A method for measuring a mud-water interface

Abstract

This invention provides a method for measuring the mud-water interface, belonging to the field of mud-water interface measurement technology. The method includes: acquiring a test echo waveform of the mud-water interface and determining the waveform characteristics of the test echo waveform; determining the confidence level of the test echo waveform based on the waveform characteristics; wherein the confidence level characterizes the reliability of the test echo waveform in predicting the mud level value, and the mud level value characterizes the distance from the bottom of the pool to the mud-water interface; determining at least one target model corresponding to the test echo waveform from multiple models based on the confidence level; wherein the multiple models include: a first model, a second model, and a third model; and outputting the mud level value corresponding to the test echo waveform based on the at least one target model. The measurement method provided by this invention improves the accuracy of mud level value prediction.

Description

Technical Field

[0001] This invention belongs to the field of mud-water interface measurement technology, and specifically relates to a method for measuring the mud-water interface. Background Technology

[0002] The sludge-water interface meter is an instrument widely used in wastewater treatment processes. It provides accurate measurement data to help operators optimize process parameters, such as adjusting chemical dosages, controlling pump speeds, and managing sludge scraper operations. Currently, sludge-water interface meters also need to adapt to the demands of remote monitoring and automated control. However, personnel often lack direct access to remote environments, making effective monitoring and adjustment of the process difficult. Summary of the Invention

[0003] In view of the above problems, embodiments of this application provide a method for measuring the mud-water interface in order to overcome or at least partially solve the above problems.

[0004] In a first aspect, this application provides a method for measuring the mud-water interface, the method comprising: The waveform of the echo at the mud-water interface is acquired, and the waveform characteristics of the echo are determined; wherein the waveform characteristics include at least: the peak position and the signal-to-noise ratio. Based on the waveform characteristics, the confidence level of the measured echo waveform is determined; wherein, the confidence level is used to characterize the reliability of the measured echo waveform in predicting the mud level value, and the mud level value is used to characterize the distance from the bottom of the pond to the mud-water interface. Based on the confidence level, at least one target model corresponding to the measured echo waveform is determined from multiple models; wherein, the multiple models include: a first model, a second model, and a third model; wherein, The first model is obtained based on the waveform features. The second model is trained based on the echo waveform and actual mud level value under known usage scenarios. The third model is trained based on the echo waveform and actual mud level value under unknown usage scenarios. The known usage scenarios include at least: radial flow sedimentation tank, high-efficiency sedimentation tank and horizontal flow sedimentation tank. The unknown usage scenarios are scenarios other than the known usage scenarios. Based on the at least one target model, output the mud level value corresponding to the measured echo waveform.

[0005] Further, determining the confidence level corresponding to the measured echo waveform based on the waveform features includes: Obtain the similarity between the waveform features and multiple waveforms of the first type and multiple waveforms of the second type, respectively; The confidence level is determined based on the similarity. The confidence level of the multiple waveforms of the first type is greater than that of the multiple waveforms of the second type, and the multiple waveforms are echo waveforms under the known usage scenario.

[0006] Further, determining the confidence level based on the similarity includes: If the similarity between the waveform feature and multiple waveforms of the first type exceeds a preset value, the first confidence level is determined as the confidence level. If the similarity between the waveform feature and multiple waveforms of the first type and multiple waveforms of the second type is lower than the preset value, the second confidence level is determined as the confidence level. If the similarity between the waveform feature and multiple waveforms of the second type exceeds the preset value, the third confidence level is determined as the confidence level. Wherein, the first confidence level is greater than the second confidence level, and the second confidence level is greater than the third confidence level.

[0007] Further, determining the confidence level corresponding to the measured echo waveform based on the waveform features includes: The waveform to be tested is input into the classification model to obtain the confidence level corresponding to the waveform to be tested; The classification model is trained on a dataset based on a variety of different echo waveforms and the corresponding labels of the echo waveforms. The labels are used to characterize whether the echo waveform can predict the mud level value.

[0008] Further, determining at least one target model corresponding to the measured echo waveform from multiple models based on the confidence level includes: When the confidence level is a first confidence level, the at least one target model is determined to be the prediction of the mud level value by the first model and the second model; When the confidence level is the second confidence level, the at least one target model is determined to be the first model and the third model predicting the mud level value; If the confidence level is the third confidence level, then it is determined to exit the prediction of the mud level value.

[0009] Further, when the target model is the first model and the second model, the step of outputting the mud level value corresponding to the measured echo waveform based on the at least one target model includes: Based on the first model, the first predicted mud level value of the echo waveform to be measured is determined; Based on the second model, the second predicted mud level value of the measured echo waveform is determined; Based on the first predicted mud level value and the second predicted mud level value, the mud level value is output.

[0010] Further, when the target model is the first model and the third model, the step of outputting the mud level value corresponding to the measured echo waveform based on the at least one target model includes: Based on the first model, the first predicted mud level value of the echo waveform to be measured is determined; Based on the third model, the third predicted mud level value of the measured echo waveform is determined; Based on the first predicted mud level value and the third predicted mud level value, the mud level value is output.

[0011] Furthermore, the first model is configured as follows: Based on the wave crest position, the travel time of the reflected wave from the mud surface is determined; wherein, the travel time is used to characterize the total time it takes for the ultrasonic signal to be emitted from the sensor, reach the mud-water interface, and be reflected back to the sensor; Based on the travel time and ultrasonic velocity, a first transition mud level value is determined; wherein, the first transition mud level value is used to characterize the distance from the sensor to the mud-water interface; Based on the first transition mud level value, the first predicted mud level value is output.

[0012] Furthermore, the second model is trained through the following steps: Acquire sampling data for the known usage scenario; wherein the sampling data includes echo waveforms and the actual mud level value; The training waveform features of the echo waveform are extracted from the sampled data, and the training waveform features are associated with the actual mud level value to construct a dataset; The second model is trained using the dataset as training samples.

[0013] Furthermore, after determining the mud level value based on the first predicted mud level value and the third predicted mud level value, the method further includes: The target use case for obtaining the third predicted mud level value is updated based on the measured echo waveform and the third predicted mud level value. The second model is updated based on the updated dataset.

[0014] According to the embodiment of this method, a method for measuring the mud-water interface includes: acquiring a test echo waveform of the mud-water interface and determining the waveform characteristics of the test echo waveform; wherein the waveform characteristics include at least: peak position and signal-to-noise ratio; determining the confidence level of the test echo waveform based on the waveform characteristics; wherein the confidence level is used to characterize the reliability of the test echo waveform in predicting the mud level value, and the mud level value is used to characterize the distance from the bottom of the pool to the mud-water interface; determining a target model corresponding to the test echo waveform from multiple models based on the confidence level; wherein the multiple models include: a first model, a second model, and a third model; wherein the first model is obtained based on the waveform characteristics, the second model is trained based on the echo waveform and the actual mud level value under known usage scenarios, and the third model is trained based on the echo waveform and the actual mud level value under unknown usage scenarios, wherein the known usage scenarios include at least: radial flow sedimentation tank, high-efficiency sedimentation tank, and horizontal flow sedimentation tank, and the unknown usage scenarios are scenarios other than the known usage scenarios; and outputting the mud level value corresponding to the test echo waveform based on the at least one target model.

[0015] Therefore, when the measurement method of this embodiment is applied to the measurement of the mud-water interface, the waveform characteristics of the echo waveform to be measured can be determined through the measured echo waveform at the mud-water interface. These waveform characteristics include key parameters such as peak position and signal-to-noise ratio. Then, based on the waveform characteristics, the confidence level of the echo waveform is determined. The confidence level reflects the reliability of the waveform data. If the confidence level indicates that the measured echo waveform is reliable, the mud level value can be further predicted; if the confidence level is low, we can choose to resample or take other measures to ensure the reliability of the acquired measured echo waveform.

[0016] When it is necessary to predict the mud level, we select at least one of the most suitable target models from the first, second and third models based on the confidence level. Through the collaborative work of at least one target model, we can accurately predict the mud level, so that the measurement results are no longer just a reference, but can be directly used for process control.

[0017] Furthermore, by integrating different target models to predict mud level values, not only is the accuracy of mud level prediction improved, but the validity of the measurement results is also enhanced. The final output mud level values ​​and confidence levels facilitate process control for operators. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram provided by related technologies, showing a clear mud layer and a high signal-to-noise ratio; Figure 2 This is another schematic diagram provided by related technologies, showing a clear mud layer and a high signal-to-noise ratio. Figure 3 This is a schematic diagram illustrating a mud layer with a clear structure but a low signal-to-noise ratio, provided in an embodiment of this application. Figure 4 This is another schematic diagram of a mud layer with a clear shape but a low signal-to-noise ratio provided in the embodiments of this application; Figure 5 This is a flowchart illustrating the steps of a method for measuring the mud-water interface provided in an embodiment of this application. Figure 6 This is a waveform diagram of a first type provided in an embodiment of this application; Figure 7 It is aimed at Figure 6 A second type of waveform diagram is provided; Figure 8 This is another waveform diagram of the first type provided in the embodiments of this application; Figure 9 It is aimed at Figure 8 A second type of waveform diagram is provided; Figure 10 This is a schematic diagram of a mud-water interface measurement system provided in an embodiment of this application; Figure 11 This is a schematic diagram of a system for measuring mud level provided in an embodiment of this application; Figure 12 This is a schematic diagram illustrating how a first model and a second model predict mud level values, as provided in an embodiment of this application. Figure 13 This is a schematic diagram illustrating how a first model and a second model predict mud level values, as provided in the embodiments of this application. Detailed Implementation

[0020] Exemplary embodiments of this application will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of this application are shown in the drawings, it should be understood that this application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of this application and to fully convey the scope of this application to those skilled in the art.

[0021] The relevant mud-water interface measurement technologies are mainly divided into ultrasonic sensors and optical sensors.

[0022] Optical sensors are mainly divided into two categories: beam scales and sludge concentration sensors. Their measurement principle is mainly based on the different transmission and reflection coefficients of light in liquids of different concentrations to identify the interface position. The advantage of optical sensors is high sensitivity. The disadvantages are: (1) the beam scale is easily contaminated in the sedimentation tank, affecting the measurement accuracy and the maintenance cost is high; (2) the sludge concentration method has a complex control system. Therefore, it is rarely used in practice.

[0023] The following will refer to Figures 1-4 Examples of different echo waveforms are shown, illustrating echo waveforms with high or low signal-to-noise ratios for mud layers. Figures 1-4 The horizontal axis represents time, and the vertical axis represents the signal-to-noise ratio. Figure 1 This is a schematic diagram provided by related technologies, showing a clear mud layer and a high signal-to-noise ratio. Figure 2 This is another schematic diagram provided by related technologies, showing a clear mud layer and a high signal-to-noise ratio. Figure 3 This is a schematic diagram illustrating a mud layer with a clear structure but a low signal-to-noise ratio, provided in an embodiment of this application. Figure 4 This is another schematic diagram of a mud layer with a clear shape but a low signal-to-noise ratio provided in the embodiments of this application.

[0024] Ultrasonic sensors are based on the principle of echo detection. While they are simple in structure and low in cost, they also have some problems in practical applications. Figure 1 and Figure 2 When the mud layer distribution and echo are relatively clear, the measurement results are accurate and reliable; however, for... Figure 3 and Figure 4 When the mud layer is clear but the echo is complex, it is difficult to accurately identify the mud position, and even identification errors may occur. When the mud layer distribution is unclear and the echo signal is cluttered, unpredictable measurement results may occur. Regardless of the accuracy of the measurement results, their validity is not indicated, and the control system cannot distinguish the validity of the measurement results, so they cannot be directly used for process control. Currently, most mud-water interface meters are used more as references and do not directly participate in control. In addition, the algorithm models of most products are fixed, which can also lead to incorrect measurement results in scenarios where the mud layer is clear but the echo is not suitable for existing algorithms.

[0025] Therefore, to solve the above problems, this embodiment provides a method for measuring the mud-water interface. When applied to the measurement of the mud-water interface, the mud level value of the measured echo waveform is predicted based on the confidence level of the waveform to be measured, combined with at least one target model formed by the fusion of the first model, the second model, and the third model. This solves the problem of inaccurate or incorrect mud level identification when the mud layer is clear and the measured echo waveform is complex.

[0026] Reference Figure 5 , Figure 5This is a flowchart illustrating the steps of a method for measuring the mud-water interface provided in an embodiment of this application. The flowchart includes: Step S501: Obtain the waveform of the echo to be measured at the mud-water interface, and determine the waveform characteristics of the waveform to be measured; wherein, the waveform characteristics include at least: the peak position and the signal-to-noise ratio.

[0027] In this embodiment, the measured echo waveform at the mud-water interface is first acquired. Figures 1 to 4 Any waveform can be used. Then, based on the information in the waveform diagram of the echo to be tested, the waveform characteristics of the echo to be tested are determined. These waveform characteristics include at least the peak position and the signal-to-noise ratio. In addition, the waveform characteristics may also include the number of peaks, peak height, peak width, baseline, excitation signal width, maximum amplitude, minimum amplitude, etc. The above waveform characteristics can be directly obtained by digital processing of the echo to be tested, which can solve the problems of signal reflection and distortion caused by analog signals after long-distance transmission, thereby improving the signal-to-noise ratio.

[0028] The peak position is a key feature of a signal waveform, representing the maximum amplitude reached by the signal in the echo waveform. The clearer the echo waveform, the easier it is to identify the peak position, and the ease of identifying the peak position directly affects the accuracy of the final mud level prediction. If the peak position is clear, the waveform feature extraction of the echo waveform will be more accurate.

[0029] Signal-to-noise ratio (SNR), an important parameter for measuring signal quality, represents the ratio of the power of the useful signal to the power of the background noise. A high SNR indicates good signal quality, while a low SNR indicates poor signal quality. Generally, the higher the SNR, the clearer the echo waveform. In summary, the clarity of the peak position is closely related to the SNR. A high SNR helps improve the clarity of the waveform, making the peak position more obvious. Therefore, the peak position is easier to identify, and echo waveforms with a high SNR are more likely to predict the mud level value.

[0030] Step S502: Based on the waveform characteristics, determine the confidence level of the waveform to be tested; wherein, the confidence level is used to characterize the reliability of the waveform to be tested in predicting the mud level value, and the mud level value is used to characterize the distance from the bottom of the pool to the mud-water interface.

[0031] In this embodiment, the stability and clarity of the waveform are determined based on the waveform characteristics in step S501. The signal-to-noise ratio (SNR), i.e., the ratio of the peak value of the signal to the standard deviation of the background noise, is calculated to assess the confidence level of the signal. Then, the similarity between the waveform to be tested and known waveforms is evaluated by comparing their waveform characteristics. Each known waveform corresponds to a specific confidence level, thus determining the confidence level of the waveform to be tested. The confidence level characterizes the reliability of the waveform to be tested in predicting the mud level. For example, a waveform feature with a high SNR and clear peaks corresponds to a higher confidence level, while a waveform feature with a low SNR and complex peaks corresponds to a lower confidence level. The mud level is used to characterize the distance from the bottom of the pond to the mud-water interface.

[0032] For example, the following will combine Figures 1-4 To elaborate, we will first extract... Figures 1-4 For each echo waveform to be measured, we first extract waveform features, such as peaks, troughs, rising edges, and falling edges. Then, we digitize these features, converting them into numerical data for further analysis and comparison.

[0033] Then, known waveform data of the same type or similar conditions as the echo waveform to be tested are collected, and the same feature extraction and digitization processing are performed. These known waveform features will serve as reference standards. The digitized features of the waveform to be tested are compared one by one with the digitized features of the known waveforms. This includes comparing key parameters such as peak position, trough position, waveform width, and periodicity. Based on the feature comparison results, the similarity between the echo waveform to be tested and the known waveforms is calculated. Based on the similarity calculation results, a confidence value is assigned to the echo waveform to be tested. This value can be a direct reflection of the similarity or obtained through a mapping function. For example, if the similarity ratio is high, the confidence value will also be relatively high. Thus, the confidence value of each echo waveform to be tested can be obtained.

[0034] Step S503: Based on the confidence level, determine at least one target model corresponding to the measured echo waveform from multiple models; wherein, the multiple models include: a first model, a second model, and a third model; wherein, the first model is obtained based on the waveform features, the second model is trained based on the echo waveform and actual mud level value under known usage scenarios, and the third model is trained based on the echo waveform and actual mud level value under unknown usage scenarios, wherein the known usage scenarios include at least: radial flow sedimentation tank, high-efficiency sedimentation tank, and horizontal flow sedimentation tank, and the unknown usage scenarios are scenarios other than the known usage scenarios.

[0035] In this embodiment, based on the confidence level, a target model corresponding to the measured echo waveform is determined from multiple models. Since known usage scenarios include at least radial flow sedimentation tanks, high-efficiency sedimentation tanks, and horizontal flow sedimentation tanks, and unknown usage scenarios are those other than the known ones, and the first model is derived based on waveform characteristics, i.e., directly based on waveform features such as peak and trough positions, to predict the sediment level. The first model utilizes a deep understanding of the system's physical processes to predict sediment level by analyzing the direct relationship between waveform characteristics and sediment level. This method does not rely on a large amount of historical data but focuses on modeling the system's intrinsic mechanisms.

[0036] The second model is trained based on echo waveforms and actual sediment levels under known usage scenarios. It employs machine learning algorithms, such as Support Vector Machines (SVM), Random Forests, or Neural Networks, to learn the complex relationship between waveform features and sediment levels from historical data. The second model is applicable to specific known scenarios, such as radial flow sedimentation tanks, high-efficiency sedimentation tanks, and horizontal flow sedimentation tanks, and can predict sediment levels from the measured echo waveforms in these scenarios.

[0037] The third model is trained on echo waveforms and actual mud level values ​​for unknown usage scenarios. It utilizes cloud computing resources to process and analyze large amounts of data, including data from both known and unknown scenarios. This model has higher generalization ability and can adapt to a wider range of scenarios. However, due to its large training dataset, its confidence level in predicting mud level values ​​for measured echo waveforms in known usage scenarios may be lower than that of the second model, which is specifically trained for these scenarios. Therefore, the confidence level of the third model in predicting mud level values ​​for measured echo waveforms in known usage scenarios will be lower than that of the second model. Furthermore, because of its larger training database, the third model can also predict mud level values ​​for measured echo waveforms in known usage scenarios.

[0038] At least one target model is selected from the first, second, and third models, either individually or in combination, based on confidence levels. Specifically, the first model directly predicts sludge level values ​​based on waveform characteristics, suitable for a basic understanding of sludge level changes, providing rapid response and preliminary prediction. The second model is trained on echo waveforms and actual sludge level values ​​under known usage scenarios, providing accurate predictions for specific scenarios, applicable to known usage scenarios such as radial flow sedimentation tanks, high-efficiency sedimentation tanks, and horizontal flow sedimentation tanks. Furthermore, known usage scenarios can also include CASS (Cyclic Activated Sludge System) process tanks. The third model is trained on echo waveforms and actual sludge level values ​​under unknown usage scenarios, applicable to scenarios other than known usage scenarios, enhancing the model's generalization ability. Unknown usage scenarios can include thickening tanks in waterworks, sedimentation tanks in mineral washing plants, and slurry sedimentation tanks. Therefore, a suitable model can be selected to predict sludge level values ​​based on confidence levels. In this embodiment, the first, second, and third models are only used to distinguish three different types of models and have no other meaning.

[0039] Step S504: Based on the at least one target model, output the mud level value corresponding to the measured echo waveform.

[0040] In this embodiment, based on the target model, the mud level value corresponding to the measured echo waveform is output. For example, if the target model is determined to be a single model predicting the mud level value based on the confidence level, then the mud level value output by that single model can be used as the mud level value corresponding to the measured echo waveform. For example, if the single model is the first model, then the mud level value output by the first model is used as the mud level value corresponding to the measured echo waveform. When it is determined that the target model is multiple models predicting the mud level value, then the average of the mud level values ​​output by the multiple models can be used as the mud level value corresponding to the measured echo waveform. Alternatively, the mud level value corresponding to the measured echo waveform can be output as a weighted average based on the confidence level ratio of each model. For example, based on the confidence level evaluation result of each model, a corresponding weight is assigned. The weight reflects the reliability of the model prediction; the higher the confidence level of the model, the higher its weight should be. For example, if the first model has a prediction confidence of 0.8 in a specific scenario, the second model has 0.6, and the third model has 0.7, then they might be assigned weights of 0.4, 0.3, and 0.3, respectively. When multiple target models are identified, these weights are used to weight the prediction results of each model. For instance, if the weights of the first and second models are 0.5 and 0.5, respectively, then the mud level value of the measured echo waveform can be a weighted average of the mud level values ​​predicted by the two models. Therefore, the confidence level is directly proportional to the weight; that is, the higher the confidence level, the greater the weight assigned. In this way, the model with higher confidence has a greater influence in the final prediction, thereby improving the overall accuracy and reliability of the prediction. In this manner, the weighting is closely related to the level of confidence, ensuring the accuracy of the prediction results and the adaptability of the model.

[0041] Furthermore, since the second model is trained on echo waveforms and actual mud level values ​​under known usage scenarios, it can provide relatively reliable predictions for these specific scenarios. Therefore, with moderate confidence levels, the weight of the second model in the weighting can be appropriately increased to leverage its advantages in specific scenarios. Meanwhile, the third model, trained on a broader dataset including echo waveforms and actual mud level values ​​for unknown usage scenarios, has better generalization ability. With low confidence levels, the third model can serve as an effective supplement, especially when the first and second models struggle to provide accurate predictions.

[0042] Therefore, the weights and confidence levels of the weighted average can be determined in the following way: First, based on the already determined confidence level of the echo waveform to be measured, the weights of the first model, the second model, and the third model in the weighted average are determined according to the confidence level.

[0043] For example, when the confidence level is in the first range, the weight of the first model is greater than the weight of the second model, and the weight of the second model is greater than the weight of the third model; When the confidence level is in the second range, the weight of the second model is greater than the weight of the first model, and the weight of the first model is greater than the weight of the third model. When the confidence level is in the third range, the weight of the third model is greater than the weight of the second model, and the weight of the second model is greater than the weight of the first model. Among them, the first range is larger than the second range, and the second range is larger than the third range.

[0044] For example, if the confidence level is high, the first model may have a larger weight; if the confidence level is medium, the second model may have a larger weight; and if the confidence level is low, the third model may have a larger weight. With medium confidence, the weight of the second model can be appropriately increased while maintaining the weight of the third model to ensure prediction accuracy and generalization ability. Finally, the predictions from the three models are weighted according to their respective weights to obtain the final mud level prediction. This method ensures that the most reliable mud level predictions are obtained at different confidence levels, while also fully utilizing the advantages of each model.

[0045] In summary, the measurement method provided in this embodiment, when measuring the mud level at the mud-water interface, can solve the problem of inaccurate or incorrect mud level identification when the mud layer is clear and the echo is complex by using a target model fused from the first, second, and third models. By training and predicting complex echo data, the measurement effectiveness in such cases is resolved. Furthermore, for situations where the mud layer distribution is unclear and the echo is cluttered, the confidence level corresponding to the mud level value can be output simultaneously, ensuring that the measurement results are not merely for reference. The control system can then perform process control based on the mud level measurement value and the confidence level data.

[0046] In one specific embodiment, determining the confidence level corresponding to the measured echo waveform based on the waveform characteristics may include the following steps: First, the similarity between the first type of multiple waveforms and the second type of multiple waveforms and the waveform features is obtained respectively; then, the confidence level is determined based on the similarity; wherein, the confidence level of the first type of multiple waveforms is greater than the confidence level of the second type of multiple waveforms, and the multiple waveforms are echo waveforms under the known usage scenario.

[0047] In this embodiment, the confidence level of the echo waveform to be tested is determined based on waveform characteristics. First, multiple waveforms of the first type and multiple waveforms of the second type are acquired. The first type of waveform refers to the echo waveform with high confidence that can be obtained under certain ideal conditions in a known usage scenario, such as high signal-to-noise ratio and clear peak positions, through normal sampling. The second type of waveform refers to the abnormal waveform obtained under certain non-ideal conditions in a known usage scenario, such as bubble interference, low signal-to-noise ratio, unclear peak positions, or other interference factors. Therefore, the confidence level of the multiple waveforms of the first type is higher than that of the multiple waveforms of the second type. Thus, the similarity between the waveform characteristics of the echo waveform to be tested and the multiple waveforms of the first and second types can be compared to determine the confidence level range of the echo waveform to be tested.

[0048] Specifically, methods for determining similarity can employ cosine similarity or cosine distance. Cosine similarity is a method that measures the directional similarity between two vectors (the waveform to be measured and various waveforms of type 1 or type 2). In waveform comparison, we can convert the waveforms into vectors and then evaluate the angle between these vectors. The smaller the angle, the more similar the waveforms. Cosine distance is a method for measuring the difference between two vectors; it is derived by calculating the angle between the two vectors (the waveform to be measured and various waveforms of type 1 or type 2). In waveform comparison, we can imagine that the larger the angle between two waveform vectors, the greater their difference and the lower their similarity.

[0049] For example, the following will refer to Figures 6-9 , Figures 6-9 The horizontal axis represents time, and the vertical axis represents the signal-to-noise ratio. Multiple waveforms of the first type and multiple waveforms of the second type are described below: Figure 6 This is a waveform diagram of a first type provided in an embodiment of this application. Figure 7 It is aimed at Figure 6 A second type of waveform diagram is provided. Figure 6 and Figure 7 For the same sampling scenario, Figure 6 To obtain a normal waveform without interference from other factors during the sampling process. Figure 7 This is an abnormal waveform observed as the sludge scraper passes through during the sampling process. Figure 8 This is another waveform diagram of the first type provided in the embodiments of this application. Figure 9 It is aimed at Figure 8 A second type of waveform diagram is provided. Figure 8 and Figure 9 For the same sampling scenario, Figure 8 To obtain a normal waveform without interference from other factors during the sampling process. Figure 9 This refers to the abnormal waveform, or invalid waveform, that occurs when the sampling device is lifted into the air during the sampling process. Figure 6 and Figure 7 For example, Figure 6 The confidence level of the corresponding waveform is set to 100. Figure 7 The confidence level of the corresponding waveform is set to 0, and the waveform characteristics of the echo waveform to be measured are compared with... Figure 6 and Figure 7 The similarity, assuming the waveform of the echo to be measured is similar to Figure 6 If the similarity is 80%, then the confidence level of the waveform to be tested is 80%. Assuming the waveform to be tested is similar to... Figure 7 If the similarity is 100%, then the confidence level of the waveform to be tested is 0.

[0050] Furthermore, confidence levels can be determined by obtaining the similarity between multiple waveforms of the first type and multiple waveforms of the second type and waveform features, respectively. Figure 6 For example, when three waveforms of the first type are obtained from multiple waveforms of the first type, and... Figure 6 The similarity of the waveforms is 30%, 50%, and 20%, respectively, and the confidence levels of the three first-type waveforms are 80%, 90%, and 60%, respectively. Then, we obtain... Figure 6 The confidence level is 81. Specifically, each similarity value is multiplied by its corresponding confidence level, then summed, and finally divided by the sum of the similarities. This method considers each first-type waveform and... Figure 6 The similarity of the waveforms was used to weight the confidence scores based on the degree of similarity. Waveforms with higher similarity scores have a greater impact on the final confidence score calculation, allowing for a more accurate assessment. Figure 6 The confidence level of the waveform, especially when there are multiple reference waveforms and different confidence levels.

[0051] In one specific embodiment, determining the confidence level based on the similarity may include the following steps; First, if the similarity between the waveform feature and multiple waveforms of the first type exceeds a preset value, a first confidence level is determined as the confidence level. Then, if the similarity between the waveform feature and multiple waveforms of the first type and multiple waveforms of the second type is lower than the preset value, a second confidence level is determined as the confidence level. Finally, if the similarity between the waveform feature and multiple waveforms of the second type exceeds the preset value, a third confidence level is determined as the confidence level. Wherein, the first confidence level is greater than the second confidence level, and the second confidence level is greater than the third confidence level.

[0052] In this embodiment, since the confidence level of the multiple waveforms of the first type is greater than that of the multiple waveforms of the second type, the confidence level range of the echo waveform to be tested can be determined by comparing the similarity of the waveform features with the waveforms of the first type and the waveforms of the second type. In addition, the multiple waveforms of the first type and the multiple waveforms of the second type are waveforms under known usage scenarios, while the echo waveform to be tested may be a waveform under an unknown usage scenario in addition to being a waveform under a known usage scenario. Therefore, determining the confidence level based on similarity can not only consider the similarity of waveforms under known usage scenarios, but also determine whether the echo waveform to be tested belongs to an unknown usage scenario.

[0053] For example, since the first confidence level is greater than the second confidence level, and the second confidence level is greater than the third confidence level, when the similarity between the waveform to be tested and one of the waveforms in the first type exceeds a preset value, it indicates that the waveform to be tested is likely obtained under ideal conditions. Therefore, we set its confidence level to the first confidence level. The preset value can be set according to the actual situation, and this embodiment does not limit it.

[0054] The similarity between the waveform to be tested and various waveforms of the first and second types is lower than the preset value. This indicates that the waveform to be tested is neither similar to the high-confidence waveform nor the low-confidence waveform, and is likely an echo waveform from an unknown usage scenario. Therefore, we set its confidence level to the second confidence level, which is lower than the first confidence level but higher than the third confidence level.

[0055] The similarity between the waveform under test and various waveforms of the second type exceeded the preset value. This indicates that the waveform under test was an abnormal waveform obtained under suboptimal conditions. Therefore, we set its confidence level to the third level. Mud level values ​​obtained from waveforms under test that meet the third confidence level are generally unreliable.

[0056] In one specific embodiment, when determining the confidence level corresponding to the waveform to be measured based on the waveform characteristics, the following steps may also be included: The waveform to be tested is input into a classification model to obtain the confidence level corresponding to the waveform to be tested; wherein, the classification model is trained on a dataset based on a variety of different echo waveforms and the labels corresponding to the echo waveforms, and the labels are used to characterize whether the echo waveform can predict the mud level value.

[0057] In this embodiment, the confidence level of the echo waveform to be tested is determined based on waveform features. Specifically, the echo waveform to be tested is input into a classification model, and the confidence level of the echo waveform is obtained through the classification model. In this embodiment, the classification waveform is obtained by training on a dataset based on different echo waveforms and their corresponding labels. Figures 6-9 To elaborate, one can Figure 6 and Figure 8 The corresponding echo waveforms are labeled as predictable waveforms, meaning waveforms that can predict mud level values. Figure 7 and Figure 9 The corresponding echo waveforms are labeled as unpredictable waveforms, meaning the waveform for which the mud level value cannot be predicted. Figures 6-9 The corresponding echo waveforms are used as datasets to train classification models. The trained classification models can be directly used to analyze the waveform features of the echo waveforms to be tested and determine the confidence level of the echo waveforms to be tested.

[0058] For example, suppose the waveform of the echo to be measured is Figure 7 Then Figure 7 The data is input into the classification model, and the classification model is determined. Figure 7 If the similarity between the waveform characteristics of the corresponding echo waveform and one of the multiple waveforms of the second type in the classification model exceeds a preset value, then it can be determined that... Figure 7 The corresponding echo waveform is a low-confidence waveform, that is... Figure 7 This corresponds to a very low confidence level and is also an unpredictable waveform.

[0059] In one specific embodiment, determining at least one target model corresponding to the measured echo waveform from multiple models based on the confidence level may include the following steps: First, when the confidence level is at the first confidence level, the at least one target model is determined to be the first model and the second model predicting the mud level value; then, when the confidence level is at the second confidence level, the at least one target model is determined to be the first model and the third model predicting the mud level value; finally, when the confidence level is at the third confidence level, it is determined to exit the prediction of the mud level value.

[0060] In this embodiment, since the first model is based on waveform features, it can output the mud level value of the echo waveform under test regardless of the scenario. The second model is trained based on the echo waveform and actual mud level value under known usage scenarios, so the first model can output a relatively accurate mud level value for the echo waveform under known usage scenarios. The third model is trained based on the echo waveform and actual mud level value under unknown usage scenarios, so it can output a relatively accurate mud level value for the echo waveform under unknown usage scenarios. Therefore, when the confidence level is at the first level, it means that the echo waveform under test is an echo waveform under known usage scenarios, and the confidence level is high. Thus, the mud level value of the echo waveform under test can be predicted using the first and second models. When the confidence level is at the second level, it means that the echo waveform under test is an echo waveform under unknown usage scenarios, and the confidence level is moderate. Thus, the target model can predict the mud level value of the echo waveform under test using the first and third models. When the confidence level is third, it means that the waveform to be tested is the waveform to be tested under a known application scenario, and the confidence level is very low. Even if the mud level value of the waveform to be tested is predicted, the data cannot be used for subsequent work and research. Therefore, in the case of a confidence level of third, in order to avoid unnecessary occupation of computing space, the target model is determined to exit the prediction of the mud level value of the waveform to be tested.

[0061] In one specific embodiment, when the target model is the first model and the second model, the step of outputting the mud level value corresponding to the measured echo waveform based on the at least one target model may include the following steps: First, based on the first model, a first predicted mud level value of the echo waveform to be measured is determined; then, based on the second model, a second predicted mud level value of the echo waveform to be measured is determined; finally, based on the first predicted mud level value and the second predicted mud level value, the mud level value is output.

[0062] In this embodiment, when the target models are the first model and the second model, the waveform to be tested is first analyzed according to the first model to obtain the first predicted mud level value of the waveform. Then, the waveform to be tested is analyzed according to the second model. Based on historical data and learning algorithms, the second model can identify the pattern and trend of the waveform to be tested and obtain the second predicted mud level value of the waveform. Finally, the mud level value to be output by the first and second predicted mud level values ​​is determined together. Specifically, the mud level value can be determined by averaging the sum of the first and second predicted mud level values, or by assigning different weights to the first and second predicted mud level values ​​and calculating a weighted average as the final mud level value. The weights can be determined based on the accuracy, reliability, or applicability of the model in a specific scenario. In some cases, they can also be determined based on the confidence level. Of course, a more complex fusion algorithm can also be used to combine the advantages of the first and second predicted mud level values ​​to obtain a comprehensive predicted mud level value.

[0063] In one specific embodiment, when the at least one target model is the first model and the third model, the step of outputting the mud level value corresponding to the measured echo waveform based on the at least one target model may include the following steps: first, determining the first predicted mud level value of the measured echo waveform based on the first model; then, determining the third predicted mud level value of the measured echo waveform based on the third model; and finally, outputting the mud level value based on the first predicted mud level value and the third predicted mud level value.

[0064] In this embodiment, when the target models are the first model and the third model, the first model is used to analyze the waveform of the echo to be tested, obtaining the first predicted mud level value of the waveform. Then, the third model is used to analyze the waveform of the echo to be tested. The third model, based on big data and machine learning technology, can process data from a wide range of scenarios, including known and unknown usage scenarios, to obtain the third predicted mud level value. The second predicted mud level value of the echo to be tested is then obtained. The first and third predicted mud level values ​​are combined to output the final mud level value. This can be achieved through simple averaging, weighted averaging, or more complex fusion algorithms. Combining the first and third models can reduce the bias that may exist in a single model and improve the accuracy of mud level prediction. When facing unknown usage scenarios, the combination of the two models can provide a more comprehensive analysis and enhance the model's ability to handle abnormal situations.

[0065] In one specific embodiment, the first model is configured as follows: First, based on the wave crest position, the travel time of the reflected wave from the mud surface is determined; wherein, the travel time is used to characterize the total time it takes for the ultrasonic signal to travel from the sensor, reach the mud-water interface, and be reflected back to the sensor; then, based on the travel time and the ultrasonic velocity, a first transition mud level value is determined; wherein, the first transition mud level value is used to characterize the distance from the sensor to the mud-water interface; finally, based on the first transition mud level value, a first predicted mud level value is output.

[0066] In this embodiment, the first model determines the first predicted mud level value by first analyzing the peak position of the measured echo waveform to determine the travel time of the reflected wave from the mud surface. Travel time refers to the total time it takes for an ultrasonic signal to travel from the sensor, reach the mud-water interface, and reflect back to the sensor. Using the determined travel time and the known velocity of ultrasound in the medium, the first model calculates the first transition mud level value. This value represents the distance from the sensor to the mud-water interface, and the calculation formula can be expressed as: First transition mud level value = travel time × ultrasonic velocity / 2.

[0067] The division by 2 is because the ultrasonic signal needs to travel from the sensor to the mud-water interface and then back to the sensor; therefore, the actual distance is half the distance the ultrasonic wave travels during its travel time. The first model, based on physical principles, provides a direct method for measuring mud level, reducing prediction inaccuracies caused by algorithm or model errors.

[0068] In one specific embodiment, the second model is trained through the following steps: First, sampling data of the known usage scenario is obtained; wherein, the sampling data includes echo waveforms and actual mud level values; then, the training waveform features of the echo waveforms are extracted from the sampling data, and the training waveform features are associated with the actual mud level values ​​to construct a dataset; finally, the second model is trained using the dataset as training samples.

[0069] In this embodiment, sampling data from a known usage scenario is first acquired. This sampling data includes echo waveforms and their corresponding actual mud level values, which are directly measured. Then, training waveform features are extracted from the sampling data. These training waveform features may include peak position, signal-to-noise ratio, number of peaks, peak height, and peak width. These features reflect the waveform characteristics of the echo waveform and have a direct or indirect relationship with the mud level value. The extracted training waveform features are then correlated with the actual mud level value to construct a dataset. This dataset serves as a sample for training a second model, establishing the relationship between the training waveform features and the mud level value. Using the constructed dataset as training samples, a machine learning algorithm is used to train the second model. This model can learn the complex relationship between the training waveform features and the mud level value and can predict the mud level value for new measured echo waveforms.

[0070] In one specific embodiment, after determining the mud level value based on the first predicted mud level value and the third predicted mud level value, the following steps may be further included: First, the target application scenario of the third predicted mud level value is obtained. Based on the measured echo waveform and the third predicted mud level value, the dataset corresponding to the second model is updated. Then, based on the updated dataset, the second model is updated.

[0071] In this embodiment, after determining the mud level value based on the first and third predicted mud level values, it is also necessary to determine the target use scenario of the third predicted mud level value predicted by the third model, and to determine whether the target use scenario is an unknown use scenario. If the target use scenario is an unknown use scenario, the data set corresponding to the second model is updated based on the measured echo waveform and the third predicted mud level value. This means that new actual observation data (the measured echo waveform and the third predicted mud level value) are added to the training dataset to ensure that the second model can adapt to new use scenarios, so that the second model is applicable to more and more use scenarios and increases the effectiveness of the mud level value output by the second model.

[0072] In this embodiment, refer to Figure 10 , Figure 10 This is a schematic diagram of a mud-water interface measurement system provided in an embodiment of this application; from Figure 10 As can be seen from the data, the system includes: an acquisition module 1001, a first determination module 1002, a second determination module 1003, and a third determination module 1004. The acquisition module 1001 is used to acquire the waveform of the echo to be measured at the mud-water interface and determine the waveform characteristics of the waveform to be measured; wherein, the waveform characteristics include at least: the peak position and the signal-to-noise ratio; The first determining module 1002 is used to determine the confidence level of the measured echo waveform based on the waveform characteristics; wherein the confidence level is used to characterize the reliability of the measured echo waveform in predicting the mud level value, and the mud level value is used to characterize the distance from the bottom of the pool to the mud-water interface. The second determining module 1003 is used to determine, based on the confidence level, at least one target model corresponding to the measured echo waveform from a plurality of models; wherein the plurality of models includes: a first model, a second model, and a third model; wherein, The first model is obtained based on the waveform features. The second model is trained based on the echo waveform and actual mud level value under known usage scenarios. The third model is trained based on the echo waveform and actual mud level value under unknown usage scenarios. The known usage scenarios include at least: radial flow sedimentation tank, high-efficiency sedimentation tank and horizontal flow sedimentation tank. The unknown usage scenarios are scenarios other than the known usage scenarios. The output module 1004 is used to output the mud level value corresponding to the measured echo waveform based on at least one of the target models.

[0073] exist Figure 10 The provided measurement system may include Figure 11 The provided method outputs the measured mud level value. Figure 11 This is a schematic diagram of a system for measuring mud level values ​​provided in an embodiment of this application. Figure 11 The system includes a classification model, a first model, a second model, and a third model. The first, classification, and second models can reside on the local computing module, while the third model resides in the cloud and can be accessed by the local computer. The input of the classification model receives the input of the measured echo waveform. Its output is connected to the inputs of the first, second, and third models. The outputs of the first, second, and third models are connected to the computer's decision fusion algorithm. This algorithm receives the predicted mud level values ​​from the first, second, and third models respectively, then fuses them to output the final mud level value. Furthermore, the third predicted mud level value output by the third model can update the training set in the second model; therefore, the third model can also be connected to the second model. The following will use... Figure 11 The functions that the measurement system provided in this embodiment can achieve are described in detail: from Figure 11 As can be seen, firstly, ultrasonic waves are emitted towards the mud-water interface, then the echoes returned are received and collected, i.e., the waveform to be measured, and the waveform to be measured is digitized. The digitized waveform to be measured is then input into a classification model. The classification model determines the confidence level of the waveform to be measured, and then a target model suitable for predicting mud level values ​​based on the confidence level is determined.

[0074] When the confidence level is determined to be the first confidence level, that is, when the waveform to be measured is a waveform with a very high confidence level, in the absence of other requirements, the average value of the superposition of the first predicted mud level value output by the first model and the second predicted mud level value output by the second model can be selected as the final mud level value. At the same time, the first confidence level can also be output as a reference for subsequent judgment of the reliability of the mud level value.

[0075] When the confidence level is determined to be the second confidence level, i.e., the waveform of the test echo is the waveform in the middle of the confidence level, under the condition that there are no other requirements, the average value of the superposition of the first predicted mud level value output by the first model and the third predicted mud level value output by the third model can be selected as the final mud level value. At the same time, the second confidence level can also be output as a reference for subsequent judgment of the reliability of this mud level value. In addition, the third predicted mud level value output by the third model can also be input into the second model to update the dataset of the second model, so that the second model can add new use cases in subsequent training and increase the generalization of the second model.

[0076] When the confidence level is determined to be the third confidence level, that is, when the waveform to be tested is a waveform with a very low confidence level, the waveform to be tested is considered an invalid waveform and the final mud level value is not output. The third confidence level can be output as a reference for subsequent judgment on the reliability of not outputting the mud level value.

[0077] For example, the following will use the field test data of the sludge-water interface meter installed in a conventional sedimentation tank of a wastewater treatment plant as an example, and refer to... Figure 12 This will illustrate the predictions of mud level values ​​by the first and second models, respectively. Figure 12 This is a schematic diagram illustrating how a first model and a second model predict mud level values, as provided in an embodiment of this application. Figure 12 The results of two models predicting mud level values ​​are shown. Figure 12 In the diagram, the curve with larger fluctuations represents the first predicted mud level value output by the first model, while the curve with smaller fluctuations represents the second predicted mud level value output by the second model. After processing by the decision system, the final output mud level value matches the curve with smaller fluctuations, meaning it conforms to the prediction result of the second model. This decision result indicates that, in this specific situation, the measurement value of the first model may exhibit jumps, while the output of the second model is more stable; therefore, the prediction provided by the second model is more stable and reliable. Thus, in some cases, the second predicted mud level value output by the second model can be directly used as the final mud level value output.

[0078] Furthermore, the following will use the field test data of the sludge-water interface meter installed in the membrane tank of an SBR (Sequencing Batch Reactor) in a wastewater treatment plant as an example, and refer to... Figure 13 This will illustrate the predictions of mud level values ​​by the first and second models, respectively. Figure 13 This is a schematic diagram illustrating how a first model and a second model predict mud level values, as provided in the embodiments of this application. Figure 13 As can be seen, the first row of curves represents the measured values ​​output by the decision fusion system, which are the final values ​​after comprehensively considering the prediction results of the first and second models. The second row of curves represents the measured values ​​output by the first model, which are the prediction results derived based on physical principles and process knowledge. The third row of curves represents the changes in dissolved oxygen levels in the tank. During the aeration stage, the dissolved oxygen level increases, while during the sludge settling stage, the dissolved oxygen level essentially returns to zero.

[0079] from Figure 13 As can be observed, during the aeration stage, the sludge-water interface is relatively fixed because the tank is filled with oxygen. Therefore, the decision fusion system outputs a fixed value, namely the depth of the tank. During the settling stage, the sludge level actually changes as the sludge settles, and the decision fusion system adjusts its output value accordingly. In contrast, if only the output of the first model (the second row of curves) is used, the measurement results may be unreliable during the settling stage because the actual sludge level changes are not fully considered. This indicates that, in this situation, relying solely on the first model may not accurately reflect the true change in sludge level, while the decision fusion system, by integrating the prediction results of the first and second models, can provide a more accurate sludge level measurement. Therefore, in this case, the first and second models can be used together to output the final sludge level value.

[0080] In summary, the method provided in this embodiment introduces a second model algorithm based on the traditional first model, solving the problem of inaccurate or incorrect mud level identification when the mud layer is clear but the echo is complex. By training and predicting on complex echo data, the measurement validity problem in this case is solved. For cases where the mud layer distribution is unclear and the echo is messy, the confidence level corresponding to the mud level value can also be output, indicating that the measurement result is unreliable.

[0081] In addition, a confidence level can be added as an output, with each mud level measurement having a corresponding confidence level; this makes the final output mud level value no longer just a reference, and the control system can perform process control based on the mud level value and confidence level data; Secondly, for scenarios where the mud layer is clear but the echo is not suitable for existing algorithms, this invention utilizes a third model to collect data from new scenarios, expand the dataset, retrain the model, update the dataset of the second model, and improve prediction accuracy and adaptability. Finally, by digitally processing the waveform of the echo to be tested, the invention solves the problems of signal reflection and distortion caused by long-distance transmission of analog signals, thereby improving the signal-to-noise ratio.

[0082] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0083] Embodiments of the present invention are described with reference to flowchart illustrations and / or block diagrams of methods and apparatus according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0084] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0085] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal equipment, causing a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0086] Although preferred embodiments of the present invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present invention.

[0087] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0088] The above provides a detailed description of the method for measuring the mud-water interface provided by the present invention. Specific examples have been used to illustrate the principle and implementation of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation and application scope based on the idea of ​​the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method of measuring a mudline interface, characterized by, The method includes: The waveform of the echo at the mud-water interface is acquired, and the waveform characteristics of the echo are determined; wherein the waveform characteristics include at least: the peak position and the signal-to-noise ratio. Based on the waveform characteristics, the confidence level of the measured echo waveform is determined; wherein, the confidence level is used to characterize the reliability of the measured echo waveform in predicting the mud level value, and the mud level value is used to characterize the distance from the bottom of the pond to the mud-water interface. Based on the confidence level, at least one target model corresponding to the measured echo waveform is determined from multiple models; wherein, the multiple models include: a first model, a second model, and a third model; wherein, The first model is obtained based on the waveform features. The second model is trained based on the echo waveform and actual mud level value under known usage scenarios. The third model is trained based on the echo waveform and actual mud level value under unknown usage scenarios. The known usage scenarios include at least: radial flow sedimentation tank, high-efficiency sedimentation tank and horizontal flow sedimentation tank. The unknown usage scenarios are scenarios other than the known usage scenarios. Based on at least one of the target models, output the mud level value corresponding to the measured echo waveform.

2. The measurement method according to claim 1, characterized in that, The step of determining the confidence level corresponding to the measured echo waveform based on the waveform features includes: Obtain the similarity between the waveform features and multiple waveforms of the first type and multiple waveforms of the second type, respectively; The confidence level is determined based on the similarity. The confidence level of the multiple waveforms of the first type is greater than that of the multiple waveforms of the second type, and the multiple waveforms are echo waveforms under the known usage scenario.

3. The measurement method according to claim 2, characterized in that, Determining the confidence level based on the similarity includes: If the similarity between the waveform feature and multiple waveforms of the first type exceeds a preset value, the first confidence level is determined as the confidence level. If the similarity between the waveform feature and multiple waveforms of the first type and multiple waveforms of the second type is lower than the preset value, the second confidence level is determined as the confidence level. If the similarity between the waveform feature and multiple waveforms of the second type exceeds the preset value, the third confidence level is determined as the confidence level. Wherein, the first confidence level is greater than the second confidence level, and the second confidence level is greater than the third confidence level.

4. The measuring method according to any one of claims 1 to 3, characterized in that, The step of determining the confidence level corresponding to the measured echo waveform based on the waveform features includes: The waveform to be tested is input into the classification model to obtain the confidence level corresponding to the waveform to be tested; The classification model is trained on a dataset based on a variety of different echo waveforms and the corresponding labels of the echo waveforms. The labels are used to characterize whether the echo waveform can predict the mud level value.

5. The measurement method according to claim 1, characterized by, The step of determining at least one target model corresponding to the measured echo waveform from multiple models based on the confidence level includes: When the confidence level is a first confidence level, the at least one target model is determined to be the prediction of the mud level value by the first model and the second model; When the confidence level is the second confidence level, the at least one target model is determined to be the first model and the third model predicting the mud level value; If the confidence level is the third confidence level, then it is determined to exit the prediction of the mud level value.

6. The measurement method according to claim 1, characterized by, When the target model is the first model and the second model, the step of outputting the mud level value corresponding to the measured echo waveform based on the at least one target model includes: Based on the first model, the first predicted mud level value of the echo waveform to be measured is determined; Based on the second model, the second predicted mud level value of the measured echo waveform is determined; Based on the first predicted mud level value and the second predicted mud level value, the mud level value is output.

7. The measurement method according to claim 1, characterized by, When the target model is the first model and the third model, the step of outputting the mud level value corresponding to the measured echo waveform based on the at least one target model includes: Based on the first model, the first predicted mud level value of the echo waveform to be measured is determined; Based on the third model, the third predicted mud level value of the measured echo waveform is determined; Based on the first predicted mud level value and the third predicted mud level value, the mud level value is output.

8. The measurement method of claim 1, wherein, The first model is configured as follows: Based on the wave crest position, the travel time of the reflected wave from the mud surface is determined; wherein, the travel time is used to characterize the total time it takes for the ultrasonic signal to be emitted from the sensor, reach the mud-water interface, and be reflected back to the sensor; Based on the travel time and ultrasonic velocity, a first transition mud level value is determined; wherein, the first transition mud level value is used to characterize the distance from the sensor to the mud-water interface; Based on the first transition mud level value, the first predicted mud level value is output.

9. The method of claim 1, wherein, The second model is obtained through the following steps: Acquire sampling data for the known usage scenario; wherein the sampling data includes echo waveforms and the actual mud level value; The training waveform features of the echo waveform are extracted from the sampled data, and the training waveform features are associated with the actual mud level value to construct a dataset; The second model is trained using the dataset as training samples.

10. The measurement method according to claim 7, characterized by, After determining the mud level value based on the first predicted mud level value and the third predicted mud level value, the method further includes: The target use case for obtaining the third predicted mud level value is updated based on the measured echo waveform and the third predicted mud level value. The second model is updated based on the updated dataset.

Images