Ultrasonic cleaning machine remote monitoring diagnosis method based on internet of things

Abstract

The application belongs to the technical field of ultrasonic cleaning machine operation monitoring, and specifically discloses a remote monitoring and diagnosing method for an ultrasonic cleaning machine based on the Internet of Things, which matches a cleaning mode according to dirt information of objects to be cleaned, and the total time length, cleaning standards and cleaning operation indexes of each cleaning stage under the corresponding mode; bubble parameters, water body parameters and dirt state parameters on the surface of the objects to be cleaned are collected in real time during cleaning; the bubble and water body parameters are analyzed cooperatively, and dirt state verification is started; if an abnormal signal is triggered or the cleaning standards are not reached, an abnormal stage is marked, and abnormal indexes and degrees are identified; if the abnormal stage is not completed or is not the final stage, the cleaning strategy is dynamically adjusted based on the abnormal indexes and degrees, and the adjustment is recorded; after all stages are completed, a cleaning efficiency diagnosis report is generated according to the position and degree of the abnormal stage, and the report is uploaded to the cloud. Thus, accurate remote diagnosis and treatment of the cleaning machine are realized.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of ultrasonic cleaning machine operation monitoring, and in particular, relates to an ultrasonic cleaning machine remote monitoring and diagnosis method based on the Internet of Things. BACKGROUND

[0002] Ultrasonic cleaning machines are widely used in the field of precision manufacturing, and their operating parameters directly affect cleaning precision and effectiveness. With the development of Internet of Things technology and intelligent manufacturing, to adapt to high-precision cleaning needs, a remote monitoring and diagnosis mode has emerged.

[0003] The remote monitoring and diagnosis mode can realize remote tracking, abnormal early warning and fault diagnosis of the cleaning machine state. The prior art, such as the ultrasonic cleaning machine operation monitoring and early warning system based on data analysis disclosed in Chinese Patent Application No. 202410149598.4, monitors the core operating parameters of the ultrasonic cleaning machine in real time, generates evaluation results through dynamic analysis. When the parameters are abnormal, the correction mechanism is automatically triggered, and the parameter recovery time is synchronously tracked. If the correction delay exceeds the set threshold, a graded warning is immediately started. Thus, the cleaning effectiveness is ensured, and the operation safety is also guaranteed.

[0004] The prior art, such as the control method and system of an ultrasonic cleaning device disclosed in Chinese Patent Application No. 202311245530.8, establishes a matching model of material weight and cleaning parameters based on historical cleaning data, and automatically selects the optimal cleaning program through real-time weighing. Environmental indicators are continuously monitored during the cleaning process, and working parameters are dynamically adjusted. This technology realizes stable and standard cleaning effect of different specifications of workpieces, optimizes energy efficiency, and sets the best cleaning parameters.

[0005] The first prior art solution focuses on monitoring the operating parameters of the equipment, and the second prior art solution focuses on matching the process parameters. Obviously, both of them ignore the physical nature of the cleaning process, and there are the following problems: 1. Both of them ignore the monitoring of the cavitation effect carrier, and do not capture key indicators such as cavitation bubble density, distribution uniformity, and collapse strength, which reflect the actual cleaning power, which may lead to incorrect judgment of the actual cleaning effectiveness of the subsequent equipment operation.

[0006] 2. Both patents regard cleaning as a single continuous process, and do not establish a staged verification mechanism, which cannot display the differentiated cleaning detail state of different stages.

[0007] 3. Both patents lack real-time attention to the state of the cleaned object, making the evaluation perspective single, the verification result reliability biased, and the subsequent cleaning effect not significantly improved, and the subsequent cleaning machine maintenance direction not clear. SUMMARY

[0008] In view of this, in order to solve the above problems, the present application provides an ultrasonic cleaning machine remote monitoring and diagnosis method based on the Internet of Things.

[0009] The object of the present application can be achieved by the following technical solutions: The present application provides an ultrasonic cleaning machine remote monitoring and diagnosis method based on the Internet of Things, which comprises: matching a cleaning mode according to the dirt information of the object to be cleaned, and calling the cleaning configuration under the cleaning mode, including the total time length of each cleaning stage, cleaning standards and cleaning operation indicators.

[0010] The cleaning mode is executed, and the bubble parameters, water body parameters and dirt state parameters on the surface of the object to be cleaned are collected in real time.

[0011] The bubble parameters and water body parameters are analyzed in coordination, and the dirt state verification is started, if the coordinated analysis triggers an abnormal signal or the dirt state verification does not pass the cleaning standard, the current stage is marked as an abnormal stage, the abnormal indicators and abnormal degree are identified, otherwise the cleaning continues.

[0012] If the abnormal stage is not completed or is not the final stage, the cleaning strategy is dynamically adjusted based on the abnormal indicators and abnormal degree, the cleaning adjustment is marked and the strategy is recorded.

[0013] After completing all stages of cleaning, the cleaning efficiency diagnosis report is generated based on the abnormal stage position and the abnormal degree, and is uploaded to the cloud through the Internet of Things gateway.

[0014] Compared with the prior art, the present application has the following advantages: (1) The present application collects bubble parameters, water body parameters and dirt state parameters on the surface of the object to be cleaned, effectively solving the problem of ignoring the monitoring of the cavitation effect directly, fully considering the influence of key indicators reflecting the actual cleaning power, and reducing the operation misjudgment rate.

[0015] (2) The present application matches the cleaning stage and analyzes the collected parameters in coordination, establishes an independent stage verification mechanism, and solves the problem that a single continuous process cannot locate the failure link, which is convenient for displaying the differentiated cleaning detail state of different stages, and also convenient for precise marking of the abnormal stage and subsequent cleaning adjustment.

[0016] (3) The present application marks the abnormal stage through the coordinated analysis of bubble parameters and water body parameters and the cross verification with dirt state parameters, solves the evaluation deviation problem caused by the lack of monitoring of the state of the object to be cleaned, forms a multi-dimensional cleaning efficiency evaluation system, and also significantly improves the subsequent cleaning effect and clearly defines the maintenance direction of the subsequent cleaning machine.

[0017] (4) The application reduces false alarm rate caused by environmental interference and load fluctuation as much as possible through dynamic tolerance boundary correction and abnormality degree grading mechanism, and realizes accurate identification and abnormality degree quantization grading of abnormal indicators.

[0018] (5) Based on abnormal stage position, residual time length, abnormal indicator and abnormal degree matching strategy library, the cleaning parameters are dynamically adjusted, the lag problem of traditional method adjustment is solved, and adaptive optimization and accurate energy consumption control of cleaning strategy are realized. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0020] Figure 1 It is the overall implementation flowchart of the application.

[0021] Figure 2 It is the specific implementation step flowchart of step 3 of the application.

[0022] Figure 3 It is the specific implementation step flowchart of step 4 of the application.

[0023] Figure 4 It is the specific implementation step flowchart of the application.

[0024] Figure 5 It is the specific implementation step flowchart of the application. DETAILED DESCRIPTION

[0025] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only some embodiments of the application, not all embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the application.

[0026] Please refer to Figure 1 As shown in the drawings, the application provides an ultrasonic cleaning machine remote monitoring and diagnosis method based on Internet of Things, which comprises the following steps: step 1, matching a cleaning mode according to the dirt information of the object to be cleaned, and calling the cleaning configuration under the cleaning mode, including the total time length, cleaning standard and cleaning operation index of each cleaning stage.

[0027] Specifically, the dirt information includes dirt type, adhesion strength, coverage area ratio and chemical properties of pollutants.

[0028] The cleaning mode includes ordinary cleaning, medium-intensity cleaning and high-intensity cleaning, and different modes correspond to preset cleaning stage division, cleaning standard, cleaning operation index and time length ratio.

[0029] It should be added that the cleaning setting index includes but is not limited to cleaning time length, cleaning agent configuration, ultrasonic frequency and cleaning water temperature.

[0030] It should be added that the cleaning mode matching is based on a preset mode database, and the database contains special modes divided according to dirt and material combination, such as metal piece heavy oil dirt mode, plastic piece dust mode and glass piece chemical residue mode, etc.

[0031] The specific matching operation can be performed by comparing the collected dirt information with the characteristic parameters of each mode, such as oil dirt corresponding to 28 kHz ultrasonic frequency and 60℃ water temperature characteristics, dust corresponding to 40 kHz ultrasonic frequency and normal temperature characteristics comparison, selecting the mode with a matching degree greater than or equal to a preset matching degree threshold. If the matching degree is insufficient, the parameter influence priority is determined according to the pre-weighted distribution rule of dirt type, degree, material, distribution, the basic parameter library containing ultrasonic frequency, power, water temperature, etc. is called, the recommended value of each parameter is calculated by linear weighted summation algorithm, and the parameter combination rationality is verified according to the preset coordination rule. If not, fix the highest two parameters and iterate and optimize other parameters or refer to historical data to correct, and finally generate a temporary cleaning mode containing parameter sequence of each stage.

[0032] The weight distribution rule is set based on the influence degree of each parameter on the cleaning effect, for example, the dirt type directly determines the highest weight of the cleaning mechanism. The dirt degree affects the weight of the cleaning intensity and time length. The material is associated with the safety threshold weight of the cleaning parameter. The dirt distribution has a certain influence on the local cleaning strategy but the overall correlation is weak, so the weight is the lowest, and the distribution is verified by a plurality of experimental data, which can guarantee the effectiveness and adaptability of the parameter combination.

[0033] The matching degree threshold can be obtained by statistical analysis of a large number of cleaning experimental data of different dirt degrees and material combinations, for example, when the matching degree is greater than or equal to 85%, the pass rate after cleaning is stable at more than 92%. If the matching degree decreases to 80%, the pass rate drops to 78%, and 85% can be selected as the matching degree threshold to represent the critical point of balancing the cleaning effect and mode adaptability.

[0034] Step 2, execute the cleaning mode, and real-time collect the bubble parameters, water body parameters and dirt state parameters of the surface of the cleaned object in the current cleaning stage.

[0035] The specific acquisition process of step 2 is as follows: the bubble generation rate, bubble collapse intensity and bubble distribution uniformity are acquired in real time by a high-frequency ultrasonic Doppler sensor array arranged at multiple positions on the inner wall of the cleaning tank, and are integrated into bubble parameters.

[0036] The water body turbidity, pollutant particle concentration and COD value are monitored by a sensor array integrated on the tank bottom, and are integrated into water body parameters.

[0037] The surface image of the object to be cleaned is acquired by a multi-angle optical imaging system, and the dirt residual area ratio, adhesion thickness and distribution density are acquired based on image analysis, and are integrated into dirt state parameters.

[0038] Understandably, the quantification method of bubble distribution uniformity is as follows: the cleaning tank is divided into several equal-area regions, the bubble density of each region is acquired, the variation coefficient is obtained by calculating the ratio of the corresponding standard deviation value to the mean value of the density data, the result obtained by subtracting the variation coefficient from 1 is taken as the uniformity index, and the bubble distribution uniformity is obtained by min-max normalization to the interval of 0-1.

[0039] Understandably, the distribution density can be calculated by using image segmentation and feature extraction algorithm, the total area of residual dirt is calculated and compared with the surface area of the object, the actual residual dirt area ratio is obtained, the surface image of the object to be cleaned is divided into several standard unit area regions, the independent dirt spots in each region are marked based on image gray threshold segmentation and morphological processing, the number of dirt spots in the unit area region is counted, and the spatial distribution density value is generated.

[0040] The embodiment of the present application effectively solves the problem of ignoring the monitoring of the direct carrier of cavitation effect by acquiring bubble parameters, water body parameters and dirt state parameters of the object to be cleaned, fully considers the influence of key indicators reflecting the actual cleaning power, and reduces the operation misjudgment rate.

[0041] Please refer to Figure 2 The bubble parameters and water body parameters are analyzed cooperatively, and the dirt state verification is started, if the cooperative analysis triggers an abnormal signal or the dirt state verification does not pass the cleaning standard, the current stage is marked as an abnormal stage, the abnormal index and the abnormal degree are identified, otherwise the cleaning is continued.

[0042] Specifically, please refer to Figure 4 The specific analysis process of the cooperative analysis includes: A1, the bubble parameter threshold interval and the water body parameter reference value of the current stage are called.

[0043] A2, based on the total time length of the stage, a continuous time window is divided, the relative deviation rate of the bubble parameter acquisition value and the median value in the threshold interval, and the deviation degree of the water body parameter acquisition value and the reference value are calculated, and the deviation degree is quantified by percentage deviation.

[0044] The response exception signal is triggered when any of the following conditions is met: the relative deviation rate of any bubble parameter exceeds the dynamic tolerance boundary.

[0045] The deviation degree of any water body parameter continuously increases for at least two time windows.

[0046] A3, otherwise, if the proportion of parameters located in the preset critical range exceeds the threshold, the part of parameters is scaled and normalized by min-max standardization.

[0047] A4, the normalized values are weighted and fused based on the preset weight to generate a comprehensive abnormal trend index, and if the index exceeds the threshold, an abnormal signal is triggered, otherwise not.

[0048] It should be noted that the preset weight is set according to the influence degree of the water body parameter on the cleaning effect, which can be an empirical value or manually imported by artificial numerical value.

[0049] Further, the specific setting process of the dynamic tolerance boundary is as follows: according to the external environment interference intensity and the cleaning standard matching tolerance baseline table, the initial tolerance baseline is output.

[0050] The load characteristics of the object to be cleaned are obtained by a plurality of source sensors, and when the load characteristics exceed the preset reference range, the initial tolerance baseline is dynamically corrected according to the load exceeding proportion to output a corrected tolerance boundary.

[0051] According to the difference proportion between the environmental interference intensity and the disturbance threshold value, the tolerance boundary expansion proportion is obtained by matching the preset expansion rule, and the tolerance boundary expansion instruction is triggered.

[0052] After the expansion instruction is triggered, the upper limit and the lower limit of the current corrected tolerance boundary are widened according to the corresponding expansion proportion, and are automatically restored to the initial tolerance baseline after the interference intensity falls within the disturbance threshold value.

[0053] It should be noted that the correction of the corrected tolerance boundary is as follows: the initial tolerance baseline is expanded in the same direction according to the load exceeding proportion, and the expansion proportion is the load exceeding proportion. For example, if the load exceeding proportion is 20%, the upper limit and the lower limit of the tolerance baseline are widened by 20% respectively, thereby generating a corrected tolerance boundary.

[0054] The specific collection process of the external environmental interference intensity is as follows: the temperature and humidity data and vibration data are collected in real time by the temperature and humidity sensor and the vibration sensor deployed outside the cleaning machine.

[0055] The sliding average filter is used to eliminate transient fluctuations to obtain smoothed real-time temperature and real-time humidity, and the window can be set to 5 seconds.

[0056] The vibration acceleration time domain signal is converted into a frequency domain signal by Fourier transform, and the vibration amplitude effective value in the 10Hz-1000Hz frequency band is extracted, that is, high-frequency noise is excluded.

[0057] The temperature reference interval of the cleaning machine is matched based on the current stage, and the humidity and vibration amplitude reference interval is called.

[0058] Based on the temperature, humidity and vibration amplitude reference range, the vibration amplitude in the real-time temperature, humidity and vibration data is compared with the reference range respectively.

[0059] The temperature, humidity and vibration amplitude are marked as each environmental parameter. If a certain environmental parameter is below the reference range or within the reference range, the interference degree of the corresponding environmental parameter is 0, otherwise the ratio of the exceeding part to the reference interval width is taken as the interference degree of the corresponding environmental parameter and the maximum value is limited to 1.

[0060] According to the cleaning standard of the current stage, the interference degree value of each environmental parameter is weighted and summed to obtain the external environment interference intensity.

[0061] For example, the rated temperature reference interval is 15℃-30℃ in the initial cleaning stage, and the rated temperature reference interval is 20℃-25℃ in the precision cleaning stage. The humidity reference interval can be set to 40%-60% according to the experience value. The vibration amplitude reference range is based on the inherent vibration level of the cleaning machine without external interference. The vibration acceleration effective value in the initial stable state is collected by the vibration sensor, and the acceleration effective value is set as the reference value. The allowed vibration amplitude reference interval is determined with reference to the reference value, which is between 80% of the reference value and 120% of the reference value.

[0062] In one embodiment, vibration has a greater impact on cleaning accuracy. For the precision cleaning stage, the vibration amplitude weight can be set to 0.6, the temperature weight to 0.2, and the humidity weight to 0.2. For the initial cleaning stage, the temperature weight and humidity weight are set to 0.3 respectively, and the vibration amplitude weight is set to 0.4.

[0063] The load characteristics are obtained by combining the total mass, surface area, material hardness and dirt coverage density of the objects to be cleaned obtained by the pressure sensor at the bottom of the cleaning tank and the image recognition device at the inlet.

[0064] Understandably, the specific settings of the preset expansion rule are as follows: when the difference ratio is 10%-30%, generate an expansion ratio of 10%; when the difference ratio is 30%-50%, generate an expansion ratio of 20%; when the difference ratio exceeds 50%, generate an expansion ratio of 30%.

[0065] It needs to be explained that when the difference ratio of the environmental interference intensity exceeding the preset disturbance threshold is 10%-30%, the interference has a slight effect on the cleaning parameters, and the extension ratio of 10% can avoid the decrease of monitoring accuracy caused by excessive relaxation of the tolerance boundary while coping with the interference. When the difference ratio is 30%-50%, the interference effect is moderately enhanced, and the extension ratio of 20% can adapt to the increase of parameter fluctuation amplitude, and can balance the fault tolerance demand and the abnormal recognition sensitivity. When the difference ratio exceeds 50%, the interference may significantly affect the parameter stability, and the extension ratio of 30% can moderately widen the tolerance boundary, which can avoid the temporary fluctuation of the parameters caused by strong interference from being misjudged as an abnormality, and will not cause the tolerance boundary to be excessively relaxed, and can ensure that the substantial deviation of the bubble parameters and the water body parameters can be captured.

[0066] Specifically, please refer to Figure 5 As shown in the figure, the specific process of the dirt state verification includes: B1, based on the current stage cleaning standard, calling each dirt state parameter threshold.

[0067] B2, comparing each dirt state parameter collected at present with its threshold one by one, if there is a dirt state parameter exceeding the threshold, the verification is failed.

[0068] B3, if all parameters do not exceed the threshold and the remaining cleaning time is less than or equal to the preset critical end time, it is determined that the verification is passed.

[0069] B4, if the remaining cleaning time is greater than the critical end time, the following sub-steps are executed:

[0070] B5, based on the cleaning efficiency decay curve and the cleaning standard, setting the dirt state parameter threshold of each window in the current stage.

[0071] B6, based on the time matching of the completed windows, the collected dirt state parameters are attributed to the corresponding window according to time.

[0072] B7, calculating the difference between the mean value of the dirt state parameters in each completed window and the threshold of the window.

[0073] B8, if the change rate of the difference of the dirt state parameters in the continuous two windows exceeds the preset value, the verification is failed, otherwise it is determined that the dirt state verification is passed.

[0074] Understandably, the critical end time is set according to the total cleaning time, for example, if the total cleaning time is 1 hour, the critical end time can be set to 10 minutes.

[0075] Understandably, the cleaning standard consists of a threshold for the proportion of residual dirt area, a threshold for adhesion thickness, and a threshold for distribution density. The threshold for the proportion of residual dirt area refers to the ratio of the total area of ​​residual dirt on the surface of the object to the total surface area of ​​the object. The threshold for adhesion thickness refers to the average thickness threshold of the residual dirt adhesion layer. The threshold for distribution density refers to the threshold for the number of residual dirt particles per unit area.

[0076] For example, the threshold for the percentage of residual dirt area in the initial cleaning stage is set to 30%, the threshold for the percentage of residual dirt area in the precision cleaning stage is set to 10%, the threshold for the percentage of residual dirt area in the rinsing stage is set to 2%, and the threshold for the adhesion thickness in the initial cleaning stage is set to 50. The adhesion thickness threshold for the precision cleaning stage is set to 20. The adhesion thickness threshold during the rinsing stage is set to 5. For example, the distribution density threshold for the initial cleaning stage is set to 5 units / The distribution density threshold for the precision cleaning stage is set to 2 units / The distribution density threshold during the rinsing stage is set to 0.5 cells / day. .

[0077] Understandably, the cleaning efficiency decay curve is generated by fitting historical data, and the curve type includes one of linear decay, exponential decay, or piecewise step decay. The specific curve type is determined based on the dirt information of the object to be cleaned.

[0078] For example, an exponential decay curve is used for grease-based dirt, where the decay rate is determined by the decay coefficient. For instance, if the initial stripping rate is 80% per minute, the rate decreases with energy decay... Regular decline, Indicates time, As a natural constant, a segmented step curve is used for oxidized layer contaminants. The threshold of each segment is based on a preset critical energy threshold. For example, when the energy is greater than or equal to 80% of the rated power, it is a rapid stripping segment, and 60%-80% is a slow stripping segment. Each segment corresponds to a fixed reduction rate. A linear decay curve is used for particulate contaminants, and the reduction rate is consistent.

[0079] In one specific embodiment, an example execution process for step B5 is as follows: Assuming the current stage is the precision cleaning stage, the threshold values ​​for the percentage of residual dirt area, the adhesion thickness, and the distribution density are 10%, 20%, and 20%, respectively. and 2 / and as the last time window threshold, and the dirt type marked in the dirt information of the object to be cleaned is grease, the adjacent window drop is limited to not more than 20%. At the same time, the total length of the precision cleaning stage is 20 minutes, the number of time windows is 4, i.e. 0-5 minutes, 5-10 minutes, 10-15 minutes, and 15-20 minutes, and the theoretical dirt state parameter threshold of each window is calculated through the associated curve function corresponding to the exponential decay curve, and the function formula is: , is the dirt state parameter threshold at the mth moment, is the initial dirt state parameter threshold, is the decay coefficient, can be 0.05.

[0080] The threshold values of the four windows can be obtained by inversely deducing the previous window threshold based on the last time window threshold, and combining the constraint that the adjacent window drop is limited to not more than 20% as follows: the residual area ratio threshold of window 1 is 25%, the attached thickness threshold is 40 , the distribution density threshold is 4 , the residual area ratio threshold of window 2 is 20%, the attached thickness threshold is 30 , the distribution density threshold is 3 , the residual area ratio of window 3 is 15%, the attached thickness is 25 , the distribution density is 2.5 , and the residual area ratio threshold of window 4 is 10%, the attached thickness threshold is 20 , and the distribution density threshold is 2 .

[0081] Understandably, based on the current complete completion window of the performed duration matching, such as in the case of known time window division interval, such as 1 window per 5 minutes, the end time of each window is taken as the judgment basis, that is, the performed duration is compared with the end time of each window in turn, and the window whose end time is less than or equal to the performed duration is the complete completion window.

[0082] Specifically, the specific identification process of the abnormal index includes: if the response abnormal signal is triggered, the bubble parameter whose relative deviation rate exceeds the dynamic tolerance boundary is marked as an abnormal bubble parameter, the water body parameter whose deviation degree continuously increases is marked as an abnormal water body parameter, and the abnormal bubble parameter and the abnormal bubble parameter are integrated into an abnormal index.

[0083] If the response abnormal signal is not triggered, the cleaning operation index is marked as an abnormal index.

[0084] ​It should be noted that the bubble parameters and the water body parameters are core process parameters affecting the cleaning effect, the cavitation of the bubbles directly determines the dirt peeling efficiency, the water body parameter state directly reflects the dissolution of the dirt, and the abnormality of the two will directly cause the dirt state parameter to be out of standard, and in the cleaning process, the abnormality of the bubble parameters and the water body parameters often occurs before the abnormality of the dirt state parameter, so it is not necessary to take the dirt state parameter on the surface of the object to be cleaned into the abnormal index identification, at this time, it is necessary to consider whether the cleaning operation setting is abnormal. And under normal circumstances, the basic abnormality also tends to be that the cleaning setting is not reasonable, such as the cleaning temperature or the cleaning time length not reaching the dirt cleaning requirement of the object to be cleaned.

[0085] The embodiment of the application reduces the false alarm rate caused by environmental interference and load fluctuation as much as possible through the dynamic tolerance boundary correction and the abnormality degree grading mechanism, and realizes accurate identification of the abnormal index and quantitative grading of the abnormality degree.

[0086] In another specific embodiment, the specific identification process of the abnormality degree includes: if the response abnormal signal is not triggered, outputting a default initial abnormality degree value.

[0087] If the response abnormal signal is triggered, the abnormal index set is generated after the normalization of the abnormal index.

[0088] If the abnormal index set contains only single-class parameters, the abnormal degree is obtained by matching the preset abnormality degree rule table according to the maximum abnormal index.

[0089] If multiple parameters are contained, when the abnormal degrees corresponding to the maximum abnormal indexes of different parameters belong to different levels by one level, the abnormal degree corresponding to the parameter with a higher preset influence weight is selected, and when the difference is greater than or equal to two levels, the highest matching abnormal degree is selected.

[0090] Understandably, the initial abnormality degree value can be based on the statistical law of the untriggered response abnormal signal scene in the historical cleaning data, by analyzing a large number of similar cleaning objects under this scene, combining expert experience calibration, the initial value is uniformly set to a mild abnormality or a corresponding quantitative value, such as 0.3.

[0091] It should be noted that the preset abnormality degree rule table can determine the corresponding relationship between different abnormal indexes and the cleaning effect decay rate through experiments, such as index 0.2 corresponding to 10% decline in efficiency, 0.5 corresponding to 30% decline, combined with the grading definition of the abnormal degree in the industry standard, the mapping relationship between the index interval and the degree level is determined after machine learning training of historical abnormal cases, and finally an abnormal degree rule table covering various parameters is formed.

[0092] Understandably, the default initial value when the trigger signal is not triggered can quickly establish a reference, and after the trigger signal, the abnormal index is normalized to unify the order of magnitude, the single-class parameter focuses on the most significant abnormality, and the multi-class parameter combines the level difference and the influence weight of different parameters to balance the influence weight, which avoids misjudgment of a single parameter and prevents secondary abnormalities from covering core problems, and finally realizes the precise matching of the abnormal degree and the actual cleaning effect, and provides a reliable basis for subsequent adjustment strategies.

[0093] Understandably, the preset influence weight is set based on the core influence mechanism of bubble parameters and water body parameters on cleaning efficiency and experimental data verification: wherein the bubble parameters are directly related to the cavitation effect of ultrasonic cleaning, and the abnormality will lead to insufficient cavitation intensity, and the direct influence weight on cleaning effect is 60%. The water body parameter mainly reflects the state of dirt dissolution and discharge, which is an indirect representation of cleaning effect, and is significantly affected by bubble parameters, and the influence weight can be set to 40%. And this value can be slightly dynamically adjusted according to the actual scene.

[0094] The embodiment of the application solves the evaluation deviation problem caused by the lack of state monitoring of the cleaned object by co-analyzing the bubble parameters and the water body parameters and cross-verifying with the dirt state parameters to mark the abnormal stage, forms a multi-dimensional cleaning efficiency evaluation system, and can also significantly improve the subsequent cleaning effect and clearly determine the subsequent cleaning machine maintenance direction.

[0095] Please refer to Figure 3 As shown in the figure, step 4, if the cleaning is not completed in the abnormal stage or it is not the final stage, the cleaning strategy is dynamically adjusted based on the abnormal index and the abnormal degree, the cleaning adjustment is marked, and the strategy is recorded.

[0096] Specifically, the specific generation of the dynamically adjusted cleaning strategy includes: C1, when the cleaning is not completed in the abnormal stage, the preliminary adjustment strategy is generated based on the abnormal stage position, the remaining time, the abnormal index and the abnormal degree.

[0097] It should be noted that the strategy library is constructed by collecting a large number of historical cleaning cases, the cases cover different combinations of abnormal stage position, remaining time, abnormal index and degree, and is combined with the corresponding relationship between the adjustment measures and the cleaning effect improvement rate determined by experiments and the adjustment after expert experience calibration and machine learning training, and the content includes specific adjustment schemes in each scene: such as slight bubble abnormality corresponding to 5% increase in ultrasonic power and 2s extension of bubble period in the initial stage of abnormal stage, sufficient time, switching pulse water flow and supplementing cleaning agent, i.e. 8% increase in concentration, in the middle stage of abnormal stage, and the like.

[0098] At the same time, the adjustment parameter threshold is determined, such as the power not exceeding the upper limit of 30%, to form a structured strategy library covering all scenarios.

[0099] C2, if the bubble parameter, water parameter and dirt state parameter are all up to standard for two consecutive times, the strategy is adopted, otherwise the next level strategy of the same abnormal index is called.

[0100] It should be noted that the next abnormal level strategy refers to the strategy of the next level when the abnormal level corresponds to the next level, for example, if the current matching strategy corresponds to M level, the M+1 level strategy is called, and if the highest abnormal level is reached, the maximum safety threshold strategy of the level is called.

[0101] C3, when the abnormal stage is completed and is not the final stage, if there is a historical cleaning adjustment mark, the change rate of the dirt state parameter before and after adjustment is compared.

[0102] C4, if the change rate does not meet the expectation, it is determined whether the adjustment direction is wrong or the adjustment amplitude is insufficient.

[0103] Understandably, the determination of whether the adjustment direction is wrong or the adjustment amplitude is insufficient includes: calculating the deviation value of the dirt state parameter change rate and the expected change rate, if the deviation value is greater than the preset threshold, it is determined that the adjustment direction is wrong, otherwise it is determined that the adjustment amplitude is insufficient.

[0104] C5, if the adjustment direction is wrong, switch the dimension and match the new strategy by downshift, if the adjustment amplitude is insufficient, superimpose the same dimension adjustment strategy of the previous mark cleaning adjustment by safety ratio, and the total amplitude does not exceed the parameter safety threshold.

[0105] It should be noted that the switching of the adjustment dimension includes: selecting a control dimension different from the original adjustment dimension from the bubble parameter, water parameter or cleaning operation index, and the safety ratio is a configurable value of 50%-80%.

[0106] Understandably, switching the adjustment dimension means switching from the original adjustment dimension such as temperature, ultrasonic power, etc. to a dimension with different action mechanism, avoiding repeated invalid intervention, and downshift matching the new strategy: selecting the adjustment amplitude by reducing one level according to the current abnormal level, such as adapting the medium level cleaning strategy corresponding to the severe level, thereby reducing the secondary risk.

[0107] Illustratively, cleaning a semiconductor wafer, the presequence dirt residual area proportion is 22%, the dirt residual area proportion threshold is 20%, the water temperature is adjusted from 40°C to 45°C, and after 3 minutes, the dirt residual area proportion rises to 25%, at which time the adjustment direction is wrong.

[0108] At this time, the conversion dimension can be switched from temperature adjustment to bubble parameter adjustment, and the mechanism changes from thermal dissolution to cavitation stripping. At the same time, the proportion of the residual area of the dirt reaches the threshold of the proportion of the residual area of the dirt marked in the heavy abnormality level, and the downshift adopts the moderate strategy, that is, the ultrasonic power is increased from 300 W to 350 W, instead of 400 W of the heavy strategy, and the bubble density is increased from 50 to 60, instead of 70. And the residual rate is reduced to 18% after 3 minutes, without causing damage to the wafer surface due to aggressive adjustment.

[0109] For another example, when the adjustment amplitude is insufficient, taking cleaning mechanical parts as an example, the same dimension adjustment strategy of the previous marked cleaning adjustment is used to adjust the cleaning agent concentration from 4% to 5%. After 5 minutes, the residual rate of the dirt is reduced from 15% to 12%, and it is expected to be reduced to 10%.

[0110] Continue to adjust the cleaning agent concentration, and the superimposed amplitude is 100% of the previous one, that is, increase by 1% again, and the concentration is increased to 6%. The safety threshold of the cleaning agent is 7%, and the total amplitude is 1%, which meets the constraint requirement. After 5 minutes, the residual rate is reduced to 9%, which meets the standard at this time, and the corrosion risk caused by too high concentration is not triggered.

[0111] C6, if there is no historical mark or it is the first stage, the total time of the next stage is used as the remaining time to match the strategy library to generate a preliminary adjustment strategy.

[0112] C7, if the bubble parameters, water parameters and dirt state parameters of the two consecutive time windows are all up to standard after execution, the default strategy of the stage is restored.

[0113] It should be noted that after the default strategy is restored, if the same abnormal index is detected again subsequently, the preliminary adjustment strategy is skipped, and the last effective final adjustment strategy is directly executed.

[0114] The embodiment of the application dynamically adjusts the cleaning parameters based on the abnormal stage position, the remaining time, the abnormal index and the abnormal degree to match the strategy library, solves the lag problem of the traditional method adjustment, and realizes adaptive optimization and precise energy consumption control of the cleaning strategy.

[0115] Step 5, after completing all stages of cleaning, a cleaning efficiency diagnosis report is generated based on the abnormal stage position and the abnormal degree, and is uploaded to the cloud through an Internet of Things gateway.

[0116] Specifically, the generation process of the cleaning efficiency diagnosis report comprises: matching a preset stage influence weight according to the abnormal stage position.

[0117] The influence weight of all abnormal stages is corrected according to the abnormal degree of the corresponding abnormal stage to obtain a final abnormal degree.

[0118] ​​The ratio of the number of abnormal stages to the total number of cleaning stages is calculated, and the cleaning efficiency index is obtained by combining the ratio and the reciprocal of the abnormality degree.

[0119] A cleaning efficiency diagnosis report is outputted, which contains the position of the abnormal stage, the final abnormality degree, and the cleaning efficiency index.

[0120] It should be noted that the stage influence weight is set based on the importance of each cleaning stage in the overall cleaning process and the contribution to the final cleaning effect, and the correction factor is the sum of the preset stage influence weight multiplied by the abnormality degree of the corresponding abnormal stage.

[0121] It should be further noted that the cleaning efficiency index is calculated using a linear weighted summation algorithm, and the weighted summation method is used to integrate the abnormality ratio and the reciprocal of the abnormality degree. The abnormality ratio reflects the range of abnormality, and the reciprocal of the abnormality degree reflects the severity of the abnormality. For example, the influence of a single severe abnormality on the efficiency is greater than that of multiple slight abnormalities, so the weight of the abnormality degree is set to be higher than that of the abnormality ratio. For example, the weights of the abnormality ratio and the reciprocal of the abnormality degree can be 0.4 and 0.6, respectively.

[0122] The above is only an example and description of the concept of the present application. Those skilled in the art can make various modifications or supplements to the described specific embodiments or use similar ways to replace them, as long as they do not deviate from the concept of the present application or exceed the scope defined by the present application, and they should belong to the protection scope of the present application.

Claims

1. A remote monitoring and diagnosis method for an ultrasonic cleaning machine based on the Internet of Things, characterized in that, The method comprises: According to the dirt information of the object to be cleaned, the cleaning mode is matched, and the cleaning configuration under the cleaning mode is called, including the total time of each cleaning stage, the cleaning standard and the cleaning operation index; The cleaning mode is executed, and the bubble parameters, water body parameters and dirt state parameters of the surface of the object to be cleaned in the current cleaning stage are collected in real time; The bubble parameters and water body parameters are analyzed cooperatively, and the dirt state verification is started. If the cooperative analysis triggers an abnormal signal or the dirt state verification does not pass the cleaning standard, the current stage is marked as an abnormal stage, the abnormal index and the abnormal degree are identified, otherwise the cleaning continues; If the abnormal stage is not completed or is not the final stage, the cleaning strategy is dynamically adjusted based on the abnormal index and the abnormal degree, the cleaning adjustment is marked and the strategy is recorded; After completing all stages of cleaning, a cleaning efficiency diagnosis report is generated based on the abnormal stage position and the abnormal degree, and is uploaded to the cloud through the Internet of Things gateway; The specific analysis process of the cooperative analysis comprises: The bubble parameter threshold interval and the water body parameter reference value of the current stage are called; Based on the total time of the stage, a continuous time window is divided, the relative deviation rate of the bubble parameter collection value and the median value of the threshold interval, and the deviation degree of the water body parameter collection value and the reference value are calculated. The deviation degree is quantified by percentage deviation; When any one of the following conditions is met, an abnormal signal is triggered: The relative deviation rate of any bubble parameter exceeds the dynamic tolerance boundary; The deviation degree of any water body parameter continuously increases for at least two time windows; Otherwise, if the proportion of parameters located in the preset critical range exceeds the threshold value, the parameters are normalized; The normalized values are weighted and fused based on the preset weight to generate a comprehensive abnormal trend index. If the index exceeds the threshold value, an abnormal signal is triggered, otherwise no signal is triggered; The specific setting process of the dynamic tolerance boundary is as follows: According to the external environmental disturbance intensity and the cleaning standard, a tolerance baseline table is matched, and an initial tolerance baseline is output; The load characteristics of the object to be cleaned are obtained through multiple source sensors. When the load characteristics exceed the preset reference range, the initial tolerance baseline is dynamically corrected according to the load exceeding proportion, and a corrected tolerance boundary is output; According to the difference proportion between the environmental disturbance intensity and the disturbance threshold value, the tolerance boundary expansion proportion is obtained by matching the preset expansion rule, and a tolerance boundary expansion instruction is triggered; After the expansion instruction is triggered, the upper limit and the lower limit of the current corrected tolerance boundary are widened by the corresponding expansion proportion, and are automatically restored to the initial tolerance baseline after the disturbance intensity falls within the disturbance threshold value.

2. The remote monitoring and diagnosis method for the Internet of Things based ultrasonic cleaning machine according to claim 1, characterized in that: The dirt information includes dirt type, adhesion strength, coverage area proportion and pollutant chemical properties; The cleaning mode includes ordinary cleaning, medium intensity cleaning and high intensity cleaning, and different modes correspond to preset cleaning stage division, cleaning standard, cleaning operation index and time ratio. 3.The IoT-based remote monitoring and diagnosing method of an ultrasonic cleaning machine according to claim 1, wherein: The specific collection process is as follows: The bubble generation rate, bubble collapse intensity and bubble distribution uniformity are obtained in real time by a high-frequency ultrasonic Doppler sensor array arranged at multiple positions on the inner wall of the cleaning tank, and are integrated into bubble parameters; The water body turbidity, pollutant particle concentration and COD value are monitored by a tank bottom integrated sensor array, and are integrated into water body parameters; The dirt state parameters are obtained by multi-angle image acquisition of the surface of the object to be cleaned by an optical imaging system and image analysis.

4. The remote monitoring and diagnosis method for the Internet of Things based ultrasonic cleaning machine according to claim 1, characterized in that: The specific process of the dirt state verification includes: thresholds of the dirt state parameters are retrieved based on the current stage cleaning standard; each dirt state parameter currently collected is compared with its threshold one by one, and if there is a dirt state parameter exceeding the threshold, the verification fails; if all parameters do not exceed the threshold and the remaining cleaning time is less than or equal to the preset critical end time, it is determined that the verification passes; if the remaining cleaning time is greater than the critical end time, the following sub-steps are performed: thresholds of the dirt state parameters of the current stage are set based on the cleaning efficiency decay curve and the cleaning standard; the dirt state parameters collected are attributed to the corresponding window according to time based on the time already spent; the difference between the mean value of the dirt state parameters in each completed window and the threshold of the window is calculated; if the change rate of the difference of the dirt state parameters in two consecutive windows exceeds the preset value, the verification fails, otherwise it is determined that the dirt state verification passes. 5.The IoT-based remote monitoring and diagnosing method of ultrasonic cleaning machine according to claim 1, wherein: The specific identification process of the abnormality index includes: if the response abnormality signal is triggered, the bubble parameter with a relative deviation rate exceeding the dynamic tolerance boundary is marked as an abnormal bubble parameter, and the water body parameter with a continuously increasing deviation degree is marked as an abnormal water body parameter, and the abnormal bubble parameter and the abnormal bubble parameter are integrated into an abnormal index; if the response abnormality signal is not triggered, the cleaning operation index is marked as an abnormal index. 6.The IoT-based remote monitoring and diagnosing method of an ultrasonic cleaning machine according to claim 1, wherein: The specific identification process of the abnormality degree includes: if the response abnormality signal is not triggered, a default initial abnormality degree value is output; if the response abnormality signal is triggered, the abnormal index is normalized to generate an abnormal index set; if the abnormal index set contains only a single type of parameter, the abnormal degree is obtained by matching the preset abnormal degree rule table according to the maximum abnormal index; if it contains multiple types of parameters, when the maximum abnormal index of different types of parameters corresponds to an abnormal degree belonging to a grade difference of one level, the abnormal degree corresponding to the parameter with a higher preset impact weight is selected, and when the difference is greater than or equal to two levels, the highest matching abnormal degree is selected.

7. The IoT-based remote monitoring and diagnostic method for ultrasonic cleaning machines as claimed in claim 1, wherein: The specific generation of the dynamically adjusted cleaning strategy includes: when the abnormal stage is not completed, a preliminary adjustment strategy is generated based on the abnormal stage position, the remaining time, the abnormal index and the abnormal degree; if the bubble parameter, the water body parameter and the dirt state parameter all meet the standard continuously twice after execution, the strategy is adopted, otherwise the strategy of the next degree level of the abnormal index is called; when the abnormal stage is completed and it is not the final stage, if there is a historical cleaning adjustment mark, the change rate of the dirt state parameter before and after adjustment is compared; if the change rate does not meet the expectation, it is determined whether the adjustment direction is wrong or the adjustment amplitude is insufficient; if the adjustment direction is wrong, the dimension is switched and a new strategy is matched according to downshift, and if the adjustment amplitude is insufficient, the same dimension adjustment strategy of the previous mark cleaning adjustment is superimposed according to a safety ratio, and the total amplitude does not exceed the parameter safety threshold; if there is no historical mark or it is the first stage, the total time of the next stage is used as the remaining time to generate a preliminary adjustment strategy based on the strategy library. If the bubble parameter, the water body parameter and the dirt state parameter of two continuous time windows are up to standard after the execution, the default strategy of the stage is restored.

8. The IoT-based remote monitoring and diagnostic method for ultrasonic cleaning machines as claimed in claim 1, wherein: The generation process of the cleaning efficiency diagnosis report comprises: Matching a preset stage influence weight according to the abnormal stage position; Correcting the influence weight of all abnormal stages according to the abnormal degree of the corresponding abnormal stage to obtain a final abnormal degree; Counting the proportion of the number of abnormal stages and the total number of cleaning stages, and comprehensively obtaining a cleaning efficiency index according to the proportion and the reciprocal of the abnormal degree; Outputting a cleaning efficiency diagnosis report containing the abnormal stage position, the final abnormal degree and the cleaning efficiency index.

Images