Cloud platform maintenance method for health state of feeding system equipment

Abstract

The invention discloses a cloud platform maintenance method for the health state of feeding system equipment, and the method comprises the following steps: S1, collecting an output signal of a weighing sensor in a unit time range, and relates to the technical field of industrial Internet of Things equipment state monitoring. According to first time sequence data formed by collecting output signals of a weighing sensor and sorting according to time, second time sequence data formed by monitoring and obtaining near-real-time geomagnetic activity data and sorting according to time, normalization and segmented aggregation approximate processing are firstly carried out on the two types of data to obtain dimension reduction data fragments, and the dimension reduction data fragments are subjected to dimension reduction processing; after combining into a joint data sample set, selecting a selected value based on cumulative probability distribution and an information entropy maximization principle, and generating a corresponding sensor symbol sequence and a geomagnetic symbol sequence; and according to symbol matching scores, mismatching score penalty and vacancy score penalty rules, a scoring matrix is constructed by adopting a dynamic programming algorithm, and common-mode interference caused by geomagnetic storm to a weighing sensor in a feeding system is identified by combining comprehensive confidence scores.

Description

Technical Field

[0001] This invention relates to the field of industrial Internet of Things (IoT) equipment status monitoring technology, and in particular to a cloud platform maintenance method for the health status of a material feeding system. Background Technology

[0002] In industrial applications such as material feeding systems, ultra-high precision load cells, such as loss-in-weight scales, are core control components. Based on the principle of electromagnetic force compensation, they are extremely sensitive to even weak electromagnetic interference. Geomagnetic storms cause drastic changes in the Earth's magnetic field, which induces extremely weak but persistent eddy currents and potential differences in the sensor's internal wiring loops and metal structures. This interference signal highly overlaps with the electrical signal generated by changes in material weight across the entire frequency band, inducing quasi-DC, global common-mode interference within the sensor that overlaps with the actual weight signal frequency band. This interference causes a consistent drift in readings from all similar sensors, manifesting as a quasi-DC drift with a frequency below 1Hz. The system may mistakenly interpret this as a genuine, synchronous change in material weight. This renders traditional redundancy verification mechanisms, such as two-out-of-three voting, ineffective, potentially leading to misjudgments in the control system and causing production accidents.

[0003] Currently, in order to mitigate the impact of geomagnetic storms, macroscopic geomagnetic activity data streams such as the Kp index are directly introduced for correlation judgment. However, the Kp index is global, and due to the existence of disturbances from the global index to the local area, it cannot accurately reflect the instantaneous magnetic field change rate of the factory location. The instantaneous magnetic field change rate is the direct physical quantity that interferes with the sensor. The Kp index has multiple uncertainties from physical disturbances to electrical signal response, resulting in poor effectiveness. The false alarm rate of alarms relying on the Kp index needs to be reduced. Summary of the Invention

[0004] To address the technical problems existing in the background art, this invention proposes a cloud platform maintenance method for the health status of feeding system equipment.

[0005] The present invention proposes a cloud platform maintenance method for the health status of a feeding system, comprising the following steps: S1. Within a unit time range: collect the output signal of the weighing sensor and arrange the output signal of the weighing sensor in chronological order to form the first time series data; monitor and acquire near real-time geomagnetic activity data and arrange the geomagnetic activity data in chronological order to form the second time series data; Near-real-time geomagnetic activity data refers to geomagnetic activity data with an update frequency of no more than 3 hours; A unit of time refers to a time range of a fixed length, such as within a day, within 6 hours, or within 3 hours. S2. Generate the corresponding sensor symbol sequence and geomagnetic symbol sequence based on the first time series data and the second time series data; S3. Perform local sequence alignment between the sensor symbol sequence and the geomagnetic symbol sequence to obtain the comprehensive confidence score; S4. When the overall confidence score exceeds the overall confidence threshold, an alarm message is generated.

[0006] Preferably, in S2, the first time-series data and the second time-series data are discretized to generate corresponding sensor symbol sequences and geomagnetic symbol sequences, as follows: S11. Normalize the first time series data and the second time series data, and perform segmented aggregation approximation processing on the normalized data to obtain the first dimensionality-reduced data segment corresponding to the first time series data and the second dimensionality-reduced data segment corresponding to the second time series data. S12. Merge the first and second dimensionality-reduced data segments to form a joint data sample set; S13. Obtain the cumulative probability distribution of all data points in the joint data sample set; S14. Predefine a candidate alphabet size set containing K candidate values, where K is an integer greater than 1, each candidate value corresponds to an integer greater than 1, and any two candidate values ​​are not equal. For each candidate value in the candidate alphabet size set, a selected value is generated based on the cumulative probability distribution of all data points in the joint data sample set. Let the selected value be M, where M is an integer greater than 1. S15. Generate sensor symbol sequence and geomagnetic symbol sequence based on the selected value.

[0007] Preferably, in S14, for each candidate value in the candidate alphabet size set, a selected value is generated based on the cumulative probability distribution of all data points in the joint data sample set, as follows: S141. For each candidate value in the candidate alphabet size set, let the value of the candidate value be N. For each candidate value, perform the following operations in sequence: S1411. Based on the cumulative probability distribution of all data points in the joint data sample set, let the number of breakpoints be N-1. The N-1 breakpoints are used to ensure that the N data intervals formed by the N-1 breakpoints on the cumulative probability distribution of all data points in the joint data sample set have equal probabilities on the cumulative probability distribution. S1412. Based on N-1 breakpoints, set up an alphabet containing N unique symbols, and assign the N unique symbols in the alphabet to N data intervals in sequence. For each data point in the joint data sample set, convert the data point into a unique symbol in the alphabet based on the data interval it corresponds to; then form a symbol sequence from the unique symbols obtained after converting all data points in the original data point order. S1413, Obtain the information entropy of the symbol sequence; S142. After traversing the K candidate values, select the candidate value corresponding to the maximum information entropy as the selected value.

[0008] Preferably, in S15, a sensor symbol sequence and a geomagnetic symbol sequence are generated based on the selected value, as follows: Get the M data intervals divided by M-1 breakpoints corresponding to the selected value, and use them as the selected data intervals. Get the alphabet containing M unique symbols corresponding to the selected value, and use it as the selected alphabet. Select M unique symbols from the alphabet and assign them sequentially to M selected data ranges; For each data point in the first dimensionality-reduced data segment, convert the data point into a unique symbol corresponding to the selected alphabet based on the selected data interval. The unique symbols obtained after converting all data points in the first dimensionality reduction data segment are used to form a sensor symbol sequence in the original data point order. Each unique symbol in the sensor symbol sequence is used as a sensor symbol; For each data point in the second dimensionality-reduced data segment, convert the data point into a unique symbol corresponding to the selected alphabet based on the selected data interval. The unique symbols obtained after converting all data points in the second dimensionality-reduced data segment are used to form a geomagnetic symbol sequence in the original data point order. Each unique symbol in the geomagnetic symbol sequence is used as a geomagnetic symbol.

[0009] Preferably, in step S3, a local sequence comparison is performed between the sensor symbol sequence and the geomagnetic symbol sequence to obtain a comprehensive confidence score, as follows: S31. According to the pre-defined comparison time window, within the comparison time window, extract the sensor symbol sequence window corresponding to the comparison time window from the sensor symbol sequence, and extract the geomagnetic symbol sequence window corresponding to the comparison time window from the geomagnetic symbol sequence. S32. Based on the sensor symbol sequence window and the geomagnetic symbol sequence window, and according to the preset symbol matching score, mismatch penalty score and empty space penalty score rules, a scoring matrix is ​​constructed using a dynamic programming algorithm. Based on the preset rules for symbol matching score, non-match penalty score and empty space penalty score, when obtaining the value of each cell in the scoring matrix, a heuristic rule is used to prune the alignment of the cells involved in the calculation to obtain the pruned scoring matrix. S33. Based on the pruned scoring matrix, obtain the cell with the largest cell value in the scoring matrix and select it as the selected cell; Based on the selected cells, obtain the alignment of the selected cells and use it as the selection alignment. Obtain the sensor symbol sequence corresponding to the selected alignment in the sensor symbol sequence window, and use it as the selected sensor symbol sequence; Get the geomagnetic symbol sequence corresponding to the selected alignment in the geomagnetic symbol sequence window, and use it as the selected geomagnetic symbol sequence; As an explanation, based on the alignment source information recorded by the dynamic programming algorithm during the construction of the scoring matrix, starting from the selected cell, we gradually trace back to its previous matching position until we reach the cell with a score of zero; the sensor symbol sequence and geomagnetic symbol sequence corresponding to this complete tracing process are used as the selected sensor symbol sequence and the selected geomagnetic symbol sequence.

[0010] S34. By selecting the sensor symbol sequence and the geomagnetic symbol sequence, a comparison sequence is formed, and a comprehensive confidence score is generated based on the comparison sequence.

[0011] Preferably, in S32, each row of the scoring matrix corresponds to a sensor symbol in the sensor symbol sequence window; In the scoring matrix, each column corresponds to a geomagnetic symbol in the geomagnetic symbol sequence window; As an explanation, in the scoring matrix, the order of the rows is exactly the same as the order in which the sensor symbol sequence windows appear in the sequence; the order of the columns is exactly the same as the order in which the geomagnetic symbol sequence windows appear in the sequence. The value of each cell in the scoring matrix is ​​used to represent the highest matching score that can be obtained among all possible alignment methods when the corresponding sensor symbol and geomagnetic symbol are used as the current matching endpoint, starting from the beginning position of the sensor symbol sequence window and the geomagnetic symbol sequence window, according to the preset symbol matching score, non-matching penalty score and empty space penalty score rules.

[0012] Preferably, in S32, heuristic rules are used to prune the alignment methods involved in the calculation, as follows: Heuristic rules include time sequence constraints and persistence constraints; The time sequence constraint rule is: in one possible alignment method, the matching score is calculated only if the time point corresponding to the geomagnetic symbol is not later than the time point corresponding to the sensor symbol to be matched; As an explanation, this means excluding alignment methods that violate the time sequence constraint rules; such alignment methods will not be included in the numerical calculation of this cell. The persistence constraint rule is as follows: In a possible alignment method, obtain the continuous matching length of the current alignment method. When the continuous matching length is lower than the preset minimum effective length, set the matching score of the alignment method to 0.

[0013] As an explanation, the continuous match length is: in one possible alignment method with the sensor symbol and geomagnetic symbol corresponding to the current cell as the match endpoint, the maximum number of symbols that can be successfully matched continuously and without interruption along the match path starting from the match endpoint; the continuous match length does not include any unmatched symbols or symbols that are interrupted by empty space penalty points.

[0014] Preferably, in S34, a comprehensive confidence score is generated based on the alignment sequence, as follows: Obtain the numerical range corresponding to the selected sensor symbol sequence in the first dimension-reduced data segment, and use it as the sensor segment; Obtain the numerical range corresponding to the selected geomagnetic symbol sequence in the second dimension-reduced data segment, and use it as the geomagnetic segment; The cosine similarity between sensor segments and geomagnetic segments is obtained as the morphological similarity. It should be noted that the cosine similarity between the sensor fragment and the geomagnetic fragment is obtained because the normalized morphological change trends of the sensor fragment and the geomagnetic fragment will show a high degree of similarity when they are disturbed by geomagnetic storms, and it is unrelated to the original physical quantities they represent. The number of sensor symbols in the selected sensor symbol sequence is obtained through statistical analysis and used as the length of the comparison sequence. As an explanation, the selected sensor symbol sequence and the selected geomagnetic symbol sequence are aligned in pairs; By using statistical analysis, if the sensor symbol and the geomagnetic symbol are the same, they are recorded as a symbol pair. The number of symbol pairs in the comparison sequence that are the same as the sensor symbol and the geomagnetic symbol is obtained, and the ratio of the number of symbol pairs to the length of the comparison sequence is used as the symbol consistency ratio. After normalizing the morphological similarity, alignment sequence length, and sign consistency ratio, a comprehensive confidence score is obtained by weighted summation. As an explanation, the weights of normalized morphological similarity, alignment sequence length, and symbol consistency ratio can be obtained based on historical data using a grid search algorithm, with the goal of maximizing the classification accuracy of the comprehensive confidence score for geomagnetic interference events; or they can be obtained experimentally based on expert experience.

[0015] The cloud platform maintenance method for the health status of the feeding system equipment proposed in this invention has the following beneficial technical effects: This application uses first time-series data (formed by collecting and sorting the output signals of weighing sensors) and second time-series data (formed by collecting and sorting near-real-time geomagnetic activity data). First, it normalizes and performs segmented aggregation to obtain dimensionality-reduced data segments, which are then merged into a joint data sample set. Based on the principles of cumulative probability distribution and maximizing information entropy, selected values ​​are chosen to generate corresponding sensor symbol sequences and geomagnetic symbol sequences. Then, a sequence window is extracted through a preset comparison time window. A scoring matrix is ​​constructed using a dynamic programming algorithm based on symbol matching scores, mismatch penalties, and gap penalties. Combining morphological similarity, comparison sequence length, and symbol consistency ratio, a comprehensive confidence score is obtained through normalized weighted calculation. This enables the identification of common-mode interference caused by geomagnetic storms on weighing sensors in the feeding system. An alarm is generated when the comprehensive confidence score exceeds a threshold. This effectively alleviates the technical problems of existing technologies where directly introducing geomagnetic activity data streams, such as the Kp index, is ineffective due to multiple uncertainties from global indices to local disturbances and from physical disturbances to electrical signal responses, and where the false alarm rate relying on the Kp index needs to be reduced. Attached Figure Description

[0016] Figure 1 This is a flowchart of a cloud platform maintenance method for the health status of a feeding system according to the present invention. Detailed Implementation

[0017] Embodiments of the present invention are described in detail below. Examples of these embodiments are illustrated in the accompanying drawings, wherein the same or similar symbols denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0018] like Figure 1 The cloud platform maintenance method for the health status of a feeding system includes the following steps: S1. Within a unit time range: collect the output signal of the weighing sensor and arrange the output signal of the weighing sensor in chronological order to form the first time series data; monitor and acquire near real-time geomagnetic activity data and arrange the geomagnetic activity data in chronological order to form the second time series data; Near-real-time geomagnetic activity data refers to geomagnetic activity data with an update frequency of no more than 3 hours; A unit of time refers to a time range of a fixed length, such as within a day, within 6 hours, or within 3 hours. S2. Generate the corresponding sensor symbol sequence and geomagnetic symbol sequence based on the first time series data and the second time series data; S3. Perform local sequence alignment between the sensor symbol sequence and the geomagnetic symbol sequence to obtain the comprehensive confidence score; S4. When the overall confidence score exceeds the overall confidence threshold, an alarm message is generated; The overall confidence threshold in S4 can be dynamically adjusted using machine learning algorithms based on historical false alarm statistics.

[0019] In an optional embodiment, in S2, the first time-series data and the second time-series data are discretized to generate corresponding sensor symbol sequences and geomagnetic symbol sequences, as follows: S11. Normalize the first time series data and the second time series data, and perform segmented aggregation approximation processing on the normalized data to obtain the first dimensionality-reduced data segment corresponding to the first time series data and the second dimensionality-reduced data segment corresponding to the second time series data. S12. Merge the first and second dimensionality-reduced data segments to form a joint data sample set; S13. Obtain the cumulative probability distribution of all data points in the joint data sample set; S14. Predefine a candidate alphabet size set containing K candidate values, where K is an integer greater than 1, each candidate value corresponds to an integer greater than 1, and any two candidate values ​​are not equal. For each candidate value in the candidate alphabet size set, a selected value is generated based on the cumulative probability distribution of all data points in the joint data sample set. Let the selected value be M, where M is an integer greater than 1. In an optional embodiment, in S14, for each candidate value in the candidate alphabet size set, a selected value is generated based on the cumulative probability distribution of all data points in the joint data sample set, as follows: S141. For each candidate value in the candidate alphabet size set, let the value of the candidate value be N. For each candidate value, perform the following operations in sequence: S1411. Based on the cumulative probability distribution of all data points in the joint data sample set, let the number of breakpoints be N-1. The N-1 breakpoints are used to ensure that the N data intervals formed by the N-1 breakpoints on the cumulative probability distribution of all data points in the joint data sample set have equal probabilities on the cumulative probability distribution. S1412. Based on N-1 breakpoints, set up an alphabet containing N unique symbols, and assign the N unique symbols in the alphabet to N data intervals in sequence. For each data point in the joint data sample set, convert the data point into a unique symbol in the alphabet based on the data interval it corresponds to; then form a symbol sequence from the unique symbols obtained after converting all data points in the original data point order. S1413, Obtain the information entropy of the symbol sequence; S142. After traversing the K candidate values, select the candidate value corresponding to the maximum information entropy as the selected value. S15. Generate sensor symbol sequence and geomagnetic symbol sequence based on the selected value; In an optional embodiment, in S15, a sensor symbol sequence and a geomagnetic symbol sequence are generated based on the selected value, as follows: Get the M data intervals divided by M-1 breakpoints corresponding to the selected value, and use them as the selected data intervals. Get the alphabet containing M unique symbols corresponding to the selected value, and use it as the selected alphabet. Select M unique symbols from the alphabet and assign them sequentially to M selected data ranges; For each data point in the first dimensionality-reduced data segment, convert the data point into a unique symbol corresponding to the selected alphabet based on the selected data interval. The unique symbols obtained after converting all data points in the first dimensionality reduction data segment are used to form a sensor symbol sequence in the original data point order. Each unique symbol in the sensor symbol sequence is used as a sensor symbol; For each data point in the second dimensionality-reduced data segment, convert the data point into a unique symbol corresponding to the selected alphabet based on the selected data interval. The unique symbols obtained after converting all data points in the second dimensionality-reduced data segment are used to form a geomagnetic symbol sequence in the original data point order. Each unique symbol in the geomagnetic symbol sequence is used as a geomagnetic symbol; By merging the first and second dimensionality-reduced data segments to form a joint data sample set, and based on its cumulative probability distribution, the candidate value that maximizes the information entropy of the transformed symbol sequence is selected from a set of candidate alphabet sizes containing K candidate values. Then, the sensor symbol sequence and the geomagnetic symbol sequence are generated using the data interval and alphabet corresponding to the selected value, respectively. This method achieves adaptive optimization of the alphabet size, effectively avoiding the problem that key feature information of sensor data and geomagnetic data may be masked or oversimplified due to a fixed preset alphabet size. This lays a foundation for subsequent sequence comparison that better reflects the distribution characteristics of the original data and improves the accuracy of geomagnetic interference identification.

[0020] In an optional embodiment, in S3, a local sequence alignment is performed between the sensor symbol sequence and the geomagnetic symbol sequence to obtain a comprehensive confidence score, as follows: S31. According to the pre-defined comparison time window, within the comparison time window, extract the sensor symbol sequence window corresponding to the comparison time window from the sensor symbol sequence, and extract the geomagnetic symbol sequence window corresponding to the comparison time window from the geomagnetic symbol sequence. S32. Based on the sensor symbol sequence window and the geomagnetic symbol sequence window, and according to the preset symbol matching score, mismatch penalty score and empty space penalty score rules, a scoring matrix is ​​constructed using a dynamic programming algorithm. In an optional embodiment, in S32, each row of the scoring matrix corresponds to a sensor symbol in the sensor symbol sequence window; In the scoring matrix, each column corresponds to a geomagnetic symbol in the geomagnetic symbol sequence window; As an explanation, in the scoring matrix, the order of the rows is exactly the same as the order in which the sensor symbol sequence windows appear in the sequence; the order of the columns is exactly the same as the order in which the geomagnetic symbol sequence windows appear in the sequence. The value of each cell in the scoring matrix is ​​used to represent the highest matching score that can be obtained among all possible alignment methods when the sensor symbol and geomagnetic symbol corresponding to that cell are used as the current matching endpoint, starting from the beginning position of the sensor symbol sequence window and the geomagnetic symbol sequence window, according to the preset symbol matching score, non-matching penalty score and empty space penalty score rules. Based on the preset rules for symbol matching score, non-match penalty score and empty space penalty score, when calculating the value of each cell in the scoring matrix, a heuristic rule is used to prune the alignment of the cells involved in the calculation to obtain the pruned scoring matrix. In an optional embodiment, in S32, heuristic rules are used to prune the alignment methods involved in the calculation, as follows: Heuristic rules include time sequence constraints and persistence constraints; The time sequence constraint rule is: in one possible alignment method, the matching score is calculated only if the time point corresponding to the geomagnetic symbol is not later than the time point corresponding to the sensor symbol to be matched; As an explanation, this means excluding alignment methods that violate the time sequence constraint rules; such alignment methods will not be included in the numerical calculation of this cell. The persistence constraint rule is as follows: In a possible alignment method, obtain the continuous matching length of the current alignment method. When the continuous matching length is lower than the preset minimum effective length, set the matching score of the alignment method to 0.

[0021] As an explanation, the continuous matching length is: in one possible alignment method with the sensor symbol and geomagnetic symbol corresponding to the current cell as the matching endpoint, the maximum number of symbols that can be successfully matched continuously and without interruption along the matching path starting from the matching endpoint; the continuous matching length does not include any unmatched symbols or symbols that are interrupted by empty space penalty points; S33. Based on the pruned scoring matrix, obtain the cell with the largest cell value in the scoring matrix and select it as the selected cell; Based on the selected cells, obtain the alignment of the selected cells and use it as the selection alignment. Obtain the sensor symbol sequence corresponding to the selected alignment in the sensor symbol sequence window, and use it as the selected sensor symbol sequence; Get the geomagnetic symbol sequence corresponding to the selected alignment in the geomagnetic symbol sequence window, and use it as the selected geomagnetic symbol sequence; As an explanation, based on the alignment source information recorded by the dynamic programming algorithm during the construction of the scoring matrix, starting from the selected cell, we gradually trace back to its previous matching position until we reach the cell with a score of zero; the sensor symbol sequence and geomagnetic symbol sequence corresponding to this complete tracing process are used as the selected sensor symbol sequence and the selected geomagnetic symbol sequence.

[0022] S34. By selecting the sensor symbol sequence and the geomagnetic symbol sequence, a comparison sequence is formed, and a comprehensive confidence score is generated based on the comparison sequence; In an optional embodiment, in S34, a comprehensive confidence score is generated based on the alignment sequence, as follows: Obtain the numerical range corresponding to the selected sensor symbol sequence in the first dimension-reduced data segment, and use it as the sensor segment; Obtain the numerical range corresponding to the selected geomagnetic symbol sequence in the second dimension-reduced data segment, and use it as the geomagnetic segment; The cosine similarity between sensor segments and geomagnetic segments is obtained as the morphological similarity. It should be noted that the cosine similarity between the sensor fragment and the geomagnetic fragment is obtained because the normalized morphological change trends of the sensor fragment and the geomagnetic fragment will show a high degree of similarity when they are disturbed by geomagnetic storms, and it is unrelated to the original physical quantities they represent. The number of sensor symbols in the selected sensor symbol sequence is obtained through statistical analysis and used as the length of the comparison sequence. As an explanation, the selected sensor symbol sequence and the selected geomagnetic symbol sequence are aligned in pairs; By using statistical analysis, if the sensor symbol and the geomagnetic symbol are the same, they are recorded as a symbol pair. The number of symbol pairs in the comparison sequence that are the same as the sensor symbol and the geomagnetic symbol is obtained, and the ratio of the number of symbol pairs to the length of the comparison sequence is used as the symbol consistency ratio. After normalizing the morphological similarity, alignment sequence length, and symbol consistency ratio, the overall confidence score is obtained by weighted summation.

[0023] By obtaining the cosine similarity between sensor segments and geomagnetic segments as morphological similarity, the similarity of their normalized morphological change trends under geomagnetic storm interference can be effectively captured. The symbol consistency ratio is used as the ratio of the number of statistical symbol pairs to the length of the comparison sequence, which accurately quantifies the matching accuracy of the symbol sequence. The persistence of the interference signal is evaluated by statistically comparing the length of the comparison sequence. Finally, the comprehensive confidence score is obtained by normalizing the morphological similarity, the length of the comparison sequence, and the symbol consistency ratio and then weighting and summing them, thereby achieving accurate identification and quantitative assessment of geomagnetic storm interference.

[0024] As an explanation, the weights of normalized morphological similarity, alignment sequence length, and symbol consistency ratio can be obtained based on historical data using a grid search algorithm, with the goal of maximizing the classification accuracy of the comprehensive confidence score for geomagnetic interference events; or they can be obtained experimentally based on expert experience.

[0025] This application uses first time-series data (formed by collecting and sorting the output signals of weighing sensors) and second time-series data (formed by collecting and sorting near-real-time geomagnetic activity data). First, it normalizes and performs segmented aggregation to obtain dimensionality-reduced data segments, which are then merged into a joint data sample set. Based on the principles of cumulative probability distribution and maximizing information entropy, selected values ​​are chosen to generate corresponding sensor symbol sequences and geomagnetic symbol sequences. Then, a sequence window is extracted through a preset comparison time window. A scoring matrix is ​​constructed using a dynamic programming algorithm based on symbol matching scores, mismatch penalties, and gap penalties. Combining morphological similarity, comparison sequence length, and symbol consistency ratio, a comprehensive confidence score is obtained through normalized weighted calculation. This enables the identification of common-mode interference caused by geomagnetic storms on weighing sensors in the feeding system. An alarm is generated when the comprehensive confidence score exceeds a threshold. This effectively alleviates the technical problems of existing technologies where directly introducing geomagnetic activity data streams, such as the Kp index, is ineffective due to multiple uncertainties from global indices to local disturbances and from physical disturbances to electrical signal responses, and where the false alarm rate relying on the Kp index needs to be reduced.

[0026] This method converts sensor and geomagnetic data into symbol sequences and performs local sequence comparison, which can more accurately identify low-frequency common-mode interference caused by geomagnetic storms hidden in normal signal fluctuations, which cannot be distinguished by traditional similarity comparison methods.

[0027] For clarification, "acquisition" in this application refers to obtaining the required content or data using existing technical means.

[0028] Furthermore, any content not described in detail in this specification is existing technology known to those skilled in the art.

[0029] In the embodiments provided by this invention, it should be understood that the disclosed system or method can be implemented in other ways. For example, the embodiments of the invention described above are merely illustrative; for instance, the division of modules is only a logical functional division, and there may be other division methods in actual implementation.

[0030] 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; 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 this embodiment according to actual needs.

[0031] Furthermore, the functional modules in the various embodiments of the present invention 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 module can be implemented in hardware or in the form of hardware plus software functional modules.

[0032] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the basic characteristics of the present invention.

[0033] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A cloud platform maintenance method for the health status of a feeding system, characterized in that, Includes the following steps: S1. Within a unit time range: collect the output signal of the weighing sensor and arrange the output signal of the weighing sensor in chronological order to form the first time series data; monitor and acquire near real-time geomagnetic activity data and arrange the geomagnetic activity data in chronological order to form the second time series data; S2. Generate the corresponding sensor symbol sequence and geomagnetic symbol sequence based on the first time series data and the second time series data; S3. Perform local sequence alignment between the sensor symbol sequence and the geomagnetic symbol sequence, and calculate a comprehensive confidence score; S4. When the overall confidence score exceeds the overall confidence threshold, an alarm message is generated.

2. The cloud platform maintenance method for the health status of the feeding system equipment according to claim 1, characterized in that, In S2, the first time-series data and the second time-series data are discretized to generate corresponding sensor symbol sequences and geomagnetic symbol sequences, as follows: S11. Normalize the first time series data and the second time series data, and perform segmented aggregation approximation processing on the normalized data to obtain the first dimensionality-reduced data segment corresponding to the first time series data and the second dimensionality-reduced data segment corresponding to the second time series data. S12. Merge the first and second dimensionality-reduced data segments to form a joint data sample set; S13. Obtain the cumulative probability distribution of all data points in the joint data sample set; S14. Predefine a candidate alphabet size set containing K candidate values, where K is an integer greater than 1, each candidate value corresponds to an integer greater than 1, and any two candidate values ​​are not equal. For each candidate value in the candidate alphabet size set, a selected value is generated based on the cumulative probability distribution of all data points in the joint data sample set. Let the selected value be M, where M is an integer greater than 1. S15. Generate sensor symbol sequence and geomagnetic symbol sequence based on the selected value.

3. The cloud platform maintenance method for the health status of the feeding system equipment according to claim 2, characterized in that, In S14, for each candidate value in the candidate alphabet size set, the selected value is generated based on the cumulative probability distribution of all data points in the joint data sample set, as follows: S141. For each candidate value in the candidate alphabet size set, let the value of the candidate value be N. For each candidate value, perform the following operations in sequence: S1411. Based on the cumulative probability distribution of all data points in the joint data sample set, let the number of breakpoints be N-1. The N-1 breakpoints are used to ensure that the N data intervals formed by the N-1 breakpoints on the cumulative probability distribution of all data points in the joint data sample set have equal probabilities on the cumulative probability distribution. S1412. Based on N-1 breakpoints, set up an alphabet containing N unique symbols, and assign the N unique symbols in the alphabet to N data intervals in sequence. For each data point in the joint data sample set, convert the data point into a unique symbol in the alphabet based on the data interval it corresponds to; then form a symbol sequence from the unique symbols obtained after converting all data points in the original data point order. S1413, Obtain the information entropy of the symbol sequence; S142. After traversing the K candidate values, select the candidate value corresponding to the maximum information entropy as the selected value.

4. The cloud platform maintenance method for the health status of the feeding system equipment according to claim 3, characterized in that, In S15, based on the selected value, a sensor symbol sequence and a geomagnetic symbol sequence are generated, as follows: Get the M data intervals divided by M-1 breakpoints corresponding to the selected value, and use them as the selected data intervals. Get the alphabet containing M unique symbols corresponding to the selected value, and use it as the selected alphabet. Select M unique symbols from the alphabet and assign them sequentially to M selected data ranges; For each data point in the first dimensionality-reduced data segment, convert the data point into a unique symbol corresponding to the selected alphabet based on the selected data interval. The unique symbols obtained after converting all data points in the first dimensionality reduction data segment are used to form a sensor symbol sequence in the original data point order. Each unique symbol in the sensor symbol sequence is used as a sensor symbol; For each data point in the second dimensionality-reduced data segment, convert the data point into a unique symbol corresponding to the selected alphabet based on the selected data interval. The unique symbols obtained after converting all data points in the second dimensionality-reduced data segment are used to form a geomagnetic symbol sequence in the original data point order. Each unique symbol in the geomagnetic symbol sequence is used as a geomagnetic symbol.

5. The cloud platform maintenance method for the health status of the feeding system equipment according to claim 1, characterized in that, In S3, a local sequence comparison is performed between the sensor symbol sequence and the geomagnetic symbol sequence to obtain a comprehensive confidence score, as follows: S31. According to the pre-defined comparison time window, within the comparison time window, extract the sensor symbol sequence window corresponding to the comparison time window from the sensor symbol sequence, and extract the geomagnetic symbol sequence window corresponding to the comparison time window from the geomagnetic symbol sequence. S32. Based on the sensor symbol sequence window and the geomagnetic symbol sequence window, and according to the preset symbol matching score, mismatch penalty score and empty space penalty score rules, a scoring matrix is ​​constructed using a dynamic programming algorithm. Based on the preset rules for symbol matching score, non-match penalty score and empty space penalty score, when obtaining the value of each cell in the scoring matrix, a heuristic rule is used to prune the alignment of the cells involved in the calculation to obtain the pruned scoring matrix. S33. Based on the pruned scoring matrix, obtain the cell with the largest cell value in the scoring matrix and select it as the selected cell; Based on the selected cells, obtain the alignment of the selected cells and use it as the selection alignment. Obtain the sensor symbol sequence corresponding to the selected alignment in the sensor symbol sequence window, and use it as the selected sensor symbol sequence; Get the geomagnetic symbol sequence corresponding to the selected alignment in the geomagnetic symbol sequence window, and use it as the selected geomagnetic symbol sequence; S34. By selecting the sensor symbol sequence and the geomagnetic symbol sequence, a comparison sequence is formed, and a comprehensive confidence score is generated based on the comparison sequence.

6. The cloud platform maintenance method for the health status of the feeding system equipment according to claim 5, characterized in that, In S32, each row of the scoring matrix corresponds to a sensor symbol in the sensor symbol sequence window; In the scoring matrix, each column corresponds to a geomagnetic symbol in the geomagnetic symbol sequence window; The value of each cell in the scoring matrix is ​​used to represent the highest matching score that can be obtained among all possible alignment methods when the corresponding sensor symbol and geomagnetic symbol are used as the current matching endpoint, starting from the beginning position of the sensor symbol sequence window and the geomagnetic symbol sequence window, according to the preset symbol matching score, non-matching penalty score and empty space penalty score rules.

7. The cloud platform maintenance method for the health status of the feeding system equipment according to claim 6, characterized in that, In S32, heuristic rules are used to prune the alignment methods involved in the calculation, as follows: Heuristic rules include time sequence constraints and persistence constraints; The time sequence constraint rule is: in one possible alignment method, the matching score is calculated only if the time point corresponding to the geomagnetic symbol is not later than the time point corresponding to the sensor symbol to be matched; The persistence constraint rule is as follows: In a possible alignment method, obtain the continuous matching length of the current alignment method. When the continuous matching length is lower than the preset minimum effective length, set the matching score of the alignment method to 0.

8. The cloud platform maintenance method for the health status of the feeding system equipment according to claim 7, characterized in that, In S34, a comprehensive confidence score is generated based on the alignment sequence, as follows: Obtain the numerical range corresponding to the selected sensor symbol sequence in the first dimension-reduced data segment, and use it as the sensor segment; Obtain the numerical range corresponding to the selected geomagnetic symbol sequence in the second dimension-reduced data segment, and use it as the geomagnetic segment; The cosine similarity between sensor segments and geomagnetic segments is obtained as the morphological similarity. The number of sensor symbols in the selected sensor symbol sequence is obtained through statistical analysis and used as the length of the comparison sequence. By using statistical analysis, if the sensor symbol and the geomagnetic symbol are the same, they are recorded as a symbol pair. The number of symbol pairs in the alignment sequence that are the same as the sensor symbol and the geomagnetic symbol is obtained, and the ratio of the number of symbol pairs to the length of the alignment sequence is used as the symbol consistency ratio. After normalizing the morphological similarity, alignment sequence length, and symbol consistency ratio, the overall confidence score is obtained by weighted summation.

Images