Liquid level indication anti-leakage tray based on weight sensor and liquid level measurement method thereof

Abstract

The invention discloses a liquid level indication anti-leakage tray based on a weight sensor. The liquid level indication anti-leakage tray is characterized in that the weight sensor detects weight signals of chemicals in a liquid medicine barrel; the signal processing module converts signals detected by the weight sensor into liquid level height, the liquid level height is converted into liquid level percentage, the liquid pumping rate of a liquid medicine pump is calibrated online in real time through RLS, the reference pump speed is input into an EKF algorithm, the liquid level height is fused through the EKF algorithm, a nonlinear kinetic model is constructed to predict the liquid level track, and the liquid level return time is calculated. Calculating safety trigger time according to the liquid level return-to-zero time, designing a dynamic calculation interval function based on the remaining time, and adaptively adjusting the execution frequency of a prediction algorithm according to the liquid level height and the liquid level change rate; and the signal output module sends a shutdown control signal to the liquid medicine pump according to the time point of the safety triggering time. The liquid level monitoring function of converting the weight into the liquid level is achieved, and when the liquid level of liquid medicine is about to be zero, the liquid medicine pump is automatically stopped working.

Description

Technical Field

[0001] The present invention relates to the technical field of liquid level measurement, and in particular to a liquid level indicating anti-leakage tray based on a weight sensor and a liquid level measurement method thereof. Background Art

[0002] Plastic barrels for wet electronic chemicals are made of UHMWPE (ultra-high molecular weight, high-density polyethylene). These barrels, with their high strength and corrosion resistance, are used for packaging liquid chemicals in the semiconductor, chemical, petroleum, and pharmaceutical industries. Some mask manufacturing processes and cleaning machines require the use of these barrels for chemical supply. These barrels contain highly purified wet electronic chemicals. To prevent chemical leaks, these barrels are placed on leak-proof trays for liquid supply. During use, it is necessary to determine the remaining liquid level in the barrels to ensure timely allocation of new barrels of liquid to the workshop and to prevent nitrogen from being drawn into the liquid supply lines by process and cleaning machines when the remaining liquid level in the barrels reaches zero, potentially causing abnormal liquid supply to the machines. There are currently two methods for liquid level judgment: The first method involves manually shaking the plastic bucket to determine the remaining liquid level. However, this method has the disadvantages of infrequent manual checks, which prevent real-time feedback on the liquid level. Furthermore, manual level determination is inaccurate, often resulting in significant discrepancies with the actual level. This results in delayed liquid allocation and compromises in the machine's liquid supply stability.

[0003] Another method is to use a liquid level gauge, which is divided into contact and non-contact types. A contact level gauge requires a portion of the sensor, such as a float, or the entire sensor to be placed in the liquid. However, the top cover of the plastic chemical packaging barrels currently used only has two openings with a diameter of 20 mm: one is the air inlet for nitrogen filling, and the other is the liquid outlet for chemical supply. Therefore, neither opening can be installed with a contact level gauge. For a non-contact level gauge, a fixing clamp and sensor need to be installed on the plastic barrel. However, the plastic barrel needs to be replaced and recycled after the chemical is used up, so a non-contact level gauge cannot be fixed on the barrel body.

[0004] Currently, there is a lack of a highly accurate liquid level detection method for wet electronic chemical plastic barrels, which does not require the installation of an online liquid level indicator device to detect the liquid level of the chemical in the barrel in real time and ensure the stability of the liquid supply. Summary of the Invention

[0005] The purpose of the present invention is to provide a liquid level indicating anti-leakage tray based on a weight sensor and a liquid level measurement method thereof, which can realize the liquid level monitoring function of converting weight to liquid level without obvious changes in the basic function and volume size of the anti-leakage tray, and automatically stop the operation of the liquid medicine pump when the liquid medicine level returns to zero.

[0006] The present invention is achieved through the following technical solutions: In a first aspect, a first embodiment of the present invention provides a liquid level indicating anti-leakage tray based on a weight sensor, comprising: a weight sensor, an anti-leakage tray, a signal processing module, and a signal output module. The weight sensor is used to detect the weight signal of the chemicals in the liquid medicine barrel and transmit the detected weight signal to the signal processing module; The anti-leakage tray is used to support the liquid medicine barrel; The signal processing module is used to convert the signal detected by the weight sensor into a liquid level height, convert it into a liquid level percentage based on the liquid level height, use the recursive least squares method to calibrate the liquid pumping rate in real time online to obtain a reference pump speed, input the reference pump speed into the extended Kalman filter algorithm, integrate the extended Kalman filter algorithm with the liquid level height, construct a nonlinear dynamic model to predict the liquid level trajectory, calculate the liquid level zeroing time based on the integral algorithm, calculate the safety trigger time based on the liquid level zeroing time, design a dynamic calculation interval function based on the remaining time, and adaptively adjust the execution frequency of the prediction algorithm according to the liquid level height and the liquid level change rate to achieve intelligent scheduling of computing resources; The signal output module is used to output a control signal for stopping the liquid medicine pump to the connected machine equipment according to the time point of the safety trigger time.

[0007] Furthermore, it also includes a liquid level indication module, which is used to display the liquid level height under the control of the signal processing module.

[0008] Furthermore, it also includes a voice broadcast module, which is used to send an alarm voice signal when the liquid level is zero and broadcast the current liquid level under the control of the signal processing module.

[0009] Furthermore, it also includes a temperature measurement module, which is used to collect the temperature of the medicine liquid in the medicine liquid barrel and send it to the signal processing module. The signal processing module corrects the dynamic density of the medicine liquid that changes due to temperature changes based on the collected medicine liquid temperature.

[0010] Furthermore, the weight sensor adopts eight full-bridge strain pressure sensors with a symmetrical eight-point layout, wherein four weight sensors are arranged at the four corners of the bottom rectangle of the anti-leakage tray, and the other four mass sensors are arranged at the midpoint of each side of the bottom rectangle of the anti-leakage tray. The anti-leakage tray is provided with eight supporting feet, wherein four supporting feet are arranged at the four corners of the bottom rectangle of the anti-leakage tray, and the other four supporting feet are arranged at the midpoint of each side of the bottom rectangle of the anti-leakage tray.

[0011] Furthermore, the signal processing module includes a liquid level calculation unit, which calculates the load weight of the weight sensor, dynamically eliminates data collected by outlier weight sensors, uses a dynamic weight distribution formula to calculate the total weight of the medicine liquid by weighted average of the remaining weight sensors, calculates the net weight of the medicine liquid based on the total weight of the medicine liquid, calculates the volume of the medicine liquid based on the net weight of the medicine liquid, converts the volume of the medicine liquid into liquid level, and converts the liquid level into liquid level percentage.

[0012] Furthermore, the signal processing module further includes a dynamic prediction unit and an extended Kalman filter prediction unit; The dynamic prediction unit sets initial parameters for the dynamic prediction phase, including an estimated initial pump speed, an RLS forgetting factor, an initial value of a covariance matrix, EKF process noise, and observation noise, constructs a pumping dynamics model, and uses a recursive least squares method to update the pumping rate of the liquid medicine pump online to obtain a reference pump speed; The extended Kalman filter prediction unit performs reference pump speed correction and liquid level state estimation based on the input reference pump speed through liquid level feedback, outputs a predicted liquid level trajectory, calculates the liquid level zeroing time based on the integration algorithm, and calculates the safety trigger time based on the liquid level zeroing time.

[0013] Furthermore, the signal processing module also includes an adaptive calculation interval unit, which calculates the liquid level change rate according to the liquid level height, calculates the remaining time constant according to the liquid level height and the liquid level change rate, designs a dynamic calculation interval function based on the remaining time, adaptively adjusts the execution frequency of the prediction algorithm according to the current liquid level state and the liquid level change speed, and adopts different control strategies according to the remaining time.

[0014] In a second aspect, another embodiment of the present invention provides a method for measuring the liquid level of an anti-leakage tray using a liquid level indicator based on a weight sensor, comprising the following steps: Get the weight signal sent by the weight sensor; The signal detected by the weight sensor is converted into liquid level height, which is then converted into liquid level percentage based on the liquid level height. The recursive least squares method is used to calibrate the liquid pump pumping rate online in real time to obtain a reference pump speed. The reference pump speed is input into the extended Kalman filter algorithm, which integrates the liquid level height. A nonlinear dynamic model is constructed to predict the liquid level trajectory. The liquid level return time is calculated based on the integral algorithm, and the safety trigger time is calculated based on the liquid level return time. A dynamic calculation interval function is designed based on the remaining time. The execution frequency of the prediction algorithm is adaptively adjusted according to the liquid level height and liquid level change rate to achieve intelligent scheduling of computing resources. A shutdown control signal is sent to the liquid medicine pump according to the time point of the safety trigger time.

[0015] Furthermore, converting the signal detected by the weight sensor into a liquid level height, and converting the liquid level height into a liquid level percentage specifically includes: Calculate the load weight of the weight sensor, dynamically remove the data collected by the outlier weight sensor, use the dynamic weight distribution formula to calculate the total weight of the remaining weight sensors as the weighted average, calculate the net weight of the medicine liquid based on the total weight of the medicine liquid, calculate the volume of the medicine liquid based on the net weight of the medicine liquid, convert the volume of the medicine liquid to the liquid level, and convert the liquid level to the liquid level percentage; The method adopts the recursive least squares method to calibrate the liquid pumping rate of the liquid pump in real time online to obtain the reference pumping speed, inputs the reference pumping speed into the extended Kalman filter algorithm, integrates the liquid level height with the extended Kalman filter algorithm, constructs a nonlinear dynamic model to predict the liquid level trajectory, calculates the liquid level zeroing time based on the integral algorithm, and calculates the safety trigger time according to the liquid level zeroing time. Specifically, the method includes: Initial parameters for the dynamic prediction phase were set, including the initial pump speed estimate, the RLS forgetting factor, the initial value of the covariance matrix, the EKF process noise, and the observation noise. A pumping dynamics model was constructed, and the recursive least squares method was used to update the pumping rate of the liquid medicine pump online to obtain the benchmark pumping speed. Based on the input reference pump speed, the reference pump speed is corrected and the liquid level state is estimated through liquid level feedback, the predicted liquid level trajectory is output, the liquid level zeroing time is calculated based on the integral algorithm, and the safety trigger time is calculated based on the liquid level zeroing time.

[0016] Compared with the prior art, the present invention has the following advantages and beneficial effects: Embodiments of the present invention provide a weight-sensor-based liquid level indicator and liquid level measurement method. This device detects the weight of liquid in a medicine barrel and converts it into a liquid level value. This not only fulfills the basic function of preventing liquid leakage but also innovatively combines liquid level monitoring with the anti-leakage tray. This allows for weight-to-level monitoring without significantly altering the basic function or volume of the anti-leakage tray. Furthermore, it predicts liquid level trends and sends a shutdown control signal to the medicine pump when the liquid level in the medicine barrel is about to reach zero. This precisely prevents air bubbles from entering the pumping line while minimizing waste of the medicine barrel, achieving microsecond dynamic response. The system utilizes a dual-core algorithm, RLS (recursive least squares) and EKF (extended Kalman filter), to achieve millisecond-level fusion processing of sensor data. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the following briefly introduces the drawings required for use in the examples. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be considered as limiting the scope. A person of ordinary skill in the art can also derive other relevant drawings based on these drawings without inventive effort. In the drawings: Figure 1 A block diagram showing the principle of a liquid level indication and anti-leakage tray based on a weight sensor provided in the first embodiment of the present invention; Figure 2 A schematic structural diagram of a liquid level indicating anti-leakage tray based on a weight sensor provided in the first embodiment of the present invention; Figure 3 A schematic diagram of the control flow of the signal processing module in the first embodiment of the present invention is provided; Figure 4 Schematic diagram of the EKF algorithm scheduling based on the adaptive calculation interval strategy of the remaining time; Figure 5 Schematic diagram of the characteristics of the adaptive calculation interval strategy based on the remaining time; Figure 6 This is a flow chart of a method for measuring the liquid level of an anti-leakage tray using a liquid level indication sensor provided by another embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with examples and drawings. The exemplary embodiments of the present invention and their descriptions are only used to explain the present invention and are not intended to limit the present invention.

[0019] like Figure 1-3As shown, a first embodiment of the present invention provides a liquid level indication anti-leakage tray based on a weight sensor, comprising: a weight sensor, an anti-leakage tray, a signal processing module, a signal output module, a liquid level indication module, a voice broadcast module, and a temperature measurement module, wherein the weight sensor is used to detect the weight signal of the chemical in the liquid medicine barrel and transmit the detected weight signal to the signal processing module; The anti-leakage tray is used to support the liquid medicine barrel. The signal processing module is used to convert the signal detected by the weight sensor into the liquid level height, and then convert it into a liquid level percentage based on the liquid level height. The recursive least squares method is used to calibrate the liquid pump's pumping rate in real time online to obtain a baseline pump speed. The baseline pump speed is input into the extended Kalman filter algorithm. The extended Kalman filter algorithm integrates the liquid level height to construct a nonlinear dynamic model to predict the liquid level trajectory. The liquid level zeroing time is calculated based on the liquid level zeroing time, and the safety trigger time is calculated based on the liquid level zeroing time. A dynamic calculation interval function is designed based on the remaining time. The execution frequency of the prediction algorithm is adaptively adjusted according to the liquid level height and liquid level change rate to achieve intelligent scheduling of computing resources. The signal output module is used to output a control signal to shut down the liquid medicine pump to the connected equipment based on the time of the safety trigger time. The liquid level indication module is used to display the liquid level height under the control of the signal processing module. The voice broadcast module is used to issue an alarm voice signal when the liquid level reaches zero and to broadcast the current liquid level under the control of the signal processing module. The temperature measurement module is used to collect the temperature of the liquid medicine in the liquid medicine barrel and send it to the signal processing module. The signal processing module corrects the dynamic density of the liquid medicine that changes due to temperature changes based on the collected liquid medicine temperature.

[0020] The weight sensor utilizes eight full-bridge strain gauge pressure sensors in a symmetrical eight-point layout. The eight pressure sensors are evenly distributed on the same horizontal plane, employing a symmetrical structure: four sensors are located at the four corners of the rectangular bottom of the anti-seepage tray; the remaining four sensors are located at the midpoints of each side of the bottom rectangle. This structure improves redundancy, evenly distributes the load, and reduces errors caused by the varying placement of the plastic barrels. The anti-seepage tray is made of polyethylene and has eight support legs: four at the four corners of the rectangular bottom; the remaining four at the midpoints of each side of the rectangular bottom. The liquid level indicator module utilizes a liquid level indicator bar, a long strip composed of multiple 10-segment LED digital tube modules. Each 10-segment LED digital tube module consists of 10 vertically arranged rectangular LED light units. The liquid level indicator bar is mounted vertically on the floor or wall, at the same height as the plastic barrel. The more LEDs illuminated, the higher the liquid level, providing a visual display of the liquid level.

[0021] The signal processing module has a built-in 8-channel AD converter, which runs a specific algorithm to convert the signals from the eight weight sensors into liquid level values, control the lighting height of the liquid level indicator bar, and directly output the liquid level signal in digital form to the PLC modules of the process and cleaning machines. The voice broadcast module uses a text-to-speech (TTS) voice broadcast module, which converts preset alarm text into a voice alarm when the liquid level reaches 0. When the user manually presses a button to trigger it, the signal processing module is requested to obtain the current liquid level value and voice broadcast the current level.

[0022] Eight sensors are symmetrically installed at the four corners (S1-S4) and the midpoints of each side (S5-S8) of the anti-leakage tray to form a double-layer redundant layout; the weight of the plastic barrel is evenly transferred to the sensors through the tray, and each sensor bears the local load. .

[0023] Full-bridge circuit principle: Each sensor has four built-in strain gauges to form a Wheatstone bridge, outputting a differential voltage : ; Where i is the sensor number, is the output voltage of the i-th sensor (unit: V), and force In a linear relationship, is the bridge excitation voltage, is the resistance change caused by the pressure strain of the plastic barrel, is the initial resistance of the sensor, is the local load borne by the i-th sensor.

[0024] Signal amplification and filtering: Each sensor It is amplified by AD8422 (gain 500x) and filtered by a second-order RC low-pass filter (cut-off frequency 10Hz) to suppress high-frequency noise.

[0025] Synchronous sampling and digitization: The ADS1258 synchronously acquires 8-channel signals at a rate of 1kSPS, with 24-bit resolution ensuring ±0.01% quantization accuracy.

[0026] Temperature compensation: The sensor integrates DS18B20 to monitor sensor temperature and dynamically correct strain gauge sensitivity drift.

[0027] The signal processing module includes a liquid level calculation unit, which calculates the load of the weight sensor, dynamically eliminates the data collected by the outlier weight sensor, uses the dynamic weight distribution formula to calculate the total weight of the remaining weight sensors by weighted average, calculates the net weight of the medicine liquid based on the total weight of the medicine liquid, calculates the volume of the medicine liquid based on the net weight of the medicine liquid, converts the volume of the medicine liquid into the liquid level, and converts the liquid level into a liquid level percentage.

[0028] The signal processing module also includes a dynamic prediction unit and an extended Kalman filter prediction unit. The dynamic prediction unit sets the initial parameters of the dynamic prediction stage, which include the initial pump speed estimate, the RLS forgetting factor, the initial value of the covariance matrix, the EKF process noise and the observation noise, and constructs a pumping dynamics model. The recursive least squares method (RLS) is used to update the pumping rate of the liquid pump online to obtain the reference pump speed. The extended Kalman filter (EKF) prediction unit corrects the reference pump speed and estimates the liquid level state based on the input reference pump speed through liquid level feedback, outputs the predicted liquid level trajectory, calculates the liquid level zeroing time based on the integral algorithm, and calculates the safety trigger time based on the liquid level zeroing time.

[0029] The signal processing module also includes an adaptive calculation interval unit, which calculates the liquid level change rate based on the liquid level height, calculates the remaining time constant based on the liquid level height and the liquid level change rate, designs a dynamic calculation interval function based on the remaining time, adaptively adjusts the execution frequency of the prediction algorithm according to the current liquid level status and the liquid level change speed, and adopts different control strategies according to the remaining time.

[0030] Specifically, the specific working method of the liquid level calculation unit is as follows: Calculate the weight of the weight sensor load: ; in, is the calibration coefficient, is the reference voltage (zero offset) of the i-th sensor in no-load (after tare), is the load weight of the i-th sensor The output voltage under the action of, unit: V, is the actual measurement value of the sensor, is the raw weight measurement value of the i-th sensor (unit: kg).

[0031] Weight sensor redundancy and fault tolerance and liquid medicine weight calculation: Dynamically remove outlier sensors to achieve high reliability: Calculate the median weight of 8 weight sensors If a sensor Deviation If the error exceeds 3 times the standard deviation, it is marked as a fault and the sensor result is discarded. The total weight is then calculated using the following dynamic weight distribution formula: , , in, is the total weight of the liquid and the barrel, is the weight coefficient of the i-th sensor (unitless, range 0-1), is the noise standard deviation of the i-th sensor (unit: kg, obtained from historical data statistics), is the raw weight measurement value of the i sensors (unit: kg), where The weight is based on the optimal estimation theory, and the weight is proportional to the inverse of the noise variance (the smaller the variance, the more reliable the data). By calibration or in-flight data statistics (calculate the standard deviation of the last 100 measurements).

[0032] Calculation of liquid volume: Calculate the net weight of the liquid medicine: , in, is the weight of the empty barrel, is the weight of the liquid medicine, is the total weight, this step refers to the tare operation, so we can get .

[0033] Convert the net weight of the liquid medicine to volume: , in, ρ ( T ) is the dynamic density. Since the density of the liquid medicine changes with temperature, the density data of the liquid medicine at the current temperature can be obtained by storing a density table in the system and performing a table lookup method. is the volume of the drug solution.

[0034] The volume of the liquid is converted to the liquid level: the binary method is used for iteration, that is, the signal processing module pre-stores the calibration data V(h), that is, the volume and height corresponding data table, and the liquid level height h in the barrel is calculated by interpolation ， By dividing the bucket into N segments (N=100), the height of each segment is , For the total height of the plastic barrel, measure the circumference of each height segment and calculate the corresponding radius , ,…, , each segment corresponds to a height interval of [0, ),[ , ),…,[ , ], and then calculate the volume of the non-standard cylindrical plastic barrel segment according to the volume approximation formula: let the current liquid level be h, and the segment number of the first k complete cylindrical segments be k, ; The height of the remaining incomplete cylindrical segment is .

[0035] Total approximate volume formula: , Among them, , which means the volume of a complete segment: that is, the first k segments are complete cylinders, and the total volume of these complete cylinders is calculated , that is, the residual volume of the current incomplete cylinder segment. The height of the k +1 incomplete cylinder segment part is . Since the liquid volume has been obtained in the liquid volume calculation, the liquid level can be calculated according to the total approximate volume formula h： (1) Set the height search interval , ], that is, [0, ]; (2) Calculate the midpoint height: ; (3) Calculate the midpoint volume : According to substituted into the total approximate volume formula, we get ; (4) Compare with the liquid volume V: If < V, it means the liquid level is higher than , update = ; If > V, it means the liquid level is lower than , update = ; (5) Repeat the iteration until | - V| < ( is the preset accuracy, = 10 -6 m 3 ); (6) Finally, output , which is the required liquid level height h ， After calculating the liquid level height h, use the following formula to convert the liquid level into a liquid level percentage: .

[0036] Use the above mathematical method to achieve the process and purpose of measuring weight, converting weight to volume, converting volume to liquid level height, and converting liquid level height to liquid level percentage.

[0037] After obtaining the value of the liquid level height, enter the dynamic prediction stage. The liquid level height value is input into the dynamic prediction algorithm to predict the liquid level zeroing time , send the liquid supply stop signal in advance, leave enough reaction time for the liquid pump and the machine, accurately prevent bubbles from entering the liquid extraction pipeline, and achieve the purpose of not wasting the liquid in the liquid barrel as much as possible.

[0038] Set the initial parameters for the dynamic prediction stage: The initial pump speed estimate is set to ; RLS forgetting factor , initial value of covariance matrix .

[0039] EKF process noise , observation noise .

[0040] The mathematical relationship between the liquid level and time is described by the equation, assuming that the liquid pump is moving at a rate of When pumping liquid, the liquid level drops to satisfy the conservation of mass: ; Cross-sectional area: , defined by table lookup or piecewise function.

[0041]

[0042] Discretization (time step ): This step divides the time into segments with fixed intervals, and uses a simplified formula to calculate the liquid level change in each segment, that is, to simplify the calculation process: .

[0043]

[0044] Online update of pumping rate using recursive least squares (RLS) This step is to calibrate the liquid pumping rate in real time , adapt to the actual flow rate changes.

[0045] Define the linear regression form: rewrite the discrete equation in the previous step as: ;

[0046] Let the observation value be , the eigenvector is , then the observation equation is: ; in, (liquid level change), is the characteristic quantity, (negative time step / cross-sectional area), is the observation noise.

[0047]

[0048] RLS iteration formula: ; in, K ( t ) represents the Kalman gain, which calculates the weight of the current data, P(t) is the updated covariance, It indicates that the confidence is improved. The lower the covariance, the more accurate the pump speed estimate. λ is the forgetting factor (0.95≤λ<1), which weakens the influence of old data. Q^pump(t) is the estimated pump speed at time t.

[0049]

[0050] The following is an example of the recursive least squares (RLS) online update pumping rate calculation process: Assume the initial state of the system: Initial estimate , provided by the equipment manufacturer or the equipment nameplate, initial covariance (large values ​​indicate low confidence), forgetting factor .

[0051] first step: Always updated.

[0052] Input data: (Liquid level drops 0.048m); (barrel cross-sectional area ); Calculate gain : ; Update pump speed estimate: ; Update covariance : ; Output: Pumping rate The estimate was fine-tuned from 0.09 to 0.0912 , the covariance changes from 1 to 0.813 (the confidence level increases and the estimate becomes more precise).

[0053] Step 2: Always updated.

[0054] New observations: ; (The drop in liquid level causes the cross-sectional area of ​​the liquid medicine barrel to change); Calculate gain : ; Update pump speed estimate: ; Update covariance : ; Output: Pumping rate The estimated value was fine-tuned to 0.0909 , the covariance drops to 0.702 (the confidence level is further improved and the estimate is more precise).

[0055] Through the continuous iteration of the RLS iterative formula, the system can track the actual pumping speed change in real time (such as the voltage fluctuation of the liquid pump or the mechanical wear of the liquid pump causing the pump speed change). The value can ensure the high prediction accuracy of the liquid pump downtime. This step outputs The value (the estimated pump speed at the current moment) is significant in: 1. Dynamically calibrating the actual pumping rate; 2. Eliminating the effects of pump speed drift (such as voltage fluctuations and mechanical wear).

[0056] The input value of the Extended Kalman Filter (EKF) prediction unit is the reference pump speed output by the dynamic prediction unit in the previous step. , the output value of the extended Kalman filter (EKF) prediction unit: EKF corrects the reference pump speed after the liquid level feedback and high-precision liquid level state estimates and predictions.

[0057] RLS and EKF synergistic advantages: 1. RLS focuses on pump speed tracking; 2. EKF focuses on liquid level state estimation and pump speed correction; 3. RLS and EKF collaborative process: RLS provides estimated pump speed dynamic parameters, which EKF uses to perform high-precision liquid level state estimation and prediction, as well as high-precision pump speed parameter correction, achieving precise liquid pump shutdown control. EKF computational goal: To integrate model predictions with real-time measurements to optimize liquid level estimation.

[0058] Define the state vector: ;

[0059] State equation (nonlinear): ; Process noise: .

[0060]

[0061] Observation equation: ; Observation noise: ;

[0062] EKF iteration steps: Prediction step: ;

[0063] ;

[0064] Update step: .

[0065]

[0066] The following is the specific calculation process of the extended Kalman filter (EKF) prediction unit: Step 0: Initialization (initialization is only performed when the calculation process is first started), that is, initializing x0; enter: Initial state estimate: ; in, ; Initial covariance: (initial uncertainty); Process noise covariance , observation noise variance (like ); Output: None.

[0067] Step 1: Prediction step (forward calculation based on the model) State prediction: subsequent moments ( t ≥1): x(t-1) is directly inherited from the output of the previous update step: Input: The optimal estimate x(t-1) at the previous moment; calculate: ; in, Provides an estimate of the pump speed for the RLS.

[0068] Output: Prior state estimate (predicted value).

[0069] Jacobian matrix calculation: enter: and barrel cross-sectional area ; calculate: ; like Segment-wise constant ( ) is simplified to: ; Output: Linearized matrix .

[0070] Covariance prediction (key output 1) Input: covariance at the previous moment , the Jacobian matrix , process noise ; calculate: ; (model uncertainty); Output: Prior covariance matrix , represents the uncertainty of the predicted state (the larger the value, the lower the credibility), Liquid level prediction noise variance, Pump speed prediction noise variance.

[0071] Step 2: Update step (fusing sensor measurements) Calculate the Kalman gain: enter: , the observation matrix (Only observe the liquid level), ; calculate: ; Output: Kalman gain ; represents the covariance between the state and the measurement; represents the variance of the prediction residual; A high value indicates that the trust measure > the trust model; A low value indicates that the trust model > the trust measure.

[0072] Status Update: enter: , , sensor measurement value ; calculate: ; Observation residuals: , the large residual indicates that the model may be biased, Correction status.

[0073] Output: Posterior state estimate (Best estimate).

[0074] Covariance Update (Key Output 2) enter: , , ; calculate: ; in, is the identity matrix.

[0075] Output: Posterior covariance matrix , represents the uncertainty of the updated state (confidence is improved after fusion measurement).

[0076] Step 3: Predicted liquid level trajectory (core output) Input: Current best estimate ; Calculation: As the starting point, recursively execute the prediction step to the target liquid level : ;

[0077] Output: Liquid level prediction trajectory ; Application: Calculate remaining time ; Predicted trajectories and locations of the covariance matrix: Predicted liquid level trajectory: After a complete EKF cycle, based on Extrapolation generation (step 3); Covariance matrix: is the prediction step output, prediction uncertainty; : Update step output, indicating the final estimated uncertainty.

[0078] Step 4: EKF time update: Update t with t+1, and the calculation process returns to step 1 (receive new liquid level data and RLS Update ), and a new cycle begins.

[0079] Zero time calculation: ;

[0080] Input: EKF output optimal estimates of predicted fluid level and pump speed , Calculated as zero time and .

[0081] Output: predicted return time , meaning: The system predicts the liquid level of the liquid tank from the current height The time required to drop to zero. Get the liquid level zero time After that, it enters the stage of sending the liquid supply stop signal.

[0082] Signal trigger (input: predicted time ): Goal: Compensate for system delays and achieve millisecond-level trigger signals.

[0083] Calculating safety trigger time The formula is: ; in, is the pump valve closing delay (actual measurement: such as 0.5 seconds), is a safety margin (such as 0.3 seconds). Real-time judgment: When the system current time When the stop signal is sent immediately, the formula means that + Send a stop signal within seconds.

[0084] The specific workflow of the adaptive calculation interval unit is as follows: Design a dynamic calculation interval function: ;

[0085] Among them, h for Current liquid level height, is the liquid level change rate, Δ t calc To dynamically calculate the interval, τ min is the minimum calculation interval, τ max is the maximum calculation interval, β is the sensitivity coefficient, To prevent zero division, prevent is 0.

[0086] Real-time response: The calculation frequency is proportional to the speed of liquid level change: Δ t calc ∝∣ ∣, the faster the liquid level changes, the more frequent the calculation.

[0087] Critical priority: The critical state means that the remaining time calculated according to the current liquid level change speed is less than the set critical value (such as 10 seconds). The calculation frequency is inversely proportional to the liquid level height: Δ t calc ∝ h 1. The lower the liquid level, the more frequent the calculation.

[0088] Dynamic stability: Ensure smooth transition through exponential function to prevent system oscillation caused by sudden changes in calculation frequency. Represents the remaining time constant, where h is the current remaining liquid level height, The residual time constant is the imprecise estimated time for the liquid level to drop to zero.

[0089] This dynamic adjustment mechanism adaptively adjusts key time parameters based on the current liquid level status and liquid level change rate: the time interval between calls to the EKF prediction algorithm and the zero time calculation (i.e., calling the T0 calculation formula), which is used to improve the operational stability of the entire equipment and save computing resource consumption.

[0090] The role of this strategy in the prediction stage is as follows Figure 4 : Execution interval T = adaptive calculation interval, so the only computing modules that are continuously running in the entire system are the liquid level calculation unit and the adaptive calculation interval unit. The dynamic prediction unit and the extended Kalman filter prediction unit, which consume the most computing resources, run according to the time specified by the adaptive calculation interval strategy to improve system stability and prevent the entire system from crashing or freezing under full load after all resources are occupied by the zero-time dynamic prediction algorithm for a long time.

[0091] The dynamic operation characteristics of this strategy are as follows: Figure 5 The critical state means that the remaining time calculated according to the current liquid level change speed is less than the set value (such as 10 seconds). The liquid level changes at a high speed: short time interval, the critical height of the liquid level: short time interval, and the stable liquid level state: long time interval.

[0092] The working process of a liquid level indicating anti-leakage tray based on a weight sensor provided by an embodiment of the present invention is as follows: 1. Equipment installation: Install the anti-leakage base on a flat ground. 2. Connect the power supply: Insert the power adapter into the interface on the back of the base. After turning on the power, the LED light bar will light up for 3 seconds (indicating that the self-test has passed). 3. Placement of wet electronic chemical plastic barrels: Slowly place an empty plastic storage tank in the center of the base. 4. Enter the empty barrel peeling mode: Make sure the storage tank is empty, short press the "mode key" to switch to the peeling state, and the TTS voice broadcast module broadcasts "peeling mode". 5. Perform the peeling operation: Press and hold the "peel key" for 3 seconds, and the signal processing module automatically collects data from 8 weight sensors and calculates 6. Weight calibration: Pour 50KG of pure water into the barrel, press and hold the "mode key" for 5 seconds, the TTS voice broadcast module will broadcast "calibrating", and the signal processing module will automatically calculate the calibration coefficient. ,in, / 8 is the weight of the pure water loaded. Dividing by 8 is because the 8 weight sensors evenly distribute the load. is the ith sensor in the calibration weight The output voltage under action is the actual measurement value of the sensor and needs to be collected in a stable state after loading pure water; = is the reference voltage (zero offset) of the i-th sensor in an unloaded (tare) state. It is automatically recorded after the tare operation and is used to eliminate the effects of the empty tank weight and the sensor's inherent offset. After removing the weight, the TTS voice announcement module announces "Current weight 0 kg." After calibration is complete, remove the plastic barrel used for calibration. 7. Chemical Density Setting: Short-press the "Mode" key to switch to density setting mode. Use the "▲ / ▼" keys to select the chemical type (e.g., "Sulfuric Acid"). 8. Liquid level monitoring: After clicking the start monitoring button, the new liquid barrel for chemical supply is moved to the anti-leakage pallet by forklift. Every time the liquid level changes by 10%, the TTS voice broadcast module will broadcast the current liquid level (such as "current liquid level 30%"). At the same time, the liquid level indicator light will light up, and the lighting height will keep consistent with the liquid level in the plastic barrel, intuitively displaying the liquid level. Mathematical methods are used to predict the liquid level trend and trigger the liquid pump shutdown signal in advance, so as to leave sufficient reaction time for the liquid pump and the machine. While accurately preventing bubbles from entering the pumping pipeline, the purpose of wasting the liquid in the liquid barrel as much as possible is achieved. At the same time, this algorithm can realize adaptive adaptation to the error fluctuations of the machine's pumping system. No matter how severe the voltage fluctuations of the entire pumping system are or how much wear and tear the liquid pump has, the trend prediction algorithm can adapt to the error and accurately predict the liquid level.

[0093] The present invention provides a liquid level indication anti-leakage tray based on a weight sensor, which detects the weight of the medicine liquid in the barrel and converts it into a liquid level value. It not only realizes the basic function of preventing the leakage of the medicine liquid, but also innovatively combines the liquid level monitoring function with the anti-leakage tray. The liquid level monitoring function of converting weight to liquid level is realized when the basic function and volume size of the anti-leakage tray do not change significantly. At the same time, it realizes the following functions: sending the liquid level data of the medicine liquid to the process and cleaning machines. When the liquid level of the medicine liquid in the chemical packaging plastic barrel is zero, the process and cleaning machines will stop extracting the medicine liquid from the chemical packaging plastic barrel after receiving the alarm, so as to avoid drawing air into the pipeline system and causing abnormal supply of process and cleaning medicine liquid. At the same time, this system can also remind the operators at the production site through liquid level display and sound and light alarms. The operator can replace the chemical liquid barrel with a new one in a timely manner to avoid the interruption of mask production due to the temporary allocation of chemicals. At the same time, the unique symmetrical eight-sensor layout scheme designed for the anti-leakage tray has the advantages of improving redundancy, evenly distributing the load, and reducing errors caused by the different placement of plastic barrels. It can intuitively display the liquid level height, and use the signal processing module to predict the liquid level trend and trigger the liquid pump shutdown signal in milliseconds in advance, so as to leave sufficient reaction time for the liquid pump and the machine. While accurately preventing bubbles from entering the liquid extraction pipeline, the purpose of wasting the liquid in the liquid barrel as much as possible is to achieve microsecond dynamic response. The dual-core algorithm of RLS (recursive least squares) and EKF (extended Kalman filter) is used to realize millisecond-level fusion processing of sensor data. Its technical indicators are shown in Table 1.

[0094]

[0095] Table 1 Technical indicators like Figure 6 As shown, another embodiment of the present invention provides a method for measuring the liquid level of a liquid level indicating anti-leakage tray based on a weight sensor, comprising the following steps: Get the weight signal sent by the weight sensor; The signal detected by the weight sensor is converted into liquid level height, which is then converted into liquid level percentage based on the liquid level height. The recursive least squares method is used to calibrate the liquid pump pumping rate online in real time to obtain a reference pump speed. The reference pump speed is input into the extended Kalman filter algorithm, which integrates the liquid level height. A nonlinear dynamic model is constructed to predict the liquid level trajectory. The liquid level return time is calculated based on the integral algorithm, and the safety trigger time is calculated based on the liquid level return time. A dynamic calculation interval function is designed based on the remaining time. The execution frequency of the prediction algorithm is adaptively adjusted according to the liquid level height and liquid level change rate to achieve intelligent scheduling of computing resources. A shutdown control signal is sent to the liquid medicine pump according to the time point of the safety trigger time.

[0096] The signal detected by the weight sensor is converted into the liquid level height, and then converted into the liquid level percentage according to the liquid level height. Specifically, the following steps are performed: Calculate the load weight of the weight sensor, dynamically remove the data collected by the outlier weight sensor, use the dynamic weight distribution formula to calculate the total weight of the remaining weight sensors as the weighted average, calculate the net weight of the medicine liquid based on the total weight of the medicine liquid, calculate the volume of the medicine liquid based on the net weight of the medicine liquid, convert the volume of the medicine liquid to the liquid level, and convert the liquid level to the liquid level percentage; The method adopts the recursive least squares method to calibrate the liquid pumping rate of the liquid pump in real time online to obtain the reference pumping speed, inputs the reference pumping speed into the extended Kalman filter algorithm, integrates the liquid level height with the extended Kalman filter algorithm, constructs a nonlinear dynamic model to predict the liquid level trajectory, calculates the liquid level zeroing time based on the integral algorithm, and calculates the safety trigger time according to the liquid level zeroing time. Specifically, the method includes: Initial parameters for the dynamic prediction phase were set, including the initial pump speed estimate, the RLS forgetting factor, the initial value of the covariance matrix, the EKF process noise, and the observation noise. A pumping dynamics model was constructed, and the recursive least squares method was used to update the pumping rate of the liquid medicine pump online to obtain the benchmark pumping speed. Based on the input reference pump speed, the reference pump speed is corrected and the liquid level state is estimated through liquid level feedback, the predicted liquid level trajectory is output, the liquid level zeroing time is calculated based on the integral algorithm, and the safety trigger time is calculated based on the liquid level zeroing time.

[0097] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A liquid level indication anti-leakage tray based on a weight sensor, characterized in that: include: Weight sensor, anti-leakage tray, signal processing module and signal output module, The weight sensor is used to detect the weight signal of the chemicals in the liquid medicine barrel and transmit the detected weight signal to the signal processing module; The anti-leakage tray is used to support the liquid medicine barrel; The signal processing module is used to convert the signal detected by the weight sensor into a liquid level height, convert it into a liquid level percentage based on the liquid level height, use the recursive least squares method to calibrate the liquid pumping rate in real time online to obtain a reference pump speed, input the reference pump speed into the extended Kalman filter algorithm, integrate the extended Kalman filter algorithm with the liquid level height, construct a nonlinear dynamic model to predict the liquid level trajectory, calculate the liquid level zeroing time based on the integral algorithm, calculate the safety trigger time based on the liquid level zeroing time, design a dynamic calculation interval function based on the remaining time, and adaptively adjust the execution frequency of the prediction algorithm according to the liquid level height and the liquid level change rate to achieve intelligent scheduling of computing resources; The signal output module is used to output a control signal for stopping the liquid medicine pump to the connected machine equipment according to the time point of the safety trigger time.

2. The liquid level indicating anti-leakage tray based on a weight sensor according to claim 1, characterized in that: It also includes a liquid level indication module, which is used to display the liquid level height under the control of the signal processing module.

3. The liquid level indicating anti-leakage tray based on a weight sensor according to claim 1, characterized in that: It also includes a voice broadcast module, which is used to send out an alarm voice signal when the liquid level is zero and broadcast the current liquid level under the control of the signal processing module.

4. The liquid level indicating anti-leakage tray based on a weight sensor according to claim 1, characterized in that: It also includes a temperature measurement module, which is used to collect the temperature of the medicine liquid in the medicine liquid barrel and send it to the signal processing module. The signal processing module corrects the dynamic density of the medicine liquid that changes due to temperature changes based on the collected medicine liquid temperature.

5. The liquid level indicating anti-leakage tray based on a weight sensor according to claim 1, characterized in that: The weight sensor adopts eight full-bridge strain pressure sensors with a symmetrical eight-point layout, wherein four weight sensors are arranged at the four corners of the bottom rectangle of the anti-leakage tray, and the other four mass sensors are arranged at the midpoint of each side of the bottom rectangle of the anti-leakage tray. The anti-leakage tray is provided with eight supporting feet, wherein four supporting feet are arranged at the four corners of the bottom rectangle of the anti-leakage tray, and the other four supporting feet are arranged at the midpoint of each side of the bottom rectangle of the anti-leakage tray.

6. The liquid level indicating anti-leakage tray based on a weight sensor according to any one of claims 1 to 5, characterized in that: The signal processing module includes a liquid level calculation unit, which calculates the load weight of the weight sensor, dynamically eliminates data collected by outlier weight sensors, uses a dynamic weight distribution formula to calculate the total weight of the medicine liquid by weighted average of the remaining weight sensors, calculates the net weight of the medicine liquid based on the total weight of the medicine liquid, calculates the volume of the medicine liquid based on the net weight of the medicine liquid, converts the volume of the medicine liquid into liquid level, and converts the liquid level into liquid level percentage.

7. The liquid level indicating anti-leakage tray based on a weight sensor according to claim 6, characterized in that: The signal processing module also includes a dynamic prediction unit and an extended Kalman filter prediction unit; The dynamic prediction unit sets initial parameters for the dynamic prediction phase, including an estimated initial pump speed, an RLS forgetting factor, an initial value of a covariance matrix, EKF process noise, and observation noise, constructs a pumping dynamics model, and uses a recursive least squares method to update the pumping rate of the liquid medicine pump online to obtain a reference pump speed; The extended Kalman filter prediction unit performs reference pump speed correction and liquid level state estimation based on the input reference pump speed through liquid level feedback, outputs a predicted liquid level trajectory, calculates the liquid level zeroing time based on the integration algorithm, and calculates the safety trigger time based on the liquid level zeroing time.

8. The liquid level indicating anti-leakage tray based on a weight sensor according to claim 7, characterized in that: The signal processing module also includes an adaptive calculation interval unit, which calculates the liquid level change rate based on the liquid level height, calculates the remaining time constant based on the liquid level height and the liquid level change rate, designs a dynamic calculation interval function based on the remaining time, adaptively adjusts the execution frequency of the prediction algorithm according to the current liquid level state and the liquid level change speed, and adopts different control strategies according to the remaining time.

9. A method for measuring the liquid level of a liquid level indicating anti-leakage tray based on a weight sensor, characterized in that: The following steps are involved: Get the weight signal sent by the weight sensor; The signal detected by the weight sensor is converted into liquid level height, which is then converted into liquid level percentage based on the liquid level height. The recursive least squares method is used to calibrate the liquid pump pumping rate online in real time to obtain a reference pump speed. The reference pump speed is input into the extended Kalman filter algorithm, which integrates the liquid level height. A nonlinear dynamic model is constructed to predict the liquid level trajectory. The liquid level return time is calculated based on the integral algorithm, and the safety trigger time is calculated based on the liquid level return time. A dynamic calculation interval function is designed based on the remaining time. The execution frequency of the prediction algorithm is adaptively adjusted according to the liquid level height and liquid level change rate to achieve intelligent scheduling of computing resources. A shutdown control signal is sent to the liquid medicine pump according to the time point of the safety trigger time.

10. The liquid level measurement method of the liquid level indication anti-leakage tray based on a weight sensor according to claim 9, characterized in that: The step of converting the signal detected by the weight sensor into the liquid level height and converting the liquid level height into a liquid level percentage specifically includes: Calculate the load weight of the weight sensor, dynamically remove the data collected by the outlier weight sensor, use the dynamic weight distribution formula to calculate the total weight of the remaining weight sensors as the weighted average, calculate the net weight of the medicine liquid based on the total weight of the medicine liquid, calculate the volume of the medicine liquid based on the net weight of the medicine liquid, convert the volume of the medicine liquid to the liquid level, and convert the liquid level to the liquid level percentage; The method adopts the recursive least squares method to calibrate the liquid pumping rate of the liquid pump in real time online to obtain the reference pumping speed, inputs the reference pumping speed into the extended Kalman filter algorithm, integrates the liquid level height with the extended Kalman filter algorithm, constructs a nonlinear dynamic model to predict the liquid level trajectory, calculates the liquid level zeroing time based on the integral algorithm, and calculates the safety trigger time according to the liquid level zeroing time. Specifically, the method includes: Initial parameters for the dynamic prediction phase were set, including the initial pump speed estimate, the RLS forgetting factor, the initial value of the covariance matrix, the EKF process noise, and the observation noise. A pumping dynamics model was constructed, and the recursive least squares method was used to update the pumping rate of the liquid medicine pump online to obtain the benchmark pumping speed. Based on the input reference pump speed, the reference pump speed is corrected and the liquid level state is estimated through liquid level feedback, the predicted liquid level trajectory is output, the liquid level zeroing time is calculated based on the integral algorithm, and the safety trigger time is calculated based on the liquid level zeroing time.

Images