Milk powder production process tracing analysis system and method

Abstract

The invention relates to the technical field of milk powder production process monitoring analysis, in particular to a milk powder production process tracing analysis system and method, and the method comprises the steps: obtaining the historical process data of milk powder production, carrying out the data processing of the historical process data, and constructing an abnormal database through the data; performing abnormal influence transmissibility analysis on the target process and the upstream process based on the abnormal database, and determining propagation characteristics between the processes; and according to the propagation characteristics between the processes, identifying and prompting novel anomalies generated by the processes. The method comprises the following steps: dividing samples according to timestamps and calculating parameter anomaly coefficients to construct an anomaly database; new anomaly recognition is realized by integrating anomaly sequences, calculating all causes and isolating influence coefficients to analyze propagation characteristics; the problems that in the prior art, the milk powder production abnormity propagation path is not clear, influences are difficult to quantify, and novel abnormity is difficult to find are solved, and the accuracy and timeliness of production abnormity tracing and early warning are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of milk powder production process monitoring and analysis, and in particular to a milk powder production process tracing and analysis system and method. Background Art

[0002] In the milk powder production process, a complex multi-process system is formed from raw material acceptance, sterilization, ingredient preparation to drying and packaging, and the process parameters of each link are closely related.

[0003] In traditional production, due to the lack of an effective abnormal parameter sequence association mechanism, when a parameter abnormality occurs in a certain link, it is difficult to accurately locate its propagation path in the upstream and downstream processes, and it is impossible to quantify the impact of the abnormality on subsequent links. There is a lack of systematic integration and in-depth analysis of historical process data. At the same time, existing technologies have weak recognition capabilities for new abnormalities and often respond passively after the abnormality spreads, resulting in the expansion of quality problems, which not only increases production costs but also poses potential risks to product safety. Therefore, there is an urgent need for a milk powder production process analysis method that can comprehensively trace the spread of abnormalities, quantify the impact, and promptly identify new abnormalities. Summary of the Invention

[0004] The present invention aims to solve the problems existing in the background technology and propose a milk powder production process traceability analysis system and method.

[0005] The technical solution of the present invention is a milk powder production process traceability analysis method, comprising the following steps: Obtain historical process data of milk powder production and process the historical process data; Determine the abnormal parameters in each sample and calculate the parameter abnormality coefficient; Identify the target process and the corresponding upstream process and obtain multiple exception propagation events; Based on the abnormal propagation event, the abnormal parameter types corresponding to the upstream process and the target process are obtained, the corresponding historical abnormal parameter sequence is constructed, and the first abnormal sequence and the second abnormal sequence are marked respectively. The first abnormal sequence and the second abnormal sequence are combined to obtain the first abnormal data; and an abnormal database is constructed based on the first abnormal data corresponding to multiple abnormal propagation events; Analyze the propagation of abnormal impacts on the target process and upstream processes based on the abnormality database to determine the propagation characteristics between processes; Based on the propagation characteristics between processes, new anomalies generated by the process are identified and prompted.

[0006] Preferably, the parameter anomaly coefficient Ai is obtained by the formula ; In the formula, Pi is the abnormal parameter value; SPi is the standard parameter value, which takes the median of the standard parameter range. The standard parameter range is calibrated based on industry big data; αi is the influence weight of the parameter, which is calculated and calibrated based on the weight of industry big data; i is the abnormal parameter number, which is a positive integer.

[0007] Preferably, the exception propagation event satisfies the requirement that after a parameter exception occurs in an upstream process, a parameter exception begins to occur in the target process.

[0008] Preferably, the abnormal impact propagation analysis includes dividing and integrating the first abnormal data according to the same first abnormal sequence in the abnormal database to obtain a plurality of similar first abnormal data, performing a union operation on the second abnormal sequences of the plurality of similar first abnormal data, and using the union operation as the propagation abnormal sequence of the first abnormal sequence; Repeat the steps to obtain a propagation abnormality sequence corresponding to each first abnormality sequence; combine the first abnormality sequence with the corresponding propagation abnormality sequence as first abnormality sequence data of the abnormal propagation event.

[0009] Preferably, the abnormal impact propagation analysis further includes dividing and integrating abnormal propagation events according to the same first abnormal sequence to obtain several similar event libraries, and calculating the all-cause abnormality coefficient Q of the upstream process and target process of each abnormal propagation event in the similar event library based on the first abnormal sequence and the propagation abnormal sequence. The calculation formula is: .

[0010] Preferably, the abnormal impact propagation analysis further includes marking the all-cause abnormal coefficients of the upstream process and the target process as and , j is the abnormal propagation event number obtained, j is a positive integer, j∈[1,n], n is the total number of abnormal propagation events in the event library; By formula Calculate the isolated impact coefficient GY of the upstream process on the target process; determine the isolated impact coefficient range based on the preset coefficient tolerance ratio; The propagation characteristics are constructed according to the first anomaly sequence, the propagation anomaly sequence and the range of the isolated influence coefficient.

[0011] Preferably, the method for identifying abnormal processes includes extracting an abnormal parameter sequence from a selected process based on process data, marking it as a first tracing sequence, and tracing the extracted result in real time; if the first tracing sequence is empty, continuing the tracing; if the first tracing sequence is not empty, identifying whether the real-time abnormal parameter sequence is the first abnormal sequence that has been discovered; If it does not belong to the first abnormal sequence that has been discovered, it indicates that a new type of process abnormality has occurred; If it belongs to the first abnormal sequence that has been discovered, the abnormal parameter sequence of the downstream process of the middle process is extracted and marked as the second tracing sequence, and the target propagation feature is retrieved based on the first sequence; the isolated impact coefficient of the selected process on the downstream process is calculated based on the first tracing sequence and the second tracing sequence and marked as the tracing coefficient; the second tracing sequence and tracing coefficient are traced based on the target propagation feature.

[0012] Preferably, tracing the second tracing sequence includes identifying whether the real-time propagation anomaly sequence is a subsequence of the propagation anomaly sequence in the target propagation feature; if the second tracing sequence is a subsequence of the propagation anomaly sequence in the target propagation feature, then continuing the tracing; otherwise, prompting the occurrence of a new process anomaly.

[0013] Preferably, the method for tracing the tracing coefficient includes determining whether the tracing coefficient falls within the range of the isolated influence coefficient in the target propagation characteristic; if the tracing coefficient falls within the range of the isolated influence coefficient in the target propagation characteristic, then continuing the tracing; otherwise, prompting a new type of process abnormality.

[0014] The present invention also discloses a milk powder production process tracing and analysis system, which applies the above-mentioned milk powder production process tracing and analysis method, and specifically includes the following steps: Data acquisition module, used to obtain historical process data of milk powder production; Data analysis module, used to process historical process data and build an exception database; The data analysis module is used to analyze the propagation of abnormal impacts on the target process and upstream processes based on the abnormal database, and determine the propagation characteristics between the upstream process and the target process; The process tracing module is used to identify new anomalies generated by the process and provide prompts based on the propagation characteristics between processes.

[0015] Compared with the prior art, the above technical solution of the present invention has the following beneficial technical effects: An anomaly database is constructed by dividing samples by timestamps and calculating parameter anomaly coefficients. New anomalies can be identified by integrating anomaly sequences, calculating all-cause and isolated impact coefficients to analyze propagation characteristics. This solves the problems in existing technologies such as unclear propagation paths of milk powder production anomalies, difficulty in quantifying impacts, and difficulty in discovering new anomalies, thereby improving the accuracy and timeliness of production anomaly tracing and early warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a layer structure diagram of the first embodiment of the present invention. DETAILED DESCRIPTION

[0017] Example 1, as Figure 1 As shown, the present invention proposes a milk powder production process traceability analysis method, comprising: Obtain historical process data of milk powder production, process the historical process data, and build an anomaly database; Get historical process data and divide it into multiple samples based on timestamps; Determine the abnormal parameters in each sample and calculate the parameter abnormality coefficient; The parameter anomaly coefficient Ai is obtained by the formula ; In the formula, Pi is the abnormal parameter value; SPi is the standard parameter value, which takes the median value of the standard parameter range, and the standard parameter range is calibrated based on industry big data; αi is the parameter influence weight, which is calculated and calibrated based on the industry big data weight; i is the abnormal parameter number, which is a positive integer; For example, the abnormal parameter types include production ratio, energy efficiency ratio, defective product rate and cycle ratio for quantification; Determine the target process and the corresponding upstream process and obtain multiple exception propagation events; after the exception propagation event satisfies the parameter exception in the upstream process, the target process begins to have parameter exceptions; Based on the abnormal propagation event, the abnormal parameter types corresponding to the upstream process and the target process are obtained, the corresponding historical abnormal parameter sequence is constructed, and the first abnormal sequence and the second abnormal sequence are marked respectively. The first abnormal sequence and the second abnormal sequence are combined to obtain the first abnormal data; and an abnormal database is constructed based on the first abnormal data corresponding to multiple abnormal propagation events; The method for processing process data further includes obtaining parameter data of each production process, performing data cleaning and normalization on the parameter data, removing abnormal data and mapping the data to the [0, 1] interval; Analyze the propagation of abnormal impacts on the target process and upstream processes based on the abnormality database to determine the propagation characteristics between processes; The analysis of the spread of abnormal impact includes the following steps: In the anomaly database, the first anomaly data is divided and integrated according to the same first anomaly sequence to obtain a plurality of first anomaly data of the same type, and a union operation is performed on the second anomaly sequences of the plurality of first anomaly data of the same type, and the second anomaly sequences are used as propagation anomaly sequences of the first anomaly sequence; Repeat the steps to obtain the propagation exception sequence corresponding to each first exception sequence; combine the first exception sequence and the corresponding propagation exception sequence as the first exception sequence data of the exception propagation event; it should be noted that the propagation exception sequence is the abnormal parameter number sequence obtained by big data collation, which is affected by the abnormal parameter type of the upstream process and propagated in the target library process, and is calculated and integrated based on the abnormal parameter type of the target process; According to the same first abnormal sequence, the abnormal propagation events are divided and integrated to obtain several similar event libraries. Based on the first abnormal sequence and the propagation abnormal sequence, the all-cause abnormality coefficient Q of the upstream process and the target process of each abnormal propagation event in the similar event library is calculated. The calculation formula is: ; The all-cause anomaly coefficients of the upstream process and the target process are marked as and , j is the abnormal propagation event number obtained, j is a positive integer, j∈[1,n], n is the total number of abnormal propagation events in the event library; By formula Calculate the isolated impact coefficient GY of the upstream process on the target process; determine the isolated impact coefficient range based on the preset coefficient tolerance ratio; Constructing propagation characteristics based on the first anomaly sequence, propagation anomaly sequence and isolated influence coefficient range; Based on the propagation characteristics between processes, identify new anomalies generated by the process and provide prompts; The method for identifying abnormal processes includes extracting an abnormal parameter sequence from a selected process based on process data, marking the sequence as a first tracing sequence, and tracing the extracted result in real time; if the first tracing sequence is empty, continuing the tracing; if the first tracing sequence is not empty, identifying whether the real-time abnormal parameter sequence is the first abnormal sequence that has been discovered; If it does not belong to the first abnormal sequence that has been discovered, it indicates that a new type of process abnormality has occurred; If it belongs to the first anomaly sequence that has been discovered, the abnormal parameter sequence of the downstream process of the intermediate process is extracted and marked as the second tracing sequence. The target propagation feature is retrieved based on the first sequence. The isolated impact coefficient of the selected process on the downstream process is calculated based on the first and second tracing sequences and marked as the tracing coefficient. The second tracing sequence and tracing coefficient are traced based on the target propagation feature. Tracing the second tracing sequence includes identifying whether the real-time propagation anomaly sequence is a subsequence of the propagation anomaly sequence in the target propagation feature; if the second tracing sequence is a subsequence of the propagation anomaly sequence in the target propagation feature, then continuing the tracing; otherwise, prompting a new type of process anomaly; The method for tracing the traceability coefficient includes determining whether the traceability coefficient falls within the range of the isolated influence coefficient in the target propagation characteristic; if the traceability coefficient falls within the range of the isolated influence coefficient in the target propagation characteristic, then continuing the tracing; otherwise, prompting the occurrence of a new process anomaly; An anomaly database is constructed by dividing samples by timestamps and calculating parameter anomaly coefficients. New anomalies can be identified by integrating anomaly sequences, calculating all-cause and isolated impact coefficients to analyze propagation characteristics. This solves the problems in existing technologies such as unclear propagation paths of milk powder production anomalies, difficulty in quantifying impacts, and difficulty in discovering new anomalies, thereby improving the accuracy and timeliness of production anomaly tracing and early warning.

[0018] In the second embodiment, a milk powder production process tracing and analysis system proposed by the present invention is applied to the milk powder production process tracing and analysis method proposed in the first embodiment, and specifically comprises the following steps: Data acquisition module, used to obtain historical process data of milk powder production; Data analysis module, used to process historical process data and build an exception database; The data analysis module is used to analyze the propagation of abnormal impacts on the target process and upstream processes based on the abnormal database, and determine the propagation characteristics between the upstream process and the target process; The process tracing module is used to identify new anomalies generated by the process and provide prompts based on the propagation characteristics between processes.

[0019] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A milk powder production process traceability analysis method, characterized in that: The following steps are involved: Obtain historical process data of milk powder production and process the historical process data; Identify the target process and the corresponding upstream process and obtain multiple exception propagation events; Based on the abnormal propagation event, the abnormal parameter types corresponding to the upstream process and the target process are obtained, the corresponding historical abnormal parameter sequence is constructed, and the first abnormal sequence and the second abnormal sequence are marked respectively. The first abnormal sequence and the second abnormal sequence are combined to obtain the first abnormal data; Building an anomaly database based on first anomaly data corresponding to a plurality of anomaly propagation events; Analyze the propagation of abnormal impacts on the target process and upstream processes based on the abnormality database to determine the propagation characteristics between processes; Based on the propagation characteristics between processes, new anomalies generated by the process are identified and prompted.

2. A milk powder production process tracing analysis method according to claim 1, characterized in that: The parameter anomaly coefficient Ai is obtained by the formula Where Pi is the abnormal parameter value; SPi is the standard parameter value, which takes the median value of the standard parameter range. The standard parameter range is calibrated based on industry big data; αi is the influence weight of the parameter, which is calculated and calibrated based on the weight of industry big data. i is the exception parameter number, which is a positive integer.

3. A milk powder production process tracing analysis method according to claim 2, characterized in that: After the exception propagation event satisfies the parameter exception in the upstream process, the target process begins to have parameter exceptions.

4. The milk powder production process tracing analysis method according to claim 2, characterized in that: The abnormal impact propagation analysis includes dividing and integrating the first abnormal data according to the same first abnormal sequence in the abnormal database to obtain a plurality of similar first abnormal data, performing a union operation on the second abnormal sequences of the plurality of similar first abnormal data, and using the union operation as the propagation abnormal sequence of the first abnormal sequence; Repeat the steps to obtain a propagation abnormality sequence corresponding to each first abnormality sequence; combine the first abnormality sequence with the corresponding propagation abnormality sequence as first abnormality sequence data of the abnormal propagation event.

5. A milk powder production process tracing analysis method according to claim 4, characterized in that: The abnormal impact propagation analysis also includes dividing and integrating abnormal propagation events according to the same first abnormal sequence to obtain several similar event libraries. Based on the first abnormal sequence and the propagation abnormal sequence, the all-cause abnormality coefficient Q of the upstream process and the target process of each abnormal propagation event in the similar event library is calculated. The calculation formula is: .

6. A milk powder production process tracing analysis method according to claim 5, characterized in that: The abnormal impact propagation analysis also includes marking the all-cause abnormal coefficients of the upstream process and the target process as and , j is the abnormal propagation event number obtained, j is a positive integer, j∈[1,n], n is the total number of abnormal propagation events in the event library; By formula Calculate the isolated impact coefficient GY of the upstream process on the target process; Determine the range of isolated influence coefficients according to the preset coefficient tolerance ratio; The propagation characteristics are constructed according to the first anomaly sequence, the propagation anomaly sequence and the range of the isolated influence coefficient.

7. A milk powder production process tracing analysis method according to claim 6, characterized in that: The method for identifying abnormal processes includes extracting an abnormal parameter sequence from a selected process based on process data, marking it as a first tracing sequence, and tracing the extracted results in real time; if the first tracing sequence is empty, tracing is continued; If the first tracing sequence is not empty, identifying whether the real-time abnormal parameter sequence is the first abnormal sequence that has been discovered; If it does not belong to the first abnormal sequence that has been discovered, it indicates that a new type of process abnormality has occurred; If it belongs to the first anomaly sequence that has been discovered, the abnormal parameter sequence of the downstream process of the mid-process is extracted and marked as the second traceability sequence and the target propagation feature is retrieved based on the first sequence; Based on the first tracing sequence and the second tracing sequence, the isolated impact coefficient of the selected process on the downstream process is calculated and marked as the tracing coefficient; the second tracing sequence and the tracing coefficient are traced based on the target propagation characteristics.

8. The milk powder production process tracing analysis method according to claim 7, characterized in that: Tracing the second tracing sequence includes identifying whether the real-time propagation anomaly sequence is a subsequence of the propagation anomaly sequence in the target propagation feature; if the second tracing sequence is a subsequence of the propagation anomaly sequence in the target propagation feature, continuing the tracing; Otherwise, a new type of process exception is prompted.

9. The milk powder production process tracing analysis method according to claim 7, characterized in that: The method for tracing back the tracing coefficient includes determining whether the tracing coefficient falls within the range of the isolated influence coefficient in the target propagation characteristic; if the tracing coefficient falls within the range of the isolated influence coefficient in the target propagation characteristic, then continuing the tracing; Otherwise, a new type of process exception is prompted.

10. A milk powder production process tracing and analysis system, applied to the milk powder production process tracing and analysis method according to any one of claims 1 to 9, characterized in that: The specific steps include: Data acquisition module, used to obtain historical process data of milk powder production; Data analysis module, used to process historical process data and build an exception database; The data analysis module is used to analyze the propagation of abnormal impacts on the target process and upstream processes based on the abnormal database, and determine the propagation characteristics between the upstream process and the target process; The process tracing module is used to identify new anomalies generated by the process and provide prompts based on the propagation characteristics between processes.

Images