Water-encountering deflation tester

Abstract

The invention provides a water deflation tester, and relates to the technical field of testers, the water deflation tester comprises a tester body and a temperature sensor, an experiment table is fixedly connected in the tester body, a display screen is fixedly connected to the inner wall of a groove formed in the side wall of one side of the tester body, and the temperature sensor is fixedly connected to the inner wall of the tester body. The device further comprises a reaction space module; an automatic water injection module; an automatic gas collection module; an interaction module; a temperature control module; according to the invention, a prediction model is trained through an LSTM algorithm, the current reaction trend is analyzed in real time, and the device reduces the situation that when an abnormality occurs, the abnormality can be found only by observing a display screen or analyzing the data afterwards, so that the detection efficiency is improved. The probability of failure of the whole group of test data or out-of-control reaction process is reduced, redundant information interference is reduced, and the model efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of testing instruments, in particular to a water-releasing testing instrument. Background Art

[0002] A water outgassing tester is an instrument used to detect whether a material or equipment will produce gas when it comes into contact with water. This test is usually used to evaluate the waterproof performance of a material or to detect whether certain chemical reactions will produce gas after contact with water.

[0003] When using the existing water-release tester, first check whether the tester body is sealed intact, without cracks or leaks, and calibrate the sensors at the same time. Use a standard pressure source to calibrate the pressure sensor, verify the accuracy of the temperature sensor through a constant temperature bath, and calibrate the gas flow meter with a soap film flow meter. Prepare the sample according to the standard, weigh and record the sample mass, and ensure that the test environment temperature is constant to avoid strong airflow in the fume hood interfering with gas collection. Connect the data acquisition system, set the sampling frequency, confirm that the display screen and interactive module are functioning normally, place the pretreated sample in the reaction vessel, fix the sealing cover, connect the water injection pipe, pressure sensor and gas collection device, check the air tightness of each interface to ensure that there is no gas leakage, and then use a peristaltic pump or precision syringe to pump gas into the reaction vessel at a set rate. Inject water and start the data acquisition system at the same time, record the start time of water injection, record key parameters during the test, and after the test, when the pressure curve tends to be stable or reaches the preset reaction time, stop water injection, close all valves, and then export the time series data in the data acquisition system, check the data integrity, remove abnormal data points, use linear interpolation to fill in short-term missing data, and finally calculate the key features, draw the "time-rate curve", analyze the reaction kinetics, and combine the Pearson correlation coefficient to screen features with high correlation with reaction trends. The "passive collection + manual judgment" method of traditional testers requires manual observation of the display screen or subsequent data analysis to discover abnormalities when they occur, which may cause the entire set of test data to fail or the reaction process to get out of control.

[0004] Therefore, it is necessary to provide a new water-release tester to solve the above-mentioned technical problems. Summary of the Invention

[0005] In order to solve the above technical problems, the present invention provides a water release tester.

[0006] The present invention provides a water-degassing tester comprising a tester body and a temperature sensor. A laboratory table is fixedly connected to the interior of the tester body, and a display screen is fixedly connected to the inner wall of a groove formed in a side wall of the tester body. The temperature sensor is fixedly connected to the inner wall of the tester body. The tester is characterized in that the tester further comprises a reaction space module for providing a reaction site and accommodating water-degassing reactants, so that the reaction can proceed within a specific space.

[0007] Automatic water injection module, used to inject water into the reaction space module;

[0008] The automatic gas collection module is responsible for collecting the gas generated by the water-releasing reactants and maintaining the gas pressure in the reaction space at the target value;

[0009] A control module, used to control the operation of the automatic water injection module and the automatic gas collection module;

[0010] An interactive module, used for displaying the test results through a display screen;

[0011] Temperature control module, used to ensure that the test substance reacts with water at a specific temperature;

[0012] The data processing module is used to process relevant data in the detection process. At the same time, it uses the test data stored in the data processing module to view historical test data or manage saved data. It then trains the prediction model through the LSTM algorithm to analyze the current reaction trend in real time.

[0013] Preferably, the method of training a prediction model by an LSTM algorithm and analyzing the current reaction trend in real time comprises the following steps:

[0014] S1, data acquisition, real-time acquisition of reaction process data from the temperature sensor set in the tester body, and setting high-frequency sampling according to the reaction rate;

[0015] S2. Data cleaning and standardization. Outlier processing: Remove sensor noise or sudden changes through statistical methods or sliding window filtering. Fill missing values ​​using linear interpolation, forward filling, or filling based on the mean of adjacent moments. Standardization / normalization: Convert data of different dimensions to the interval [-1, 1] or [0, 1] to prevent the model from being affected by the feature scale.

[0016] S3, Feature Extraction and Sequence Construction, key feature extraction, extracting the water injection rate, real-time pressure change rate, gas generation rate, and temperature change trend features. The Pearson correlation coefficient is used to screen features with high correlation with the reaction trend. At the same time, time series samples are constructed, dividing historical data into windows of fixed length.

[0017] S4. Design an LSTM network architecture. The network layer configuration includes an input layer, an LSTM layer, and a fully connected layer. The input layer can determine the input dimension based on the number of features. The LSTM layer is designed to have 1-2 layers of LSTM units. Each layer contains an appropriate number of neurons to capture long-term dependencies in the time series. The fully connected layer connects the outputs of the LSTM layer and maps them to the prediction target through an activation function.

[0018] S5. Model training and data set division: divide the data into training set, validation set, and test set in chronological order to prevent future data from leaking into the training phase. Use the test set to calculate indicators and verify the generalization ability of the model.

[0019] S6. Real-time analysis and deployment: deploy the trained model to the data processing module or edge computing device of the test instrument. At the same time, establish a real-time data interface to connect to the sensor data stream, obtain the current reaction data at a fixed frequency, and standardize it according to the preprocessing method during training.

[0020] Preferably, the gas automation collection module includes a gas micro-pressure sensor, a digital-to-analog conversion circuit, a gas collection motor encoder, a gas encoder signal conditioning unit and a gas collection stepper motor, wherein the gas micro-pressure sensor is a sensitive element for measuring tiny gas pressure changes, the digital-to-analog conversion circuit is an electronic circuit that converts digital signals into analog signals, the gas collection motor encoder is a position / speed detection device integrated in the gas collection device drive motor, and converts the motor rotation angle, displacement or speed into an electrical signal, the gas encoder signal conditioning unit is the intermediate link connecting the encoder and the control system, and the gas collection stepper motor is used to accurately control the gas flow, valve opening or sampling pump speed.

[0021] Preferably, the test results include a time-rate curve, a current deflation rate, a current deflation volume, a maximum deflation rate, and the test duration.

[0022] Preferably, the temperature control module includes a heater, a compressor, an air inlet in the constant temperature box and a circulating fan. The heater is used to convert electrical energy, thermal energy or other energy into thermal energy. The compressor is a device used to increase the gas pressure through mechanical work. The air inlet in the constant temperature box is used to affect the key indicators of temperature and humidity uniformity and gas composition stability in the box. The circulating fan is used to achieve temperature uniformity control and airflow organization optimization through forced convection.

[0023] Preferably, the control of the automatic water injection module comprises the following steps:

[0024] S10, system initialization and parameter setting, setting water injection target amount, water injection rate, trigger adjustment, and input parameters through the control display screen, and checking the accuracy of the water level sensor and flow sensor;

[0025] S20, signal detection and triggering, the sensor monitors the liquid level in the container or the system operating status in real time, and sends a signal to the controller when the detection value reaches the preset trigger condition;

[0026] S30, water injection execution control, the controller outputs a command to open the water injection solenoid valve or water pump, adjusts the opening according to the preset rate, and uses the flow sensor to provide real-time feedback on the water injection flow. If the deviation from the set value is large, the controller automatically adjusts the valve opening or pump speed. At the same time, when the liquid level approaches the upper limit, the controller gradually reduces the water injection rate. When the upper limit is reached, the valve / pump is immediately closed to avoid overflow;

[0027] S40, abnormality handling and alarm, monitoring abnormalities such as water pump overload, valve stagnation, sensor failure, etc., alarming through indicator lights or system logs. At the same time, in case of serious faults, the controller forcibly cuts off the water injection circuit and triggers sound and light alarms;

[0028] S50, data recording and end, water injection data storage, recording water injection time, total amount and rate. After water injection is completed, the system resets to standby state and waits for the next trigger.

[0029] Preferably, the operation of the gas automatic collection module component includes the following steps:

[0030] S100: Preparation and parameter setting before collection: determine the gas collection volume, collection method and gas type, and confirm the collection capacity sealing, pressure sensor range, and vacuum pump / compressor status;

[0031] S200, gas generation / introduction trigger, source trigger: if the gas comes from the reaction device, when the reaction reaches the gas production condition, the controller opens the gas introduction valve; if it is an external gas source, the introduction is triggered by the pressure switch or flow sensor, and the controller starts the collection process after receiving the gas production signal or manual start signal;

[0032] S300, gas collection execution control, switches the gas flow direction through the solenoid valve, and enters the collection container through the drying tube and filter. The pressure sensor monitors the pressure in the container in real time. When it exceeds the set value, the controller adjusts the pressure relief valve or gas source valve to maintain stable pressure. If the gas collection is sensitive to temperature, the controller adjusts the collection parameters based on the temperature sensor data;

[0033] S400, safety and abnormality control, overpressure protection. When the container pressure exceeds the safety threshold, the emergency pressure relief valve automatically opens and stops the gas supply to prevent the container from exploding. The air tightness sensor monitors pipeline leakage and alarms and cuts off the gas supply when an abnormality is found.

[0034] S500, collection completion and subsequent processing: After collection is completed, close the inlet and outlet valves and record the collection time, gas volume, and pressure / temperature data.

[0035] Compared with related technologies, the water-release tester provided by the present invention has the following beneficial effects:

[0036] The present invention trains a prediction model through the LSTM algorithm and analyzes the current reaction trend in real time. First, the reaction process data is obtained in real time from the temperature sensor and other equipment set in the test instrument body, and high-frequency sampling is set according to the reaction rate to improve data accuracy. Then, statistical methods or sliding window filtering are used to eliminate sensor noise and abnormal mutation values. Missing data are supplemented by linear interpolation, forward filling or adjacent time mean filling. Then, key features such as water injection rate, real-time pressure change rate, gas generation rate, temperature change trend, etc. are extracted, and the Pearson correlation coefficient is used to screen out variables that are highly correlated with the reaction trend. At the same time, historical data are used to calculate the correlation coefficient of the reaction process. The system divides the time into fixed-length time windows, constructs sequence samples suitable for LSTM input, and then designs the LSTM network architecture. Finally, the trained LSTM model is deployed to the data processing module or edge computing device of the test instrument, and a real-time data interface is established to connect to the sensor data stream. The current reaction data is obtained at a fixed frequency and input into the model after standardization according to the preprocessing method of the training phase. This device reduces the need for manual observation of display screens or subsequent data analysis when an anomaly occurs, which may cause the entire set of test data to fail or the reaction process to get out of control. It also reduces redundant information interference and improves model efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 A structural block diagram of a water-releasing tester provided by the present invention;

[0038] Figure 2 A flowchart of the process of training a prediction model using the LSTM algorithm provided by the present invention;

[0039] Figure 3 A flowchart of the control of the automatic water injection module provided by the present invention;

[0040] Figure 4 This is a flowchart of the working process of the gas automatic collection module components provided by the present invention. DETAILED DESCRIPTION

[0041] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0042] Please refer to Figure 1 、 Figure 2 、 Figure 3 as well as Figure 4 ,in, Figure 1 A structural block diagram of a water-releasing tester provided by the present invention; Figure 2 A flowchart of the process of training a prediction model using the LSTM algorithm provided by the present invention; Figure 3 A flowchart of the control of the automatic water injection module provided by the present invention; Figure 4This is a flowchart of the working process of the gas automatic collection module components provided by the present invention.

[0043] In the specific implementation process, Figures 1 to 4 As shown, the apparatus comprises a tester body and a temperature sensor. A laboratory table is fixedly connected to the interior of the tester body, and a display screen is fixedly connected to the inner wall of a groove provided in a side wall of the tester body. The temperature sensor is fixedly connected to the inner wall of the tester body. The apparatus also comprises a reaction space module for providing a reaction site and accommodating reactants that release gas when exposed to water, so that the reaction can proceed within a specific space.

[0044] It should be noted that the reaction space module in the water outgassing tester is the core area where the sample contacts water and triggers the gas production reaction. Its design must take into account sealing, controllability and safety.

[0045] Moreover, the core components of the reaction space module include the main container structure, sealing and safety components, sample and water contact control mechanism, and sensor and monitoring system. Among them, the material selection of the main container structure: corrosion resistance, using 316L stainless steel, polytetrafluoroethylene (PTFE) or glass to prevent the corrosive gas generated by the reaction of samples (such as metals, chemicals) and water from eroding the container; pressure resistance, if the reaction may produce high-pressure gas, the container must meet the pressure requirements (such as design pressure ≥ 1.5 times the maximum expected pressure) and pass the pressure test (such as water pressure test);

[0046] The sealing structure of the sealing and safety components includes flange sealing and dynamic sealing. The flange sealing adopts metal spiral wound gaskets or O-rings (made of fluororubber) and is evenly tightened with bolts to ensure airtightness. Dynamic sealing, such as the agitator shaft seal, adopts a magnetic coupling to avoid gas leakage when the shaft rotates. The safety components include pressure relief valves, explosion-proof membranes and temperature fuses. Among them, the pressure relief valve: sets the opening pressure, and automatically exhausts when the pressure in the reaction space exceeds the threshold to prevent explosion. The explosion-proof membrane: serves as auxiliary protection for the pressure relief valve, and ruptures to release pressure when overpressure occurs. The rupture pressure needs to be accurately calibrated. The temperature fuse: if the reaction exothermicity causes the temperature to be too high, the fuse will melt and cut off the heating power supply;

[0047] The triggering methods of the sample and water contact control mechanism include automatic triggering and manual triggering. Automatic triggering: PLC controls the solenoid valve or stepper motor to release water according to a preset program (such as timing, liquid level signal) to achieve automatic reaction. Manual triggering: manually release water to the sample area through a knob or lever mechanism, which is suitable for simple laboratory devices. The contact area adjustment method includes sample tray design and water flow rate control. The sample tray design: can be raised or rotated to adjust the contact area between the sample and water and control the reaction rate. Water flow rate control: the water injection rate is adjusted by a flow meter and a solenoid valve to avoid splashing caused by violent reaction.

[0048] The key sensors in the sensor and monitoring system include pressure sensors, temperature sensors, and liquid level sensors. The pressure sensor has a range of 0-1MPa (adjusted according to the reaction gas pressure) and an accuracy of ±0.5% FS, which monitors the pressure changes in the reaction space in real time.

[0049] Temperature sensor: PT100 platinum resistance or thermocouple, temperature measurement range -20℃~200℃, accuracy ±0.5℃, used to monitor reaction temperature (exothermic reaction requires special monitoring);

[0050] Liquid level sensor: ultrasonic or capacitive, monitors the water level in the reaction space to prevent overfilling;

[0051] Automatic water injection module, used to inject water into the reaction space module;

[0052] It should be noted that the automatic water injection module includes a water injection motor encoder, a water injection encoder signal conditioning unit, a water injection stepping motor and a water injection motor zero position;

[0053] The water injection motor encoder is connected to the control module through the water injection encoder signal conditioning unit;

[0054] The water injection stepping motor and the water injection motor zero position are respectively connected to the control module;

[0055] The automatic gas collection module is responsible for collecting the gas generated by the water-releasing reactants and maintaining the gas pressure in the reaction space at the target value;

[0056] It should be noted that the automatic gas collection module in the water outgassing tester is the core component for achieving accurate collection, measurement and analysis of reaction gases. Its design must meet the requirements of "air tightness, measurement accuracy, and anti-interference" to ensure the reliability of gas volume and composition data;

[0057] The gas automation collection module includes a gas micro-pressure sensor, a digital-to-analog conversion circuit, a gas collection motor encoder, a gas encoder signal conditioning unit, and a gas collection stepper motor. The gas micro-pressure sensor is a sensitive element used to measure tiny gas pressure changes. The digital-to-analog conversion circuit is an electronic circuit that converts digital signals into analog signals. The gas collection motor encoder is a position / speed detection device integrated into the gas collection device drive motor, and converts the motor rotation angle, displacement, or speed into an electrical signal. The gas encoder signal conditioning unit is the intermediate link between the encoder and the control system. The gas collection stepper motor is used to accurately control the gas flow, valve opening, or sampling pump speed.

[0058] Maintaining the target air pressure value includes the following steps:

[0059] Step 1: Collect air pressure data in the reaction space at a high frequency (e.g., 1 time / second) to ensure that small fluctuations are captured in a timely manner, and the sensor converts the air pressure signal into an electrical signal and transmits it to the controller via a cable or wirelessly;

[0060] Step 2: Data processing and deviation judgment: The controller receives the real-time air pressure value, compares it with the preset target value, and calculates the deviation value;

[0061] Step 3: Execution of air pressure regulation. Adjustment when the air pressure is too high: the controller sends an opening signal to the exhaust valve, and the gas is discharged into the collection tank or external safe area through the pipeline. At the same time, the gas buffer tank absorbs part of the gas, slowing down the exhaust speed and avoiding a sudden drop in air pressure. The pressure sensor continuously monitors. When the air pressure approaches the target value, the controller gradually closes the exhaust valve. Adjustment when the air pressure is too low: the controller starts the gas supply device (such as the inert gas cylinder pressure reducing valve) to fill the reaction space with gas, and the gas buffer tank releases the reserve gas to assist in quickly increasing the air pressure. When it approaches the target value, the controller reduces the gas supply flow until the air pressure stabilizes.

[0062] Step 4: Closed-loop feedback control: Through the closed-loop process of "monitoring-judgment-execution", the pressure deviation is continuously corrected to form adaptive control;

[0063] Step 5: Safety protection and abnormal handling: When the air pressure exceeds a certain threshold (such as 10%) of the target value, the emergency exhaust valve is forced to open to prevent equipment damage and an underpressure alarm is sounded. When the air pressure continues to be lower than the target value, the system alarms and suspends the reaction (such as stopping water injection) to check for leaks.

[0064] A control module, used to control the operation of the automatic water injection module and the automatic gas collection module;

[0065] It should be noted that the work of controlling the automatic water injection module includes the following steps:

[0066] S10, system initialization and parameter setting, setting water injection target amount, water injection rate, trigger adjustment, and input parameters through the control display screen, and checking the accuracy of the water level sensor and flow sensor;

[0067] S20, signal detection and triggering, the sensor monitors the liquid level in the container or the system operating status in real time, and sends a signal to the controller when the detection value reaches the preset trigger condition;

[0068] S30, water injection execution control, the controller outputs a command to open the water injection solenoid valve or water pump, adjusts the opening according to the preset rate, and uses the flow sensor to provide real-time feedback on the water injection flow. If the deviation from the set value is large, the controller automatically adjusts the valve opening or pump speed. At the same time, when the liquid level approaches the upper limit, the controller gradually reduces the water injection rate. When the upper limit is reached, the valve / pump is immediately closed to avoid overflow;

[0069] S40, abnormality handling and alarm, monitoring abnormalities such as water pump overload, valve stagnation, sensor failure, etc., alarming through indicator lights or system logs. At the same time, in case of serious faults, the controller forcibly cuts off the water injection circuit and triggers sound and light alarms;

[0070] S50, data recording and end, water injection data storage, recording water injection time, total amount and rate. After water injection is completed, the system resets to standby state and waits for the next trigger.

[0071] The working of the gas automatic collection module components includes the following steps:

[0072] S100: Preparation and parameter setting before collection: determine the gas collection volume, collection method and gas type, and confirm the collection capacity sealing, pressure sensor range, and vacuum pump / compressor status;

[0073] S200, gas generation / introduction trigger, source trigger: if the gas comes from the reaction device, when the reaction reaches the gas production condition, the controller opens the gas introduction valve; if it is an external gas source, the introduction is triggered by the pressure switch or flow sensor, and the controller starts the collection process after receiving the gas production signal or manual start signal;

[0074] S300, gas collection execution control, switches the gas flow direction through the solenoid valve, and enters the collection container through the drying tube and filter. The pressure sensor monitors the pressure in the container in real time. When it exceeds the set value, the controller adjusts the pressure relief valve or gas source valve to maintain stable pressure. If the gas collection is sensitive to temperature, the controller adjusts the collection parameters based on the temperature sensor data;

[0075] S400, safety and abnormality control, overpressure protection. When the container pressure exceeds the safety threshold, the emergency pressure relief valve automatically opens and stops the gas supply to prevent the container from exploding. The air tightness sensor monitors pipeline leakage and alarms and cuts off the gas supply when an abnormality is found.

[0076] S500, collection completion and subsequent processing: After collection is completed, the inlet and outlet valves are closed, and the collection time, gas volume, and pressure / temperature data are recorded;

[0077] An interactive module, used for displaying the test results through a display screen;

[0078] It should be noted that the interactive module of the water-release tester is centered on the display screen, which is responsible for the visualization of test data, operation guidance, and result presentation. Its design must take into account data accuracy, interface usability, and safety.

[0079] The test results include time-rate curve, current deflation rate, current deflation volume, maximum deflation rate and test duration;

[0080] Temperature control module, used to ensure that the test substance reacts with water at a specific temperature;

[0081] The temperature control module includes a heater, a compressor, an air inlet inside the constant temperature box, and a circulating fan. The heater is used to convert electrical energy, thermal energy, or other energy into thermal energy. The compressor is a device used to increase gas pressure through mechanical work. The air inlet inside the constant temperature box is used to influence the key indicators of temperature and humidity uniformity and gas composition stability inside the box. The circulating fan is used to achieve temperature uniformity control and optimize airflow organization through forced convection.

[0082] The data processing module is used to process relevant data during the detection process. It also uses the test data stored in the data processing module to view historical test data or manage saved data. It then uses the LSTM algorithm to train the prediction model and analyze the current reaction trend in real time.

[0083] It should be noted that LSTM (Long Short-Term Memory) is a special recurrent neural network (RNN) structure designed specifically to address the "vanishing gradient" and "exploding gradient" problems encountered by traditional RNNs when processing long time series data. It can effectively capture long-term dependencies in time series and is therefore widely used in natural language processing, speech recognition, time series forecasting, financial modeling and other fields.

[0084] Using the LSTM algorithm to train a prediction model and analyze current reaction trends in real time involves the following steps:

[0085] S1, data acquisition, real-time acquisition of reaction process data from the temperature sensor set in the tester body, and setting high-frequency sampling according to the reaction rate;

[0086] It should be noted that the reaction process data includes: time series data, water injection rate, reaction temperature, pressure change, gas generation, solution color value (through image recognition module), environmental parameters: ambient temperature, humidity (if there is an environmental control module), test configuration parameters: sample type, water injection volume, initial conditions;

[0087] S2. Data cleaning and standardization. Outlier processing: Remove sensor noise or sudden changes through statistical methods or sliding window filtering. Fill missing values ​​using linear interpolation, forward filling, or filling based on the mean of adjacent moments. Standardization / normalization: Convert data of different dimensions to the interval [-1, 1] or [0, 1] to prevent the model from being affected by the feature scale.

[0088] It should be noted that the commonly used method is: minimum-maximum standardization:

[0089]

[0090] Z-score normalization:

[0091]

[0092] S3, Feature Extraction and Sequence Construction, key feature extraction, extracting the water injection rate, real-time pressure change rate, gas generation rate, and temperature change trend features. The Pearson correlation coefficient is used to screen features with high correlation with the reaction trend. At the same time, time series samples are constructed, dividing historical data into windows of fixed length.

[0093] It should be noted that the derived characteristics, pressure-temperature coupling characteristics: P×T (reflecting the intensity of the reaction), gas generation acceleration: △ 2 V / △t 2 (predicting reaction growth rate), historical trend characteristics: mean, variance, and slope of the past n time steps;

[0094] Moreover, each window corresponds to a prediction target which is input and output. Input: [x t-L+1 , x t-L+2 ,…x t ], where L is the eigenvector of the time step;

[0095] Output: x t+1 , the target value for the next time step, such as the pressure value in the next 1 second;

[0096] The Pearson Correlation Coefficient (Pearson Correlation Coefficient), also known as the Pearson product-moment correlation coefficient, is a statistic that measures the degree of linear correlation between two continuous variables. It is represented by the symbol r. In response trend research, it is often necessary to screen out the feature with the strongest linear correlation with the target variable from among many features.

[0097] S4. Design an LSTM network architecture. The network layer configuration includes an input layer, an LSTM layer, and a fully connected layer. The input layer can determine the input dimension based on the number of features. The LSTM layer is designed to have 1-2 layers of LSTM units. Each layer contains an appropriate number of neurons to capture long-term dependencies in the time series. The fully connected layer connects the outputs of the LSTM layer and maps them to the prediction target through an activation function.

[0098] S5. Model training and data set division: divide the data into training set, validation set, and test set in chronological order to prevent future data from leaking into the training phase. Use the test set to calculate indicators and verify the generalization ability of the model.

[0099] S6. Real-time analysis and deployment: deploy the trained model to the data processing module or edge computing device of the test instrument. At the same time, establish a real-time data interface to connect to the sensor data stream, obtain the current reaction data at a fixed frequency, and standardize it according to the preprocessing method during training.

[0100] The circuits and controls involved in the present invention are all prior art and will not be described in detail here.

[0101] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention's description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A water-degassing tester, comprising a tester body and a temperature sensor, wherein a laboratory table is fixedly connected to the interior of the tester body, and a display screen is fixedly connected to the inner wall of a groove provided on one side wall of the tester body, and the temperature sensor is fixedly connected to the inner wall of the tester body, characterized in that: It also includes a reaction space module for providing a reaction site and accommodating reactants that release gas when in contact with water, so that the reaction can proceed in a specific space; Automatic water injection module, used to inject water into the reaction space module; The automatic gas collection module is responsible for collecting the gas generated by the water-releasing reactants and maintaining the gas pressure in the reaction space at the target value; A control module, used to control the operation of the automatic water injection module and the automatic gas collection module; An interactive module, used for displaying the test results through a display screen; Temperature control module, used to ensure that the test substance reacts with water at a specific temperature; The data processing module is used to process relevant data in the detection process. At the same time, it uses the test data stored in the data processing module to view historical test data or manage saved data. It then trains the prediction model through the LSTM algorithm to analyze the current reaction trend in real time.

2. The water release tester according to claim 1, characterized in that: The method of training the prediction model by the LSTM algorithm and analyzing the current reaction trend in real time includes the following steps: S1, data acquisition, real-time acquisition of reaction process data from the temperature sensor set in the tester body, and setting high-frequency sampling according to the reaction rate; S2. Data cleaning and standardization. Outlier processing: Remove sensor noise or sudden changes through statistical methods or sliding window filtering. Fill missing values ​​using linear interpolation, forward filling, or filling based on the mean of adjacent moments. Standardization / normalization: Convert data of different dimensions to the interval [-1, 1] or [0, 1] to prevent the model from being affected by the feature scale. S3, Feature Extraction and Sequence Construction, key feature extraction, extracting the water injection rate, real-time pressure change rate, gas generation rate, and temperature change trend features. The Pearson correlation coefficient is used to screen features with high correlation with the reaction trend. At the same time, time series samples are constructed, dividing historical data into windows of fixed length. S4. Design an LSTM network architecture. The network layer configuration includes an input layer, an LSTM layer, and a fully connected layer. The input layer can determine the input dimension based on the number of features. The LSTM layer is designed to have 1-2 layers of LSTM units, each containing an appropriate number of neurons to capture long-term dependencies in the time series. The fully connected layer connects the outputs of the LSTM layer and maps them to the prediction target through an activation function. S5. Model training and data set division: divide the data into training set, validation set, and test set in chronological order to prevent future data from leaking into the training phase. Use the test set to calculate indicators and verify the generalization ability of the model. S6. Real-time analysis and deployment: deploy the trained model to the data processing module or edge computing device of the test instrument. At the same time, establish a real-time data interface to connect to the sensor data stream, obtain the current reaction data at a fixed frequency, and standardize it according to the preprocessing method during training.

3. The water release tester according to claim 2, characterized in that: The gas automation collection module includes a gas micro-pressure sensor, a digital-to-analog conversion circuit, a gas collection motor encoder, a gas encoder signal conditioning unit and a gas collection stepper motor, wherein the gas micro-pressure sensor is a sensitive element used to measure tiny gas pressure changes, the digital-to-analog conversion circuit is an electronic circuit that converts digital signals into analog signals, the gas collection motor encoder is a position / speed detection device integrated into the gas collection device drive motor, and converts the motor rotation angle, displacement or speed into an electrical signal, the gas encoder signal conditioning unit is the intermediate link connecting the encoder and the control system, and the gas collection stepper motor is used to accurately control the gas flow, valve opening or sampling pump speed.

4. The water release tester according to claim 3, characterized in that: The test results include a time-rate curve, current deflation rate, current deflation volume, maximum deflation rate, and test duration.

5. The water release tester according to claim 4, characterized in that: The temperature control module includes a heater, a compressor, an air inlet in a constant temperature box and a circulating fan. The heater is used to convert electrical energy, thermal energy or other energy into thermal energy. The compressor is a device used to increase gas pressure through mechanical work. The air inlet in the constant temperature box is used to influence the key indicators of temperature and humidity uniformity and gas composition stability in the box. The circulating fan is used to achieve temperature uniformity control and airflow organization optimization through forced convection.

6. The water release tester according to claim 5, characterized in that: The control of the automatic water injection module comprises the following steps: S10, system initialization and parameter setting, setting water injection target amount, water injection rate, trigger adjustment, and input parameters through the control display screen, and checking the accuracy of the water level sensor and flow sensor; S20, signal detection and triggering, the sensor monitors the liquid level in the container or the system operating status in real time, and sends a signal to the controller when the detection value reaches the preset trigger condition; S30, water injection execution control, the controller outputs a command to open the water injection solenoid valve or water pump, adjusts the opening according to the preset rate, and uses the flow sensor to provide real-time feedback on the water injection flow. If the deviation from the set value is large, the controller automatically adjusts the valve opening or pump speed. At the same time, when the liquid level approaches the upper limit, the controller gradually reduces the water injection rate. When the upper limit is reached, the valve / pump is immediately closed to avoid overflow; S40, abnormality handling and alarm, monitoring abnormalities such as water pump overload, valve stagnation, sensor failure, etc., alarming through indicator lights or system logs. At the same time, in case of serious faults, the controller forcibly cuts off the water injection circuit and triggers sound and light alarms; S50, data recording and end, water injection data storage, recording water injection time, total amount and rate. After water injection is completed, the system resets to standby state and waits for the next trigger.

7. The water release tester according to claim 6, characterized in that: The working of the gas automatic collection module component includes the following steps: S100: Preparation and parameter setting before collection: determine the gas collection volume, collection method and gas type, and confirm the collection capacity sealing, pressure sensor range, and vacuum pump / compressor status; S200, gas generation / introduction trigger, source trigger: if the gas comes from the reaction device, when the reaction reaches the gas production condition, the controller opens the gas introduction valve; if it is an external gas source, the introduction is triggered by the pressure switch or flow sensor, and the controller starts the collection process after receiving the gas production signal or manual start signal; S300, gas collection execution control, switches the gas flow direction through the solenoid valve, and enters the collection container through the drying tube and filter. The pressure sensor monitors the pressure in the container in real time. When it exceeds the set value, the controller adjusts the pressure relief valve or gas source valve to maintain stable pressure. If the gas collection is sensitive to temperature, the controller adjusts the collection parameters based on the temperature sensor data; S400, safety and abnormality control, overpressure protection. When the container pressure exceeds the safety threshold, the emergency pressure relief valve automatically opens and stops the gas supply to prevent the container from exploding. The air tightness sensor monitors pipeline leakage and alarms and cuts off the gas supply when an abnormality is found. S500, collection completion and subsequent processing: After collection is completed, close the inlet and outlet valves and record the collection time, gas volume, and pressure / temperature data.

Images