Intelligent foam monitoring control method and device in lubricating oil filling process

Abstract

The invention discloses an intelligent foam monitoring control method and device in a lubricating oil filling process, and relates to the technical field of monitoring control, and the method comprises the steps: carrying out the real-time collection of the state data of lubricating oil in a filling container, and obtaining a multi-mode foam feature data set; performing foam feature fusion evaluation based on the multi-modal foam feature data set to generate a foam state influence score; lubricating oil filling control analysis is carried out according to the foam state influence score, a foam filling risk grade is generated, and a defoaming control instruction is constructed according to the foam filling risk grade; and the defoaming control instruction is sent to a filling execution mechanism, a defoaming device is controlled in a linkage mode to conduct defoaming operation, and closed-loop inhibition of foam in the filling process is achieved. The technical problem that foam generation and expansion in the filling process cannot be accurately recognized and inhibited in real time in the prior art is solved, and the technical effects of improving foam recognition precision and achieving foam dynamic regulation and control and closed-loop inhibition in the filling process are achieved.

Description

Technical Field

[0001] The present invention relates to the field of monitoring and control technology, and in particular to a method and device for intelligently monitoring and controlling foam in a lubricating oil filling process. Background Art

[0002] During the lubricant filling process, large amounts of foam are easily generated due to fluid disturbances, changes in filling speed, and the physical properties of the liquid itself. The continuous generation and unstable expansion of foam not only affects filling accuracy but can also lead to problems such as misjudgment of the liquid level, overflow, and poor sealing. Currently, in actual filling production lines, foam detection often relies on a single sensor signal, such as a liquid level sensor or image analysis, which fails to comprehensively reflect the foam's true dynamic state. Furthermore, the lack of intelligent control strategies that are linked to changes in foam state leads to a delayed response to foam regulation during the filling process, making effective real-time control impossible. Summary of the Invention

[0003] The present application provides a method and device for intelligent monitoring and control of foam during lubricating oil filling, which is used to solve the technical problem in the prior art that it is impossible to accurately identify and suppress the generation and expansion of foam during the filling process in real time.

[0004] In view of the above problems, the present application provides a method and device for intelligent monitoring and control of foam during the lubricating oil filling process.

[0005] In a first aspect of the present application, a method for intelligently monitoring and controlling foam during lubricating oil filling is provided, the method comprising:

[0006] The lubricating oil status data in the filling container is collected in real time to obtain a multimodal foam feature data set; a foam feature fusion assessment is performed based on the multimodal foam feature data set to generate a foam status impact score; a lubricating oil filling control analysis is performed based on the foam status impact score to generate a filling foam risk level, and a defoaming control instruction is constructed based on the filling foam risk level; the defoaming control instruction is sent to the filling actuator, which controls the defoaming device to perform a defoaming operation, thereby achieving closed-loop suppression of foam during the filling process.

[0007] In a second aspect of the present application, a device for intelligently monitoring and controlling foam during lubricating oil filling is provided, the device comprising:

[0008] A data acquisition module is used to collect lubricating oil status data in the filling container in real time to obtain a multimodal foam feature data set; an evaluation module is used to perform foam feature fusion evaluation based on the multimodal foam feature data set to generate a foam status impact score; a control analysis module is used to perform lubricating oil filling control analysis based on the foam status impact score, generate a filling foam risk level, and construct a defoaming control instruction based on the filling foam risk level; a defoaming operation module is used to send the defoaming control instruction to the filling actuator, and control the defoaming device to perform a defoaming operation to achieve closed-loop suppression of foam during the filling process.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] The present application collects the lubricating oil status data in the filling container in real time to obtain a multimodal foam feature data set; performs foam feature fusion evaluation based on the multimodal foam feature data set to generate a foam status impact score; performs lubricating oil filling control analysis based on the foam status impact score to generate a filling foam risk level, and constructs a defoaming control instruction based on the filling foam risk level; sends the defoaming control instruction to the filling actuator, and controls the defoaming device to perform a defoaming operation, thereby achieving closed-loop suppression of foam during the filling process. The present invention solves the technical problem in the prior art that it is impossible to accurately identify and suppress the generation and expansion of foam in the filling process in real time. Through the real-time collection and fusion evaluation of multimodal foam feature data, a foam status impact score is generated, and a defoaming control instruction is dynamically constructed based on the score to achieve linkage control with the defoaming device, thereby achieving the technical effect of improving foam recognition accuracy and achieving dynamic regulation and closed-loop suppression of foam during the filling process. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0012] Figure 1 A schematic flow chart of a method for intelligently monitoring and controlling foam during lubricating oil filling provided in an embodiment of the present application;

[0013] Figure 2 Schematic diagram of the structure of the intelligent foam monitoring and control device during the lubricating oil filling process provided in an embodiment of the present application.

[0014] Description of reference numerals: data acquisition module 11 , evaluation module 12 , control and analysis module 13 , defoaming operation module 14 . DETAILED DESCRIPTION

[0015] The present application provides an intelligent foam monitoring and control method and device during the lubricating oil filling process, aiming to solve the technical problem in the prior art that it is impossible to accurately identify and suppress the generation and expansion of foam in real time during the filling process. Through the real-time collection and fusion evaluation of multimodal foam characteristic data, a foam state impact score is generated, and defoaming control instructions are dynamically constructed based on the score to achieve linkage control with the defoaming device, thereby achieving the technical effect of improving foam identification accuracy and realizing dynamic regulation and closed-loop suppression of foam during the filling process.

[0016] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only some of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0017] It should be noted that any variations of the terms "include" and "have" are intended to cover non-exclusive inclusions. For example, a process, method, apparatus, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or are inherent to these processes, methods, products or devices.

[0018] Example 1, as Figure 1 As shown, the present application provides a method for intelligent monitoring and control of foam during lubricating oil filling, the method comprising:

[0019] Step S100: collecting the lubricating oil status data in the filling container in real time to obtain a multimodal foam characteristic data set.

[0020] In this embodiment, to achieve real-time data collection of lubricant status within a filling container, a multi-source sensor array is triggered before filling begins to synchronously collect data from the filling container, acquiring a multimodal dynamic filling dataset containing pressure fluctuations, dielectric constant changes, and foam image sequences. Subsequently, a three-dimensional spatial coordinate system is constructed based on the container's geometric parameters. Filling speed data is retrieved, and the various sensor data are dynamically calibrated in time and space using the coordinate system to form a structured calibration dataset. Multimodal characteristic parameters are then extracted from the calibration data, ultimately resulting in a multimodal foam characteristic dataset.

[0021] Furthermore, in the method provided in the embodiment of the application, real-time data on the state of the lubricating oil in the filling container is collected to obtain a multimodal foam characteristic data set, which also includes:

[0022] Before the lubricating oil is filled, the multi-source sensor group is triggered to synchronously collect data from the filling container to obtain a multi-mode dynamic filling data set, which includes filling pressure fluctuation data, filling dielectric constant change data, and filling foam image sequence data; a three-dimensional spatial coordinate system of the filling container is constructed according to geometric parameters; the filling speed data is retrieved, and the filling pressure fluctuation data, the filling dielectric constant change data, and the filling foam image sequence data are dynamically calibrated in time and space based on the three-dimensional spatial coordinate system and the filling speed data to obtain a calibration data set; multi-modal feature extraction is performed based on the calibration data set to obtain the multi-modal foam feature data set.

[0023] In this embodiment, before the lubricant is filled, a multi-source sensor system consisting of a piezoelectric pressure sensor, a capacitive sensor, and an industrial vision camera is activated to synchronously and real-timely collect the state of the lubricant in the filling container. The piezoelectric pressure sensor is used to capture pressure changes caused by foam disturbances during the filling process, generating filling pressure fluctuation data; the capacitive sensor senses changes in the dielectric constant of the lubricant medium, characterizing the degree of bubble doping and generating filling dielectric constant change data; and the industrial vision camera continuously captures image sequences at a fixed frame rate to record the foam generation and morphological changes, generating filling foam image sequence data. Through this acquisition process, a multimodal dynamic filling dataset is obtained.

[0024] Next, the filling container's three-dimensional coordinate system is constructed based on its geometric parameters. This process begins by reading the container's product design drawings to obtain its actual geometric parameters, including its inner diameter, height, opening diameter, radius of curvature, and wall angle. Based on these parameters, a parametric modeling approach is employed, with the center of the container's bottom serving as the coordinate origin. The vertical Z axis represents the axis of liquid level rise, while the X and Y axes correspond to the container's horizontal cross-section. This completes the construction of the container's three-dimensional coordinate system.

[0025] The filling speed data is then retrieved—the volumetric flow rate of lubricant injected into the container per unit time. Based on this filling speed data and the constructed three-dimensional spatial coordinate system of the filling container, the rate of rise of the lubricant level is calculated. Furthermore, the spatial height information and sampling frame rate of each sensor are combined to create a timestamp mapping table that reflects the temporal positional correspondence between the liquid level and the sensor observation points. Using this timestamp mapping table, the filling pressure fluctuation data, the filling dielectric constant change data, and the filling foam image sequence data are time-aligned and spatially matched, completing the spatiotemporal dynamic calibration of each modal data and outputting the calibration dataset.

[0026] Multimodal feature extraction is then performed based on the calibration dataset. Specifically, the calibrated data is first analyzed item by item to obtain filling pressure fluctuation correction data, filling dielectric constant change correction data, and filling foam image sequence correction data. Then, based on the filling pressure fluctuation correction data, a sliding window variance algorithm is used to calculate the pressure variance data. Based on the filling dielectric constant change correction data, the first-order difference method is used to extract the rate of change per unit time to obtain the dielectric gradient data. Based on the filling foam image sequence correction data, a threshold segmentation method is used to extract the foam area, calculate the foam coverage and average bubble diameter, and further combine the two to derive the dynamic expansion rate of the foam.

[0027] Finally, the pressure variance data, dielectric gradient data, and foam dynamic expansion rate are fused and integrated in a unified time series to construct a multimodal foam characteristic dataset for characterizing foam evolution behavior.

[0028] Furthermore, in the method provided in the embodiment of the application, the filling speed data is retrieved, and the filling pressure fluctuation data, the filling dielectric constant change data, and the filling foam image sequence data are subjected to spatiotemporal dynamic calibration based on the three-dimensional spatial coordinate system and the filling speed data to obtain a calibration data set, further comprising:

[0029] The multi-source sensor group is position-labeled based on the three-dimensional spatial coordinate system to determine multiple sensor position coordinates; the lubricating oil liquid level rise rate is calculated based on the filling speed data, the liquid level rise rate is mapped to the frame rate of the multi-source sensor group, and a timestamp mapping table is constructed; the filling pressure fluctuation data, the filling dielectric constant change data, and the filling foam image sequence data are subjected to spatiotemporal synchronous matching correction according to the timestamp mapping table and the multiple sensor position coordinates to obtain the calibration data set.

[0030] In an embodiment of the present application, the position of the multi-source sensor group is first marked based on the three-dimensional spatial coordinate system. Specifically, the installation height and direction information of each sensor are read, and the rectangular coordinate mapping method is used to convert the physical installation point of each sensor into a spatial coordinate with the center of the bottom of the container as the origin, and multiple sensor position coordinates are determined.

[0031] Next, we retrieve the filling speed data—the volumetric flow rate of the lubricant injected into the container per unit time—and calculate the rate of change of the liquid level over time based on the container's bottom area. This step uses a linear volume-to-height conversion method, using the formula h / t = v / A, where h is the liquid level rise, t is the unit time, v is the filling speed, and A is the cross-sectional area of ​​the filling container. This calculates the rate of change of the liquid level per unit time, or the liquid level rise rate.

[0032] The obtained liquid level rise rate is then matched to the sampling frequency of the multi-source sensor, and a timestamp mapping table is constructed using linear time interpolation. Specifically, a time-liquid level comparison relationship is established, with each sampling moment as the horizontal coordinate and the change in liquid level as the vertical coordinate. For various types of sensors (such as industrial vision cameras that sample 30 frames per second and capacitive sensors that sample every 0.1 seconds), the sampling time of each frame is accurately matched to the liquid level based on its frame rate, ensuring that different types of data frames can be associated with the same liquid level state.

[0033] Finally, based on the constructed timestamp mapping table and the calibrated sensor position coordinates, a joint calibration process is performed on the collected filling pressure fluctuation data, filling dielectric constant change data, and filling foam image sequence data. This step uses a consistency registration method based on spatiotemporal mapping. That is, by comparing the liquid level corresponding to the sampling time of each data frame with the spatial position of its sensor observation point, data points that meet the spatial liquid level intersection condition are selected as valid frames. For deviated or discontinuous data frames, linear interpolation is used to fill them on the time axis to ensure that the multimodal data is synchronized and consistent in time and space. Finally, a calibration dataset is output. This calibration dataset contains the synchronization information of the three types of modalities: filling pressure fluctuation correction data, filling dielectric constant change correction data, and filling foam image sequence correction data under the same liquid level state.

[0034] Furthermore, in the method provided in the embodiment of the application, multimodal feature extraction is performed based on the calibration data set to obtain the multimodal foam feature data set, further comprising:

[0035] The calibration data set is parsed to extract filling pressure fluctuation correction data, filling dielectric constant change correction data, and filling foam image sequence correction data; pressure variance data is calculated based on the filling pressure fluctuation correction data; dielectric constant change correction data is calculated to determine dielectric gradient data; foam coverage and average bubble diameter are extracted based on the filling foam image sequence correction data; foam dynamic expansion rate is calculated based on the foam coverage and the average bubble diameter; the pressure variance data, the dielectric gradient data, and the foam dynamic expansion rate are correlated and integrated in time series to construct the multimodal foam feature data set.

[0036] In an embodiment of the present application, the calibration data set is first parsed, the data that has completed time and space alignment is classified and standard decoded according to the data source, and three types of corrected modal data are extracted to obtain filling pressure fluctuation correction data, filling dielectric constant change correction data and filling foam image sequence correction data.

[0037] For the filling pressure fluctuation correction data, the sliding window variance method is used to perform disturbance intensity analysis. The pressure data in the time series is divided into multiple time windows of fixed length (for example, each 0.5 second is a window), and the variance of the pressure value is calculated in each window to quantify the degree of liquid disturbance in that period and obtain the pressure variance data.

[0038] For the filling dielectric constant change correction data, the first-order difference method is used to analyze the dielectric response change. By performing difference calculations on the dielectric constant values ​​at adjacent time points, the rate of change of the dielectric properties per unit time is calculated to characterize the degree of interference on the dielectric response after the foam is mixed into the lubricating oil, and to obtain continuous dielectric gradient data.

[0039] When processing the corrected data from a series of filling foam images, a fixed-threshold image segmentation method is first applied to extract the foam region within the image. By setting grayscale or brightness thresholds, the foam region in the image is effectively distinguished from the liquid surface background. Pixel region statistics are then used to calculate the area of ​​the segmented results, determining the proportion of the foam region to the total image area and extracting this as the foam coverage. Contour recognition and geometric measurement are also used to identify the foam edges within the image. The equivalent diameter of each bubble outline is measured, and the average value of multiple bubbles is calculated to obtain the average bubble diameter.

[0040] Based on the obtained foam coverage and average bubble diameter, the spatial evolution rate of the foam was analyzed using the time-difference expansion rate calculation method. This method calculates the foam expansion rate by comparing the change in the product of the foam coverage and the average bubble diameter between two consecutive image frames and dividing it by the time interval between the frames. This gives the dynamic foam expansion rate.

[0041] Finally, the pressure variance data, dielectric gradient data and foam dynamic expansion rate are sequence fused based on timestamps. The time series feature fusion method is used to align and merge the three types of modal features into a multidimensional feature vector in a unified format, and then organize them in chronological order to form a complete multimodal foam feature dataset.

[0042] Step S200: performing foam feature fusion evaluation based on the multimodal foam feature dataset to generate a foam state impact score.

[0043] In this embodiment, foam feature fusion assessment is performed based on a multimodal foam feature dataset. Pressure variance data, dielectric gradient data, and foam dynamic expansion rate are first normalized to construct a standard feature vector. Subsequently, a weighted fusion is performed on each feature to extract foam variation trends and generate a preliminary assessment result. Based on this, filling stage information is incorporated to modify the assessment result, ultimately outputting a foam state impact score that comprehensively reflects the degree of foam state variation.

[0044] Furthermore, in the method provided in the embodiment of the application, performing foam feature fusion evaluation based on the multimodal foam feature dataset to generate a foam state impact score further includes:

[0045] The pressure variance data, the medium gradient data, and the foam dynamic expansion rate are normalized to construct a standard feature vector; attention weighted fusion is performed based on the standard feature vector to obtain multiple feature dynamic weight parameters; foam growth trend is predicted based on the multiple feature dynamic weight parameters to generate a primary foam index, filling stage information is introduced, and spatiotemporal context compensation is performed according to the filling stage information to generate a compensation result; the primary foam index is corrected according to the compensation result to generate a foam correction index, and nonlinear score mapping is performed based on the foam correction index to obtain a foam state impact score.

[0046] In the embodiment of the present application, the pressure variance data, the dielectric gradient data, and the dynamic expansion rate of the foam are first normalized. This process adopts the minimum-maximum normalization method, that is, based on the maximum and minimum value intervals in the feature history samples, the feature values ​​at the current moment are linearly mapped and compressed to the standard interval of [0, 1] to complete the standard scale conversion and obtain the standard feature vector.

[0047] Next, attention-weighted fusion is performed based on the standard feature vector. During this process, dynamic weighting is performed based on the change amplitude of each standard feature within a sliding time window (e.g., 1 second as a time frame). Specifically, the time series variance value of each feature is calculated within the current time window, and then the inverse is taken and normalized to obtain three dynamic weighting parameters of the features. For example, if the variance of the pressure variance within the current window is 0.05, the dielectric gradient is 0.01, and the foam expansion rate is 0.03, the inverses are 20, 100, and 33.3, respectively. After normalization, the attention weights are 0.14, 0.69, and 0.17, and the sum of the three is 1.

[0048] Subsequently, the foam growth trend is predicted based on multiple characteristic dynamic weight parameters. This process uses a weighted linear combination method to multiply the standard characteristic vector with the corresponding dynamic weight parameter and sum them to obtain the primary foam index. The filling stage information is then introduced, and spatiotemporal context compensation is performed according to the filling stage information. In the filling context compensation stage, the liquid level filling rate is calculated by the ratio of the liquid level height to the container height in the real-time filling process, and the current filling stage information is identified based on this. According to the preset rules, the filling rate is divided into initial filling (<30%), mid-term filling (30% to 80%) and late filling (≥80%), and the context compensation coefficients are configured as 0.8, 1.0, and 1.5 respectively. This compensation coefficient is used as the compensation result.

[0049] The primary foam index is then corrected based on the compensation results. Combined with information from the current filling phase and based on a pre-set risk adjustment mechanism, the response strength of the foam index is dynamically adjusted to generate a modified foam index that reflects the actual filling risk environment. A nonlinear segmented mapping operation is then performed on the modified foam index, making the value more sensitive in high-risk areas. This results in a quantified foam impact score.

[0050] Furthermore, in the method provided in the embodiment of the application, the primary foam index is corrected according to the compensation result to generate a foam correction index, and a nonlinear score mapping is performed based on the foam correction index to obtain a foam state impact score, which also includes:

[0051] When the filling stage information is initial filling, the primary foam index is risk suppressed according to the compensation result to generate a first foam correction index; when the filling stage information is mid-term filling, the primary foam index is risk benchmarked and analyzed according to the compensation result to generate a second foam correction index; when the filling stage information is late filling, the primary foam index is risk amplified according to the compensation result to generate a third foam correction index; the first foam correction index, the second foam correction index, or the third foam correction index is segmented and calculated to obtain the foam state impact score.

[0052] In an embodiment of the present application, when the filling stage information is determined to be the initial filling stage (i.e., the liquid level filling rate is less than 30%), it is considered that the container space is sufficient and the risk of foam overflow is low. Therefore, a risk suppression mechanism is adopted, that is, the primary foam index is multiplied by the aforementioned context compensation coefficient (set to 0.8) to obtain a first foam correction index.

[0053] If the current filling stage information belongs to the mid-term filling stage (filling rate is between 30% and 80%), it is assumed that the foam risk in this stage is in a relatively balanced state. Therefore, the risk benchmark analysis mechanism is executed, that is, the primary foam index is directly multiplied by the compensation coefficient 1.0 without amplifying or compressing the original value, so as to obtain the second foam correction index.

[0054] When the filling stage is identified as late (fill rate greater than or equal to 80%), the foam's upward movement is limited, significantly increasing the probability of overflow. Therefore, the risk amplification mechanism is activated. The primary foam index is then multiplied by the contextual compensation factor of 1.5 to generate the third foam correction index.

[0055] Then, the first foam correction index, the second foam correction index, or the third foam correction index is calculated as a segmented index. Specifically, if the correction index is ≤0.3, the score = correction index × 50, which increases linearly; if the correction index is in the range of (0.3, 0.7], the score = 15 + 100 × (correction index - 0.3), which enhances the median sensitivity; if the correction index is >0.7, the score = 50 + 150 × (correction index - 0.7). Through this process, the foam state impact score is calculated.

[0056] Step S300: performing lubricating oil filling control analysis according to the foam state impact score, generating a filling foam risk level, and constructing a defoaming control instruction according to the filling foam risk level.

[0057] In this embodiment, the current filling state is first dynamically risk-graded based on the foam impact score, generating a corresponding filling foam risk level. Subsequently, historical filling process logs are retrieved and reinforcement learning training is performed based on this data to generate a defoaming decision library containing response strategies for various scenarios. The current risk level is input into this decision library for strategy matching, resulting in several optional defoaming action combinations. This is then combined with the current filling environment parameters for multi-parameter optimization, resulting in the most appropriate defoaming control instructions for the current operating conditions.

[0058] Furthermore, in the method provided in the embodiment of the application, lubricant filling control analysis is performed based on the foam state impact score, a filling foam risk level is generated, and defoaming control instructions are constructed based on the filling foam risk level, further comprising:

[0059] Dynamic risk grading is performed based on the foam state impact score to divide the filling foam risk level; historical filling process record logs are retrieved, reinforcement learning is performed based on the historical filling process record logs, and a defoaming decision library is constructed; the filling foam risk level is synchronized to the defoaming decision library for query to obtain multiple defoaming action combinations; dynamic optimization is performed based on the multiple defoaming action combinations in combination with filling environment parameters to construct the defoaming control instructions.

[0060] In the embodiment of the present application, when dynamic risk grading is performed based on the foam state impact score, the interval classification method is used to map the foam state impact score to a risk level label. Specifically, the foam state impact score is divided into three level intervals. If the score value is less than or equal to 30, it is identified as a low risk level, indicating that the foam behavior can be ignored; if the score value is in the interval of (30, 70], it is identified as a medium risk level, indicating that the foam may interfere with filling but is still controllable; if the score value is higher than 70, it is identified as a high risk level. Through this process, the filling foam risk level is determined.

[0061] Next, in the process of retrieving the historical filling process log, the log database containing the four-tuple of "score-risk level-control action combination-response feedback" is retrieved from the historical records to extract the historical record sequence related to the current filling foam risk level. Subsequently, the Q-value update mechanism is used to construct the state-action mapping relationship, and the Q-Learning algorithm is used to take each risk level as the state input and the control action combination as the action output. The immediate reward value is calculated based on the foam reduction rate and filling error after the corresponding action, and the Q value is iteratively updated. For example, when the risk level is "medium risk", the historical action "injection rate -5% + stop flow delay 1.5s" produces a 15% reduction in foam and a reduction in error to 0.5ml, and the corresponding Q value is increased. Through this reinforcement learning method, a defoaming decision library containing multiple sets of state-action value functions is constructed, and a defoaming decision library for multi-level foam states is obtained.

[0062] The filling foam risk level is then synchronized to the defoaming decision database for query, and the currently identified filling foam risk level is used as the state input item to retrieve multiple high Q value action combinations under the corresponding state. Each set of action combinations contains at least two operating parameter dimensions, such as "liquid injection rate adjustment ratio", "stop flow delay" and "auxiliary jet intensity", and is accompanied by historical execution return values ​​for response comparison. Based on the Q value ranking, the top N candidate action paths are output. For example, under high risk levels, combinations such as "liquid injection rate -10% + stop flow delay 2.5s + high-frequency aerosol spray 1.2s" are returned to obtain multiple defoaming action combinations.

[0063] Finally, dynamic optimization is performed based on multiple defoaming action combinations combined with filling environmental parameters. Specifically, first, based on the filling stage information, the filling environment temperature parameters and the filling lubricating oil temperature parameters are collected through temperature sensing means, and the deviation between the two is calculated to generate a temperature deviation value. Then, the environmental impact mapping factor is established in combination with the deviation value, and the environmental compensation factor is obtained according to the grading rules and added to the filling environmental parameters. Subsequently, the filling environmental parameters are used as constraints to dynamically adapt and optimize the multiple defoaming action combinations obtained from the defoaming decision library. For each set of action combinations, the parameter compensation method is first used to perform environmental correction processing on the three key parameters of the injection rate adjustment ratio, the flow stop delay and the duration of the high-frequency aerosol spray. The correction process is based on the environmental compensation factor. For example, if the current environmental compensation factor is 0.85, the combination with the original injection rate adjusted to -10% will be corrected to -12% (-10% / 0.85) to enhance the anti-foaming effect under high temperature and high foam risk conditions. Subsequently, a multi-factor weighted evaluation method was used to score the environmental adaptability of each combination, setting three evaluation dimensions. Parameter consistency refers to whether the corrected parameters meet the equipment's execution upper limit and operating tolerance, with a score range of 0-10 points. The filling beat matching assesses whether the combined action causes excessive delay to the standard beat, with a beat difference of <1 second set based on the actual beat offset. The historical feedback effectiveness is normalized based on the foam suppression success rate of the combination in similar environments. Finally, a weighted average algorithm was used to aggregate the scores of the three dimensions with weights of 0.4, 0.3, and 0.3 to obtain the total adaptability score of the combination in the current environment. For example, a combination scored 9, 8, and 10 in the three dimensions, respectively, with a final score of 0.4×9+0.3×8+0.3×10=8.9. After completing the scoring and sorting of all combinations, the action combination with the highest total score was selected as the final output defoaming control instruction.

[0064] Furthermore, in the method provided in the embodiment of the application, the process of constructing the filling environment parameters further includes:

[0065] Temperature sensing is performed based on the filling stage information to obtain filling environment temperature parameters and filling lubricating oil temperature parameters; deviation calculation is performed based on the filling environment temperature parameters and the filling lubricating oil temperature parameters to generate a temperature deviation value; filling foam correlation impact analysis is performed based on the temperature deviation value to construct an environmental impact mapping factor; graded compensation is performed based on the environmental impact mapping factor, and an environmental compensation factor is constructed and added to the filling environment parameters.

[0066] In this embodiment, temperature acquisition is first performed based on information from the current filling phase. A temperature sensor located above the filling area measures the ambient temperature, while an oil temperature sensor installed on the inner wall of the filling line measures the real-time temperature of the lubricating oil being filled. These parameters, respectively, are used to determine the ambient temperature and the lubricating oil temperature.

[0067] Next, a deviation calculation is performed based on the filling environment temperature parameter and the filling lubricating oil temperature parameter, that is, the temperature deviation value is obtained by subtracting the filling environment temperature parameter from the filling lubricating oil temperature parameter.

[0068] Then, based on the temperature deviation value, the filling foam correlation impact analysis was carried out, and the temperature difference interval lookup table method was used to construct the environmental impact mapping factor. This process relies on a preset rule table, which divides the temperature difference into several intervals (such as: [0–10]℃, (10–20]℃, (20–30]℃) and gives the corresponding mapping factors for each interval (such as: 0.95, 0.85, 0.72). If the temperature deviation value is 21.5℃, the corresponding mapping factor is 0.72, indicating that the environmental conditions have a significant trend of improving foam activity.

[0069] After obtaining the environmental impact mapping factor, a graded compensation is performed based on the environmental impact mapping factor to construct an environmental compensation factor. The grading rule uses a range-corresponding mechanism, categorizing the environmental impact mapping factor into high-risk segments (0.6–0.75), medium-risk segments (0.75–0.9), and low-risk segments (0.9–1.0). The corresponding environmental compensation factors are 0.85, 0.92, and 1.00, respectively. In the above example, the environmental impact mapping factor of 0.72 falls into the high-risk segment, so the matching environmental compensation factor is 0.85. Finally, the resulting environmental compensation factor is added to the filling environmental parameters, which include the original measured ambient temperature, oil temperature, and the environmental compensation factor.

[0070] Step S400: sending the defoaming control instruction to the filling actuator, and controlling the defoaming device to perform defoaming operation, thereby realizing closed-loop suppression of foam during the filling process.

[0071] In an embodiment of the present application, after the defoaming control instruction is sent to the filling actuator, the linkage control process is started to drive the corresponding defoaming device to operate, including operation links such as liquid injection rate adjustment, flow stop control and high-frequency aerosol spraying. After the defoaming action is executed, the defoaming status data of the lubricating oil after defoaming is continuously collected, such as indicators such as foam height, bubble density and dissipation time, and the defoaming status data is fed back and verified to calculate the foam suppression effect. Subsequently, the suppression effect is graded to determine the current foam level index. If the foam level index does not drop to the preset safety threshold, the defoaming control instruction used last time is iteratively updated, the execution parameters are adjusted and a defoaming control update instruction is formed. The updated instruction is sent to the filling actuator again to drive the defoaming device to adaptively optimize on the original basis to achieve enhanced execution of the foam suppression action. Through this closed-loop feedback and instruction update mechanism, the foam level is dynamically controlled to ensure the stability of the filling process and the safe control of the liquid level.

[0072] Furthermore, in the method provided in the embodiment of the application, after the defoaming device is controlled to operate, the method further includes:

[0073] Continuously collect defoaming status data of the lubricating oil after defoaming, verify based on the defoaming status data, and obtain the foam suppression effect; perform grading based on the foam suppression effect, determine the foam level index, and if the foam level index does not drop to the safety threshold, iteratively update the defoaming control instruction to obtain a defoaming control update instruction; execute the defoaming control update instruction to perform control recording, and adaptively optimize the control defoaming device to achieve closed-loop suppression of foam during the filling process.

[0074] In the present embodiment, after a lubricant has completed a defoaming operation, post-defoaming lubricant status data is continuously collected, including data on foam coverage, foam height, average bubble diameter, and foam duration. This data is used to form defoaming status data. This defoaming status data is compared with pre-defoaming foam characteristic data to calculate the extent of the reduction in various foam characteristics, thereby determining the current degree of foam removal, or the foam suppression effect. For example, if the foam height decreases from 8.2 mm to 3.5 mm and the coverage decreases from 35% to 14%, the foam suppression effect can be quantified.

[0075] Next, a classification is performed based on the foam suppression effect. Specifically, the foam coverage, foam height, bubble diameter, and dissipation time are normalized to the interval [0, 1], and a weighted average is calculated according to the preset weights (e.g., 25% each) to serve as the foam level index. For example, at a certain moment, the normalized values ​​of each parameter are 0.32, 0.40, 0.28, and 0.36, respectively, resulting in a calculated foam level index of 0.34. The safety threshold is set at 0.30. If the current level index exceeds this threshold, it is considered that the foam state still has potential risks.

[0076] When the foam level index rises above the safety threshold, the control command update process is initiated, adjusting the parameters of the previously executed defoaming control command. This control command includes three key parameters: the injection rate adjustment range, the stop delay time, and the jet duration. Based on the degree of insufficient foam suppression, the command is fine-tuned, for example, increasing the injection rate from -8% to -10%, extending the stop delay time from 1.5 seconds to 1.8 seconds, and increasing the jet duration from 1.0 second to 1.2 seconds. These new parameters are then combined into the defoaming control update command.

[0077] After executing the updated defoaming control update command, the control action, parameters used, execution time, and corresponding foam level index are simultaneously recorded in the control record database. If a similar foam level situation occurs again, the historical records can be used to prioritize the parameter combinations that have been verified to be effective. Furthermore, based on a comprehensive comparison of each execution feedback and parameter results, different combinations are prioritized to achieve gradual optimization of the control strategy, achieving the goal of adaptive optimization of control commands as the foam state changes. Through the continuous cycle of this monitoring, verification, updating, recording, and optimization process, closed-loop foam suppression control is achieved during the lubricant filling process.

[0078] In the embodiments of the present application, in summary, the embodiments of the present application have at least the following technical effects:

[0079] The present application collects the lubricating oil status data in the filling container in real time to obtain a multimodal foam feature data set; performs foam feature fusion evaluation based on the multimodal foam feature data set to generate a foam status impact score; performs lubricating oil filling control analysis based on the foam status impact score to generate a filling foam risk level, and constructs a defoaming control instruction based on the filling foam risk level; sends the defoaming control instruction to the filling actuator, and controls the defoaming device to perform a defoaming operation, thereby achieving closed-loop suppression of foam during the filling process. The present invention solves the technical problem in the prior art that it is impossible to accurately identify and suppress the generation and expansion of foam in the filling process in real time. Through the real-time collection and fusion evaluation of multimodal foam feature data, a foam status impact score is generated, and a defoaming control instruction is dynamically constructed based on the score to achieve linkage control with the defoaming device, thereby achieving the technical effect of improving foam recognition accuracy and achieving dynamic regulation and closed-loop suppression of foam during the filling process.

[0080] Example 2, based on the same inventive concept as the method for intelligent monitoring and controlling foam during lubricating oil filling in the above embodiment, Figure 2 As shown, the present application provides an intelligent foam monitoring and control device for the lubricating oil filling process. The device and method embodiments in the present application are based on the same inventive concept. The device includes:

[0081] The data acquisition module 11 is used to collect the lubricating oil status data in the filling container in real time to obtain a multimodal foam feature data set; the evaluation module 12 is used to perform foam feature fusion evaluation based on the multimodal foam feature data set to generate a foam status impact score; the control analysis module 13 is used to perform lubricating oil filling control analysis based on the foam status impact score, generate a filling foam risk level, and construct a defoaming control instruction based on the filling foam risk level; the defoaming operation module 14 is used to send the defoaming control instruction to the filling actuator, and control the defoaming device to perform a defoaming operation to achieve closed-loop suppression of foam during the filling process.

[0082] Furthermore, the device is also used to implement the following functions:

[0083] Before the lubricating oil is filled, the multi-source sensor group is triggered to synchronously collect data from the filling container to obtain a multi-mode dynamic filling data set, which includes filling pressure fluctuation data, filling dielectric constant change data, and filling foam image sequence data; a three-dimensional spatial coordinate system of the filling container is constructed according to geometric parameters; the filling speed data is retrieved, and the filling pressure fluctuation data, the filling dielectric constant change data, and the filling foam image sequence data are dynamically calibrated in time and space based on the three-dimensional spatial coordinate system and the filling speed data to obtain a calibration data set; multi-modal feature extraction is performed based on the calibration data set to obtain the multi-modal foam feature data set.

[0084] Furthermore, the device is also used to implement the following functions:

[0085] The multi-source sensor group is position-labeled based on the three-dimensional spatial coordinate system to determine multiple sensor position coordinates; the lubricating oil liquid level rise rate is calculated based on the filling speed data, the liquid level rise rate is mapped to the frame rate of the multi-source sensor group, and a timestamp mapping table is constructed; the filling pressure fluctuation data, the filling dielectric constant change data, and the filling foam image sequence data are subjected to spatiotemporal synchronous matching correction according to the timestamp mapping table and the multiple sensor position coordinates to obtain the calibration data set.

[0086] Furthermore, the device is also used to implement the following functions:

[0087] The calibration data set is parsed to extract filling pressure fluctuation correction data, filling dielectric constant change correction data, and filling foam image sequence correction data; pressure variance data is calculated based on the filling pressure fluctuation correction data; dielectric constant change correction data is calculated to determine dielectric gradient data; foam coverage and average bubble diameter are extracted based on the filling foam image sequence correction data; foam dynamic expansion rate is calculated based on the foam coverage and the average bubble diameter; the pressure variance data, the dielectric gradient data, and the foam dynamic expansion rate are correlated and integrated in time series to construct the multimodal foam feature data set.

[0088] Furthermore, the device is also used to implement the following functions:

[0089] The pressure variance data, the medium gradient data, and the foam dynamic expansion rate are normalized to construct a standard feature vector; attention weighted fusion is performed based on the standard feature vector to obtain multiple feature dynamic weight parameters; foam growth trend is predicted based on the multiple feature dynamic weight parameters to generate a primary foam index, filling stage information is introduced, and spatiotemporal context compensation is performed according to the filling stage information to generate a compensation result; the primary foam index is corrected according to the compensation result to generate a foam correction index, and nonlinear score mapping is performed based on the foam correction index to obtain a foam state impact score.

[0090] Furthermore, the device is also used to implement the following functions:

[0091] When the filling stage information is initial filling, the primary foam index is risk suppressed according to the compensation result to generate a first foam correction index; when the filling stage information is mid-term filling, the primary foam index is risk benchmarked and analyzed according to the compensation result to generate a second foam correction index; when the filling stage information is late filling, the primary foam index is risk amplified according to the compensation result to generate a third foam correction index; the first foam correction index, the second foam correction index, or the third foam correction index is segmented and calculated to obtain the foam state impact score.

[0092] Furthermore, the device is also used to implement the following functions:

[0093] Dynamic risk grading is performed based on the foam state impact score to divide the filling foam risk level; historical filling process record logs are retrieved, reinforcement learning is performed based on the historical filling process record logs, and a defoaming decision library is constructed; the filling foam risk level is synchronized to the defoaming decision library for query to obtain multiple defoaming action combinations; dynamic optimization is performed based on the multiple defoaming action combinations in combination with filling environment parameters to construct the defoaming control instructions.

[0094] Furthermore, the device is also used to implement the following functions:

[0095] Temperature sensing is performed based on the filling stage information to obtain filling environment temperature parameters and filling lubricating oil temperature parameters; deviation calculation is performed based on the filling environment temperature parameters and the filling lubricating oil temperature parameters to generate a temperature deviation value; filling foam correlation impact analysis is performed based on the temperature deviation value to construct an environmental impact mapping factor; graded compensation is performed based on the environmental impact mapping factor, and an environmental compensation factor is constructed and added to the filling environment parameters.

[0096] Furthermore, the device is also used to implement the following functions:

[0097] Continuously collect defoaming status data of the lubricating oil after defoaming, verify based on the defoaming status data, and obtain the foam suppression effect; perform grading based on the foam suppression effect, determine the foam level index, and if the foam level index does not drop to the safety threshold, iteratively update the defoaming control instruction to obtain a defoaming control update instruction; execute the defoaming control update instruction to perform control recording, and adaptively optimize the control defoaming device to achieve closed-loop suppression of foam during the filling process.

[0098] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0099] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

[0100] This specification and drawings are merely illustrative of the present application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art may make various modifications and variations to this application without departing from the scope of this application. Thus, this application is intended to include such modifications and variations as fall within the scope of this application and its equivalents.

Claims

1. Intelligent foam monitoring and control method during lubricating oil filling process, characterized in that: The method comprises: Real-time data collection of lubricating oil status in the filling container is performed to obtain a multimodal foam feature dataset; performing a foam feature fusion evaluation based on the multimodal foam feature dataset to generate a foam state impact score; performing lubricant filling control analysis based on the foam state impact score, generating a filling foam risk level, and constructing a defoaming control instruction based on the filling foam risk level; The defoaming control instruction is sent to the filling actuator, and the defoaming device is controlled in conjunction to perform the defoaming operation, thereby realizing closed-loop suppression of foam during the filling process.

2. The method for intelligent monitoring and control of foam during lubricating oil filling according to claim 1, characterized in that: Real-time data collection of lubricating oil status in a filling container is performed to obtain a multimodal foam characteristic dataset, the method comprising: Before the lubricating oil is filled, the multi-source sensor group is triggered to synchronously collect data from the filling container to obtain a multi-mode dynamic filling data set, which includes filling pressure fluctuation data, filling dielectric constant change data, and filling foam image sequence data; Construct a three-dimensional space coordinate system of the filling container according to geometric parameters; Retrieving filling speed data, and performing spatiotemporal dynamic calibration on the filling pressure fluctuation data, the filling dielectric constant change data, and the filling foam image sequence data based on the three-dimensional spatial coordinate system and the filling speed data to obtain a calibration data set; Multimodal feature extraction is performed based on the calibration data set to obtain the multimodal foam feature data set.

3. The method for intelligent monitoring and control of foam during lubricating oil filling according to claim 2, characterized in that: Retrieving filling speed data, and performing spatiotemporal dynamic calibration on the filling pressure fluctuation data, the filling dielectric constant change data, and the filling foam image sequence data based on the three-dimensional spatial coordinate system and the filling speed data to obtain a calibration data set, the method comprising: Marking the positions of the multi-source sensor group based on the three-dimensional spatial coordinate system to determine multiple sensor position coordinates; Calculating a liquid level rising rate of the lubricating oil according to the filling speed data, mapping the liquid level rising rate to a frame rate of a multi-source sensor group, and constructing a timestamp mapping table; The filling pressure fluctuation data, the filling dielectric constant change data, and the filling foam image sequence data are subjected to spatiotemporal synchronization matching correction according to the timestamp mapping table and the multiple sensor position coordinates to obtain the calibration data set.

4. The method for intelligent monitoring and controlling foam during lubricating oil filling according to claim 2, characterized in that: Performing multimodal feature extraction based on the calibration data set to obtain the multimodal foam feature data set, the method comprising: Parsing the calibration data set to extract filling pressure fluctuation correction data, filling dielectric constant change correction data, and filling foam image sequence correction data; Performing calculation based on the filling pressure fluctuation correction data to determine pressure variance data; Calculating based on the filling dielectric constant change correction data to determine dielectric gradient data; Extracting foam coverage and average bubble diameter based on the corrected data of the filling foam image sequence; Calculating the dynamic expansion rate of foam according to the foam coverage and the average diameter of bubbles; The pressure variance data, the medium gradient data, and the foam dynamic expansion rate are correlated and integrated according to a time series to construct the multimodal foam feature data set.

5. The method for intelligent monitoring and controlling foam during lubricating oil filling according to claim 4, characterized in that: Performing a foam feature fusion evaluation based on the multimodal foam feature dataset to generate a foam state impact score, the method comprising: Normalizing the pressure variance data, the medium gradient data, and the foam dynamic expansion rate to construct a standard feature vector; Performing attention weighted fusion based on the standard feature vector to obtain multiple feature dynamic weight parameters; Predicting foam growth trends based on the multiple characteristic dynamic weight parameters to generate a primary foam index, introducing filling stage information, performing spatiotemporal context compensation based on the filling stage information, and generating a compensation result; The primary foam index is corrected according to the compensation result to generate a foam correction index, and a nonlinear score mapping is performed based on the foam correction index to obtain a foam state impact score.

6. The method for intelligent monitoring and controlling foam during lubricating oil filling according to claim 5, characterized in that: The primary foam index is corrected according to the compensation result to generate a foam correction index, and a nonlinear score mapping is performed based on the foam correction index to obtain a foam state impact score, the method comprising: When the filling stage information is initial filling, risk suppression is performed on the primary foam index according to the compensation result to generate a first foam correction index; When the filling stage information is mid-term filling, performing a risk benchmark analysis on the primary foam index according to the compensation result to generate a second foam correction index; When the filling stage information indicates late filling, the primary foam index is risk-amplified according to the compensation result to generate a third foam correction index; The first foam correction index, the second foam correction index, or the third foam correction index is subjected to segmented index calculation to obtain the foam state impact score.

7. The method for intelligent monitoring and controlling foam during lubricating oil filling according to claim 5, characterized in that: Performing lubricant filling control analysis based on the foam state impact score to generate a filling foam risk level, and constructing a defoaming control instruction based on the filling foam risk level, the method comprising: Performing dynamic risk grading according to the foam state impact score to divide the filling foam risk level; Retrieving historical filling process record logs, performing reinforcement learning based on the historical filling process record logs, and building a defoaming decision library; Synchronizing the filling foam risk level to the defoaming decision database for query to obtain multiple defoaming action combinations; The defoaming control instruction is constructed based on the dynamic optimization of the multiple defoaming action combinations combined with the filling environment parameters.

8. The method for intelligent monitoring and controlling foam during lubricating oil filling according to claim 7, characterized in that: The construction process of filling environment parameters includes: Perform temperature sensing based on the filling stage information to obtain filling environment temperature parameters and filling lubricating oil temperature parameters; Perform deviation calculation based on the filling environment temperature parameter and the filling lubricating oil temperature parameter to generate a temperature deviation value; Performing filling foam correlation impact analysis based on the temperature deviation value and constructing an environmental impact mapping factor; Gradual compensation is performed based on the environmental impact mapping factors, and environmental compensation factors are constructed and added to the filling environmental parameters.

9. The method for intelligent monitoring and controlling foam during lubricating oil filling according to claim 1, characterized in that: After the linkage control defoaming device is activated, the method includes: Continuously collecting defoaming status data of the lubricating oil after defoaming, and performing verification based on the defoaming status data to obtain a foam suppression effect; performing a classification based on the foam suppression effect to determine a foam level index, and if the foam level index does not drop to a safety threshold, iteratively updating the defoaming control instruction to obtain a defoaming control update instruction; The defoaming control update instruction is executed to perform control recording, and the control defoaming device is adaptively optimized to achieve closed-loop suppression of foam during the filling process.

10. Intelligent foam monitoring and control device during lubricating oil filling process, characterized in that: The device is used to implement the method for intelligently monitoring and controlling foam in the lubricating oil filling process as claimed in any one of claims 1 to 9, and the device comprises: A data acquisition module is used to collect the lubricating oil status data in the filling container in real time to obtain a multimodal foam characteristic data set; an evaluation module, configured to perform a foam feature fusion evaluation based on the multimodal foam feature dataset and generate a foam state impact score; a control analysis module, configured to perform lubricant filling control analysis based on the foam state impact score, generate a filling foam risk level, and construct a defoaming control instruction based on the filling foam risk level; The defoaming operation module is used to send the defoaming control instruction to the filling actuator, and control the defoaming device to perform the defoaming operation in a linked manner to achieve closed-loop suppression of foam during the filling process.

Images