Processing and filling monitoring method based on multi-source sensor data

Abstract

The present application relates to the field of data processing technology, and in particular to a method for monitoring processing and filling based on multi-source sensor data. The method comprises: weighted fusion of sensor data from the processing and filling process based on the correlation between sensor data and filling quality to obtain a real-time sequence of fused data; calculation of the local deviations between the real-time sequence and the standard sequence in each time period to obtain the noise probability of each time period in the real-time sequence; and weighted summation of the local deviations in each time period based on the noise probability to obtain a global deviation. If the global deviation is greater than a global threshold, the monitoring result is abnormal; otherwise, the monitoring result is normal. The technical solution of the present application can eliminate the influence of noise on the monitoring results and accurately obtain the monitoring results of processing and filling.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a processing and filling monitoring method based on multi-source sensor data. Background Art

[0002] The processing and filling process is a key step in the production of chemical reagents, directly related to the production quality of the product packaging process. To ensure the production quality of chemical reagents, how to accurately monitor key parameters during processing and filling has become a pressing issue.

[0003] At present, the patent application document with publication number CN116933210A discloses a food processing filling equipment monitoring method and system based on multi-dimensional sensors, wherein the method includes: obtaining monitoring data of food processing filling equipment, and constructing a filling area monitoring matrix using the monitoring data; obtaining the intrinsic deviation coefficient of the filling area according to the filling area monitoring matrix; obtaining the appearance matching coefficient of the filling area according to the filling area monitoring matrix; obtaining a comprehensive quality deviation index according to the intrinsic deviation coefficient and the appearance matching coefficient of the filling area; obtaining a comprehensive deviation similarity according to the comprehensive quality deviation index; obtaining a filling abnormality risk index according to the comprehensive quality deviation index; obtaining a filling area abnormality confidence coefficient according to the filling abnormality risk index and the comprehensive deviation similarity; and completing the monitoring of the filling equipment according to the filling area abnormality confidence coefficient.

[0004] The above method obtains the intrinsic deviation coefficient and the appearance consistency coefficient based on the monitoring data, and realizes the monitoring of the filling equipment by combining the intrinsic deviation coefficient and the appearance consistency coefficient, wherein the intrinsic deviation coefficient is obtained based on the weight data and the liquid level height data, and the appearance consistency coefficient is obtained based on the colorimetric data; however, the working environment of the processing and filling equipment is complex, and in the process of obtaining the monitoring data, noise data irrelevant to the monitoring process will be introduced, resulting in inaccurate monitoring results of the processing and filling. Summary of the Invention

[0005] In order to solve the technical problem of inaccurate monitoring results of processing and filling, the present application provides a processing and filling monitoring method based on multi-source sensor data, which can eliminate the influence of noise on the monitoring results and accurately obtain the monitoring results of processing and filling.

[0006] In a first aspect, the present application provides a processing and filling monitoring method based on multi-source sensor data, the monitoring method comprising: weighted fusion of sensor data in the processing and filling process according to the correlation between the sensor data and the filling quality to obtain a real-time sequence of the fused data; calculating the local deviation of the real-time sequence and the standard sequence in each time period, and obtaining the noise probability of each time period in the real-time sequence, including: obtaining a sample sequence and filling quality of the fused data in the historical canning process, and dividing the sample sequence into multiple time periods at equal intervals; calculating the local deviation of each time period between the sample sequence and the standard sequence, and taking the time period in which the local deviation is greater than the local threshold as a difference segment, and in response to the number of difference segments being 1, the timestamp of the difference segment is replaced by and the difference sequence between the sample sequence and the standard sequence in the difference segment as a set of training data; in response to the absolute value of the difference between the filling quality and the standard quality being greater than the difference threshold, the noise label of the training data is 0, otherwise, the noise label of the training data is 1; the training data is input into the binary classification model, the noise probability is output, and the binary classification model is trained based on the cross entropy function between the noise probability and the noise label; the timestamp and difference sequence of each time period in the real-time sequence are input into the trained binary classification model, and the noise probability of each time period in the real-time sequence is output; the local deviation of each time period is weighted and summed according to the noise probability to obtain the global deviation, and in response to the global deviation being greater than the global threshold, the monitoring result is abnormal, otherwise the monitoring result is normal.

[0007] During the processing and filling process, multiple sensor data can be collected at each moment. The sensor data at each moment are weighted according to the correlation between each sensor data and the filling quality to obtain the fused data at each moment of the processing and filling process, and then the real-time sequence of the fused data is obtained; the real-time sequence of the fused data is divided into multiple time periods with equal intervals, and the local deviations between the real-time sequence and the standard sequence in each time period, as well as the noise probability of each time period in the real-time sequence are calculated. The noise probability can reflect the degree of influence of the fluctuation of the real-time sequence in the corresponding time period on the filling quality. The greater the noise probability, the smaller the influence of the fluctuation of the real-time sequence in the corresponding time period on the filling quality, and the smaller the noise probability, the greater the influence of the fluctuation of the real-time sequence in the corresponding time period on the filling quality; the local deviations of each time period are weighted and summed according to the noise probability to obtain the global deviation. The global deviation eliminates the influence of noise data irrelevant to canning quality monitoring on the monitoring results. By comparing the global deviation and the global threshold, the monitoring results of processing and filling are accurately obtained.

[0008] Preferably, the sensor data The calculation method of the correlation between the filling quality and the sensor data includes: In addition, the DTW distance between the historical sequences of each sensor data is used as the clustering distance, and each sample data is clustered to obtain multiple clusters. The sample data includes the historical sequence of each sensor data in the historical canning process and the canning quality; within any cluster, the sensor data is calculated. The average sequence of each historical sequence, sensor data The variance of the DTW distance between each historical sequence and the average sequence is taken as the first variance, and the absolute value of the difference between the first variance and the canned quality variance in the cluster is calculated; the absolute value of the difference of each cluster is weighted summed according to the proportion of sample data in the cluster, and the sensor data The correlation between filling quality and weighted summation is negatively correlated.

[0009] Based on sensor data In addition, the DTW distance between the historical sequences of each sensor data is used to cluster the sample data to obtain multiple clusters. The sensor data in the clusters The historical sequences of sensor data are basically the same; the sensor data are calculated in the cluster Correlation between filling quality and excluding sensor data The influence of other sensor data is eliminated to ensure the accuracy of the correlation.

[0010] Preferably, the sensor data Correlation with filling quality for:

[0011] , is any minimum value, For sensor data The weighted summation result of .

[0012] Preferably, the sensor data The weighted sum of Satisfies the relationship:

[0013] ; is the total number of clusters, Cluster The number of sample data in , is the number of all sample data, and Clusters The first variance and the canned quality variance.

[0014] Each cluster can obtain an absolute value of the difference between the first variance and the canned mass variance. The larger the sample data in the cluster, the more the absolute value of the difference can truly reflect the sensor data. Therefore, the absolute value of the difference between each cluster is weighted by the number of sample data in the cluster to ensure the accuracy of the correlation.

[0015] Preferably, the time Fusion data for:

[0016] ; is the type of sensor data, For sensor data The correlation, is the sum of the correlations of all sensor data, For the moment Sensor data The dimensionless value of .

[0017] Preferably, before obtaining the real-time sequence of fused data, the monitoring method further comprises: performing denoising processing on the sensor data by using Gaussian filtering.

[0018] Preferably, the method for obtaining the standard sequence includes: taking sample sequences with filling qualities greater than a quality threshold as candidate sequences, calculating the average value of all candidate sequences at any time, and obtaining the standard sequence; the standard quality is the average filling quality of each candidate sequence.

[0019] Preferably, the global bias Satisfies the relationship: , is the number of time periods, Time period in real-time sequence The noise probability, For time period local deviation.

[0020] Global deviation eliminates the impact of noise data irrelevant to canning quality monitoring on the monitoring results, so that the global deviation can accurately reflect the difference between the real-time sequence and the standard sequence of the fused data. The greater the difference, the greater the possibility of abnormality in the filling process.

[0021] Preferably, the training of the binary classification model based on the cross-entropy function between the noise probability and the noise label includes: calculating the proportion of the number of training data in each time period in the total amount of training data, and taking the difference between 1 and the proportion as the normalized weight of the corresponding time period; calculating the cross-entropy function between the noise probability and the noise label, and training the binary classification model based on the product of the cross-entropy function and the normalized weight of the time period in the training data.

[0022] To ensure that the binary classification model can accurately obtain the noise probability of any time period, it is necessary to pay attention to whether the amount of training data in different time periods is balanced during the training process, and assign larger normalized weights to the training data corresponding to time periods with a smaller proportion, to alleviate the problem of unbalanced amount of training data in each time period, so that the binary classification model can accurately obtain the noise probability of any time period.

[0023] Preferably, the binary classification model is a fully connected neural network or a logistic classification model.

[0024] The technical solution of this application has the following beneficial technical effects:

[0025] During the processing and filling process, multiple sensor data can be collected at each moment. The sensor data at each moment are weighted according to the correlation between each sensor data and the filling quality to obtain the fused data at each moment of the processing and filling process, and then the real-time sequence of the fused data is obtained; the real-time sequence of the fused data is divided into multiple time periods with equal intervals, and the local deviations between the real-time sequence and the standard sequence in each time period, as well as the noise probability of each time period in the real-time sequence are calculated. The noise probability can reflect the degree of influence of the fluctuation of the real-time sequence in the corresponding time period on the filling quality. The greater the noise probability, the smaller the influence of the fluctuation of the real-time sequence in the corresponding time period on the filling quality, and the smaller the noise probability, the greater the influence of the fluctuation of the real-time sequence in the corresponding time period on the filling quality; the local deviations of each time period are weighted and summed according to the noise probability to obtain the global deviation. The global deviation eliminates the influence of noise data irrelevant to canning quality monitoring on the monitoring results. By comparing the global deviation and the global threshold, the monitoring results of processing and filling are accurately obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a flow chart of a processing and filling monitoring method based on multi-source sensor data according to an embodiment of the present application. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by technicians in this field without making creative efforts are within the scope of protection of this application.

[0028] According to a first aspect of the present application, the present application provides a processing and filling monitoring method based on multi-source sensor data. Figure 1 FIG. 1 is a flow chart of a method for monitoring processing and filling based on multi-source sensor data according to an embodiment of the present application. Figure 1As shown, the processing and filling monitoring method based on multi-source sensor data includes steps S101 to S103, which are described in detail below.

[0029] S101 , weighted fusion of sensor data during the filling process is performed based on the correlation between the sensor data and the filling quality to obtain a real-time sequence of fused data.

[0030] In one embodiment, after a historical canning process is completed, a historical sequence of sensor data in the historical canning process is collected, and the canning quality of the historical canning process is manually marked. The canning quality adopts a scoring system, and the canning quality value ranges from 0 to 10. The larger the value, the better the canning quality.

[0031] The sensor data includes filling speed, filling pressure, filling weight, etc.

[0032] In this way, each historical sequence of sensor data corresponds to a canning quality, and the canning quality is scored from 0 to 10. The historical sequence of each sensor data and the canning quality in a historical canning process are used as sample data, and the correlation between the sensor data and the filling quality can be calculated based on the sample data.

[0033] Specifically, sensor data The calculation method of the correlation between the filling quality and the sensor data includes: In addition, the DTW distance between the historical sequences of each sensor data is used as the clustering distance, and each sample data is clustered to obtain multiple clusters; within any cluster, the sensor data is calculated The average sequence of each historical sequence, sensor data The variance of the DTW distance between each historical sequence and the average sequence is taken as the first variance, and the absolute value of the difference between the first variance and the canned quality variance in the cluster is calculated; the absolute value of the difference of each cluster is weighted summed according to the proportion of sample data in the cluster, and the sensor data The correlation between filling quality and weighted summation is negatively correlated.

[0034] Among them, the Kmeans clustering algorithm can be used to cluster the sample data to obtain multiple clusters, and a cluster includes at least one sample data; it can be understood that among the multiple sample data in a cluster, the sensor data The first variance is used to characterize the sensor data in the cluster. The degree of difference in the historical sequence, the larger the first variance, the smaller the sensor data in the cluster The greater the difference between historical sequences, the greater the difference between canned quality variance is. Canned quality variance is used to characterize the difference in filling quality within the cluster. When the first variance is small and the canned quality variance is large, it means that the difference in filling quality is not caused by sensor data. The difference in historical sequence caused by sensor data The correlation with filling quality is small; when the first variance is large and the canning quality variance is small, it means that the sensor data Differences in historical sequences will not cause differences in filling quality, sensor data The correlation with filling quality is also small; when the first variance is basically consistent with the canned quality variance, it means that the sensor data Differences in historical sequences can cause differences in filling quality, sensor data The correlation with filling quality is large, so the correlation is negatively correlated with the absolute value of the difference between the first variance and the can quality variance.

[0035] Specifically, sensor data Correlation with filling quality for: , is any minimum value, For sensor data The weighted sum of . Among them, the minimum To avoid the situation where the denominator is 0, in the embodiment of the present application, the minimum value The value of is 0.1.

[0036] Among them, sensor data The weighted sum of Satisfies the relationship:

[0037] ; is the total number of clusters, Cluster The number of sample data in , is the number of all sample data, and Clusters The first variance and the canned quality variance.

[0038] So, when calculating sensor data and filling quality, based on sensor data In addition, the DTW distance between the historical sequences of each sensor data is used to cluster the sample data to obtain multiple clusters. The sensor data in the clusters The historical sequences of sensor data are basically the same; the sensor data are calculated in the cluster Correlation between filling quality and excluding sensor data The influence of other sensor data is eliminated to ensure the accuracy of the correlation.

[0039] In one embodiment, after obtaining the correlation between each sensor data and the filling quality, the sensor data at any time in the processing and filling process are weighted and fused according to the correlation to obtain the fused data at any time, and then obtain a real-time sequence of the fused data in the processing and filling process.

[0040] Specifically, the moment Fusion data for: ; is the type of sensor data, For sensor data The correlation, is the sum of the correlations of all sensor data, For the moment Sensor data The dimensionless value of the sensor data can be obtained by normalization or standardization to eliminate the dimension of the sensor data.

[0041] It should be noted that since the fused data is a weighted sum of the sensor data, and the sensor data will inevitably introduce environmental noise during the acquisition process, in order to ensure that the fused data can accurately represent the filling quality, before obtaining the real-time sequence of the fused data, the monitoring method also includes: using Gaussian filtering to denoise the sensor data.

[0042] In this way, the sensor data at any time in the filling process are weighted and fused according to the correlation to obtain the fused data at any time. The fused data can retain the sensor data that effectively characterizes the filling quality and obtain a real-time sequence of the fused data.

[0043] S102 , calculating the local deviation between the real-time sequence and the standard sequence in each time period, and obtaining the noise probability of each time period in the real-time sequence.

[0044] In one embodiment, since the real-time sequence is the result of weighted fusion of sensor data, changes in sensor data can cause fluctuations in the real-time sequence. However, not all fluctuations in the real-time sequence will affect filling quality. For example, in the early stages of chemical processing, since the chemical reagents in the can are relatively low, even large fluctuations in the filling speed will not significantly reduce the filling quality. Therefore, these fluctuations in filling speed will not affect filling quality. In other words, relative to filling quality monitoring, these fluctuations in filling speed are considered noise data. However, in the later stages of filling, since the chemical reagents in the can have essentially met the filling standards, even small fluctuations in filling speed will significantly reduce the filling quality. Therefore, these fluctuations in filling speed have a significant impact on filling quality. In other words, relative to filling quality monitoring, these fluctuations in filling speed are considered non-noise data.

[0045] In order to avoid the influence of noise data in the real-time sequence on canning quality inspection, it is necessary to detect the real-time sequence to obtain the noise level in each time period of the implementation sequence.

[0046] Specifically, obtaining the noise probability of each time period in the real-time sequence includes: obtaining a sample sequence and filling quality of fused data from a historical canning process, and dividing the sample sequence into multiple time periods at equal intervals; calculating the local deviation of each time period between the sample sequence and the standard sequence, and taking the time period in which the local deviation is greater than the local threshold as a difference segment; in response to the number of difference segments being 1, taking the timestamp of the difference segment and the difference sequence between the sample sequence and the standard sequence in the difference segment as a set of training data; in response to the absolute value of the difference between the filling quality and the standard quality being greater than the difference threshold, the noise label of the training data is 0, otherwise, the noise label of the training data is 1; inputting the training data into a binary classification model, outputting the noise probability, and training the binary classification model based on the cross-entropy function between the noise probability and the noise label; inputting the timestamp and difference sequence of each time period in the real-time sequence into the trained binary classification model, and outputting the noise probability of each time period in the real-time sequence.

[0047] The sample data consists of a historical sequence of sensor data and filling quality from a single filling process. Based on the correlation between sensor data and filling quality, the historical sequences are weighted and fused to produce a sample sequence of fused data. This means that a single fused data sample sequence can be collected from a single filling process. Because a filling process lasts the same length, each sample sequence has the same length as the real-time sequence and contains the same number of time segments.

[0048] The local deviation is the Euclidean distance or DTW distance between the sample sequence and the standard sequence in the corresponding time period; the local threshold and the difference threshold are both 1; compared with the standard sequence, only the difference segment exists in the sample sequence, and when the absolute value of the difference between the filling quality and the standard quality is greater than the difference threshold, it means that the difference segment can cause a change in the filling quality, and the difference segment is non-noise data with respect to filling quality monitoring. Therefore, when the absolute value of the difference between the filling quality and the standard quality is greater than the difference threshold, the noise label of the training data is 0; similarly, when the absolute value of the difference between the filling quality and the standard quality is not greater than the difference threshold, it means that the difference segment does not cause a change in the filling quality, and the difference segment is noise data with respect to filling quality monitoring, and the noise label of the training data is 1.

[0049] The binary classification model is used to map the timestamp and difference sequence of any time period into a classification result of whether it is noise or not. The binary classification model can be an existing binary classification model such as a fully connected neural network or a logistic classification model. To ensure that the binary classification model can accurately obtain the noise probability of any time period, it is necessary to pay attention to whether the number of training data in different time periods is balanced during the training process. For example, if the difference segments in all training data are within the time period , then the trained binary classification model can accurately obtain the time period The noise probability, and for the time period However, the accuracy of the noise probability cannot be guaranteed for other time periods outside the time period. Therefore, the difference in the amount of training data in different time periods needs to be considered in the process of training the binary classification model. The binary classification model training based on the cross-entropy function between the noise probability and the noise label includes: calculating the proportion of the number of training data in each time period in the total amount of training data, and taking the difference between 1 and the proportion as the normalized weight of the corresponding time period; calculating the cross-entropy function between the noise probability and the noise label, and training the binary classification model based on the product of the cross-entropy function and the normalized weight of the time period in the training data.

[0050] In this way, when training the binary classification model, a larger normalized weight is assigned to the training data corresponding to the time period with a smaller proportion, which alleviates the problem of uneven number of training data in each time period and enables the binary classification model to accurately obtain the noise probability of any time period.

[0051] In other embodiments, sample data can be obtained by experimental methods to obtain training data for each time period. For example, Training data, adjust the time period in the standard sequence sensor data, so that the sample sequence is in the time period Fluctuations occur, i.e. time periods is the difference segment, thus obtaining the time period training data in each time period, thereby alleviating the problem of unbalanced amount of training data in each time period.

[0052] It should be noted that the standard sequence is a time series of fused data corresponding to canning quality that meets quality requirements; it can be selected by those skilled in the art from historical canning processes, and the standard quality is the filling quality corresponding to the standard sequence. In other embodiments, the method for obtaining the standard sequence includes: selecting sample sequences with filling quality greater than a quality threshold as candidate sequences, calculating the average value of all candidate sequences at any time, and obtaining the standard sequence; the standard quality is the average filling quality of each candidate sequence.

[0053] In this way, the noise probability of each time period in each real-time sequence is obtained based on the timestamp of each time period and the difference sequence between the real-time sequence and the standard sequence within the time period, so as to achieve accurate measurement of the noise probability of each time period. The larger the noise probability, the smaller the impact of the fluctuation of the corresponding time period on the filling quality. When obtaining the monitoring results, the local deviation of the corresponding time period should be ignored to improve the accuracy of the monitoring results.

[0054] S103 , weighted summing of the local deviations in each time period according to the noise probability to obtain a global deviation. If the global deviation is greater than a global threshold, the monitoring result is abnormal, otherwise the monitoring result is normal.

[0055] In one embodiment, the global deviation Satisfies the relationship: , is the number of time periods, Time period in real-time sequence The noise probability, For time period local deviation.

[0056] In this way, the global deviation eliminates the impact of noise data irrelevant to canning quality monitoring on the monitoring results, so that the global deviation can accurately reflect the difference between the real-time sequence and the standard sequence of the fused data. The greater the difference, the greater the possibility of abnormality in the processing and filling process.

[0057] The global deviation is greater than the global threshold. If the global deviation is greater than the global threshold, the greater the degree to which the real-time sequence of the fused data deviates from the standard sequence during the filling process, the monitoring result is abnormal. Otherwise, the monitoring result is normal. The global threshold is 2.

[0058] In this way, the canning quality during the processing and filling process is monitored. When the monitoring result is abnormal, it means that the canning equipment needs to be repaired to ensure that better filling quality can be obtained in the next processing and filling process.

[0059] It should be noted that a person skilled in the art may make a number of modifications and improvements without departing from the concept of the present application, and these modifications and improvements are all within the scope of protection of the present application. Therefore, the scope of protection of the patent application shall be based on the appended claims.

Claims

1. A processing and filling monitoring method based on multi-source sensor data, characterized in that: The monitoring method includes: weighted fusion of sensor data during the filling process based on the correlation between the sensor data and filling quality to obtain a real-time sequence of fused data; Sensor data The calculation method of the correlation between filling quality and filling quality includes: Sensor data In addition, the DTW distance between the historical sequences of each sensor data is used as the clustering distance to cluster each sample data to obtain multiple clusters. The sample data includes the historical sequence of each sensor data in the historical canning process and the canning quality; Within any cluster, calculate sensor data The average sequence of each historical sequence, sensor data The variance of the DTW distance between each historical sequence and the average sequence is taken as the first variance, and the absolute value of the difference between the first variance and the canned quality variance in the cluster is calculated; The absolute value of the difference of each cluster is weighted summed according to the proportion of sample data in the cluster. The correlation between filling quality and weighted summation results is negatively correlated; Calculating the local deviation of the real-time sequence and the standard sequence in each time period, and obtaining the noise probability of each time period in the real-time sequence, including: obtaining the sample sequence and filling quality of the fused data of the historical canning process, and dividing the sample sequence into multiple time periods at equal intervals; calculating the local deviation of each time period between the sample sequence and the standard sequence, and taking the time period in which the local deviation is greater than the local threshold as a difference segment, and in response to the number of difference segments being 1, taking the timestamp of the difference segment and the difference sequence between the sample sequence and the standard sequence in the difference segment as a set of training data; in response to the absolute value of the difference between the filling quality and the standard quality being greater than the difference threshold, the noise label of the training data is 0, and conversely, the noise label of the training data is 1; inputting the training data into a binary classification model, outputting the noise probability, and training the binary classification model based on the cross entropy function between the noise probability and the noise label; inputting the timestamp and difference sequence of each time period in the real-time sequence into the trained binary classification model, and outputting the noise probability of each time period in the real-time sequence; The global deviation is obtained by weighted summing of the local deviations in each time period according to the noise probability. If the global deviation is greater than the global threshold, the monitoring result is abnormal, otherwise the monitoring result is normal.

2. The processing and filling monitoring method based on multi-source sensor data according to claim 1 is characterized in that: Sensor data Correlation with filling quality for: , is any minimum value, For sensor data The weighted summation result of .

3. The processing and filling monitoring method based on multi-source sensor data according to claim 1 is characterized in that: Sensor data The weighted sum of Satisfies the relationship: ; is the total number of clusters, Cluster The number of sample data in , is the number of all sample data, and Clusters The first variance and the canned quality variance.

4. The processing and filling monitoring method based on multi-source sensor data according to claim 1 is characterized in that: time Fusion data for: ; is the type of sensor data, For sensor data The correlation, is the sum of the correlations of all sensor data, For the moment Sensor data The dimensionless value of .

5. The processing and filling monitoring method based on multi-source sensor data according to claim 1 is characterized in that: Before obtaining the real-time sequence of fused data, the monitoring method further includes: Gaussian filtering is used to denoise the sensor data.

6. The processing and filling monitoring method based on multi-source sensor data according to claim 1 is characterized in that: The method for obtaining the standard sequence includes: The sample sequences whose filling quality is greater than the quality threshold are taken as candidate sequences, and the average value of all candidate sequences at any time is calculated to obtain the standard sequence; The standard quality is the average filling quality of each candidate sequence.

7. The processing and filling monitoring method based on multi-source sensor data according to claim 1 is characterized in that: Global bias Satisfies the relationship: , is the number of time periods, Time period in real-time sequence The noise probability, For time period local deviation.

8. The processing and filling monitoring method based on multi-source sensor data according to claim 1 is characterized in that: The training of the binary classification model based on the cross entropy function between noise probability and noise labels includes: Calculate the proportion of training data in each time period to the total training data, and use the difference between 1 and the proportion as the normalized weight of the corresponding time period; Calculate the cross entropy function between the noise probability and the noise label, and train a binary classification model based on the product of the cross entropy function and the normalized weight of the time segment in the training data.

9. The processing and filling monitoring method based on multi-source sensor data according to claim 1 is characterized in that: The binary classification model is a fully connected neural network or a logical classification model.

Images