Control method and system for maintaining vacuum state in shutdown vacuum preservation state
By using GM(1,1) model and rolling modeling to predict air pressure changes during shutdown vacuum storage, and dynamically adjusting the vacuum pump control scheme, the problem of poor air pressure control in traditional methods is solved, achieving more efficient and stable vacuum environment maintenance.
Patent Information
- Application Number
- CN202411960710.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-26
- Publication Date
- 2025-05-02
Smart Images
Figure CN119914509A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of battery baking, and in particular to a control method for maintaining a vacuum state in a shutdown vacuum storage state and a system for maintaining a vacuum state in a shutdown vacuum storage state. Background Art
[0002] In the current battery post-processing and storage process, in order to keep the battery surface dry and avoid secondary intrusion of moisture or impurities in the air, it is often necessary to vacuum store the baked batteries in the shutdown state. Traditional shutdown vacuum storage mainly relies on a simple upper and lower threshold control strategy: when the air pressure in the baking chamber cavity gradually rises to a certain upper limit due to a small air leak in the equipment after shutdown, the vacuum pump is started to reduce the air pressure to the set lower limit, and then the vacuum pump is turned off to wait for the next upper limit trigger point. However, this control method has many technical difficulties and shortcomings.
[0003] Traditional shutdown vacuum preservation can only intervene passively when the air pressure reaches the upper threshold, and lacks prediction of the trend of air pressure changes over time. When the environmental conditions or equipment status change (such as aging of the sealing ring, room temperature changes, etc.) and the leakage rate becomes unstable, reactive control that relies solely on fixed upper and lower limits will cause the vacuum pump to start and stop frequently. Not only does it increase energy consumption and equipment wear, it also causes the air pressure to fluctuate within a large range, affecting the quality maintenance of the battery in the shutdown state. The existing solution does not perform trend analysis or predictive modeling on the air pressure data. Due to the lack of predictive capabilities, the system can only take remedial measures when a problem occurs (the air pressure reaches the upper limit), but cannot take proactive preventive actions before the air pressure approaches the upper limit. This lag makes the air pressure control lack refined management, and it is difficult to maintain a stable vacuum environment within a smaller fluctuation range.
[0004] In addition, current control methods often lack dynamic update and adaptive mechanisms. After the initial setting, the general scheme is not optimized or adjusted as the real-time data changes. As a result, when the actual working conditions, device aging or environmental factors change, the system still uses the original parameters and cannot meet the precise control requirements under the new conditions. The result may be that the control effect gradually deviates from the ideal state, and the vacuum stability and energy saving are affected.
[0005] In summary, in order to solve the above problems in the shutdown vacuum preservation process, a prediction model and dynamic control mechanism based on real-time data is needed, so that the system can predict the trend of air pressure changes in advance and actively intervene at the most appropriate time, thereby reducing the frequent start and stop of the vacuum pump, reducing energy consumption, extending the life of the equipment, and continuously providing the battery with a better and more stable vacuum environment in the shutdown state. Summary of the invention
[0006] The purpose of the embodiments of the present invention is to provide a control method and system for maintaining a vacuum state in a shutdown vacuum preservation state, so as to at least solve the problems of low maintenance accuracy and insufficient response speed in existing vacuum state maintenance solutions.
[0007] In order to achieve the above-mentioned objectives, the first aspect of the present invention provides a control method for maintaining a vacuum state in a shutdown vacuum preservation state, which is applied to the control of maintaining the vacuum state in a shutdown vacuum preservation state after battery baking. The method comprises: in response to a shutdown vacuum preservation trigger signal, performing air pressure information collection of a baking chamber at multiple time points within an initialization cycle; constructing a GM (1, 1) model for predicting the air pressure information at subsequent moments based on the air pressure information of the baking chamber at each time point as an initial prediction model; performing model verification on the initial prediction model based on the air pressure information of the baking chamber at multiple time points within a test cycle, and using the prediction model that passes the verification as the air pressure prediction model; during the shutdown vacuum preservation process, performing rolling modeling based on the air pressure prediction model, and using the real-time collected air pressure information of the baking chamber and all previous air pressure information as input parameters of the air pressure prediction model at the current moment, performing air pressure information prediction at the next moment, and generating and executing a control scheme for the vacuum pump based on the air pressure prediction value.
[0008] Optionally, the air pressure information of the baking chamber at multiple time points is executed within the initialization cycle, including: starting from the triggering moment of the shutdown vacuum preservation trigger signal, triggering the system initialization, and identifying the corresponding initialization cycle; performing air pressure information collection once at a predetermined time interval within the initialization cycle, and arranging the collection moments in chronological order to obtain an air pressure timing information set; performing clustering on the air pressure timing information set to identify the number of discrete points, and re-executing initialization when the data of the discrete points is greater than a preset threshold, until the data of the discrete points is less than the preset threshold, and outputting the air pressure information set.
[0009] Optionally, the GM(1,1) model for predicting the air pressure information at subsequent moments based on the air pressure information of the baking chamber at each time point is constructed as an initial prediction model, including: based on the GM(1,1) model rules, performing air pressure accumulation at each moment based on a time sequence to obtain an accumulation sequence; constructing a first-order linear differential equation approximation model based on the accumulation sequence to determine the model parameters to be determined; discretizing the first-order linear differential equation approximation model, and solving each parameter to be determined based on the discretized first-order linear differential equation approximation model; constructing a corresponding GM(1,1) model based on the solved parameters, and constructing a corresponding initial prediction model.
[0010] Optionally, the pressure accumulation rule at each moment is:
[0011]
[0012] Where n is the number of sampling points; x(1) (k) is the accumulated value of the air pressure at the kth sampling point sequence; x (0) (i) is the collected air pressure value at the position of the i-th sampling point sequence; the first-order linear differential equation approximate model is expressed as:
[0013]
[0014] Among them, a and b are the parameters to be determined for the model.
[0015] Optionally, the discretization of the first-order linear differential equation approximate model and solving each parameter to be determined based on the discretized first-order linear differential equation approximate model include: discretizing the first-order linear differential equation approximate model based on the background value sequence to obtain a discretized model; converting the discretized model into a matrix form, and solving the discretized model in matrix form based on the least squares method to obtain parameter values of each parameter to be determined; wherein the discretized model is expressed as:
[0016] x(0)(k)+az (1) (k) = b, k = 1, 2, ..., n
[0017] in, Represents the background value sequence of the kth sampling point sequence position.
[0018] Optionally, the initial prediction model is expressed as:
[0019] x (0) (n+1)=x (1) (n+1)-x (1) (n)
[0020] Among them, x (0) (n+1) is the predicted value of air pressure at time n+1; x (1) (n+1) and x (1) (n) are the accumulated air pressure values at the n+1th and nth moments generated based on the GM (1, 1) model.
[0021] Optionally, the initial prediction model is verified based on the air pressure information of the baking chamber at multiple time points within the test cycle, and the prediction model that passes the verification is used as the air pressure prediction model, including: within the test cycle, based on the initial prediction model, the air pressure value prediction of each preset moment is performed in sequence to obtain the air pressure prediction value of each preset moment; when each preset moment is reached, the actual air pressure value at each preset moment is compared with the air pressure prediction value at the corresponding preset moment, and when the absolute value of the difference between the two is less than a preset difference threshold, a valid prediction is recorded; when the valid predictions are recorded continuously for a preset number of times, it is determined that the initial prediction model has passed the verification; conversely, if the absolute value of the difference between the two is not less than the preset difference threshold, the air pressure information set within the initialization cycle is updated based on the actual air pressure value at the current moment to update the initial prediction model, until the valid predictions are recorded continuously for the preset number of times, and it is determined that the initial prediction model has passed the verification.
[0022] Optionally, during the shutdown vacuum preservation process, rolling modeling is performed based on the air pressure prediction model, and the real-time acquired air pressure information of the baking chamber and all previous air pressure information are used as input parameters of the air pressure prediction model at the current moment, and the air pressure information prediction at the next moment is performed, and the control scheme of the vacuum pump is generated and executed based on the air pressure prediction value, including: each time a new actual air pressure value is obtained, the air pressure information set is updated based on the actual air pressure value, and the air pressure prediction model is updated based on the air pressure information set as the air pressure prediction model for the next moment; the air pressure information at the current moment and all previous air pressure information are used as input parameters of the air pressure prediction model corresponding to the current moment, and the air pressure information prediction at the next moment is performed to obtain the air pressure prediction value for the next moment; based on the difference between the air pressure prediction value at the next moment and the preset standard air pressure value for shutdown vacuum preservation, the adjustment scheme of the vacuum pump is determined, and the corresponding control instruction is generated based on the adjustment scheme for the vacuum pump to execute the control instruction.
[0023] The second aspect of the present invention provides a control system for maintaining a vacuum state in a shutdown vacuum preservation state, which is applied to the vacuum state maintenance control in a shutdown vacuum preservation state after battery baking. The system includes: an acquisition unit, which is used to respond to a shutdown vacuum preservation trigger signal and perform air pressure information acquisition of a baking chamber at multiple time points within an initialization cycle; a model generation unit, which is used to construct a GM (1, 1) model for predicting the air pressure information at subsequent moments based on the air pressure information of the baking chamber at each time point, as an initial prediction model; a model verification unit, which is used to perform model verification on the initial prediction model based on the air pressure information of the baking chamber at multiple time points within a test cycle, and use the prediction model that passes the verification as the air pressure prediction model; a control unit, which is used to perform rolling modeling based on the air pressure prediction model during the shutdown vacuum preservation process, and use the real-time collected air pressure information of the baking chamber and all previous air pressure information as input parameters of the air pressure prediction model at the current moment, perform air pressure information prediction at the next moment, and generate and execute a control scheme for the vacuum pump based on the air pressure prediction value.
[0024] On the other hand, the present invention provides a computer-readable storage medium having instructions stored thereon, which, when executed on a computer, enables the computer to execute the above-mentioned control method for maintaining a vacuum state in a shutdown vacuum storage state.
[0025] Through the above technical solution, the solution of the present invention realizes dynamic and forward-looking control of the air pressure in the baking chamber during the shutdown vacuum preservation process. First, the GM (1, 1) model is constructed by initially collecting air pressure information at multiple time points, and the model is verified during the test cycle to ensure the prediction accuracy and reliability of the model. Subsequently, in the formal shutdown vacuum preservation, the system continuously updates the prediction model, combining the real-time collected air pressure data with historical data to improve the prediction ability. In this way, the trend of the air pressure can be accurately predicted before the next moment of the rise, thereby triggering the vacuum pump to work in advance or in time. This effectively reduces the energy waste and equipment wear caused by the frequent start and stop of the vacuum pump, helps to maintain a more stable fluctuation of the vacuum degree, and ensures that the quality of the battery in the shutdown state is not affected by changes in air pressure, thereby achieving higher process stability and reliability.
[0026] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS
[0027] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:
[0028] Figure 1It is a flowchart of the steps of a control method for maintaining a vacuum state in a shutdown vacuum storage state provided by an embodiment of the present invention;
[0029] Figure 2 It is a system structure diagram of a control system for maintaining a vacuum state in a shutdown vacuum preservation state provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0030] The specific implementation of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the present invention, and is not used to limit the present invention.
[0031] Figure 1 1 is a flow chart of a method for controlling a vacuum state in a shutdown vacuum storage state provided by an embodiment of the present invention. Figure 1 As shown, an embodiment of the present invention provides a control method for maintaining a vacuum state in a shutdown vacuum preservation state, the method comprising:
[0032] Step S10: In response to the shutdown vacuum preservation trigger signal, the air pressure information of the baking chamber is collected at multiple time points within the initialization cycle.
[0033] Specifically, starting from the triggering moment of the shutdown vacuum preservation trigger signal, the initialization is triggered and the corresponding initialization cycle is identified; the air pressure information is collected once at a predetermined time interval during the initialization cycle, and the air pressure time series information set is obtained based on the chronological order of the collection moments; clustering is performed on the air pressure time series information set to identify the number of discrete points, and when the data of the discrete points is greater than a preset threshold, initialization is re-executed until the data of the discrete points is less than the preset threshold, and the air pressure information set is output.
[0034] In the embodiment of the present invention, at the time of shutdown vacuum preservation triggering, the initialization cycle is immediately identified and started. At this time, the equipment will periodically collect the air pressure information in the baking chamber at a predetermined time interval. This predetermined time interval can be reasonably set according to the actual process requirements, equipment performance and historical data of the leakage rate. For example, the air pressure value is collected every 1 second to ensure that enough air pressure data points are obtained in a short period of time to form an initial air pressure change curve.
[0035] After the initialization cycle is over, a set of pressure data sequences arranged in order in time is obtained, namely, the pressure time series information set. Then, when processing these data, cluster analysis technology is introduced to control and screen the data quality. The core goal of clustering is to group the pressure data points according to similarity, so as to identify abnormal points and discrete data points. Discrete points usually refer to data samples that deviate too much from the main trend on the pressure curve. They may be caused by instantaneous interference of sensors, data errors or accidental factors. If there are too many discrete points, it means that the initial data set cannot fully reflect the normal state, which will have an adverse effect on the reliability and accuracy of the subsequent prediction model.
[0036] When the cluster analysis identifies that the number of discrete points exceeds the preset threshold, the initialization step will be re-executed. This means entering an initialization cycle again to re-collect and screen the air pressure data until the proportion of discrete points drops to the ideal range. This cyclic mechanism ensures that the final output air pressure information set is a data set with small deviation, high quality, and can represent the current real state, laying a good foundation for subsequent modeling, prediction, and control. When the amount of discrete data is controlled below the threshold, the air pressure information set is output as the final initialization result, providing a reliable data source for the next step of predictive model construction (GM (1, 1) model establishment).
[0037] Based on the scheme of the present invention, cluster analysis is used to screen out too many discrete points, so that the air pressure information set finally obtained is more stable and stable, and a meaningful benchmark is constructed for subsequent trend analysis and prediction. High-quality initialization data can significantly reduce the prediction model parameter estimation error, thereby improving the prediction accuracy and reliability. When GM (1, 1) or other prediction models are used in the future, these high-quality data can better fit the actual air pressure change trend. Due to the dynamic detection and feedback of the discrete data volume, when the environment or equipment status changes (such as equipment aging, reduced sealing, etc.), it can still adapt to the data characteristics under the new conditions by reinitialization, thereby maintaining the effectiveness of the prediction model. Supported by high-quality data, the subsequent prediction model can provide more accurate input for the vacuum pump control scheme. It is possible to more flexibly determine when to turn on or off the vacuum pump during downtime, thereby reducing energy consumption, reducing equipment wear and ensuring the continuous stability of the vacuum degree.
[0038] Step S20: constructing a GM(1,1) model for predicting the air pressure information at subsequent moments based on the air pressure information of the baking chamber at each time point as an initial prediction model.
[0039] Specifically, the GM (1, 1) model is the most basic and most commonly used prediction model in grey theory. It models and predicts one-dimensional data sequences, and is particularly suitable for situations where the amount of data is small, the information is incomplete, and the internal laws are not easy to directly observe. Based on the GM (1, 1) model rules, the pressure accumulation at each moment is performed based on the time sequence to obtain the accumulation sequence; a first-order linear differential equation approximate model is constructed based on the accumulation sequence to determine the parameters to be determined for the model; the first-order linear differential equation approximate model is discretized, and the parameters to be determined are solved based on the discretized first-order linear differential equation approximate model; the corresponding GM (1, 1) model is constructed based on the solved parameters, and the corresponding initial prediction model is constructed.
[0040] Furthermore, the pressure accumulation rule at each moment is:
[0041]
[0042] Where n is the number of sampling points; x (1) (k) is the accumulated value of the air pressure at the kth sampling point sequence; x (0) (i) is the collected air pressure value at the position of the i-th sampling point sequence; the first-order linear differential equation approximate model is expressed as:
[0043]
[0044] Among them, a and b are the parameters to be determined for the model.
[0045] Specifically, the discretization of the first-order linear differential equation approximate model and solving each parameter to be determined based on the discretized first-order linear differential equation approximate model include: discretizing the first-order linear differential equation approximate model based on the background value sequence to obtain a discretized model; converting the discretized model into a matrix form, and solving the discretized model in matrix form based on the least squares method to obtain the parameter value of each parameter to be determined; wherein the discretized model is expressed as:
[0046] x (0) (k)+az (1) (k) = b, k = 1, 2, ..., n
[0047] in, Represents the background value sequence of the kth sampling point sequence position.
[0048] The solution rule is to convert the discretized model into a matrix form. The corresponding matrix form is:
[0049]
[0050] Then replace it with the following rules:
[0051]
[0052] Then we have:
[0053]
[0054] Solve using the least squares method:
[0055]
[0056] Furthermore, the initial prediction model is expressed as:
[0057] x (0) (n+1)=x (1) (n+1)-x (1) (n)
[0058] Among them, x (0) (n+1) is the predicted value of air pressure at time n+1; x (1) (n+1) and x (1) (n) are the accumulated air pressure values at the n+1th and nth moments generated based on the GM (1, 1) model.
[0059] The calculation rules for the cumulative value at each moment are:
[0060]
[0061] In the implementation of the present invention, before constructing the GM (1, 1) model for prediction, the original air pressure data sequence needs to be specially processed. This processing step is called "accumulation generation". The so-called "accumulation" can be understood as adding the data in the initial time series from the starting point in sequence. Suppose you have a series of air pressure data points arranged in time, which are measured at different times. The accumulation process starts from the first data point and records it as the starting point of the accumulation result; then at the second moment, add the values of the first two data points to get the second value after accumulation; then at the third moment, add the values of the first three data points to get the third accumulated value... and continue in this way until all the data points in this series are accumulated. After accumulation, a smoother sequence is obtained. Because this sequence gradually accumulates the data of all previous moments over time, its changes will not be as drastic and jumpy as the original sequence, showing a trend towards smoother changes.
[0062] With this accumulated smooth sequence, we can create better conditions for subsequent predictive modeling. The next problem to be solved is: find a simplified model suitable for describing its development trend from this smooth data. At this time, two key parameters are needed. These two parameters determine the change speed and final stable trend of the prediction model. These two parameters can be imagined as two scales describing the characteristics of a smooth curve. One scale controls the upward or downward offset of the curve, and the other scale determines how the curve gradually approaches a certain equilibrium value over time.
[0063] In order to obtain these two parameters, a set of logical steps is needed to compare the accumulated data with the original data, and with the help of certain symmetry and trend analysis methods, these two parameters can best describe the future trend of the pressure curve. Without formula description, this process can be understood as: constantly trying to fit a simple smooth curve on the changes in the accumulated and smoothed pressure data, adjusting the two parameters to make this smooth curve as close as possible to the trend of the data points. After such continuous adjustment and adaptation, the two parameters finally determined will make the curve neither deviate too much from the past data distribution, nor give reasonable prediction values in the future.
[0064] Once the parameters are determined, the model has the ability to predict future air pressure values. The prediction logic here is: the pair of parameters that have been found is like a set of rules. If the current cumulative trend is like this, then the cumulative value at the next moment will continue to move forward according to the rules set by these two parameters. When the original air pressure value needs to be predicted, the predicted cumulative result will be compared with the cumulative result of the previous moment to get the predicted value at the next moment. This is like having mastered a rule - after each period of time, the overall change in air pressure will follow a certain trend and amplitude.
[0065] Step S30: performing model verification on the initial prediction model based on the air pressure information of the baking chamber at multiple time points in the test cycle, and using the prediction model that passes the verification as the air pressure prediction model.
[0066] Specifically, within the test cycle, the air pressure value prediction for each preset moment is performed in sequence based on the initial prediction model to obtain the air pressure prediction value for each preset moment; when each preset moment is reached, the actual air pressure value at each preset moment is compared with the air pressure prediction value at the corresponding preset moment, and when the absolute value of the difference between the two is less than the preset difference threshold, a valid prediction is recorded; when the valid predictions are recorded continuously for a preset number of times, it is determined that the initial prediction model has been verified; conversely, if the absolute value of the difference between the two is not less than the preset difference threshold, the air pressure information set within the initialization cycle is updated based on the actual air pressure value at the current moment to update the initial prediction model, until the valid predictions are recorded continuously for the preset number of times, it is determined that the initial prediction model has been verified.
[0067] In the implementation of the present invention, after entering the test cycle, the air pressure values at a set of pre-set time points will be predicted based on the initial prediction model. These moments are called "preset moments", and each preset moment corresponds to an actual measurable air pressure value. During the test cycle, the initial prediction model is used to generate the air pressure prediction value at each preset moment before the preset moment. When the time actually reaches this preset moment, the actual air pressure value is read from the baking chamber and compared with the previous prediction value. Through this comparison, the prediction accuracy of the model at that specific moment can be evaluated.
[0068] In order to determine whether the prediction performance of the model is satisfactory, a preset difference threshold needs to be defined. This threshold is equivalent to an accuracy standard, which indicates the allowable deviation range between the actual air pressure value and the predicted air pressure value. When the absolute value of the difference between the actual air pressure and the predicted air pressure is less than this threshold, it means that the prediction at this moment is valid and the accuracy meets the standard. This will be recorded and called an "effective prediction". Once a certain number of consecutive effective predictions are accumulated (there is a preset continuous number standard here, for example, the prediction accuracy of multiple consecutive moments is within the threshold range), it means that the current initial prediction model has passed the verification. This means that the model has a high prediction accuracy to a certain extent and can be put into use as the final air pressure prediction model.
[0069] However, if the deviation between the predicted value and the actual value at a preset moment exceeds the preset threshold, then the prediction at this time is judged to be unsatisfactory. At this time, the model needs to be adjusted and updated. During the update, the actual air pressure value at that moment is included in the data set of the initialization cycle to form a more representative and accurate set of data. On this basis, the initial prediction model will use the updated air pressure information set to re-estimate and optimize the parameters, thereby gradually correcting the prediction error and improving the overall performance and stability of the prediction model.
[0070] This process of dynamic adjustment and repeated verification makes the initial prediction model no longer static and fixed, but an "adaptive" model that can be continuously improved based on new data feedback. Through this continuous verification and optimization process, the model can filter errors and improve accuracy in a relatively short period of time, and finally stabilize after multiple consecutive successful predictions. In this way, when the model enters the formal shutdown vacuum preservation control strategy stage, the air pressure prediction model is determined after a period of rigorous testing and dynamic correction, and it has a more accurate grasp of the actual air pressure change trend.
[0071] Based on the scheme of the present invention, by verifying and dynamically updating the initial prediction model during the test cycle, this mechanism lays a solid reliability foundation for the subsequent air pressure prediction and vacuum pump control. First, model verification can ensure that inaccurate prediction logic is screened out before formal use, and only models that meet the accuracy requirements can enter the next link, providing a reliable basis for the vacuum pump control strategy. Secondly, the dynamic update process enables the model to adapt to changing environmental factors and leakage rates. When there are subtle changes in sealing conditions, ambient temperature or other factors affecting air pressure, the model can correct the deviation in time without always relying on the initial set of parameters that may not be accurate enough. Finally, the air pressure prediction model that has passed the verification can provide a more accurate prediction of the air pressure trend in actual operation, providing decision support for the early or delayed intervention of the vacuum pump. This prediction mechanism based on data feedback loop correction improves the intelligence and adaptability of the shutdown vacuum preservation process. It not only reduces the frequent start and stop of the vacuum pump, reduces energy consumption and equipment wear, but also keeps the air pressure more stable, which is conducive to ensuring that the quality of the battery in the shutdown state is not affected by fluctuations, thereby meeting higher standards of production process and quality control requirements.
[0072] Step S40: During the shutdown vacuum preservation process, rolling modeling is performed based on the air pressure prediction model, and the real-time collected air pressure information of the baking chamber and all previous air pressure information are used as input parameters of the air pressure prediction model at the current moment, and the air pressure information prediction at the next moment is performed, and the control plan of the vacuum pump is generated and executed based on the air pressure prediction value.
[0073] Specifically, each time a new actual air pressure value is obtained, the air pressure information set is updated based on the actual air pressure value, and the air pressure prediction model is updated based on the air pressure information set as the air pressure prediction model for the next moment; the air pressure information at the current moment and all previous air pressure information are used as input parameters of the air pressure prediction model corresponding to the current moment, and the air pressure information prediction for the next moment is performed to obtain the air pressure prediction value for the next moment; based on the difference between the air pressure prediction value for the next moment and the preset standard air pressure value for shutdown vacuum storage, an adjustment plan for the vacuum pump is determined, and a corresponding control instruction is generated based on the adjustment plan, so that the vacuum pump can execute the control instruction.
[0074] In the embodiment of the present invention, in the actual application scenario of vacuum preservation during shutdown, the changes in the air pressure environment are often not static. As time goes by, small leaks, changes in ambient temperature and humidity, and aging of equipment components will affect the air pressure in the cavity. In order to ensure that the ideal vacuum degree is maintained throughout the shutdown process, it is obviously not enough to build a fixed prediction model only in the initial stage. At this time, the strategy of rolling modeling is particularly important.
[0075] Rolling modeling means that at each time step of the operation, the prediction model is dynamically updated based on the latest actual air pressure value. In other words, the prediction model is no longer static, but evolves with the arrival of each new data point. When a new actual air pressure value is obtained, the data is integrated with all the air pressure information retained at the previous moment to obtain a continuously expanded or optimized air pressure information set. Based on this latest information set, the parameters in the prediction model will be readjusted and optimized so that the model always fits the air pressure trend and characteristics of the current period.
[0076] This dynamic adaptation process ensures that the prediction model can track the trajectory of air pressure changes, rather than being limited to parameter estimates obtained under earlier conditions. For example, if the sealing performance of the cavity decreases slightly during operation, causing the rate of increase in air pressure to accelerate, the rolling modeling method will automatically correct the model parameters after receiving several high actual pressure values to better predict the air pressure value at the next moment. In this way, the model is more responsive to actual conditions, making the air pressure prediction more timely and accurate.
[0077] After predicting the air pressure value at the next moment, compare this predicted value with the preset shutdown vacuum preservation standard air pressure. The standard air pressure can be understood as the ideal pressure value inside the cavity, which is the best range to ensure that the battery is not damaged by humidity or impurities when it is shut down. By performing a difference analysis between the predicted value and the standard value, it can be determined whether the vacuum pump needs to be intervened and adjusted in the next period. For example, if the prediction shows that the air pressure at the next moment will be significantly higher than the standard value, you can react in advance and send a control instruction to the vacuum pump to start it in advance to reduce the air pressure in the cavity; conversely, if the prediction shows that the air pressure at the next moment is close to or lower than the standard value, the start-up frequency of the vacuum pump can be delayed or reduced, thereby saving energy and reducing equipment wear.
[0078] This control solution based on rolling modeling and real-time prediction is not only forward-looking, but also makes the control strategy more sophisticated and flexible. During the entire shutdown vacuum preservation process, the chamber pressure is no longer just passively waiting for over-limit adjustment, but through the prediction and prejudgment of the trend to achieve the effect of active intervention.
[0079] Based on the solution of the present invention, since the model parameters are continuously updated according to the latest data, a high prediction accuracy can be maintained in a dynamic environment, and subtle changes in air pressure trends can be discovered in a timely manner. Based on the prediction, the air pressure adjustment can be planned in advance to avoid frequent passive vacuuming due to exceeding the upper limit. The intervention of the vacuum pump becomes more reasonable, which is conducive to reducing the number of pump operations, reducing energy consumption and extending the service life of the equipment. Through forward-looking prediction and preventive control, air pressure fluctuations are better constrained within the target range, which helps to provide a more stable shutdown vacuum storage environment for the battery. Regardless of whether the air pressure change trend changes due to equipment aging or environmental factors, the rolling modeling strategy can respond quickly and dynamically adjust the prediction model to a new state. This makes it no longer limited to the fixed logic of the initial condition design, and significantly improves the adaptability to uncertain environments.
[0080] Figure 2 1 is a system structure diagram of a control system for maintaining a vacuum state in a shutdown vacuum storage state provided by an embodiment of the present invention. Figure 2 As shown, an embodiment of the present invention provides a control system for maintaining a vacuum state in a shutdown vacuum preservation state, the system comprising: an acquisition unit, for responding to a shutdown vacuum preservation trigger signal, and performing air pressure information acquisition of a baking chamber at multiple time points within an initialization cycle; a model generation unit, for constructing a GM (1, 1) model for predicting the air pressure information at subsequent moments based on the air pressure information of the baking chamber at each time point, as an initial prediction model; a model verification unit, for performing model verification on the initial prediction model based on the air pressure information of the baking chamber at multiple time points within a test cycle, and using the prediction model that passes the verification as the air pressure prediction model; a control unit, for performing rolling modeling based on the air pressure prediction model during the shutdown vacuum preservation process, and using the real-time acquired air pressure information of the baking chamber and all previous air pressure information as input parameters of the air pressure prediction model at the current moment, performing air pressure information prediction at the next moment, and generating and executing a control scheme for the vacuum pump based on the air pressure prediction value.
[0081] The embodiment of the present invention further provides a computer-readable storage medium, on which instructions are stored, which, when executed on a computer, enable the computer to execute the above-mentioned control method for maintaining a vacuum state in a shutdown vacuum storage state.
[0082] Those skilled in the art will understand that all or part of the steps in the method for implementing the above-mentioned embodiments can be completed by instructing the relevant hardware through a program, and the program is stored in a storage medium, including several instructions for making a single-chip microcomputer, a chip or a processor (processor) perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.
[0083] The optional embodiments of the present invention are described in detail above in conjunction with the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details in the above embodiments. Within the technical concept of the embodiments of the present invention, the technical scheme of the embodiments of the present invention can be subjected to a variety of simple modifications, and these simple modifications all belong to the protection scope of the embodiments of the present invention. It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the embodiments of the present invention will not further describe various possible combinations.
[0084] In addition, various embodiments of the present invention may be arbitrarily combined, and as long as they do not violate the concept of the embodiments of the present invention, they should also be regarded as the contents disclosed in the embodiments of the present invention.
Claims
1. A control method for maintaining a vacuum state in a shutdown vacuum storage state, which is applied to the vacuum state maintenance control in a shutdown vacuum storage state after battery baking, characterized in that: The method comprises: In response to a shutdown vacuum preservation trigger signal, collecting air pressure information of the baking chamber at multiple time points during an initialization cycle; Based on the air pressure information of the baking chamber at each time point, a GM (1, 1) model for predicting the air pressure information at subsequent times is constructed as an initial prediction model; Performing model verification on the initial prediction model based on the air pressure information of the baking chamber at multiple time points in the test cycle, and using the prediction model that passes the verification as the air pressure prediction model; During the shutdown vacuum preservation process, rolling modeling is performed based on the air pressure prediction model, and the real-time collected air pressure information of the baking chamber and all previous air pressure information are used as input parameters of the air pressure prediction model at the current moment, and the air pressure information prediction at the next moment is performed, and the control plan of the vacuum pump is generated and executed based on the air pressure prediction value.
2. The method according to claim 1, characterized in that The gas pressure information of the baking chamber at multiple time points during the initialization cycle includes: Starting from the triggering moment of the shutdown vacuum preservation trigger signal, the system initialization is triggered and the corresponding initialization cycle is identified; Perform air pressure information collection once every predetermined time in the initialization cycle, and obtain an air pressure time series information set based on the order of the collection moments; Clustering is performed on the air pressure time series information set to identify the number of discrete points, and when the data of the discrete points is greater than a preset threshold, reinitialization is performed until the data of the discrete points is less than the preset threshold, and the air pressure information set is output.
3. The method according to claim 2, characterized in that The GM (1, 1) model for predicting the air pressure information at subsequent moments is constructed based on the air pressure information of the baking chamber at each time point as an initial prediction model, including: Based on the GM (1, 1) model rule, the pressure accumulation at each moment is performed based on the time sequence to obtain the accumulation sequence; Construct a first-order linear differential equation approximate model based on the cumulative sequence and determine the parameters to be determined; Discretizing the first-order linear differential equation approximate model, and solving various parameters to be determined based on the discretized first-order linear differential equation approximate model; The corresponding GM (1, 1) model is constructed based on the solved parameters, and the corresponding initial prediction model is constructed.
4. The method according to claim 3, characterized in that The rule for accumulating air pressure at each moment is: Where n is the number of sampling points; x (1) (k) is the accumulated value of the air pressure at the kth sampling point sequence position; x (0) (i) is the collected air pressure value at the i-th sampling point sequence position; The first-order linear differential equation approximate model is expressed as: Among them, a and b are the parameters to be determined in the model.
5. The method according to claim 4, characterized in that Discretizing the first-order linear differential equation approximate model, and solving each parameter to be determined based on the discretized first-order linear differential equation approximate model, includes: Discretizing the first-order linear differential equation approximate model based on the background value sequence to obtain a discretized model; The discretization model is converted into a matrix form, and the discretization model in the matrix form is solved based on the least squares method to obtain the parameter values of each parameter to be determined; wherein, The discretization model is expressed as: x (0) (k)+az (i) (k)=b,k=1,2,…,n in, Represents the background value sequence of the kth sampling point sequence position.
6. The method according to claim 3, characterized in that The initial prediction model is expressed as: x (0) (n+1)=x (1) (n+1)-x (1) (n) Among them, x (0) (n+1) is the predicted air pressure value at time n+1; x (1) (n+1) and x (1) (n) are the accumulated air pressure values at the n+1th and nth moments generated based on the GM (1, 1) model.
7. The method according to claim 1, characterized in that The performing of model verification on the initial prediction model based on the air pressure information of the baking chamber at multiple time points in the test cycle, and using the prediction model that passes the verification as the air pressure prediction model, comprises: During the test period, the air pressure value prediction at each preset time is performed in sequence based on the initial prediction model to obtain the air pressure prediction value at each preset time; When each preset time is reached, the actual air pressure value at each preset time is compared with the air pressure prediction value at the corresponding preset time. When the absolute value of the difference between the two is less than the preset difference threshold, a valid prediction is recorded; When a preset number of valid predictions are recorded continuously, it is determined that the initial prediction model verification has passed; On the contrary, if the absolute value of the difference between the two is not less than the preset difference threshold, the air pressure information set within the initialization period is updated based on the actual air pressure value at the current moment to update the initial prediction model until the preset number of valid predictions are recorded continuously, and it is determined that the initial prediction model verification has passed.
8. The method according to claim 1, characterized in that In the process of vacuum preservation during shutdown, rolling modeling is performed based on the air pressure prediction model, and the air pressure information of the baking chamber collected in real time and all previous air pressure information are used as input parameters of the air pressure prediction model at the current moment, and the air pressure information prediction at the next moment is performed, and a control scheme of the vacuum pump is generated and executed based on the air pressure prediction value, including: Whenever a new actual air pressure value is obtained, the air pressure information set is updated based on the actual air pressure value, and the air pressure prediction model is updated based on the air pressure information set as the air pressure prediction model for the next moment; The current air pressure information and all previous air pressure information are used as input parameters of the air pressure prediction model corresponding to the current moment, and the air pressure information prediction of the next moment is performed to obtain the air pressure prediction value of the next moment; Based on the difference between the predicted air pressure value at the next moment and the preset standard air pressure value for shutdown vacuum preservation, an adjustment plan for the vacuum pump is determined, and a corresponding control instruction is generated based on the adjustment plan so that the vacuum pump executes the control instruction.
9. A control system for maintaining a vacuum state in a shutdown vacuum storage state, which is applied to the vacuum state maintenance control in a shutdown vacuum storage state after battery baking, characterized in that: The system comprises: A collection unit, for collecting air pressure information of the baking chamber at multiple time points during an initialization cycle in response to a shutdown vacuum preservation trigger signal; A model generation unit, used to construct a GM (1, 1) model for predicting the air pressure information at a subsequent time point based on the air pressure information of the baking chamber at each time point, as an initial prediction model; A model verification unit, configured to perform model verification on the initial prediction model based on the air pressure information of the baking chamber at multiple time points in a test cycle, and use the prediction model that passes the verification as the air pressure prediction model; The control unit is used to perform rolling modeling based on the air pressure prediction model during the shutdown vacuum preservation process, and use the real-time collected air pressure information of the baking chamber and all previous air pressure information as input parameters of the air pressure prediction model at the current moment, perform air pressure information prediction at the next moment, and generate and execute the control plan of the vacuum pump based on the air pressure prediction value.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the control method for maintaining a vacuum state in a shutdown vacuum storage state as described in any one of claims 1 to 8.