Dehydration device condition monitoring method and system
By using data segmentation, fusion, and clustering methods, a state prediction model for different operating conditions was established, which solved the problem of low state detection accuracy of the dehydration device, achieved adaptive high-precision state detection, and enhanced safety risk prevention.
Patent Information
- Application Number
- CN202011415064.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2020-12-04
- Publication Date
- 2025-10-31
- Estimated Expiration
- 2040-12-04
AI Technical Summary
In existing technologies, the accuracy of status detection for dehydration devices is not high, and they cannot adapt to their complex operating conditions, resulting in insufficient prevention of safety risks.
By employing data segmentation, fusion, and clustering methods, and by acquiring the operating condition types, a state prediction model for different operating conditions is established. The state of the dehydration device is determined by the similarity between the new data and the already classified data.
It improves the accuracy of the status detection of the dehydration device, can adaptively determine the operating conditions, solves the problem of inaccurate detection accuracy, and enhances the ability to prevent safety risks.
Smart Images
Figure CN114596458B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of safety risk detection technology, and in particular to a method and system for detecting the status of a dehydration device. Background Technology
[0002] Dehydration units are key equipment in gas field water transmission pipelines. Monitoring the condition of dehydration units can prevent safety risks, and data-driven monitoring of dehydration units has received widespread attention.
[0003] In related technologies, a predictive model for the state of a dehydration device is established using existing data, and then the model is directly used to detect the state of the dehydration device. However, because the state of a dehydration device during operation is complex, a single model cannot be used to generalize, resulting in low detection accuracy for these technologies. Summary of the Invention
[0004] This application provides a method and system for detecting the state of a dehydration device, which is used to detect the state of the dehydration device and thus prevent safety risks.
[0005] On one hand, embodiments of this application provide a method for detecting the state of a dehydration device. The method includes: initializing and segmenting existing data to obtain segmented data fragments; obtaining fused data fragments based on the segmented data fragments; obtaining operating condition types based on the fused data fragments; and detecting the state of the dehydration device based on the operating condition types.
[0006] In one possible implementation, the initial segmentation of existing data to obtain segmented data fragments includes: standardizing the existing data to obtain processed data feature vectors; performing mean filtering on the processed data feature vectors to obtain filtered data; and segmenting the filtered data to obtain segmented data fragments.
[0007] In one possible implementation, obtaining the fused data segment based on the segmented data segments includes: obtaining a loss function for each data segment based on the segmented data segments; and fusing the data segments based on the loss function of each data segment to obtain the fused data segment.
[0008] In one possible implementation, obtaining the operating condition type based on the fused data segments includes: obtaining the distance between the fused data segments; clustering the fused data segments based on the distance between the fused data segments to obtain clustered data segments; and obtaining the operating condition type based on the clustered data segments.
[0009] In one possible implementation, detecting the state of the dehydration device based on the operating condition type includes: establishing a state prediction model for each operating condition type; acquiring central data for each operating condition type; inputting a new data segment, which is collected by the dehydration device; acquiring a state prediction model for the new data segment based on the new data segment, the central data for each operating condition type, and the state prediction model for each operating condition type; and detecting the state of the dehydration device based on the state prediction model for the new data segment.
[0010] On the other hand, embodiments of this application provide a status detection system for a dehydration device. The system includes: an initialization module for initializing and segmenting existing data to obtain segmented data fragments; a fusion module for obtaining fused data fragments based on the segmented data fragments; a clustering module for obtaining operating condition types based on the fused data fragments; and a prediction module for detecting the status of the dehydration device based on the operating condition types.
[0011] In one possible implementation, the initialization module is used to standardize existing data to obtain a processed data feature vector; perform mean filtering on the processed data feature vector to obtain filtered data; and perform data segmentation on the filtered data to obtain segmented data fragments.
[0012] In one possible implementation, the fusion module is configured to obtain a loss function for each data segment based on the segmented data segments; and to fuse the data segments based on the loss function of each data segment to obtain a fused data segment.
[0013] In one possible implementation, the clustering module is used to obtain the distance between the fused data segments; based on the distance between the fused data segments, cluster the fused data segments to obtain clustered data segments; and based on the clustered data segments, obtain the operating condition type.
[0014] In one possible implementation, the prediction module is configured to: establish a state prediction model for each operating condition type based on the operating condition type; acquire central data for each operating condition type; input a new data segment, which is collected by the dehydration device; acquire a state prediction model for the new data segment based on the new data segment, the central data for each operating condition type, and the state prediction model for each operating condition type; and detect the state of the dehydration device based on the state prediction model for the new data segment.
[0015] This application also provides a computer-readable storage medium having at least one program instruction or code, which, when loaded and executed by a processor, enables the computer to implement the state detection method of the dehydration device as described above.
[0016] The technical solution provided in this application includes at least the following beneficial effects: 1. It uses data segmentation technology, and on this basis, uses fusion and clustering methods to obtain the working condition type, ensuring that the working condition can be adaptively determined by calculation after data clustering without specifying the working condition. 2. It establishes models for different working conditions. When new data is acquired, the model used is determined according to the similarity between the new data and the already classified data, thereby determining the state of the dehydration device, solving the problem of complex working states and inaccurate detection accuracy of the dehydration device. Attached Figure Description
[0017] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.
[0018] Figure 1 This is a flowchart of a state detection method for a dehydration device provided in one embodiment of this application;
[0019] Figure 2 This is a schematic diagram of the status detection system of a dehydration device provided in one embodiment of this application. Detailed Implementation
[0020] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be described in further detail below with reference to the accompanying drawings.
[0021] Please refer to Figure 1 The diagram illustrates a flowchart of a state detection method for a dehydration apparatus according to an embodiment of this application. The method includes the following steps:
[0022] Step 101: Initialize the segmentation of the existing data and obtain the segmented data fragments.
[0023] In one possible implementation, the existing data is initialized and segmented, and the segmented data fragments are obtained, including but not limited to the following sub-steps:
[0024] 1011. Standardize the existing data and obtain the processed data feature vector.
[0025] Here, "existing data" refers to data that has already been obtained. Optionally, existing data can be obtained through sensors installed in the dehydration device. Optionally, based on the existing data, a vector composed of the features of the existing data is obtained, i.e., the existing data feature vector. Further, optionally, the existing data feature vector is processed according to the following formula to calculate the processed data feature vector:
[0026]
[0027] Where, x old Given the existing data feature vector, x new Let μ be the mean of the existing data vector, and σ be the standard deviation of the existing data vector. The standard deviation is the square root of the arithmetic mean (i.e., variance) of the squared deviations from the mean, representing the dispersion of a dataset. The standard deviation is calculated based on the existing data vector.
[0028] 1012. Apply mean filtering to the processed data feature vectors to obtain filtered data.
[0029] Optionally, the processed data feature vector can be subjected to mean filtering using a sliding window to obtain filtered data.
[0030] 1013. Perform data segmentation on the filtered data to obtain the segmented data fragments.
[0031] Optionally, a bottom-up data segmentation method is used to segment the filtered data into multiple small segments until the number of segments reaches the first segment size, thus obtaining the segmented data segments. Further, optionally, the number of segments in the first segment is set based on experience.
[0032] Step 102: Based on the segmented data fragments, obtain the fused data fragments.
[0033] In one possible implementation, obtaining the fused data fragment based on the segmented data fragments includes, but is not limited to, the following sub-steps:
[0034] 1021. Based on the segmented data fragments, obtain the loss function for each data fragment.
[0035] Optionally, for each segmented data fragment, the loss function is calculated using the following formula:
[0036]
[0037] Where cost represents the loss function for each data segment. Q This represents the matrix reconstruction error corresponding to each data segment. This represents the Hotelling statistical loss for each data segment. Further, optionally, the matrix reconstruction error cost for each data segment is determined using the following formula. Q :
[0038]
[0039]
[0040] Where I represents a unit vector, U i,p This represents the eigenvector matrix obtained by treating each segmented data point as a multivariate time series segment and reducing the dimensionality of the autocovariance matrix of the multivariate time series segment. For example, Principal Component Analysis (PCA) can be used as a dimensionality reduction method. i (a i b i ) represents each segmented data piece, i.e., each segmented multivariate time series, a i b i This indicates that the segmented multivariate time series is composed of the a-th... i From point b to the bth i One point.
[0041] Optionally, the Hotelling statistical loss cost for each data segment can be determined according to the following formula. T 2 :
[0042]
[0043] Among them, S i (a i b i ) represents each segmented data piece, i.e., each segmented multivariate time series, a i b i This indicates that the segmented multivariate time series is composed of the a-th... i From point b to the bth i One point. Determined according to the following formula.
[0044]
[0045] Where, x k This represents the data vector collected at the k-th time point in each segmented multivariate time series. This means dividing the matrix by its eigenvalues in each dimension.
[0046] 1022. Based on the loss function of each data segment, fuse the data segments to obtain the fused data segment.
[0047] Optionally, based on the loss function of each data segment, the data segments are fused to obtain fused data segments, including but not limited to:
[0048] 1022-1. Based on the loss function of each data segment, search for the data segment with the smallest loss function and select it as the first data segment;
[0049] 1022-2. Based on the data segment with the smallest loss function, compare the loss functions of the two segments to the left and right of the data segment, and take the data segment with the smaller loss function as the second data segment;
[0050] 1022-3. Merge the first data segment with the second data segment to obtain the merged partial data segment and the unmerged remaining data segment;
[0051] 1022-4. Repeat steps 1022-1 to 1022-3 for the merged data segments and the remaining unmerged data segments until they are merged into the second number of segments, resulting in merged data segments. Optionally, the second number of segments represents the total number of final merged data segments, set based on experience.
[0052] Step 103: Obtain the operating condition type based on the fused data fragments.
[0053] In one possible implementation, the operating condition type is obtained based on the fused data fragments, including but not limited to the following sub-steps:
[0054] 1031. Obtain the distance between the merged data segments.
[0055] Optionally, the distance d(X) between the fused data segments can be obtained according to the following formula. n ,Y n ):
[0056]
[0057] Among them, X n Y n These represent two fused data segments. Alternatively, they can be calculated using the following formula:
[0058]
[0059] in, Representing X n Y n The feature values θ of the two fused data segments i,j Represents X n The i-th principal direction and Y nThe angle between the j-th principal directions.
[0060] Alternatively, other appropriate distance measurement methods can be selected based on the characteristics of different data segments, and this application does not limit this.
[0061] 1032. Based on the distance between the fused data segments, cluster the fused data segments to obtain clustered data segments.
[0062] The fused data segments are treated as multivariate time series. Optionally, density clustering methods for multivariate time series can be used to cluster the fused data segments.
[0063] For example, based on the distance between the fused data fragments, a density-based clustering method with noise (DBSCAN) is used to cluster the fused data fragments to obtain clustered data fragments.
[0064] 1033. Obtain the working condition type based on the clustered data fragments.
[0065] Each clustered data segment is treated as a working condition type, resulting in different working condition types.
[0066] Step 104: Detect the status of the dehydration device based on the operating condition type.
[0067] In one feasible method, the status of the dehydration device is detected based on the operating condition type, including but not limited to the following sub-steps:
[0068] 1041. Based on the operating condition type, establish a state prediction model for each operating condition type.
[0069] Optionally, different state prediction models can be established for different operating conditions. In one example, the established state prediction model is a Hidden Markov Model, which can predict the value of each feature at the next moment based on the value of each feature at the previous moment.
[0070] Optionally, depending on actual needs, the state prediction model can also be a fault diagnosis model or other prediction model, such as various prediction and diagnosis models based on neural networks or other machine learning methods, etc. This application embodiment does not limit this.
[0071] 1042. Obtain central data for each operating condition type.
[0072] In one example, obtaining the central data for each working condition type includes: within the same working condition, performing a cubic spline dynamic time warping (CDTW) algorithm on the same feature at different time periods to obtain the central data for each working condition type.
[0073] Optionally, this example includes, but is not limited to, the following sub-steps:
[0074] 1042-1 Under the same working condition, according to the time sequence, the data segments corresponding to the first two time periods are taken as the first data segment and the second data segment, respectively;
[0075] 1042-2 uses the dynamic time warping (DTW) algorithm on the same feature of the first and second data segments to obtain a two-dimensional coordinate path;
[0076] 1042-3 Take the average value of the two-dimensional coordinate path, set a new sequence, use the average value of the obtained two-dimensional coordinate path as the sequence index, and use the average value of the corresponding value of the sequence index as the sequence value to obtain the new sequence;
[0077] 1042-4. The new sequence is interpolated using cubic spline curves, and then sampled starting from 0, incrementing by 1 each time, with the number of sampling points matching the number of indices, to obtain new sampled data points;
[0078] 1042-5 takes the newly sampled data points as the first data segment, takes the data segments corresponding to other time periods within the same working condition as the second data segment, and performs CDTW operation in the same way as above and repeats it until only a univariate sequence is obtained. For each feature, CDTW operation is also performed in the same way as above until a unique multivariate time series is obtained, which is called the central data. Each working condition corresponds to one such central data.
[0079] For example, for two sequences with the same feature: x = [2,0,1,1,2,4,2,1,2,0] and y = [1,1,2,4,2,1,2,0], performing a DTW operation on these two sequences yields a two-dimensional coordinate path: [(0,0), (1,0), (2,0), (3,1), (4,2), (5,3), (6,4), (7,5), (8,6), (9,7)]. This two-dimensional coordinate path starts at (0,0). Taking the average of this two-dimensional coordinate path yields: [0,0.5,1,2,3,4,5,6,7,8]. Using this average as the sequence index, the mean of the corresponding values is: [1.5,0.5,1,1,2,4,2,1,2,0]. The new sequence is interpolated using cubic spline curves. Then, starting from 0, sampling is performed by incrementing by 1 each time, with the number of sampling points matching the number of indices. That is, the values corresponding to [0, 1, 2, 3, 4, 5, 6, 7, 8, 9] are sampled to obtain new sampled data points. These new sampled data points are used as the first data segment. Data segments corresponding to other time periods within the same working condition are taken as the second data segment. The same CDTW operation is performed using the same method and repeated until only a univariate sequence is obtained. The same CDTW operation is performed on each feature until a unique multivariate time series is obtained. In this way, the central data for a certain working condition is obtained.
[0080] Optionally, the mean multivariate sequence method can also be used to obtain the center data for each working condition type, but this application embodiment does not limit this.
[0081] 1043. Input a new data segment. The new data segment is collected by the dehydration device.
[0082] 1044. Based on the new data segment, the central data of each working condition type, and the state prediction model of each working condition type, obtain the state prediction model of the new data segment.
[0083] Optionally, the distance between the new data segment and the central data of each working condition type is obtained in the same way as described above, wherein the length of the new data segment is the average of the lengths of all central data segments; based on the distance between the new data segment and the central data of each working condition type, the new data segment is classified, the working condition to which the new data segment belongs is obtained, and the state prediction model of the working condition is selected as the state prediction model of the new data segment.
[0084] 1045. Based on the state prediction model of the new data segment, detect the state of the dehydration device.
[0085] Optionally, based on the state prediction model of the new data segment, the new data segment is predicted to determine the predicted state of the new data segment, and the predicted state of the new data segment is used as the state detection result of the dehydration device.
[0086] This application uses PCA data segmentation technology and, based on this, employs a density-based clustering method to obtain the operating condition type. This ensures that after data clustering, the operating condition does not need to be specified, and the operating condition can be adaptively determined through calculation. This application also establishes state prediction models for different operating conditions. When new data is acquired, the similarity between the new data and the already classified data is calculated, and then the prediction model to be used is determined to ascertain the state of the dewatering device. This solves the problem of complex operating states and inaccurate detection accuracy of the dewatering device.
[0087] Please refer to Figure 2 The diagram illustrates a state detection system 20 for a dehydration apparatus according to an embodiment of this application. The system includes the following modules:
[0088] The initialization module 201 is used to initialize the segmentation of existing data and obtain the segmented data fragments. Optionally, the initialization module 201 is used for, but not limited to: standardizing the existing data to obtain the processed data feature vector; performing mean filtering on the processed data feature vector to obtain filtered data; and segmenting the filtered data to obtain the segmented data fragments.
[0089] The fusion module 202 is used to obtain fused data segments based on the segmented data segments. Optionally, the fusion module 202 is used for, but not limited to: obtaining the loss function of each data segment based on the segmented data segments; and fusing the data segments based on the loss function of each data segment to obtain the fused data segments.
[0090] Clustering module 203 is used to obtain the operating condition type based on the fused data fragments. Optionally, clustering module 203 is used for, but not limited to: obtaining the distance between the fused data fragments; clustering the fused data fragments based on the distance between the fused data fragments to obtain clustered data fragments; and obtaining the operating condition type based on the clustered data fragments.
[0091] The prediction module 204 is used to detect the status of the dewatering device based on the operating condition type. Optionally, the prediction module 204 is used for, but not limited to: establishing a status prediction model for each operating condition type; acquiring central data for each operating condition type; inputting a new data segment, which is collected by the dewatering device; acquiring a status prediction model for the new data segment based on the new data segment, the central data for each operating condition type, and the status prediction model for each operating condition type; and detecting the status of the dewatering device based on the status prediction model for the new data segment.
[0092] It should be understood that the above Figure 2The system provided is illustrated using the above-described division of functional modules as an example. In practical applications, the functions described above can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. Furthermore, the system and method embodiments provided in the above examples belong to the same concept, and their specific implementation processes are detailed in the method embodiments, which will not be repeated here.
[0093] In an exemplary embodiment, a computer-readable storage medium is also provided, which stores at least one piece of program code, which is loaded and executed by a processor to enable a computer to implement the state detection method of any of the above-described dehydration apparatuses.
[0094] Optionally, the aforementioned computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), magnetic tape, floppy disk, and optical data storage device, etc.
[0095] In an exemplary embodiment, a computer program or computer program product is also provided, which stores at least one computer instruction, which is loaded and executed by a processor to enable the computer to implement the state detection method of any of the above-described dehydration devices.
[0096] In the embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices or modules, or may be electrical, mechanical or other forms of connection.
[0097] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of the embodiments of this application, depending on actual needs.
[0098] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.
[0099] It should also be understood that, in the various embodiments of this application, the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.
[0100] In this application, the term "at least one" means one or more, and the term "multiple" means two or more. For example, multiple data means two or more data.
[0101] It should be understood that the terminology used in the description of the various examples herein is for the purpose of describing particular examples only and is not intended to be limiting. As used in the description of the various examples and the appended claims, the singular forms “a” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise.
[0102] The above are merely exemplary embodiments of this application and are not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application shall be included within the protection scope of this application.
Claims
1. A method for detecting the state of a dehydration device, characterized in that, The method includes: Initialize the segmentation of existing data and obtain the segmented data fragments; Based on the segmented data fragments, obtain the fused data fragments; Based on the fused data fragments, the operating condition type is obtained; Based on the aforementioned operating condition type, detect the status of the dehydration device; The initial segmentation of existing data to obtain segmented data fragments includes: The existing data is standardized according to the following formula to obtain the processed data feature vector; Where, x old Given the existing data feature vector, x new Let μ be the mean of the existing data vector, and σ be the standard deviation of the existing data vector. The standard deviation is the square root of the arithmetic mean of the squares of the deviations from the mean, representing the dispersion of a dataset. The standard deviation is calculated based on the existing data vector. The processed data feature vector is subjected to mean filtering to obtain filtered data; The filtered data is segmented to obtain segmented data fragments; The detection of the status of the dehydration device based on the operating condition type includes: Based on the aforementioned operating conditions, state prediction models for each operating condition type are established. Within the same working condition, according to the time sequence, the data segments corresponding to the first two time periods are taken as the first data segment and the second data segment, respectively. A two-dimensional coordinate path is obtained by applying a dynamic time planning algorithm to the same feature of the first data segment and the second data segment; Take the average value of the two-dimensional coordinate path, set a new sequence, use the average value of the two-dimensional coordinate path as the sequence index, and use the mean value of the corresponding value of the sequence index as the sequence value to obtain the new sequence; The new sequence is interpolated using a cubic spline curve, and then sampled starting from 0, incrementing by 1 each time, with the number of sampling points matching the number of indices, to obtain new sampled data points; The newly sampled data points are taken as the first data segment, and the data segments corresponding to other time periods within the same working condition are taken as the second data segment. The cubic spline dynamic time programming algorithm is used and repeated until only one univariate sequence is obtained. For each feature, the cubic spline dynamic time programming algorithm is used until a unique multivariate time series is obtained as the central data. Each working condition corresponds to one central data. A new data segment is input, which is acquired through a dehydration device; Based on the new data segment, the central data of each working condition type, and the state prediction model of each working condition type, obtain the state prediction model of the new data segment. The state prediction model based on the new data segment is used to detect the state of the dehydration device.
2. The method according to claim 1, characterized in that, The step of obtaining the fused data segment based on the segmented data segment includes: Based on the segmented data fragments, obtain the loss function for each data fragment; Based on the loss function of each data segment, the data segments are fused to obtain the fused data segment.
3. The method according to claim 1, characterized in that, The process of obtaining the operating condition type based on the fused data fragments includes: Obtain the distance between the fused data segments; Based on the distance between the fused data segments, the fused data segments are clustered to obtain clustered data segments; Based on the clustered data fragments, the operating condition type is obtained.
4. A status monitoring system for a dehydration device, characterized in that, The system includes: The initialization module is used to initialize and segment existing data and obtain the segmented data fragments. The fusion module is used to obtain fused data segments based on the segmented data segments; The clustering module is used to obtain the working condition type based on the fused data fragments; The prediction module is used to detect the status of the dehydration device based on the operating condition type. The initialization module is used to standardize the existing data according to the following formula to obtain the processed data feature vector; Where, x old Given the existing data feature vector, x new Let μ be the mean of the existing data vector, and σ be the standard deviation of the existing data vector. The standard deviation is the square root of the arithmetic mean of the squares of the deviations from the mean, representing the dispersion of a dataset. The standard deviation is calculated based on the existing data vector. The processed data feature vector is subjected to mean filtering to obtain filtered data; The filtered data is segmented to obtain segmented data fragments; The prediction module is used for: Based on the aforementioned operating conditions, state prediction models for each operating condition type are established. Within the same working condition, according to the time sequence, the data segments corresponding to the first two time periods are taken as the first data segment and the second data segment, respectively. A two-dimensional coordinate path is obtained by applying a dynamic time planning algorithm to the same feature of the first data segment and the second data segment; Take the average value of the two-dimensional coordinate path, set a new sequence, use the average value of the two-dimensional coordinate path as the sequence index, and use the mean value of the corresponding value of the sequence index as the sequence value to obtain the new sequence; The new sequence is interpolated using a cubic spline curve, and then sampled starting from 0, incrementing by 1 each time, with the number of sampling points matching the number of indices, to obtain new sampled data points; The newly sampled data points are taken as the first data segment, and the data segments corresponding to other time periods within the same working condition are taken as the second data segment. The cubic spline dynamic time programming algorithm is used and repeated until only one univariate sequence is obtained. For each feature, the cubic spline dynamic time programming algorithm is used until a unique multivariate time series is obtained as the central data. Each working condition corresponds to one central data. A new data segment is input, which is acquired through a dehydration device; Based on the new data segment, the central data of each working condition type, and the state prediction model of each working condition type, obtain the state prediction model of the new data segment. The state prediction model based on the new data segment is used to detect the state of the dehydration device.
5. The system according to claim 4, characterized in that, The fusion module is used to obtain a loss function for each data segment based on the segmented data segments; and to fuse the data segments based on the loss function of each data segment to obtain a fused data segment.
6. The system according to claim 4, characterized in that, The clustering module is used to obtain the distance between the fused data segments; based on the distance between the fused data segments, the fused data segments are clustered to obtain clustered data segments; and based on the clustered data segments, the operating condition type is obtained.