a central processing unit, the central processing unit is used for data transmission between modules and support for processing, calculation and storage; a user interface, the user interface is used for displaying system parameters and analysis results, real-time monitoring of the material level, temperature and material reaction state data in the reaction kettle through the monitoring module, obtaining the reaction state-temperature correlation feature vector through processing and analysis, and comparing it with the preset reaction parameters, so as to determine whether the temperature should be increased or decreased and the control of the increase and decrease of the material inlet and outlet at this time, so as to timely monitor the material reaction and ensure the stable and good progress of the reaction.

[0052] Referring to Figure 1 and Figure 3 , the exhaust gas treatment module includes at least two coolers, the multiple coolers are connected in sequence, and the inlet of the frontmost cooler is connected to the oil and gas outlet of the reaction kettle through a pipeline, the outlet of the end cooler is connected to the inlet of the filter through a pipeline, the liquid outlets of the coolers are all connected to the liquid collector through pipelines, the outlet of the filter is connected to the discharge port, and a valve is installed at the oil and gas outlet of the reaction kettle. The condensable components in the exhaust gas are cooled and collected by the multiple coolers, and the non-condensable gas is discharged after being filtered by the filter, ensuring the cleanliness of the gas discharged into the air and avoiding environmental pollution.

[0053] Further, the cooler uses circulating cooling water, the water temperature is room temperature, and the flow rate is 0.5-1 m / s.

[0054] Optionally, the monitoring module includes: a camera and multiple laser sources, both the camera and the laser sources are installed at the top of the inner cavity of the reactor for collecting material data inside the reactor; a thermometer installed inside the reactor for collecting temperature data inside the reactor; a pressure sensor installed at the top of the reactor for monitoring the pressure inside the reactor; a gas flow meter installed on the pipeline between the filter and the discharge port for monitoring the flow rate of the discharged gas. Through the combined monitoring of the laser source and the camera, the material level and the corresponding reaction process can be monitored simultaneously based on the image and brightness change of the material. The pressure sensor and the gas flow meter can adjust the discharge of the tail gas and control the pressure in the gas flow channel, so as to keep the pressure inside the reactor stable.

[0055] Referring to Figure 1 , a pressure relief valve is also installed at the top of the reactor for pressure relief and pressure adjustment. An electric heating wire is installed inside the reactor. A feed pump for feeding and discharging materials is also installed on the side and bottom of the reactor. The pressure relief valve, the electric heating wire and the feed pump are all electrically connected to the control module. The pressure safety is ensured through the pressure relief valve. The electric heating wire is used to control the temperature inside the reactor. The feed pump is used to control the height of the material level.

[0056] Referring to Figure 4 , the processing and analysis module includes: a material level calculation unit for calculating the height of the material to obtain the volume data of the material; a temperature processing unit for processing the temperature data to obtain the eigenvector of the temperature time series; a state confirmation unit for confirming the process of the material reaction state and extracting the eigenvector of the dynamic change of the material state time series; a fusion analysis unit for performing fusion analysis on the eigenvector of the temperature time series and the eigenvector of the dynamic change of the material state time series to confirm whether the temperature should increase or decrease at the current time point and the control of the feeding and discharging of the material. Through the processing of the temperature time series, the calculation of the material level of the material and the confirmation of the material reaction state, the characteristic relationship between the dynamic reaction of the material and the dynamic change of the temperature is established, so as to control the material and the temperature and ensure the reaction process.

[0057] Referring to Figure 4 and Figure 5, the material calculation unit includes highlight confirmation, width calculation, and height calculation. For highlight confirmation, a laser source irradiates the material, and a camera captures an image to confirm the position of the laser highlight on the image. The position and range of the highlight are confirmed by the change in grayscale in the captured image. That is, if the difference between the grayscale value and the surrounding grayscale is greater than a preset difference value, it is the range of the laser highlight area, and the grayscale difference value is 20 - 40. Width calculation determines the highlight width based on the position and range of the laser highlight on the image. For height calculation, when the position of the laser source is constant, there is a one-to-one correspondence between the distance from the top of the material height to the laser source and the highlight width. Thus, the material height is obtained based on the highlight width, and the height and volume data of the material in the reaction kettle are confirmed. Through the dual monitoring of the camera and the laser source, not only can the material level be confirmed by the change and range of the laser source highlight, but also, based on the change in grayscale irradiated by the laser source and the image processing of the captured image, the image feature comparison corresponding to the reaction process of the material can be synchronously confirmed, thereby assisting in confirming the reaction state of the material.

[0058] Refer to Figure 4 , the temperature processing unit performs time series encoding on the collected temperature data through a temperature time series feature encoder to obtain a temperature time series, and then performs cleaning and processing, dealing with missing values and outliers, normalizing the temperature time series vector based on the maximum value, and then applying the normalized data to the time series feature vector. The normalized feature vector is used for model training and optimization to obtain the feature vector of the temperature time series. The reaction temperature is adjusted in a timely manner according to the temperature time change series, facilitating the control of the reaction temperature at different material levels.

[0059] Refer to Figure 4 and Figure 6 , the status confirmation unit extracts key frames from the video data set collected by the camera at fixed time intervals, preprocesses the extracted key frames by resizing the image, normalizing, and enhancing the contrast, uses the global average pooling layer of a pre-trained CNN model to extract features, and for the extracted sequence of feature vectors, uses a recurrent neural network (RNN) model for sequence modeling to capture the dynamic information in the video and the temporal dependence relationship between frames, fuses the static features and dynamic features to obtain the dynamic feature vector of the final video key frame, thereby obtaining the dynamic change feature vector of the material status time series. Through the dual monitoring of the camera and the laser source, a monitoring video set is obtained, and through the processing of the monitoring video, the dynamic reaction features of the material are obtained to compare and confirm the reaction process.

[0060] Refer to Figure 4 , the fusion analysis fuses the dynamic change feature vector of the material status time series with the feature vector of the temperature time series to obtain a reaction state - temperature correlation feature vector, and compares it with the preset reaction parameters to determine whether the temperature should be increased or decreased at this time and the control of the increase and decrease of the material in and out.

[0061] A chemical machinery system with level monitoring includes the following steps:

[0062] S1. After the material is added to the reaction kettle, the electric heating wire is controlled to heat up according to the preset process. The laser source irradiates the material, and the position of the laser bright spot on the image is confirmed by the camera. The position and range of the bright spot are confirmed by the change in gray scale in the captured image, that is, if the difference between the gray scale value and the surrounding gray scale is greater than the preset difference value, it is the range of the laser bright spot area. The width of the bright spot is confirmed according to the position and range of the laser bright spot on the image. Based on the constant position of the laser source, there is a one-to-one correspondence between the distance between the top of the material height and the laser source and the width of the bright spot. Therefore, the height of the material is obtained according to the width of the bright spot, and the height and volume data of the material in the reaction kettle are confirmed, so as to facilitate timely feedback of the material level when controlling the increase and decrease of the material;

[0063] S2. The collected temperature data is encoded in time series by the temperature time series feature encoder to obtain the temperature time series, and then it is cleaned and processed to handle missing values and outliers. The temperature time series vector is normalized based on the maximum value, and then the normalized data is applied to the time series feature vector. The normalized feature vector is used for model training and optimization to obtain the feature vector of the temperature time series;

[0064] S3. The video data set collected by the camera is used to extract key frames at fixed time intervals. The extracted key frames are preprocessed by adjusting the image size, normalizing, and enhancing the contrast. The global average pooling layer of the pre-trained CNN model is used to extract features. For the extracted feature vector sequence, the recurrent neural network RNN model is used for sequence modeling to capture the dynamic information in the video and capture the time-dependent relationship between frames. The static features and dynamic features are fused to obtain the dynamic feature vector of the final video key frame, so as to obtain the dynamic change feature vector of the material state time series;

[0065] S4. The dynamic change feature vector of the material state time series is fused with the feature vector of the temperature time series to obtain the reaction state-temperature correlation feature vector, which is compared with the preset reaction parameters, so as to determine whether the temperature should increase or decrease at this time and the control of the increase and decrease of the material in and out, so as to monitor the material reaction in time and ensure the stable and good progress of the reaction;

[0066] S5. The tail gas of the reaction is discharged after treatment, and the tail gas and the reaction pressure are monitored synchronously to ensure the stable and environmental protection of the reaction.

[0067] Working principle: After the material is added to the reaction kettle, the heating wire is heated and the temperature is increased according to the preset process. The laser source irradiates the material, and the position of the laser spot on the image is confirmed by the camera. The position and range of the spot are confirmed by the gray-scale change in the captured image, that is, if the difference between the gray-scale value and the surrounding gray-scale is greater than the preset difference, it is the range of the laser spot area. The width of the spot is confirmed according to the position and range of the laser spot on the image. Based on the constant position of the laser source, there is a one-to-one correspondence between the distance between the top of the material height and the laser source and the width of the spot. Therefore, the material height is obtained according to the width of the spot, and the material height and volume data in the reaction kettle are confirmed, so as to facilitate timely feedback of the material level when controlling the increase and decrease of the material. The collected temperature data is encoded in time series by the temperature time series feature encoder to obtain the temperature time series, and then it is cleaned and processed to handle missing values and outliers. The temperature time series vector is normalized based on the maximum value, and then the normalized data is applied to the time series feature vector. The normalized feature vector is used for model training and optimization to obtain the feature vector of the temperature time series. The video data set collected by the camera is used to extract key frames at fixed time intervals. The extracted key frames are preprocessed by adjusting the image size, normalizing, and enhancing the contrast. The global average pooling layer of the pre-trained CNN model is used to extract features. For the extracted feature vector sequence, the recurrent neural network (RNN) model is used for sequence modeling to capture the dynamic information in the video and capture the time dependence between frames. The static features and dynamic features are fused to obtain the dynamic feature vector of the final video key frame, so as to obtain the dynamic change feature vector of the material state time series. The dynamic change feature vector of the material state time series is fused with the feature vector of the temperature time series to obtain the reaction state-temperature correlation feature vector, which is compared with the preset reaction parameters to determine whether the temperature should be increased or decreased at this time and the control of the increase and decrease of the material in and out, so as to monitor the material reaction in time and ensure the stable and perfect progress of the reaction. The tail gas of the reaction is discharged after treatment, and the tail gas and the reaction pressure are monitored synchronously to ensure the stable and environmental protection of the reaction.

Claims

1. A chemical machinery system with material level monitoring, characterized in that: Include: An exhaust gas treatment module, the exhaust gas treatment module is connected to the oil and gas outlet of the reactor and is used for processing the reaction oil and gas; The tail gas treatment module comprises at least two coolers, and the multiple coolers are connected in sequence, and the inlet of the front cooler is connected to the oil and gas outlet of the reactor through a pipeline, and the outlet of the end cooler is connected to the inlet of the filter through a pipeline, and the liquid outlets of the coolers are connected to the liquid collector through pipelines, and the outlet of the filter is connected to the discharge port, and a valve is installed at the oil and gas outlet of the reactor; The condensable components in the tail gas are cooled and collected through multiple coolers, and the non-condensable gas is filtered through filters before being discharged; A monitoring module, which is connected to the reactor and the tail gas treatment module and is used to detect the material level, reaction state, temperature change and tail gas data during the reaction; A processing and analysis module, which is used to process and analyze the data collected by the monitoring module, determine the material state during the reaction, and adjust the input and output materials and temperature data in a timely manner; A control module, the control module is used to timely adjust system parameters according to the analysis results of the processing and analysis module; Central processing unit, which is used for data transmission between modules and processing, calculation and storage support; A user interface is used to display system parameters and analysis results.

2. A chemical machinery system with material level monitoring according to claim 1, characterized in that: The monitoring module includes: A camera and a plurality of laser sources are installed on the top of the inner cavity of the reactor to collect material data in the reactor.

3. A chemical machinery system with material level monitoring according to claim 2, characterized in that: The monitoring module also includes: A thermometer, which is installed in the reactor and is used to collect temperature data in the reactor; A pressure sensor, which is installed on the top of the reactor and is used to monitor the pressure inside the reactor; A gas flow meter, which is installed on the pipeline between the filter and the exhaust port and is used to monitor the flow rate of the exhaust gas; Through the combined monitoring of the laser source and the camera, the material level and the corresponding reaction progress can be monitored simultaneously based on the image and brightness changes of the material. The pressure sensor and the gas flow meter can adjust the exhaust gas emission and control the pressure of the gas flow channel, thereby keeping the pressure in the reactor stable.

4. A chemical machinery system with material level monitoring according to claim 3, characterized in that: A pressure relief valve is also installed on the top of the reactor for pressure relief and pressure adjustment. A heating wire is installed in the reactor. A material pump connected to the inlet and outlet of the material is also installed on the side and bottom of the reactor. The pressure relief valve, the heating wire and the material pump are all electrically connected to the control module.

5. A chemical machinery system with material level monitoring according to claim 3, characterized in that: The processing and analysis module comprises: A material level calculation unit, which is used to calculate the material height, thereby obtaining material volume data; A temperature processing unit, wherein the temperature processing unit is used to process the temperature data to obtain a characteristic vector of a temperature time series; A state confirmation unit, which is used to confirm the material reaction state process and extract the material state time series dynamic change feature vector; The fusion analysis unit is used to fuse and analyze the characteristic vector of the temperature time series and the characteristic vector of the dynamic change of the material state time series to confirm whether the temperature at the current time point should be increased or decreased and the material inlet and outlet control.

6. A chemical machinery system with material level monitoring according to claim 5, characterized in that: The material calculation unit includes bright spot confirmation, width calculation and height calculation. The bright spot confirmation is performed by irradiating the material with a laser source and confirming the laser bright spot position on the image through a camera. The bright spot position and range are confirmed by the grayscale change in the camera image. That is, if the difference between the grayscale value and the surrounding grayscale is greater than a preset difference, it is the laser bright spot area range, and the grayscale difference is 20-40; The width calculation determines the bright spot width according to the laser bright spot position and range on the image; The height calculation is based on the fact that the laser source position is constant, and the material height top, the distance between the laser source and the bright spot width have a one-to-one correspondence, so that the material height is obtained according to the bright spot width, thereby confirming the material height and volume data in the reactor.

7. A chemical machinery system with material level monitoring according to claim 5, characterized in that: The temperature processing unit performs time series encoding on the collected temperature data through a temperature time series feature encoder to obtain a temperature time series, and then performs cleaning and processing to handle missing values ​​and abnormal values, and normalizes the temperature time series vector based on the maximum value, and then applies the normalized data to the time series feature vector, and uses the normalized feature vector for model training and optimization to obtain a feature vector of the temperature time series.

8. A chemical machinery system with material level monitoring according to claim 5, characterized in that: The state confirmation unit extracts key frames from the video data set collected by the camera according to a fixed time interval, performs preprocessing of adjusting image size, normalization, and contrast enhancement on the extracted key frames, uses the global average pooling layer of the pre-trained CNN model to extract features, and uses a recurrent neural network (RNN) model to perform sequence modeling on the extracted feature vector sequence to capture dynamic information in the video and the time dependency between frames, fuses static features and dynamic features, and obtains the dynamic feature vector of the final video key frame, thereby obtaining the feature vector of the material state time series dynamic change.

9. A chemical machinery system with material level monitoring according to claim 8, characterized in that: The fusion analysis fuses the material state time series dynamic change feature vector with the temperature time series feature vector to obtain the reaction state-temperature correlation feature vector, and compares it with the preset reaction parameters to determine whether the temperature should be increased or decreased and the material inlet and outlet increase and decrease control.

10. A chemical machinery system with material level monitoring according to any one of claims 1 to 9, characterized in that: The following steps are involved: S1. After the material is added to the reactor, the heating wire is heated and heated according to the preset process control, and the laser source is irradiated on the material. The position of the laser bright spot on the image is confirmed by the camera, and the position and range of the bright spot are confirmed by the grayscale change in the camera image, that is, the difference between the grayscale value and the surrounding grayscale is greater than the preset difference, which is the laser bright spot area range. The bright spot width is confirmed according to the position and range of the laser bright spot on the image. Based on the constant position of the laser source, the material height top and the distance between the laser source and the bright spot width have a one-to-one correspondence, so the material height is obtained according to the bright spot width, thereby confirming the material height and volume data in the reactor, so as to facilitate the timely feedback of the material level when controlling the increase or decrease of the material; S2. The collected temperature data is time-series encoded through a temperature time series feature encoder to obtain a temperature time series, and then cleaned and processed to handle missing values ​​and abnormal values, and the temperature time series vector is normalized based on the maximum value, and then the normalized data is applied to the time series feature vector, and the normalized feature vector is used for model training and optimization to obtain the feature vector of the temperature time series; S3. Extract key frames from the video data set collected by the camera according to fixed time intervals, perform preprocessing of adjusting image size, normalization, and contrast enhancement on the extracted key frames, use the global average pooling layer of the pre-trained CNN model to extract features, and use the recurrent neural network RNN ​​model to perform sequence modeling on the extracted feature vector sequence to capture dynamic information in the video and the time dependency between frames, fuse static features and dynamic features, and obtain the dynamic feature vector of the final video key frame, thereby obtaining the feature vector of the material state time series dynamic change; S4. The dynamic change feature vector of the material state time series is integrated with the feature vector of the temperature time series to obtain the reaction state-temperature correlation feature vector, and compared with the preset reaction parameters to determine whether the temperature should be increased or decreased and the increase and decrease control of the material in and out, so as to monitor the material reaction in time and ensure the stable and intact reaction; S5. The tail gas of the reaction is discharged through treatment, and the tail gas and reaction pressure are monitored simultaneously to ensure a stable and environmentally friendly reaction.

Images