Food vacuumizing control system based on intelligent chip

Abstract

The invention relates to the technical field of intelligent control and food packaging, in particular to a food vacuumizing control system based on an intelligent chip, and the system comprises a data collection module which obtains the instantaneous air pressure of an air path and a motor load current sequence; the benchmark modeling module is used for generating an ideal benchmark data sequence based on hydrodynamic force; the simulation and residual calculation module is used for injecting abnormal working condition parameters to generate a simulation data sequence, and calculating a first residual data sequence and a second residual data sequence of reality and simulation relative to the benchmark respectively; the strategy decision module is used for carrying out similarity matching on the residual error sequence and outputting a control strategy signal according to the matching degree; the problems of false vacuum misjudgment and excessive pumping pressure loss are effectively solved.

Description

Technical Field

[0001] This invention relates to the field of intelligent control and food packaging technology, specifically to a food vacuum control system based on a smart chip. Background Technology

[0002] With the widespread application of intelligent control technology in the field of food preservation, vacuum packaging systems have become an important piece of equipment for extending the shelf life of food. In order to achieve efficient and non-destructive preservation, modeling analysis and precise control of the vacuuming process have become particularly important.

[0003] In existing vacuum control technologies, systems typically rely on absolute values ​​returned by pressure sensors to determine the vacuum level. However, the physical process of pumping air involves complex fluid dynamics changes, and the single sensor data dimension makes it difficult for the control system to perceive the true state inside the container in real time. In real-world scenarios, such as nozzle blockage caused by packaging bag adsorption or volume collapse caused by deformation of soft packaging, the pressure change curves exhibit nonlinearity and similarity. In particular, the volume of soft packaging will nonlinearly shrink as the pressure decreases under negative pressure, resulting in a dynamic reduction in the effective pumping volume. This leads to a pressure drop rate characteristic that is significantly different from that of rigid containers, which is difficult to distinguish effectively using traditional threshold logic. This can easily lead to false vacuum misjudgment or damage to soft foods due to excessive pumping.

[0004] Therefore, how to perform in-depth processing of the collected time-series data, construct a mathematical model that conforms to the fluid operation mechanism, and accurately identify different abnormal operating conditions by the difference between the actual operating data and the ideal model data, and then construct corresponding adaptive control strategies, is crucial to ensuring the intelligence and reliability of the food vacuum system. Summary of the Invention

[0005] To solve the above-mentioned technical problems, the present invention provides a food vacuum control system based on a smart chip. Specifically, the technical solution of the present invention includes:

[0006] As the core of the computing process, the data processing chip is communicatively connected to a data acquisition module, a benchmark modeling module, a simulation and residual calculation module, and a strategy decision-making module.

[0007] The data acquisition module is used to acquire and output a time series dataset characterizing the operating status of the vacuum system. The time series dataset includes at least the instantaneous gas pressure sequence of the gas path and the motor load current sequence.

[0008] The benchmark modeling module is used to perform modeling calculations based on preset fluid dynamics principles and ideal system parameters, and generate and output an ideal benchmark data sequence representing the ideal vacuuming process.

[0009] The simulation and residual calculation module is used to: inject at least one set of preset abnormal operating condition parameters into the model used by the benchmark modeling module, perform simulation calculations, and generate at least one abnormal operating condition simulation data sequence; calculate a first residual data sequence between the time series dataset and the ideal benchmark data sequence; and calculate a second residual data sequence between the at least one abnormal operating condition simulation data sequence and the ideal benchmark data sequence.

[0010] The strategy decision module is used to: perform similarity matching calculation on the first residual data sequence and the second residual data sequence based on waveform morphology features or time sequence features to obtain at least one matching degree value; and execute predefined decision logic according to the comparison result of the at least one matching degree value and a preset threshold, and output the corresponding vacuuming condition judgment result and control strategy signal.

[0011] Preferably, the process by which the data acquisition module acquires the time series dataset includes:

[0012] The system receives air pressure sampling data streams from the air pressure sensor via a preset data interface and formats them into an instantaneous air pressure sequence for the air path.

[0013] The current sampling value stream from the current detection unit is received synchronously through a preset data interface and formatted into the motor load current sequence.

[0014] Preferably, the process by which the benchmark modeling module generates the ideal benchmark data sequence includes:

[0015] Recall the stored initial system parameters, which include at least the nominal volume of the pumping container, the nominal cross-sectional area of ​​the flow channel, and the rated power of the vacuum pump;

[0016] Based on the initial system parameters, iterative calculations are performed using the ideal gas law and fluid motion equations to generate a time-varying sequence of theoretical gas pressure values ​​and a sequence of theoretical motor load values, which together constitute the ideal reference data sequence.

[0017] Preferably, the process by which the simulation and residual calculation module generates the abnormal operating condition simulation data sequence includes:

[0018] Obtain the first set of abnormal operating condition parameters. The first set of parameters is used to simulate the reduction of the flow channel cross-sectional area. Based on this, the model parameters in the benchmark modeling module are corrected and the first simulation calculation is performed to generate the first simulation data sequence.

[0019] A second set of abnormal operating condition parameters is obtained. The second set of parameters is used to simulate the nonlinear relationship between container volume and air pressure. Based on this, the model in the benchmark modeling module is corrected and a second simulation calculation is performed to generate a second simulation data sequence.

[0020] Preferably, the process by which the simulation and residual calculation module calculates the first residual data sequence and the second residual data sequence is a point-by-point subtraction operation or a least squares fitting difference operation performed in the time series data domain.

[0021] Preferably, the process of similarity matching calculation performed by the strategy decision module includes:

[0022] Calculate the cross-correlation coefficient or dynamic time warping distance between the first residual data sequence and the second residual data sequence corresponding to the first simulation data sequence, as the first matching degree;

[0023] The cross-correlation coefficient or dynamic time warping distance between the first residual data sequence and the second residual data sequence corresponding to the second simulation data sequence is calculated as the second matching degree.

[0024] Preferably, the process of the strategy decision module outputting the control strategy signal is configured to execute the following logical closed loop:

[0025] If the first matching degree is greater than the first preset threshold, the working condition judgment result representing the airway blockage and the first control strategy signal corresponding to the start of the backflush procedure are output.

[0026] If the first matching degree is not greater than the first preset threshold and the second matching degree is greater than the second preset threshold, then output the working condition judgment result representing the deformation of the soft packaging and the second control strategy signal corresponding to the early termination of vacuuming and sealing.

[0027] If both the first matching degree and the second matching degree are lower than the third preset threshold, and the final state value of the instantaneous gas pressure sequence of the gas path meets the preset vacuum condition, then output the working condition judgment result representing the vacuum achievement and the third control strategy signal corresponding to the execution of standard sealing.

[0028] If any of the above conditions are not met, a signal to maintain the current operating state will be output.

[0029] The third preset threshold is less than the first preset threshold and the second preset threshold.

[0030] Compared with the prior art, the present invention has the following beneficial effects:

[0031] 1. This system constructs a mathematical model based on fluid dynamics principles within the chip, uses synthetic analysis to generate ideal baseline data and abnormal simulation data in real time, and performs residual topology comparison to achieve a deep deconstruction of physical conditions. It achieves the effect of no longer relying solely on the absolute value of the pressure sensor to determine the vacuum level. Compared with the defects of existing technologies, such as false vacuum misjudgment caused by the nozzle being sucked up, this invention can accurately identify the system operating status by comparing the difference between the actual collected data and the ideal baseline data, thus solving the failure problem of traditional control logic under complex fluid changes.

[0032] 2. This system synchronously receives air pressure and current signals through the high-precision interface of the data acquisition module, and performs spatiotemporal alignment, filtering and noise reduction, and physical dimension standardization processing to achieve accurate fusion of multi-source heterogeneous data; it achieves the effect of constructing a multi-dimensional state space reflecting the fluid state and pump working state; compared with the shortcomings of existing technologies where a single data dimension is difficult to distinguish between air path blockage and actual load, this invention provides physically complete data support for subsequent logic through the joint analysis of current and air pressure, and solves the problem of ambiguous working condition identification caused by a single data dimension.

[0033] 3. This system employs a parameter correction method through simulation and residual calculation modules to actively inject fault parameters simulating channel cross-sectional reduction and nonlinear deformation of container volume into the model, achieving constitutive-level simulation of two physical phenomena: channel blockage and flexible packaging deformation. It achieves the effect of generating theoretical residual sequences of abnormal operating conditions with high physical fidelity. Compared to existing technologies that cannot identify the characteristics of flexible foods, this invention can accurately distinguish different anomaly types by comparing the similarity between actual residuals and theoretical fault residuals, solving the problem of excessive pressure loss due to the inability to identify flexible packaging. Attached Figure Description

[0034] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0035] Figure 1 This is a structural diagram of the system of the present invention. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0037] Example 1:

[0038] Please see Figure 1 The food vacuum control system based on smart chips includes a data processing chip as the computing core, which is connected to a data acquisition module, a benchmark modeling module, a simulation and residual calculation module, and a strategy decision-making module.

[0039] The data acquisition module is used to acquire and output a time series dataset that characterizes the operating status of the vacuum system. The time series dataset includes at least the instantaneous gas pressure sequence of the gas path and the motor load current sequence.

[0040] The benchmark modeling module is used to perform modeling calculations based on preset fluid dynamics principles and ideal system parameters, and generate and output an ideal benchmark data sequence representing the ideal vacuuming process;

[0041] The simulation and residual calculation module is used to: inject at least one set of preset abnormal operating condition parameters into the model used by the baseline modeling module, perform simulation calculations, and generate at least one abnormal operating condition simulation data sequence; calculate the first residual data sequence between the time series dataset and the ideal baseline data sequence; and calculate the second residual data sequence between at least one abnormal operating condition simulation data sequence and the ideal baseline data sequence.

[0042] The strategy decision module is used to: perform similarity matching calculations based on waveform morphology features or time sequence features on the first residual data sequence and the second residual data sequence to obtain at least one matching degree value; and execute predefined decision logic based on the comparison result of at least one matching degree value and a preset threshold, and output the corresponding vacuuming condition judgment result and control strategy signal.

[0043] This embodiment elaborates on the core architecture and operating mechanism of the above-mentioned food vacuum control system based on intelligent chips. The system uses the synthetic analysis concept to build a fluid dynamics model inside the chip, generates ideal benchmark and abnormal simulation data in real time, and compares the residual topology with the actual collected data. The system is based on a data processing chip, which adopts a microcontroller MCU or edge computing SoC with a floating-point unit (FPU) to support high-frequency fluid dynamics calculation.

[0044] The data acquisition module converts analog signals from the physical world into digital time series data, outputting a time series dataset. Subscript here Explicitly defined as input, this means that the dataset serves as the raw input data for subsequent algorithmic processes. This dataset contains instantaneous gas pressure sequences reflecting real-time pressure changes within the vacuum chamber or gas path. And the motor load current sequence reflecting the torque characteristics of the vacuum pump motor under different loads. ;

[0045] The benchmark modeling module performs real-time modeling calculations within the chip based on preset fluid dynamics principles, such as the ideal gas law and Bernoulli's equation, as well as ideal parameters for a leak-free and unblocked system, generating and outputting an ideal benchmark data sequence. This sequence represents the pressure drop curve and current change curve that the system should exhibit under perfect physical conditions. Based on this, the simulation and residual calculation module executes active fault injection logic, injecting preset abnormal operating condition parameters into the mathematical model used by the baseline modeling module, such as the flow resistance coefficient simulating blockage or the volumetric deformation modulus simulating flexible packaging, generating at least one abnormal operating condition simulation data sequence. The difference between the actual collected data and the ideal benchmark data is calculated, which is the first residual data sequence. And the difference between the abnormal simulation data and the ideal baseline data, i.e., the second residual data sequence. Furthermore, the strategy decision-making module will address deviations from reality. Deviation from the theory of failure A comparison is performed, and the matching degree value is calculated using algorithms based on waveform morphology features or time series features. And accordingly, perform closed-loop control;

[0046] This embodiment introduces a dual comparison mechanism of ideal benchmark and abnormal simulation, no longer relying solely on the absolute value of the air pressure sensor to determine the vacuum level. This effectively solves the problem of false vacuum misjudgment caused by the nozzle being sucked by the packaging bag, as well as the problem of excessive air extraction pressure loss caused by the inability to identify soft foods, and achieves in-depth deconstruction and accurate identification of physical working conditions.

[0047] The process by which the data acquisition module obtains the time series dataset includes:

[0048] The system receives air pressure sampling data streams from the air pressure sensor via a preset data interface and formats them into an instantaneous air pressure sequence.

[0049] The current sampling value stream from the current detection unit is received synchronously through the preset data interface and formatted into a motor load current sequence.

[0050] This embodiment further clarifies the spatiotemporal alignment of the data acquisition module's acquisition of multi-source heterogeneous data; the data acquisition module is equipped with a high-precision analog-to-digital converter (ADC), which receives sampled data streams from piezoresistive or capacitive barometric pressure sensors via a preset I2C or SPI data interface; the module operates at a sampling frequency... Discretize the data and perform a moving average filter; the filter window length is defined here. The sampling period is 5 to 10 times, which is specified here. The frequency at which the system samples the analog signal, such as 1kHz. The number of sampling points contained in the sliding window used for noise smoothing is calculated using the following formula:

[0051]

[0052] in, express or At any moment The original sampled values, The value is obtained after applying a moving average filter to suppress high-frequency quantization noise, and then formatted as a sequence of instantaneous air pressure in the gas path. It should be noted here that, in order to ensure consistency with the ideal gas equation of state model in the benchmark modeling module in terms of the physical domain, if the sensor output is relative gauge pressure, the data acquisition module is configured to add a preset local atmospheric pressure constant to the filtered sampled value. For example, 101.3 kPa, thus making The data is standardized into an absolute pressure sequence. Simultaneously, through a preset analog input interface, it synchronously receives voltage mapping values ​​from a current detection unit (either a shunt resistor or a Hall sensor). These values ​​are then linearly transformed and formatted into a motor load current sequence. The transformation formula is:

[0053]

[0054] in, Zero current reference voltage, To calibrate the conversion coefficients; the data acquisition module performs the calibration before output. and Perform timestamp synchronization to ensure the same index. The air pressure and current values ​​correspond to the same physical moment.

[0055] This embodiment constructs a multi-dimensional state space by synchronously collecting air pressure and current data. It utilizes the characteristics that air pressure reflects the fluid state and current reflects the pump's work state to provide the necessary data foundation for distinguishing between low air pressure drop and current caused by air path blockage and high air pressure drop and current caused by actual load, thereby ensuring the physical completeness of the judgment logic from the source.

[0056] The process by which the benchmark modeling module generates an ideal benchmark data sequence includes:

[0057] Recall the stored initial system parameters, which include at least the nominal volume of the pumping container, the nominal cross-sectional area of ​​the flow channel, and the rated power of the vacuum pump;

[0058] Based on the initial system parameters, iterative calculations are performed using the ideal gas law and fluid motion equations to generate time-varying sequences of theoretical gas pressure and theoretical motor load, which together constitute an ideal reference data sequence.

[0059] This embodiment details how the baseline modeling module constructs a mathematical twin of the system, and focuses on solving the explicit coupling problem between the flow channel geometry parameters and the fluid dynamics equations; the module calls pre-stored initial system parameters. This parameter set includes the nominal initial volume of the evacuation container or vacuum bag. The nominal cross-sectional area of ​​the flow channel, i.e., the nozzle and tubing. and the nominal pumping rate curve corresponding to the rated power of the vacuum pump. It should be clarified here that, given the difference in physical dimensions between rated power and pumping rate, in order to ensure the accuracy of the physical quantities required for modeling, this embodiment will use parameters... Strictly defined as the pumping rate curve, and for code-level reproducibility, the nominal pumping rate curve... It is instantiated in the chip memory as a sheet containing A lookup table for each node, or LUT for short, is denoted as . ,in, For node pressure, For the corresponding extraction speed, look up the table node. Typically, a non-uniformly spaced distribution is used, with more frequent sampling in low-pressure areas to improve accuracy. A typical value... The range is 20 to 50, specifically defined herein, and the symbol used in this embodiment is... Refers to the volumetric pumping rate, in units of It is not the gas flux;

[0060] To ensure that the nominal cross-sectional area of ​​the flow channel truly participates in the calculation, the module calculates the aerodynamic conductance of the air path system based on the flow resistance principle in the fluid motion equation. The calculation formula is: To correct the dimensions and clarify the physical meaning, the following definition is provided: For the unit Volumetric flow conductance of the gas path system For the unit The nominal cross-sectional area of ​​the flow channel, For the unit The characteristic conductivity velocity coefficient, which physically characterizes the equivalent transport rate of gas molecules under a specific pipeline geometry; that is... This can be considered as the equivalent average velocity of gas flow within the flow channel under typical operating pressure differential; parameters The following offline calibration method was used: In a standard laboratory environment, the system gas path was connected to a calibration platform equipped with a mass flow controller and a high-precision differential pressure transmitter. The vacuum pump was controlled to run at its rated speed, and the inlet flow rate was adjusted until the pressure difference across the pipeline stabilized at a typical operating point, such as 10 kPa. The standard volumetric flow rate at this point was recorded. and pressure difference Calculate the measured flow conductance And thus, it can be deduced For common food vacuum packaging machine nozzle tubing, Typical empirical values ​​range from 30 m / s to 120 m / s, thus effectively eliminating model errors caused by pipeline manufacturing tolerances; the module calculates the effective pumping rate acting on the container. This rate is determined by the theoretical pumping speed of the vacuum pump. With flow conduction The series connection is determined by the reciprocal superposition rule in vacuum technology:

[0061]

[0062] Among them, the function Specific execution is based on Linear interpolation operation: ,satisfy To ensure the robustness of numerical calculations under excessive pressure, the interpolation logic also includes boundary saturation handling: when At that time, take ;when At that time, take Based on this, iterative calculations of gas pressure are performed using the discretized ideal gas equation of state to obtain the ideal gas pressure value. The iterative formula is as follows: During execution Before iterative calculations at each time step, the system initializes and assigns values ​​to the state variables, setting... That is, the preset local atmospheric pressure, such as At the same time, the initial theoretical current is set. Based on time step Perform the following recursive calculation:

[0063]

[0064] In particular, to prevent negative calculated air pressure values ​​due to excessively large step sizes during discrete iteration, the system enforces constraints on the calculation step size. Satisfying the CFL-like, i.e., Courant-Friedrichs-Lewy stability condition:

[0065]

[0066] in, A preset safety factor, for example, 0.05, is used to ensure that the single-step pressure drop does not exceed 5% of the current pressure; The empirical value range is typically from 0.01 to 0.1, and its selection principle needs to balance computational stability and real-time performance; among which For lookup table The pumping velocity set in;

[0067] Theoretical motor load value sequence Based on the pump's load characteristic curve Generation, i.e. Mapping function Specifically, it is constructed as a third-order polynomial model.

[0068]

[0069] in, The unit is The coefficients are determined by experimental fitting. For conventional rotary vane vacuum pumps, the coefficients are... Typically, it takes a positive value and is on the order of magnitude approximately... To reflect the increase in nonlinear load under high pressure, the constant term Corresponding to the no-load current under ultimate vacuum;

[0070] This embodiment introduces... Intermediate variables, explicit LUT interpolation logic, and strict step size constraints establish the parameters. , With output sequence Strict mathematical mapping and numerical stability guarantees between them.

[0071] The process by which the simulation and residual calculation module generates simulation data sequences for abnormal operating conditions includes:

[0072] The first set of abnormal operating condition parameters is obtained. The first set of parameters is used to simulate the reduction of the flow channel cross-sectional area. Based on this, the model parameters in the benchmark modeling module are corrected and the first simulation calculation is performed to generate the first simulation data sequence.

[0073] The second set of abnormal operating condition parameters is obtained. The second set of parameters is used to simulate the nonlinear relationship between container volume and air pressure. Based on this, the model in the benchmark modeling module is corrected and the second simulation calculation is performed to generate the second simulation data sequence.

[0074] This embodiment illustrates the specific process of actively simulating physical faults using a parameter correction method, covering two typical scenarios: flow channel blockage and soft packaging deformation. For flow channel blockage simulation, the module acquires the first set of abnormal operating condition parameters, namely the blockage factor with values ​​ranging from 0 to 1. This parameter is used to simulate the reduction of the flow channel cross-sectional area; to ensure consistency between the simulation logic and the physical mechanism, the module directly modifies the geometric parameters in the baseline model. Perform a correction operation to define the equivalent cross-sectional area under blocked conditions. ;

[0075] The system call's flow guidance calculation logic recalculates the blocking flow guidance. And update the effective pumping rate accordingly. ;Will Substituting into the air pressure iteration formula generates the first simulation data sequence. This process accurately simulates the increased air resistance and decreased pumping speed caused by cross-sectional reduction. Simultaneously, for the deformation simulation of flexible packaging, the module acquires a second set of abnormal operating condition parameters, namely the flexible deformation volume coefficient. This parameter is used to simulate the nonlinear collapse of the container volume as the air pressure decreases. To address the dimension matching problem and ensure the rigor of the physical model, this embodiment constructs a corrected volume model based on reference pressure normalization, as shown in the following formula:

[0076]

[0077] in, The volume of the dynamic container during the simulation process, in units of... ; Atmospheric pressure; Unit reference pressure, set as This is used to eliminate the physical dimensions of the base of the exponentiation operation, so that the formula satisfies the principle of dimensional consistency. The volume coefficient for flexible deformation, thanks to the above normalization process, has a constant unit of volume. No longer dependent on index Material deformation index, a dimensionless constant; This characterizes the nonlinear hardening or softening properties of packaging materials under negative pressure, with values ​​typically ranging from 0.5 to 1.5; different materials... The values ​​vary; for example, the typical empirical value for PE bags is around 1.2, while the typical empirical value for aluminum foil bags is closer to 0.8. The method for determining this parameter is as follows: In an offline calibration environment, a gradient vacuum test is performed on packaging bags of specific materials and thicknesses. The actual internal volume data points of the bag under different air pressures are recorded using the displacement method or laser scanning method. Construct the objective function The Levenberg-Marquardt nonlinear optimization algorithm is used for iterative fitting until the objective function converges, thereby accurately obtaining the specific packaging material. Value and value;

[0078] This embodiment introduces explicitly into the floating-point unit (FPU) of the chip. Normalizing division eliminates the physical unit of the base in exponentiation, making the parameters... In a physical sense, it reverts to a pure volume quantity; when writing code, technicians can directly... Defined as a double-precision floating-point volume variable, this resolves the ambiguity in parameter assignment; it should be noted that, to ensure the physical solvability of the mathematical model, the base of the exponentiation operation is adjusted during the calculation. Enforcing nonnegativity constraints, i.e., taking and the calculation results Implement minimum volume limit to ensure To prevent iterative divergence caused by parameter distortion; based on this dynamic volume Substituting into the iterative formula, the second simulation data sequence is generated. In this process, in order to accurately reproduce the physical phenomenon of reduced air extraction efficiency when the flexible packaging is deflated, the simulation algorithm forces the use of the dynamic volume calculated at the current moment when performing each step of the air pressure iteration calculation. The fixed initial volume parameter in the replacement principle model formula That is, execution This ensures that the simulation data includes the characteristic distortions introduced by the nonlinear volume shrinkage; this embodiment is not merely a passive detection method, but rather modifies the geometric parameters at the underlying level of the mathematical model. and volume parameters The definition method enables constitutive-level simulation of two physical phenomena, obstruction and deformation, ensuring that the generated simulation data has a high degree of physical fidelity.

[0079] The process of the simulation and residual calculation module calculating the first residual data sequence and the second residual data sequence is a point-by-point subtraction operation or a least squares fitting difference operation performed in the time series data domain.

[0080] This embodiment specifies the calculation method for residual data sequences, aiming to extract pure anomaly features; the simulation and residual calculation module performs operations in the time-series data domain, for the first residual data sequence This refers to the actual residual. To comprehensively utilize the dual characteristics of air pressure and current, this embodiment defines it as the normalized weighted fusion residual, calculated using the following formula:

[0081]

[0082] in, and These are preset weighting coefficients, for example, both set to 0.5. These two weights can be adjusted according to the actual system characteristics. For example, if the barometric pressure sensor has high accuracy and low noise, they can be appropriately increased. ; As a reference pressure constant, the local atmospheric pressure is used. , As a reference current constant, the rated current value of the motor is taken; similarly, for the subsequent theoretical residual sequence... and Perform the same weighted fusion calculation, that is:

[0083]

[0084] in, and These correspond to the motor load current sequences generated in the first and second simulation calculations, respectively, and their calculation methods reuse the load characteristic curves. ;

[0085] For the second residual data sequence, in the specific scenario of this embodiment, it is not a single-dimensional array, but refers to the set of residuals generated by simulation calculation; in order to accurately distinguish the fault type, this embodiment specifically instantiates this set into two independent sub-sequences: the first theoretical residual sequence corresponding to flow channel blockage. and the second theoretical residual sequence corresponding to the deformation of the flexible packaging ;in, and The numerical calculation strictly follows the above weighted fusion formula, which includes deviation information in both the pressure dimension and the current dimension.

[0086] The system performs point-by-point subtraction, or in the preferred mode, to smooth out noise, it uses least squares fitting difference, which calculates the root mean square error sequence between the measured curve and the reference curve within a sliding window. The specific algorithm is as follows:

[0087]

[0088] in, The width of the preset sliding window, for example, 20 sampling points. The value should be selected based on the noise frequency, and a balance needs to be struck between noise smoothing effect and fault response speed. It is usually taken as 1-2 times the system's main frequency period. Reference or , Reference , This refers to the index of the sampling points within the sliding window;

[0089] This embodiment removes the common-mode signal of the normal pumping process through differential operation, leaving residuals that purely reflect abnormal features such as sudden spikes or gradual shifts. This processing method greatly improves the signal-to-noise ratio, making minute physical anomalies clearly visible in the data, thereby enhancing the robustness of the system under complex operating conditions.

[0090] The similarity matching calculation process performed by the strategy decision module includes:

[0091] Calculate the cross-correlation coefficient or dynamic time warping distance between the first residual data sequence and the second residual data sequence corresponding to the first simulation data sequence, as the first matching degree;

[0092] The cross-correlation coefficient or dynamic time warping distance between the first residual data sequence and the second residual data sequence corresponding to the second simulation data sequence is calculated as the second matching degree.

[0093] This embodiment details the process of quantifying fault feature similarity using statistical tools and defines the time domain range of the algorithm execution to ensure real-time performance. To determine whether the actual residuals conform to a certain theoretical fault characteristic, the module calculates two scalar indices. The definitions of the relevant sequence and its subscript are clarified here: The first simulation data sequence Compared with ideal benchmark The difference corresponds to the flow channel blockage model; For the second simulation data sequence Compared with ideal benchmark The difference corresponds to the soft-pack deformation model; in order to adapt to the limited memory resources of the MCU and meet the real-time control requirements, the similarity matching calculation is not performed on the full historical data, but on a sliding observation window of fixed length. For example, it is executed within the last 64 sampling points; system maintenance is targeted at... , and The circular buffer (RingBuffer) triggers computation only when the buffer is full, and performs FIFO updates as new data arrives; for blocking conditions, the system computation window... and The cross-correlation coefficient is used as the first matching degree. The calculation formula, with the addition of a small regularization term, is as follows:

[0094]

[0095] in, To prevent tiny positive numbers with a denominator of zero, for example The principle for selecting this value is that it should be much smaller than... and Typical variance of the sequence; simultaneously, for soft packaging deformation, the system calculates the variance within the window. and The transformed value of the dynamic time-warped distance (DTW) between them is used as the second degree of matching. ; Module build size is Distance matrix ,element Using dynamic programming equations Recursively calculate the cumulative distance matrix To prevent division by zero due to a perfect match, and to standardize the units of measurement, this embodiment uses the following improved normalized similarity conversion formula for calculation. :

[0096]

[0097] In order to obtain accurate The system starts from the end of the cumulative distance matrix. Execute the greedy backtracking algorithm: Initialize the current coordinates and path counting Check in the loop Conditions, each selection , and The index update corresponding to the minimum of the three. and will Increment by 1 until traversing back to the starting point. If limited by chip computing power, a simplified approximation formula can also be used:

[0098]

[0099] in, This is an approximation of the equivalent geometric length of the backtracking path. For example, a preset dimensionless normalization constant. Its purpose is to map the DTW distance to a similarity interval compatible with the dimensions of the correlation coefficient, and its value is usually selected as the typical operating pressure difference of the system.

[0100] This embodiment reduces the computational complexity that would otherwise increase over time by explicitly defining a sliding window mechanism. or Constraints are of constant order This ensures the feasibility and operational stability of the algorithm on embedded chips.

[0101] Example 2:

[0102] The process of the strategy decision module outputting control strategy signals is configured to execute the following logical closed loop:

[0103] If the first matching degree is greater than the first preset threshold, the operating condition judgment result representing the airway blockage and the first control strategy signal corresponding to the start of the backflush procedure are output.

[0104] If the first matching degree is not greater than the first preset threshold and the second matching degree is greater than the second preset threshold, then output the working condition judgment result representing the deformation of the soft packaging and the second control strategy signal corresponding to the early termination of vacuuming and sealing.

[0105] If both the first matching degree and the second matching degree are lower than the third preset threshold, and the final state value of the instantaneous gas pressure sequence of the gas path meets the preset vacuum condition, then output the working condition judgment result representing the vacuum achievement and the third control strategy signal corresponding to the execution of the standard seal.

[0106] If any of the above conditions are not met, a signal to maintain the current operating state will be output.

[0107] The third preset threshold is less than the first preset threshold and the second preset threshold.

[0108] This embodiment constructs an intelligent decision-making closed loop based on multi-threshold judgment to achieve differentiated processing for different working conditions; the module is configured with three preset thresholds, namely the blocking judgment threshold. Deformation determination threshold and normal judgment threshold And satisfy and The numerical relationship is described here; the method for determining the threshold and typical values ​​are specifically disclosed to ensure the feasibility of the solution: blocking determination threshold. Based on the statistical significance level setting, by collecting no less than 1000 sets of normal operating condition data in the offline phase, calculating the cross-correlation coefficient distribution between the data and the blocking model, and taking the upper limit of the 99% confidence interval, the typical setting value is 0.85; deformation judgment threshold. The standard is based on the experimentally determined boundary of material deformation characteristics, specifically the normalized DTW similarity measured at the known physical critical point where soft packaging collapses, with a typical set value of 0.75; the normal judgment threshold. The threshold is used as the noise floor tolerance of the system. It is the maximum accidental matched noise level when the system is running under no-load conditions. The typical setting value is 0.3. It should be noted that the above threshold can be used as a configurable system parameter. It needs to be factory calibrated or adaptively adjusted for each machine or each major packaging type to construct a clear judgment boundary in the numerical space.

[0109] The system executes the hierarchical judgment logic; responding to the first matching degree Greater than This indicates that the actual residual highly fits the blockage model, the system determines that the air path is blocked, and outputs the first control strategy signal to instruct the motor controller to execute the start-up backflush procedure; in response to and This indicates that the actual residual conforms to the characteristics of soft deformation. The system determines it to be soft packaging deformation and outputs a second control strategy signal to instruct the system to execute early termination of evacuation and sealing; in response to and All below And detected The final value has reached the preset vacuum level, indicating that the actual operating curve is very close to the ideal reference curve. The system determines that the vacuum has been achieved, that is, it confirms that the vacuuming process of Example 1 has been successfully completed and is in a stable final state, and outputs the third control strategy signal to instruct the heating bar to perform standard sealing; if the above conditions are not met, the system outputs a signal to maintain the current operating state and continue to sample the next frame of data.

[0110] Furthermore, in order to meet the deterministic requirement of system operation and prevent [further issues], Being in the decision dead zone for a long time, i.e. The logical closed loop, which causes the system to enter an infinite loop state, also includes a timeout circuit breaker mechanism: the system maintains a global runtime timer. ,like Exceeding the preset maximum safe pumping time For example, if 60 seconds have passed and none of the above judgment conditions are met, a control signal indicating an abnormal timeout and a stop alarm will be forcibly output. The timeout parameters and stability thresholds mentioned above are configurable and can be preset at the factory according to typical application scenarios. The preset vacuum condition is specifically defined as: the instantaneous gas pressure sequence in the gas path. The recent Each sampling point, for example moving average Below a preset absolute pressure threshold, for example, 20 kPa, which corresponds to approximately The gauge pressure vacuum degree, and the absolute value of its rate of change. Less than the preset stability threshold .

[0111] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A food vacuum control system based on a smart chip, characterized in that, It includes a data processing chip as the core of the computing process, which is communicatively connected to a data acquisition module, a benchmark modeling module, a simulation and residual calculation module, and a strategy decision-making module. The data acquisition module is used to acquire and output a time series dataset characterizing the operating status of the vacuum system. The time series dataset includes at least the instantaneous gas pressure sequence of the gas path and the motor load current sequence. The benchmark modeling module is used to perform modeling calculations based on preset fluid dynamics principles and ideal system parameters, and generate and output an ideal benchmark data sequence representing the ideal vacuuming process. The simulation and residual calculation module is used to: inject at least one set of preset abnormal operating condition parameters into the model used by the benchmark modeling module, perform simulation calculations, and generate at least one abnormal operating condition simulation data sequence; calculate a first residual data sequence between the time series dataset and the ideal benchmark data sequence; and calculate a second residual data sequence between the at least one abnormal operating condition simulation data sequence and the ideal benchmark data sequence. The strategy decision module is used to: perform similarity matching calculation on the first residual data sequence and the second residual data sequence based on waveform morphology features or time sequence features to obtain at least one matching degree value; and execute predefined decision logic according to the comparison result of the at least one matching degree value and a preset threshold, and output the corresponding vacuuming condition judgment result and control strategy signal.

2. The food vacuum control system based on a smart chip according to claim 1, characterized in that, The process by which the data acquisition module acquires the time series dataset includes: The system receives air pressure sampling data streams from the air pressure sensor via a preset data interface and formats them into an instantaneous air pressure sequence for the air path. The current sampling value stream from the current detection unit is received synchronously through a preset data interface and formatted into the motor load current sequence.

3. The food vacuum control system based on a smart chip according to claim 1, characterized in that, The process by which the benchmark modeling module generates the ideal benchmark data sequence includes: Recall the stored initial system parameters, which include at least the nominal volume of the pumping container, the nominal cross-sectional area of ​​the flow channel, and the rated power of the vacuum pump; Based on the initial system parameters, iterative calculations are performed using the ideal gas law and fluid motion equations to generate a time-varying sequence of theoretical gas pressure values ​​and a sequence of theoretical motor load values, which together constitute the ideal reference data sequence.

4. The food vacuum control system based on a smart chip according to claim 1, characterized in that, The process by which the simulation and residual calculation module generates simulation data sequences for abnormal operating conditions includes: Obtain the first set of abnormal operating condition parameters. The first set of parameters is used to simulate the reduction of the flow channel cross-sectional area. Based on this, the model parameters in the benchmark modeling module are corrected and the first simulation calculation is performed to generate the first simulation data sequence. A second set of abnormal operating condition parameters is obtained. The second set of parameters is used to simulate the nonlinear relationship between container volume and air pressure. Based on this, the model in the benchmark modeling module is corrected and a second simulation calculation is performed to generate a second simulation data sequence.

5. The food vacuum control system based on a smart chip according to claim 4, characterized in that, The process by which the simulation and residual calculation module calculates the first residual data sequence and the second residual data sequence is a point-by-point subtraction operation or a least squares fitting difference operation performed in the time series data domain.

6. The food vacuum control system based on a smart chip according to claim 5, characterized in that, The process of similarity matching calculation performed by the strategy decision module includes: Calculate the cross-correlation coefficient or dynamic time warping distance between the first residual data sequence and the second residual data sequence corresponding to the first simulation data sequence, as the first matching degree; The cross-correlation coefficient or dynamic time warping distance between the first residual data sequence and the second residual data sequence corresponding to the second simulation data sequence is calculated as the second matching degree.

7. The food vacuum control system based on a smart chip according to claim 6, characterized in that, The process of the strategy decision module outputting control strategy signals is configured to execute the following logical closed loop: If the first matching degree is greater than the first preset threshold, the working condition judgment result representing the airway blockage and the first control strategy signal corresponding to the start of the backflush procedure are output. If the first matching degree is not greater than the first preset threshold and the second matching degree is greater than the second preset threshold, then output the working condition judgment result representing the deformation of the soft packaging and the second control strategy signal corresponding to the early termination of vacuuming and sealing. If both the first matching degree and the second matching degree are lower than the third preset threshold, and the final state value of the instantaneous gas pressure sequence of the gas path meets the preset vacuum condition, then output the working condition judgment result representing the vacuum achievement and the third control strategy signal corresponding to the execution of standard sealing. If any of the above conditions are not met, a signal to maintain the current operating state will be output. The third preset threshold is less than the first preset threshold and the second preset threshold.

Images