A photovoltaic panel coating film self-adaptive control method based on interface humidity inference

Abstract

The present application relates to the technical field of photovoltaic panel coating process, and specifically discloses a photovoltaic panel coating self-adaptive control method based on interface humidity inference. The method collects multi-source operation parameters in real time through an embedded edge computing unit, obtains predicted values of each parameter by using a prediction model independently generated based on historical trends and current state, compares the deviation between real-time measurement values and predicted values, dynamically adjusts the fusion weight of abnormal parameters in combination with a physical continuity threshold, and mainly infers the current humidity state of the contact interface between the coating device and the photovoltaic panel by using the predicted values. Furthermore, the method generates and outputs corresponding coating control instructions locally according to a pre-stored humidity-process mapping relationship, and is mainly used to improve the uniformity and process stability of photovoltaic panel coating. The method also contains a dynamic memory management mechanism based on parameter importance and periodicity, which ensures efficient use of system resources.

Description

Technical Field

[0001] This invention belongs to the field of photovoltaic panel coating technology, specifically an adaptive control method for photovoltaic panel coating based on interface humidity inference. Background Technology

[0002] In continuous photovoltaic panel coating operations, the coating quality is highly sensitive to the state of the interface between the coating device and the photovoltaic panel surface. In particular, the actual humidity at this interface has a decisive impact on the coating spread behavior, evaporation rate, and final film uniformity. Excessive interface humidity leads to overspreading of the coating and uneven film thickness; excessively low interface humidity results in rapid evaporation of the coating, discontinuous film formation, or decreased adhesion. However, the humidity at the interface between the coating device and the photovoltaic panel has the following significant characteristics: First, the interface is small in scale and in a dynamic contact state. Second, the interface humidity changes rapidly, influenced by multiple factors such as liquid supply, evaporation, and material surface properties. Third, the interface area constantly changes with the movement of the coating device, making it impossible to deploy stable sensors. Furthermore, the interface humidity cannot be directly, continuously, and reliably measured using existing humidity sensors.

[0003] Existing coating control technologies typically only adjust directly controllable parameters such as liquid flow rate and operating speed, employing empirical or rule-based control. These approaches ignore changes in environmental parameters under complex operating conditions and fail to effectively correlate operating parameters with film quality. They often rely on multiple experimental adjustments to find optimal parameter settings, and during operation, only conventional parameters can be measured and controlled, resulting in unsatisfactory coating effects. Furthermore, because the critical process state of interface humidity cannot be directly observed, existing technologies cannot detect risks in advance when interface state deviations occur. They can only passively correct after film quality has deteriorated, leading to poor coating consistency and large quality fluctuations. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide an adaptive control method for photovoltaic panel coatings based on interface humidity inference. This method can reliably infer the unmeasurable humidity state of the contact interface at resource-constrained edge points, and thereby achieve low-latency, highly robust adaptive control. This solves the technical problems of poor coating uniformity and large batch-to-batch fluctuations caused by the lack of real-time interface humidity sensing in existing technologies. The specific solution is as follows:

[0005] In a first aspect, this application provides an adaptive control method for photovoltaic panel coating based on interface humidity inference, applied to the embedded edge computing unit of a photovoltaic panel coating equipment, comprising the following steps:

[0006] The embedded edge computing unit collects multi-source operating parameters of the photovoltaic panel coating process in real time.

[0007] Based on the multi-source operating parameters, a preset prediction model is used to generate predicted values ​​for each parameter. The prediction model is generated independently based on the historical evolution trend of each parameter and the current operating status.

[0008] Calculate the deviation between the real-time measured value and the predicted value of each multi-source operating parameter, and compare the deviation with a preset physical continuity threshold;

[0009] When the deviation between the real-time measured value and the predicted value of the operating parameter exceeds the preset threshold, it is determined that the real-time measured value of the parameter is abnormal, the fusion weight of the abnormal parameter is reduced, and the predicted value is used as the main basis for inferring the humidity state of the interface between the coating device and the photovoltaic panel.

[0010] Based on the weighted parameters, the current humidity state of the contact interface is inferred;

[0011] Based on the current humidity status, the edge computing unit generates coating control commands locally and outputs them to the photovoltaic panel coating equipment.

[0012] Furthermore, the prediction model includes:

[0013] For periodic operating parameters, a periodic autoregressive model is used to generate the predicted value for the current time based on the value at the corresponding time of the previous period.

[0014] For non-periodic operating parameters, an exponential smoothing model is used to generate the predicted value for the current moment based on the numerical sequence within the historical time window.

[0015] Furthermore, when the deviation between the real-time measured value and the predicted value is within the preset threshold range, the real-time measured value is used as the basis for inferring the humidity state.

[0016] Furthermore, the reduction of the fusion weights of real-time measurements and the use of predicted values ​​as the primary basis for state inference specifically includes:

[0017] A weighted fusion algorithm is used to determine the weight of each parameter. When the deviation exceeds a preset threshold, the weight coefficient of the predicted value is adjusted to a first preset value, and the weight coefficient of the real-time measured value is adjusted to a second preset value, wherein the first preset value is greater than the second preset value.

[0018] Furthermore, the inference of the current humidity state at the interface between the coating device and the photovoltaic panel specifically includes: inputting the weighted operating parameters as input variables into the interface humidity dynamic estimation model pre-built and running locally in the embedded edge computing unit, and outputting the quantified value of the current humidity state through the interface humidity dynamic estimation model.

[0019] Furthermore, the control decision is generated based on a pre-stored interface humidity state-process parameter mapping table, which includes at least the following mapping rules: when the current humidity state is higher than a first preset threshold, a first control instruction is generated to increase the coating operation speed and / or reduce the liquid supply; when the current humidity state is lower than a second preset threshold, a second control instruction is generated to decrease the coating operation speed and / or increase the liquid supply.

[0020] Furthermore, before acquiring the multi-source operating parameters of the photovoltaic panel coating process in real time, an initialization calibration step is also included:

[0021] The photovoltaic panel coating equipment is controlled to operate under preset standard operating conditions, which include preset travel speed, liquid supply volume and environmental conditions.

[0022] Collect multi-source operating parameters under the standard operating condition, and use the prediction model to generate corresponding baseline prediction values;

[0023] Calculate the deviation between the baseline predicted value and the theoretical value corresponding to the standard operating condition;

[0024] Based on the deviation, the initial parameters of the prediction model are corrected to compensate for the influence of sensor zero-point drift on state inference.

[0025] Furthermore, it also includes dynamic memory management steps:

[0026] Based on the periodic characteristics of each operating parameter and its importance to humidity state inference, the corresponding historical data storage length is configured for different parameters;

[0027] When it is detected that the first parameter needs to increase the length of historical data storage, the system identifies the second parameter whose storage length is greater than the preset value and whose state changes steadily, reduces the length of historical data storage for the second parameter, and allocates the freed memory space to the first parameter to maintain a constant total system memory usage.

[0028] Furthermore, the historical data storage length is selected from a predefined set of discrete lengths consisting of short-period, medium-period, and long-period levels.

[0029] Secondly, the present invention provides a photovoltaic panel coating adaptive control system based on interface humidity state inference, comprising:

[0030] The data acquisition module is used to collect multi-source operating parameters of the photovoltaic panel coating process in real time through an embedded edge computing unit;

[0031] The dual-channel processing module generates predicted values ​​for each parameter based on the multi-source operating parameters and using a preset prediction model. The prediction model is generated independently based on the historical evolution trend of the parameter and the current operating state. The module calculates the deviation between the real-time measured value and the predicted value of each parameter and compares the deviation with a preset physical continuity threshold. When the deviation between the real-time measured value and the predicted value of an operating parameter exceeds the preset threshold, it determines that the real-time measured value of the parameter is abnormal and reduces the fusion weight of the abnormal parameter. At the same time, the predicted value is used as the main basis for inferring the humidity state of the interface between the coating device and the photovoltaic panel.

[0032] The state inference module infers the current humidity state of the contact interface based on the weighted parameters.

[0033] The adaptive control module generates coating control commands locally in the edge computing unit based on the current humidity status and outputs them to the photovoltaic panel coating equipment.

[0034] Thirdly, embodiments of this specification provide an electronic device, including a processor and a memory; the processor is connected to the memory; the memory is used to store executable program code; the processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the steps in a photovoltaic panel coating adaptive control method based on interface humidity inference.

[0035] Fourthly, embodiments of this specification provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a photovoltaic panel coating adaptive control method based on interface humidity inference.

[0036] The beneficial effects of the present invention's adaptive control method for photovoltaic panel coating based on interface humidity inference include at least the following:

[0037] Multi-source operating parameters during the photovoltaic panel coating process are collected in real time by an embedded edge computing unit. Predictive models, independently generated based on the historical evolution trends and current operating status of each parameter, are used to generate predicted values ​​for each parameter. By calculating the deviation between the real-time measured values ​​and the predicted values ​​and comparing them with a preset physical continuity threshold, automatic identification and dynamic weight adjustment of abnormal parameters are achieved. When an abnormality is detected in a parameter, its fusion weight is reduced. Simultaneously, the predicted value is used as the primary basis for inferring the humidity state of the contact interface, effectively improving the robustness of humidity state inference under conditions of sensor data fluctuations or transmission interference. Based on the weighted parameters, a pre-built and locally running interface humidity dynamic estimation model outputs a quantified value of the current humidity state. Corresponding coating control commands are then generated locally at the edge based on this humidity state, achieving end-to-end edge processing from data acquisition and state inference to control command generation, ensuring the real-time performance and reliability of the control commands. Furthermore, a dynamic memory management mechanism dynamically allocates the length of historical data storage based on the periodic characteristics of each parameter and its importance to humidity state inference. This optimizes memory usage efficiency in the resource-constrained embedded edge computing unit, ensuring stability during long-term continuous operation. This application can effectively solve the technical problems of poor coating uniformity and large batch-to-batch fluctuations caused by the lack of real-time sensing and closed-loop feedback mechanism for interface humidity in the prior art, and realize high-precision, low-delay, and adaptive control of the photovoltaic panel coating process. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of this specification, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 This is a schematic diagram of existing technical solutions for photovoltaic panel coating.

[0040] Figure 2 This is a schematic diagram of the current work process.

[0041] Figure 3 This is a schematic diagram of the interface humidity-driven coating control process in a photovoltaic panel coating adaptive control method based on interface humidity inference provided in the embodiments of this specification.

[0042] Figure 4 This is a schematic flowchart of an adaptive control method for photovoltaic panel coating based on interface humidity inference, provided as an embodiment of this specification.

[0043] Figure 5 This is a schematic diagram of the periodic parameter history management structure provided for the embodiments of this specification.

[0044] Figure 6 This is a schematic diagram illustrating the change in light transmittance after film formation during stable operation, as provided in the embodiments of this specification.

[0045] Figure 7 This is a schematic diagram illustrating the change in light transmittance after film formation under a sudden change in a certain parameter, as provided in the embodiments of this specification.

[0046] Figure 8 This is a schematic diagram illustrating the historical memory length changes with multiple parameters and lengths.

[0047] Figure 9 This is a schematic diagram of an electronic device provided in this embodiment. Detailed Implementation

[0048] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. The terms "first," "second," "third," etc., in the specification, claims, and accompanying drawings are used to distinguish different objects, not to describe a specific order. In addition, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0049] In the following description, terms such as “inner,” “outer,” “upper,” “lower,” “left,” and “right” are used only to facilitate the description of the embodiments and to simplify the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this specification.

[0050] Before introducing the technical solutions described in this manual, the application scenarios and related technologies of the technical solutions will be introduced.

[0051] This invention is primarily applied to the continuous coating process in photovoltaic panel manufacturing. In this process, the coating equipment moves at a constant speed along the surface of the photovoltaic panel. The coating solution is delivered to the coating head via a liquid supply device, spreads at the dynamic contact interface between the coating head and the photovoltaic panel surface, and rapidly evaporates, ultimately forming a uniform solid film layer. The thickness uniformity, adhesion, and optical properties of this film layer directly determine the power generation efficiency and lifespan of the photovoltaic panel.

[0052] In the aforementioned coating process, the humidity state at the interface between the coating device and the photovoltaic panel surface is a key factor affecting film quality. Specifically: when the interface humidity is too high, the coating solution evaporates too slowly, easily leading to coating solution accumulation, uneven film thickness, or even sagging; when the interface humidity is too low, the coating solution evaporates too quickly, easily leading to insufficient coating solution spreading, discontinuous film formation, or decreased adhesion. Therefore, achieving precise control of interface humidity is crucial for ensuring coating consistency and stability.

[0053] However, the interface humidity presents the following technical challenges: First, the interface is in a dynamic contact state, its scale is small and changes with the movement of the coating device, making it impossible to install any physical sensors for direct measurement; second, the interface humidity is affected by multiple factors such as liquid supply volume, travel speed, ambient temperature and humidity, and the characteristics of the photovoltaic panel surface material, resulting in rapid and highly nonlinear changes; third, the photovoltaic panel coating site is usually located outdoors or in a semi-open environment with unstable communication conditions, making it impossible to rely on cloud or remote servers for real-time calculation and control.

[0054] In the prior art, please refer to the appendix. Figure 1 and attached Figure 2 Operators typically adjust the liquid supply and travel speed based solely on experience, or passively correct parameters after significant deterioration in film quality, leading to difficulties in ensuring coating consistency and yield. To address these issues, this invention proposes an adaptive coating control method based on interfacial humidity state inference, with attached... Figure 3 This illustrates the process of interface humidity-driven coating control. This method does not rely on physical humidity sensors. Instead, it collects measurable multi-source operating parameters, utilizes a dual-channel data acquisition mechanism and physical continuity constraints, and infers the implicit state of interface humidity in real time within an embedded edge computing unit. Based on the inference results, it adaptively adjusts coating parameters to achieve high-quality, high-consistency coating in complex field environments. Please refer to the appendix. Figure 4 The method includes at least the following steps:

[0055] Step S1: Real-time acquisition of multi-source operating parameters. During the coating process, the embedded edge computing unit acquires multi-source operating parameters of the photovoltaic panel coating process in real time by connecting various sensors, such as flow sensors, encoders, temperature sensors, and position sensors. These multi-source operating parameters include, but are not limited to: coating device operating status parameters (such as motor current and vibration amplitude), coating liquid supply behavior parameters (such as liquid supply pump speed and liquid supply pressure), operating speed parameters (such as travel speed and acceleration), spatial position parameters (such as X / Y / Z coordinates), and photovoltaic operating environment sampling parameters (such as weather, air quality, and mass). In practical applications, the system can select at least one type of parameter for acquisition based on the sensor configuration of the coating device. For example, in some embodiments, only the travel speed and liquid supply flow rate are acquired, while in other embodiments, all of the above parameters are acquired simultaneously. The specific method can be determined according to the actual situation. These parameters are all directly measurable physical quantities used to indirectly characterize the interface humidity state, which cannot be directly observed. Therefore, the system does not attempt to place humidity sensors at the interface between the coating device and the photovoltaic panel, but instead defines the interface humidity as an implicit process state.

[0056] In this embodiment, before real-time acquisition of multi-source operating parameters of the photovoltaic panel coating process, an initialization calibration step is included: firstly, the photovoltaic panel coating equipment is controlled to operate under preset standard conditions, wherein the standard conditions include preset travel speed, liquid supply volume, and environmental conditions; then, multi-source operating parameters under the standard conditions are acquired, and the corresponding baseline predicted value is generated using the prediction model; the deviation between the baseline predicted value and the theoretical value corresponding to the standard conditions is calculated; based on the deviation, the initial parameters of the prediction model are corrected to compensate for the influence of sensor zero-point drift on state inference.

[0057] Step S2: Generate Independent Predicted Values. While acquiring real-time data, this embodiment uses a preset prediction model to independently generate predicted values ​​for each parameter based on its historical evolution trend and current operating status. The input is the parameter sequence within the historical time window and the current operating status, such as the current speed and liquid delivery rate. The output is the predicted parameter value at the current moment. This step constructs a dual-channel data architecture consisting of a real-time measurement channel and a model prediction channel. It should be noted that the generation process of these predicted values ​​does not depend on the real-time measurement values ​​themselves, that is, it is independent of the real-time measurement values ​​directly collected by the sensors, thus ensuring that the predicted values ​​and the measured values ​​are two independent sources of information. In a preferred embodiment, for parameters with obvious periodicity, such as the travel position along the long side of the photovoltaic panel, whose period is related to the length of the photovoltaic panel, the prediction model can adopt a periodic autoregressive model, using the value of the same phase point of the previous period as the prediction basis. For example, if the photovoltaic panel is 2 meters long and the system travels at a speed of 0.5 meters per second, the period is 4 seconds. The system records the location parameters at each time point in the previous cycle. At the same time in the current cycle, it directly uses the value from the previous cycle as the predicted value. For non-periodic parameters, the prediction model can use exponential smoothing or autoregressive moving average models, extrapolating the predicted value based on several recent historical data points. For example, if the ambient temperature has risen at a rate of 0.1℃ per second over the past 10 seconds, the predicted temperature at the next moment will increase by 0.1℃ from the current value.

[0058] S3: Calculate the deviation and compare it with the physical continuity threshold. The embedded edge computing unit calculates the deviation between the real-time measured value and the corresponding predicted value of each operating parameter in real time, which can be, for example, absolute deviation or relative deviation. Then, it compares the deviation with a preset physical continuity threshold. This physical continuity threshold is pre-calibrated based on the physical laws of interface humidity change. The physical basis of the physical continuity constraint is that the humidity change at the interface between the coating device and the photovoltaic panel is affected by the supply of coating liquid, evaporation, and material surface properties. Its change process has temporal continuity and does not have the physical conditions for instantaneous large jumps. Therefore, for real humidity changes that conform to physical laws, the measured values ​​of its associated parameters will not change abruptly in a very short time. Accordingly, the physical continuity threshold can be set as a reasonable upper limit of the maximum possible change of the parameter per unit time. For example, the threshold for travel speed can be set as the rate of change corresponding to the maximum acceleration.

[0059] S4: Anomaly Detection and Weight Adjustment. When the deviation between the real-time measured value and the predicted value of a certain operating parameter exceeds a preset physical continuity threshold, the embedded edge computing unit determines that the real-time measured value of that parameter is abnormal—this abnormality is not a real physical change, but may be caused by sensor noise, electromagnetic interference, or transient disturbances. In this case, the system reduces the fusion weight of this abnormal parameter in subsequent humidity state inference, while using the predicted value of this parameter as the primary basis for inferring the interface humidity state. When the deviation between the real-time measured value and the predicted value is within the preset threshold range, the system considers the real-time measured value reliable and uses it as the primary basis for humidity state inference without weight adjustment. In this case, the predicted value is only used as a backup reference, does not participate in fusion, or has a very low weight, to ensure the real-time performance and accuracy of state inference under normal operating conditions. In one specific implementation, the system uses a weighted fusion algorithm to determine the final input value of each parameter. Let the real-time measurement value be M, the predicted value be P, and the preset deviation threshold be T. When the deviation between the real-time measurement value and the predicted value is within the preset threshold, the weight coefficient of the real-time measurement value is 0.9, and the weight of the predicted value is 0.1. When the deviation between the real-time measurement value and the predicted value exceeds the preset threshold, the weight of the real-time measurement value is set to 0.2, and the weight of the predicted value is set to 0.8, thereby realizing the switching of the data source.

[0060] In another preferred embodiment, the weight adjustment is continuously correlated with the magnitude of the deviation; when the deviation exceeds a preset threshold, the real-time measurement value drops to 0. Regardless of the weight adjustment rule used, the core principle is to reduce the weight of the measured value and increase the weight of the predicted value when the deviation exceeds the threshold, so that the predicted value dominates the humidity inference. This weight adjustment can be step-wise or a continuous adjustment positively correlated with the magnitude of the deviation: the larger the deviation, the smaller the weight of the real-time measurement value and the larger the weight of the predicted value.

[0061] Step S5: Infer the current humidity state. Based on all operating parameters after weight adjustment, including the fused values ​​after the weights of the original measurements without anomalies remain unchanged and the weights of the abnormal parameters are redistributed, the embedded edge computing unit infers the current humidity state of the interface between the coating device and the photovoltaic panel. This humidity state is an implicit process state that cannot be directly measured; in other words, the system does not rely on any physical humidity sensor but is indirectly estimated through multi-source parameters. Specifically, the system adopts an interface humidity dynamic estimation model, which, after extensive training, can map these measurable explicit parameters to the current humidity state of the interface between the coating device and the photovoltaic panel. This can be constructed based on the physical dynamics equations of the coating process. For example, the contact interface can be considered as a lumped parameter system with liquid inflow, evaporation outflow, and spreading consumption, and a humidity state equation can be established: Where H is the interfacial humidity, and Q is the interfacial humidity. in Let v be the liquid flow rate, v be the travel speed, and H be the flow rate. a For ambient humidity, , , These are empirical coefficients, and the parameters (such as Q) are adjusted for weights. in v, H a Using multiple sources (such as Kalman filter or particle filter) as input, the H value is estimated in real time. This model is pre-trained or calibrated offline in an embedded edge computing unit, and only performs forward computation at runtime, resulting in low computational cost and suitability for embedded environments. In other embodiments, a data-driven neural network model, such as a lightweight BP neural network or LSTM, can be used, taking weighted multi-source parameters as input and directly outputting a quantified value of the current humidity state (e.g., a relative humidity index of 0–100%). This model also needs to be pre-built and run locally in an embedded edge computing unit, without cloud involvement.

[0062] Step S6: Generate edge-side local control commands. In terms of control decisions, the system does not rely on complex online optimization algorithms, but instead pre-stores an interface humidity state-process parameter mapping table. This table, obtained through offline experimental calibration, divides the humidity state into three regions: excessively dry, suitable, and excessively wet. Each region corresponds to a specific liquid supply adjustment range and speed adjustment strategy. This table lookup method greatly reduces the computational burden on the edge computing unit and results in a fast response speed. Specifically, based on the inferred current humidity state, the system queries the pre-stored interface humidity state-process parameter mapping table to generate corresponding coating control commands. When the inferred current humidity state is higher than a first preset threshold (e.g., 80%), a first control command is generated to increase the coating operation speed and / or decrease the liquid supply; when it is lower than a second preset threshold (e.g., 40%), a second control command is generated to decrease the coating operation speed and / or increase the liquid supply. The first and second preset thresholds can be the same or different; for example, they can both be set to 50%, employing a symmetrical control strategy. Control commands can be one or more combinations of adjusting the liquid supply volume, changing the liquid supply position (e.g., by moving the replenishment nozzle), modifying the replenishment strategy (e.g., intermittent or continuous replenishment), and adjusting the coating operation speed. These control commands directly act on the field actuators, forming an adaptive control strategy centered on the interface humidity state. Through this closed loop, the system achieves end-to-end adaptive control, progressing from "unmeasurable interface humidity" to "state inference based on multi-source parameter dual-channel consistency verification," and finally to "edge-side autonomous control."

[0063] Because the total memory of embedded edge computing units is limited, and the coating process may require continuous operation for several hours, the accumulation of historical data may lead to memory overflow. This embodiment, based on the aforementioned method, further includes dynamic memory management steps to ensure the long-term stability of the system. Traditional data caching typically uses a fixed-length circular buffer, but this may result in the loss of important parameter data or excessive memory consumption by unimportant parameters. This embodiment proposes a dynamic allocation scheme; please refer to the appendix. Figure 5 Specifically, the system allocates corresponding historical data storage lengths for different operating parameters based on their periodicity characteristics (e.g., high-frequency changing flow parameters require longer historical windows, while low-frequency changing environmental parameters require shorter windows) and their importance to humidity state inference. For example, location parameters have strong periodicity and are highly important for humidity inference, so they are allocated a longer storage length, such as storing data from the most recent 10 periods; environmental temperature parameters have weak periodicity and are less important, so they are allocated a shorter storage length, such as storing data from the most recent 100 sampling points.

[0064] The system monitors the storage space requirements of each parameter in real time. When it detects that the first parameter (such as the feed flow rate during a sudden process change) needs to increase its historical data storage length (for example, its periodicity becomes more obvious or its importance assessment increases), the system will automatically identify the second parameter whose current storage length is greater than the preset value and whose status change is relatively stable, such as a long-term constant ambient background temperature, shorten its historical data storage length (for example, from 10 cycles to 5 cycles), and reallocate the freed memory space to the first parameter, thereby keeping the total system memory usage constant.

[0065] To simplify calculations, the historical data storage length is selected from a predefined set of discrete lengths, such as a short-period level (storing the most recent 10 data points), a medium-period level (storing the last 50 data points), and a long-period level (storing the last 200 data points). Different parameters can be assigned different levels. When the storage length needs to be increased, a parameter can be upgraded from a short-period level to a medium-period level, or vice versa, while another parameter is downgraded in the opposite direction. Even if the job duration increases, the system's memory usage remains constant, preventing performance degradation or system crashes due to the accumulation of historical data. Through this dynamic memory management, the system achieves efficient utilization of memory resources without increasing hardware costs.

[0066] This invention also provides an adaptive control system for photovoltaic panel coatings based on interface humidity state inference. The system includes:

[0067] Data acquisition module: Used to collect multi-source operating parameters of the photovoltaic panel coating process in real time through an embedded edge computing unit, including but not limited to coating device operating status parameters, coating liquid supply behavior parameters, operation speed parameters, spatial position parameters, etc.

[0068] The dual-channel processing module is used to generate independent predicted values ​​for each parameter based on the multi-source operating parameters using a preset prediction model, and to compare the predicted values ​​with the real-time measured values ​​for consistency and adjust their weights. Specifically, when the deviation exceeds a preset physical continuity threshold, the measured value is determined to be abnormal, and the weight of the measured value is reduced while the weight of the predicted value is increased.

[0069] State inference module: This module infers the current humidity state of the interface between the coating device and the photovoltaic panel based on weighted parameters. This inference can be accomplished using a pre-built and locally running dynamic estimation model of interface humidity (such as the aforementioned state-space equation or neural network model).

[0070] Adaptive control module: Based on the current humidity status, it generates coating control commands locally in the edge computing unit and outputs them to the actuator of the photovoltaic panel coating equipment to adjust the liquid supply volume, liquid supply position, liquid replenishment strategy and / or coating operation speed.

[0071] The functions of each of the above modules are implemented in software or firmware within the embedded edge computing unit, without relying on cloud computing power or network communication. This achieves localization of data processing and control, reduces dependence on external networks, and ensures stable operation even under conditions of unstable communication at the photovoltaic panel site. (Appendix) Figure 6 The figure shows the change in light transmittance after film formation during stable system operation.

[0072] To further verify the effectiveness of this method, the inventors conducted experimental verification of the method in continuous coating operations on photovoltaic panels under different scenarios.

[0073] Scenario 1: Real-world abrupt environmental changes. During photovoltaic panel coating operations, sudden changes in ambient humidity or temperature can alter the coating solution's evaporation rate. The system identifies abnormal trends in interface humidity by analyzing multi-source operational data and generates control and adjustment strategies before significant changes in film quality occur. The system maintains film quality stability under abrupt environmental conditions without relying on precise humidity sensor readings. Figure 7 The figure shows the change in light transmittance after film formation under a sudden change in a certain parameter.

[0074] Scenario 2: Long-term memory stability scenarios. Please refer to the appendix. Figure 8The figure shows the variation of historical memory length with multiple parameters and lengths. In this embodiment, during the continuous coating operation of multiple photovoltaic panels, the embedded system parameter memory usage remains constant, and the historical length allocation tends to stabilize with running time. There is no performance degradation caused by the accumulation of historical data, which verifies the applicability of the method of the present invention to the long-term operation of embedded edge devices.

[0075] Scenario 3: Independent Operation Scenario with Communication Interruption. In this embodiment, the coating robot experiences a communication interruption during on-site operation, and the cloud system cannot provide real-time support. The embedded system continues to perform interface humidity state inference and coating control decisions based on the method of this invention, without any control interruption. This verifies that the method of this invention can still operate stably without relying on real-time computing power and data from the cloud, and is suitable for on-site coating conditions of photovoltaic panels.

[0076] Please see Figure 9 The diagram shown is a schematic representation of an electronic device for an adaptive control system of photovoltaic panel coating based on interface humidity state inference, as provided in an embodiment of this specification.

[0077] like Figure 9 As shown, the electronic device 600 may include at least one processor 610, at least one network interface 640, a user interface 630, a memory 650, and at least one communication bus 620.

[0078] The communication bus 620 can be used to realize the connection and communication of the above components.

[0079] The user interface 630 may include buttons, and the optional user interface may also include a standard wired interface or a wireless interface.

[0080] The network interface 640 may include, but is not limited to, Bluetooth modules, NFC modules, ZigBee modules, and UWB modules.

[0081] The processor 610 may include one or more processing cores. The processor 610 connects to various parts within the electronic device 600 using various interfaces and lines. It performs various functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 650, and by calling data stored in the memory 650. Optionally, the processor 610 may be implemented using at least one hardware form selected from DSP, FPGA, and PLA. The processor 610 may integrate one or more combinations of CPU and GPU. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display on the screen.

[0082] The memory 650 may include RAM or ROM. Optionally, the memory 650 may include a non-transitory computer-readable medium. The memory 650 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 650 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for at least one function, such as touch functionality, sound playback functionality, image playback functionality, etc., and instructions for implementing the various method embodiments described above. The data storage area may store data involved in the various method embodiments described above. Optionally, the memory 650 may also be at least one storage device located remotely from the aforementioned processor 610. As a computer storage medium, the memory 650 may include an operating system, a communication module, a user interface module, and a photovoltaic panel coating adaptive control application program based on interface humidity state inference. The processor 610 may be used to call the photovoltaic panel coating adaptive control application program based on interface humidity state inference stored in the memory 950 and execute the steps of the photovoltaic panel coating adaptive control based on interface humidity state inference mentioned in the foregoing embodiments.

[0083] The above description is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined in the claims, they should all fall within the protection scope of the present invention.

Claims

1. A photovoltaic panel coating adaptive control method based on interface humidity inference, applied to the embedded edge computing unit of a photovoltaic panel coating equipment, characterized in that, Includes the following steps: The embedded edge computing unit collects multi-source operating parameters of the photovoltaic panel coating process in real time. Based on the multi-source operating parameters, a preset prediction model is used to generate predicted values ​​for each parameter. The prediction model is generated independently based on the historical evolution trend of each parameter and the current operating status. Calculate the deviation between the real-time measured value and the predicted value of each multi-source operating parameter, and compare the deviation with a preset physical continuity threshold; When the deviation between the real-time measured value and the predicted value of the operating parameter exceeds the preset threshold, it is determined that the real-time measured value of the parameter is abnormal, the fusion weight of the abnormal parameter is reduced, and the predicted value is used as the main basis for inferring the humidity state of the interface between the coating device and the photovoltaic panel. Reducing the fusion weight of real-time measurements and using predicted values ​​as the main basis for state inference specifically includes: using a weighted fusion algorithm to determine the weight of each parameter; when the deviation exceeds a preset threshold, adjusting the weight coefficient of the predicted value to a first preset value and adjusting the weight coefficient of the real-time measurement value to a second preset value, wherein the first preset value is greater than the second preset value; Based on the weighted parameters, the current humidity state of the contact interface is inferred; Based on the current humidity state, a coating control command is generated locally in the edge computing unit and output to the photovoltaic panel coating equipment. The control command is generated based on a pre-stored interface humidity state-process parameter mapping table, which includes at least the following mapping rules: when the current humidity state is higher than a first preset threshold, a first control command is generated to increase the coating operation speed and / or reduce the liquid supply; when the current humidity state is lower than a second preset threshold, a second control command is generated to decrease the coating operation speed and / or increase the liquid supply.

2. The photovoltaic panel coating adaptive control method based on interface humidity inference according to claim 1, characterized in that, The prediction model includes: For periodic operating parameters, a periodic autoregressive model is used to generate the predicted value for the current time based on the value at the corresponding time of the previous period. For non-periodic operating parameters, an exponential smoothing model is used to generate the predicted value for the current moment based on the numerical sequence within the historical time window.

3. The photovoltaic panel coating adaptive control method based on interface humidity inference according to claim 1, characterized in that, When the deviation between the real-time measured value and the predicted value is within the preset threshold range, the real-time measured value is used as the basis for inferring the humidity status.

4. The photovoltaic panel coating adaptive control method based on interface humidity inference according to claim 1, characterized in that, The inference of the current humidity state at the interface between the coating device and the photovoltaic panel specifically includes: inputting the weighted operating parameters as input variables into the interface humidity dynamic estimation model pre-built and running locally in the embedded edge computing unit, and outputting the quantitative value of the current humidity state through the interface humidity dynamic estimation model.

5. The photovoltaic panel coating adaptive control method based on interface humidity inference according to claim 1, characterized in that, Before the real-time acquisition of multi-source operating parameters during the photovoltaic panel coating process, an initialization calibration step is also included: The photovoltaic panel coating equipment is controlled to operate under preset standard operating conditions, which include preset travel speed, liquid supply volume and environmental conditions. Collect multi-source operating parameters under the standard operating condition, and use the prediction model to generate corresponding baseline prediction values; Calculate the deviation between the baseline predicted value and the theoretical value corresponding to the standard operating condition; Based on the deviation, the initial parameters of the prediction model are corrected to compensate for the influence of sensor zero-point drift on state inference.

6. The photovoltaic panel coating adaptive control method based on interface humidity inference according to claim 1, characterized in that, It also includes dynamic memory management steps: Based on the periodic characteristics of each operating parameter and its importance to humidity state inference, the corresponding historical data storage length is configured for different parameters; When it is detected that the first parameter needs to increase the length of historical data storage, the system identifies the second parameter whose storage length is greater than the preset value and whose state changes steadily, reduces the length of historical data storage for the second parameter, and allocates the freed memory space to the first parameter to maintain a constant total system memory usage.

7. The photovoltaic panel coating adaptive control method based on interface humidity inference according to claim 6, characterized in that, The length of the historical data storage is selected from a predefined set of discrete lengths consisting of short-cycle, medium-cycle, and long-cycle levels.

8. A photovoltaic panel coating adaptive control system based on interface humidity inference, characterized in that, include: The data acquisition module is used to collect multi-source operating parameters of the photovoltaic panel coating process in real time through an embedded edge computing unit; The dual-channel processing module generates predicted values ​​for each parameter based on the multi-source operating parameters and using a preset prediction model. The prediction model is generated independently based on the historical evolution trend of the parameters and the current operating state. The module calculates the deviation between the real-time measured value and the predicted value of each parameter and compares the deviation with a preset physical continuity threshold. When the deviation between the real-time measured value and the predicted value of an operating parameter exceeds a preset threshold, it is determined that the real-time measured value of the parameter is abnormal, and the fusion weight of the abnormal parameter is reduced. At the same time, the predicted value is used as the main basis for inferring the humidity state of the interface between the coating device and the photovoltaic panel. Reducing the fusion weight of the real-time measured value and using the predicted value as the main basis for state inference specifically includes: using a weighted fusion algorithm to determine the weight of each parameter; when the deviation exceeds the preset threshold, adjusting the weight coefficient of the predicted value to a first preset value and adjusting the weight coefficient of the real-time measured value to a second preset value, wherein the first preset value is greater than the second preset value. The state inference module infers the current humidity state of the contact interface based on the weighted parameters. The adaptive control module generates coating control instructions locally in the edge computing unit based on the current humidity state and outputs them to the photovoltaic panel coating equipment. The control instructions are generated based on a pre-stored interface humidity state-process parameter mapping table, which includes at least the following mapping rules: when the current humidity state is higher than a first preset threshold, a first control instruction is generated to increase the coating operation speed and / or reduce the liquid supply; when the current humidity state is lower than a second preset threshold, a second control instruction is generated to decrease the coating operation speed and / or increase the liquid supply.

Images