Manufacturing process monitoring apparatus
A manufacturing process monitoring apparatus capable of determining a manufacturing process is anomaly, without requiring any threshold value for determining the as anomaly is provided. The manufacturing process monitoring apparatus includes a data conversion unit configured to convert process data of a manufacturing facility, a feature value analysis unit configured to analyze the converted data based on information on feature values, a data restoration unit configured to restore data for each of a plurality of categories based on the information on the feature values and information on the analyzed result, a similarity calculation unit configured to calculate a similarity for each of the plurality of categories based on the data used when being analyzed and the restored data, a category determination unit configured to determine a category of the data based on the similarity for each of the plurality of categories, a category classification unit configured to classify the category to which the process data belongs, and a process state diagnostic unit configured to diagnose a state of the manufacturing process based on a result of comparison between the determined category and the classified category.
1. A manufacturing process monitoring apparatus comprising:
circuitry configured to
periodically collect data on a material to be rolled at an entry side of a rolling mill and a delivery side of the rolling mill;
store information on feature values of a material processed under a normal operational condition of the rolling mill during a manufacturing process for each of a plurality of categories, each category regarding a different characteristic of the material, the plurality of categories including at least two of material grade, material thickness, material width, and material temperature, classified by the material grade, material thickness, and material width in advance;
convert current process data including operational data of the rolling mill or measured data from sensors in the rolling mill and feature values of a current material being processed;
analyze converted current process data based on stored information on the feature values;
restore data for each of the plurality of categories based on the stored information on the feature values and analyzed converted current process data;
calculate a similarity for each of the plurality of categories based on the converted current process data and restored data;
determine a category of the converted current process data based on the similarity for each of the plurality of categories;
classify the category to which the current process data belongs;
diagnose a state of the manufacturing process based on result of comparison between the determined category and the classified category; and
on condition that the state of the manufacturing process is diagnosed as an anomaly, make an operational condition of the rolling mill different than a normal operational condition.
2. The manufacturing process monitoring apparatus according to claim 1 , wherein the circuitry is configured to determine that the manufacturing process is anomaly when the determined category is different from the classified category or when the determined category is different from the classified category a preset number of times within a preset period.
3. The manufacturing process monitoring apparatus according to claim 1 , wherein the circuitry is configured to determine that the manufacturing process is anomaly when the determined category and the classified category do not satisfy a preset relation or when the determined category and the classified category do not satisfy the preset relation a preset number of times within a preset period.
4. A manufacturing process monitoring method, comprising:
periodically collecting data on a material to be rolled at an entry side of a rolling mill and a delivery side of the rolling mill;
storing information on feature values of a material processed under a normal operational condition of the rolling mill during a manufacturing process for each of a plurality of categories, each category regarding a different characteristic of the material, the plurality of categories including at least two of material grade, material thickness, material width, and material temperature, classified by the material grade, material thickness, and material width in advance;
converting current process data including operational data of the rolling mill or measured data from sensors in the rolling mill and feature values of a current material being processed;
analyzing converted current process data based on stored information on the feature values;
restoring data for each of the plurality of categories based on the stored information on the feature values and converted current process data;
calculating a similarity for each of the plurality of categories based on the converted current process data and restored data;
determining a category of the converted current process data based on the similarity for each of the plurality of categories;
classifying the category to which the current process data belongs;
diagnosing a state of the manufacturing process based on result of comparison between the determined category and the classified category; and
on condition that the state of the manufacturing process is diagnosed as an anomaly, making an operational condition of the rolling mill different than a normal operational condition.
5. The method according to claim 4 , further comprising determining that the manufacturing process is anomaly when the determined category is different from the classified category or when the determined category is different from the classified category a preset number of times within a preset period.
6. The method according to claim 4 , further comprising determining that the manufacturing process is anomaly when the determined category and the classified category do not satisfy a preset relation or when the determined category and the classified category do not satisfy the preset relation a preset number of times within a preset period.
7. A non-transitory computer readable storage device having computer readable instructions that when executed by circuitry cause the circuitry to:
periodically collect data on a material to be rolled at an entry side of a rolling mill and a delivery side of the rolling mill;
store information on feature values of a material processed under a normal operational condition of the rolling mill during a manufacturing process for each of a plurality of categories, each category regarding a different characteristic of the material, the plurality of categories including at least two of material grade, material thickness, material width, and material temperature, classified by the material grade, material thickness, and material width in advance;
convert current process data including operational data of the rolling mill or measured data from sensors in the rolling mill and feature values of a current material being processed;
analyze converted current process data based on stored information on the feature values;
restore data for each of the plurality of categories based on the stored information on the feature values and converted current process data;
calculate a similarity for each of the plurality of categories based on the converted current process data and restored data;
determine a category of the converted current process data based on the similarity for each of the plurality of categories;
classify the category to which the current process data belongs;
diagnose a state of the manufacturing process based on result of comparison between the determined category and the classified category; and
on condition that the state of the manufacturing process is diagnosed as an anomaly, make an operational condition of the rolling mill different than a normal operational condition.
8. The non-transitory computer readable storage device according to claim 7 , wherein the circuitry is configured to determine that the manufacturing process is anomaly when the determined category is different from the classified category or when the determined category is different from the classified category a preset number of times within a preset period.
9. The non-transitory computer readable storage device according to claim 7 , wherein the circuitry is configured to determine that the manufacturing process is anomaly when the determined category and the classified category do not satisfy a preset relation or when the determined category and the classified category do not satisfy the preset relation a preset number of times within a preset period.