JP2019159365A - State fluctuation detection apparatus and program for state fluctuation detection - Google Patents

Abstract

To detect that there is state fluctuation only from data when the state fluctuation does not occur or that there is no state fluctuation only from data when the state fluctuation occurs.SOLUTION: A state fluctuation detection apparatus includes prediction model generation means 31 for generating a prediction model using teacher data, teacher prediction data acquisition means 32 for performing prediction using the teacher data in the prediction model, measurement prediction data acquisition means 34 for performing prediction using each of measurement data in the prediction model, error acquisition means 35 for acquiring an error that is a difference between an error obtained from each of the measurement data and a standard error obtained from the teacher data, and state fluctuation detection means 36 for obtaining a dissociation between the standard error and the error and detecting state fluctuation based on the dissociation degree.SELECTED DRAWING: Figure 2

Description

  The present invention relates to a state change detection device and a state change detection program.

  For example, when measuring a current, voltage, and vibration of a machine to detect a state change such as an abnormality, the abnormality is generally determined based on past experience by a person. That is, the state change threshold is set by a person, and it is almost impossible for a machine that does not have abnormal data to change the state.

  Patent Document 1 discloses a highly versatile abnormality detection device that does not depend on the type and number of objects and events that detect abnormal situations and the space (location, time zone, etc.) for monitoring those objects and events. ing.

  Specifically, it is a device that detects that an abnormality has occurred in a target event. In this apparatus, a time series of an input pattern which is data that changes depending on an event is acquired, characteristics of transitions in the time series of the acquired input patterns are analyzed, and characteristics of the analyzed transitions and The determined reference values are compared, and if they are not approximated within a certain range, it is determined that an abnormality has occurred in the event. As described above, when it is determined that an abnormality has occurred as a result of the comparison, an output unit that performs an output operation to that effect is provided.

  In other words, a time series using data that changes according to changes in events such as the movement of a person or object to be monitored as an input pattern is acquired, and the feature amount in the transition is extracted by focusing on the pattern transition in that time series Then, the occurrence of an abnormal situation is detected by comparing with that in a normal case. Here, “event” is a phenomenon that can be expressed as computer-processable data by a signal from a device sensor or the like, a report from a person (data input), or the like. It is also applicable to.

  Cited Document 2 discloses a data prediction method or a data prediction system that uses a prediction model that realizes highly accurate prediction and that can understand the prediction process.

  The invention of this patent document 2 is a data prediction method for predicting the future based on past results. The prediction model includes a first prediction model that outputs a prediction value for at least one feature amount on the prediction target date; The second prediction model includes a prediction value output from the first prediction model as an input factor, and outputs a prediction value for each predetermined time on the prediction target day.

  Then, a prediction model construction means for constructing a prediction model using the collected past performance data and past performance data, and executing prediction by inputting the input data for prediction to the constructed prediction model and obtaining a prediction value Means, a prediction error calculation means for calculating a prediction error or a model error from the collected data of the latest results and a prediction value, a correction coefficient or a correction amount for correcting the prediction value based on the prediction error or the model error, and a correction This is realized by predicted value correction means for obtaining a predicted value.

JP 2003-256957 A JP, 2015-127914, A

  However, as described above, the conventional prediction model is aimed at improving the accuracy of prediction and the like, and a state fluctuation detection device capable of accurately capturing state fluctuations only from data when there is no state fluctuation is known. It wasn't.

  The present invention has been made in view of the current state of the state change detection device as described above, and its purpose is only data having no state change or only data having a state change. To provide a state variation detection device and a state variation detection program that can detect the absence of state variation.

  A state variation detection apparatus according to the present invention includes a prediction model that predicts an objective variable from explanatory variables by machine learning, a teacher objective variable that is an objective variable of a fixed value, and a teacher explanatory variable that is an explanatory variable of a predetermined value, A teacher prediction data acquisition means for obtaining teacher prediction data by inputting the teacher explanatory variable to the prediction model using the teacher data constituted by: a standard error that is a difference between the teacher prediction data and the teacher objective variable The measurement explanatory variable is input to the prediction model using measurement data composed of a standard error acquisition means for acquiring a measurement objective variable that is a measured objective variable and a measurement explanatory variable that is a measured explanatory variable. Measurement prediction data acquisition means for obtaining measurement prediction data, error acquisition means for acquiring an error that is a difference between the measurement prediction data and the measurement objective variable, Calculated dissociation degree of a quasi-error and the error, characterized by comprising a status change detecting means for detecting a state change on the basis of the degree of dissociation.

  The state variation detection apparatus according to the present invention includes a prediction model creating unit that creates the prediction model using the teacher data.

  In the state variation detection device according to the present invention, the state variation detection means includes a ratio of the standard error to the error, a magnification of the error with respect to the standard error, and a dissociation that is a value obtained by standardizing the ratio or the magnification. It is characterized by detecting a state variation based on the degree.

  In the state change detection device according to the present invention, the state change detection means has a dissociation degree threshold value, and determines that there is a state change when the obtained dissociation degree exceeds the dissociation degree threshold value.

  In the state change detection device according to the present invention, the state change detection means has a dissociation degree abnormality threshold value to be determined as abnormal, and determines that an abnormality has occurred when the obtained dissociation degree exceeds the dissociation degree abnormality threshold value. It is characterized by that.

  In the state variation detection device according to the present invention, the teacher data and the measurement data are composed of a plurality of data corresponding to the measurement times, and the standard error acquisition unit obtains an average of the standard errors of the measurement times to obtain a standard. An error is obtained, and the error acquisition means obtains an average of the errors of the measurement times and takes it as an error.

  The state change detection program according to the present invention includes a computer that predicts an objective variable from explanatory variables by machine learning, a teacher objective variable that is an objective variable of a fixed value, and a teacher that is an explanatory variable of a predetermined value. Teacher prediction data acquisition means for obtaining teacher prediction data by inputting the teacher explanation variable to the prediction model using teacher data composed of explanatory variables, a difference between the teacher prediction data and the teacher objective variable A standard error acquisition means for acquiring a standard error, measurement data composed of a measurement objective variable that is a measured objective variable and a measurement explanatory variable that is a measured explanatory variable, and the measurement explanatory variable to the prediction model Measurement prediction data acquisition means for obtaining measurement prediction data by inputting, an error for acquiring an error that is a difference between the measurement prediction data and the measurement objective variable And resulting unit obtains the degree of dissociation and the standard error and the error, characterized in that to function as the state variation detection means for detecting a state change on the basis of the degree of dissociation.

  In the state variation detection program according to the present invention, the computer is further caused to function as a prediction model creating unit that creates the prediction model using the teacher data.

  The state change detection program according to the present invention is obtained by standardizing the ratio between the standard error and the error, the magnification of the error with respect to the standard error, the ratio or the magnification, using the computer as the state change detection means. It is made to function so that a state change may be detected based on the dissociation degree which is a value.

  In the state change detection program according to the present invention, the computer is used as the state change detection means, and has a dissociation degree threshold value, and when the obtained dissociation degree exceeds the dissociation degree threshold value, it is determined that there is a state change. It is made to function.

  In the state variation detection program according to the present invention, the computer is used as the state variation detection means, and has a dissociation degree abnormality threshold value to be determined as abnormal, and an abnormality occurs when the obtained dissociation degree exceeds the dissociation degree abnormality threshold value. It is made to function so that it may determine with generation | occurrence | production.

  In the state variation detection program according to the present invention, the teacher data and the measurement data are composed of a plurality of data, and the standard error is obtained by obtaining an average of the standard errors of the measurement times using the computer as the standard error acquisition means. And the computer functions as the error acquisition means so as to obtain an error by calculating an average of errors of the measurement times.

  According to the present invention, it is possible to detect that there is a state change only from data when there is no state change, or that there is no state change only from data in which the state change occurs.

The block diagram which shows the structure of the computer system which implement | achieves the state variation detection apparatus which concerns on this invention. The figure which shows the means implement | achieved by the program for a state change detection with which the computer system which implement | achieves the state change detection apparatus which concerns on this invention is equipped. The flowchart for demonstrating operation | movement of the state variation detection apparatus which concerns on this invention. The figure which shows the value of the teacher data used for the state variation detection apparatus which concerns on this invention, the prediction data obtained by applying it to a prediction model, and a standard error. The figure which shows the procedure until it produces a prediction model using teacher data in the state variation detection apparatus which concerns on this invention. The figure which shows the procedure after throwing in teacher data into a prediction model in the state variation detection apparatus based on this invention, and obtaining a standard error. The figure which shows the measurement data of three times used for the state variation detection apparatus which concerns on this invention, and the prediction data and error value which are obtained by applying them to a prediction model. The figure which shows the procedure after inputting the measurement data of three times into a prediction model in the state variation detection apparatus which concerns on this invention, and obtaining an error. The figure which showed notionally the magnification of the error with respect to the standard error obtained from the measurement data of three times in the state variation detection apparatus according to the present invention. FIG. 5 is a graph showing that a peak difference as a degree of dissociation is obtained using a standard error distribution and an error distribution in the state variation detection device according to the present invention. The figure explaining state change detection based on the dissociation degree and deviation value which are the values obtained by standardizing the error for every measurement time in the state change detection apparatus which concerns on this invention.

  Embodiments of a state change detection device and a state change detection program according to the present invention will be described below with reference to the accompanying drawings. In the drawings, the same components are denoted by the same reference numerals and redundant description is omitted. The state variation detection apparatus according to the embodiment of the present invention can be configured by, for example, a personal computer, a workstation, or other computer system as shown in FIG. This computer system operates as a state change detection device by controlling each unit based on a program or data stored in the main memory 11 or read into the main memory 11 and executing necessary processing. is there.

  An external storage interface 13, an input interface 14, a display interface 15, and a data input interface 16 are connected to the CPU 10 via a bus 12. Connected to the external storage interface 13 is an external storage device 23 in which programs such as a state change detection program and necessary data are stored. An input device 24 such as a keyboard as an input device for inputting commands and data and a mouse 22 as a pointing device are connected to the input interface 14.

  A display device 25 having a display screen such as an LED or LCD is connected to the display interface 15. Sensors 26-1, 26-2,..., 26-m for obtaining measurement data are connected to the data input interface 16. The computer system may be provided with other configurations, and the configuration of FIG. 1 is only an example.

  In the above, the CPU 10 implements the means shown in FIG. 2 by the state variation detection program in the external storage device 23. That is, the prediction model creation means 31, the prediction model 30, the teacher prediction data acquisition means 32, the standard error acquisition means 33, the measurement prediction data acquisition means 34, the error acquisition means 35, and the state variation detection means 36 are realized. In addition, teacher data is stored in the external storage device 23. The prediction model creation means 31 creates the prediction model 30 using teacher data. The prediction model 30 predicts an objective variable from explanatory variables by machine learning. Here, as an algorithm for machine learning, a random forest of pattern mining can be cited, but other than this, an algorithm such as regression analysis or regression tree may be adopted.

  As for explanatory variables and objective variable prediction models, y = f (x1, x2,..., Xi,..., Xn) where y is an objective variable, xi is an explanatory variable, and f is a prediction model. Can be represented by The teacher prediction data acquisition means 32 uses the teacher data composed of a teacher objective variable that is an objective variable of a fixed value and a teacher explanatory variable that is an explanatory variable of a fixed value to predict the teacher explanatory variable. Input to the model 30 to obtain teacher prediction data. The standard error acquisition unit 33 acquires a standard error that is a difference between the teacher prediction data and the teacher objective variable.

  The measurement prediction data acquisition means 34 inputs the measurement explanatory variable to the prediction model 30 using the measurement data constituted by the measurement objective variable that is the measured objective variable and the measurement explanatory variable that is the measured explanatory variable. Thus, measurement prediction data is obtained. The measurement data can be data obtained by the sensors 26-1, 26-2, ..., 26-m. The error acquisition means 35 acquires an error that is a difference between the measurement prediction data and the measurement objective variable.

  The state fluctuation detecting means 36 obtains the degree of dissociation between the standard error and the error, and detects the state fluctuation based on the degree of dissociation.

  Since the state variation detection apparatus configured by the above-described means and the like performs processing operations according to the flowchart shown in FIG. 3, the operation will be described with reference to this flowchart. First, a prediction model 30 is created using teacher data (STEP 1). For example, teacher data includes objective variable data to be obtained by sensor A, first explanatory variable data obtained by sensor B, and second explanation obtained by sensor C, as shown in FIG. And variable data. The number of data acquisitions (the number of data in the vertical direction in FIG. 4) is arbitrary. As shown in FIG. 5, the prediction model creating means 31 creates the prediction model 30 using the teacher data T as shown in FIG.

  Next, as shown in the flowchart of FIG. 3 and the operation sequence diagram of FIG. 6, explanatory variables (data of sensor B and data of sensor C in FIG. 4) are input to the prediction model 30 using the teacher data T, and prediction is performed. Data (teacher prediction data TF (teacher data prediction result in FIG. 6)) is obtained, and an error (standard error) that is a difference between the prediction data (teacher prediction data TF) and the objective variable of the teacher data T is obtained ( (Step 2). Here, the explanatory variables are the data of the sensors B and C in the teacher data of FIG. As shown in FIG. 6, if the objective variable of the teacher data T is TT and this objective variable TT is represented by a perfect circle, for example, the teacher prediction data TF is distorted with respect to the objective variable TT. Therefore, the part that protrudes or retracts from the perfect circle is an error. Actually, the prediction data (teacher prediction data) and the error (standard error STER) are obtained as numerical values as shown in the column of “predicted value standard / error of teacher data” in FIG. Since these are obtained by calculating as many times as the number of times of data acquisition, the same quantity as the number of times of acquisition can be obtained, so the average of these is calculated and held as the average standard error (AVSTER).

  Next, as shown in the flowchart of FIG. 3, the explanatory variables M1 and M2 are input to the prediction model 30 using the measurement data M from the sensors B and C to obtain prediction data (measurement prediction data). An error which is a difference between the (measurement prediction data) and the measured objective variable of the measurement data M is obtained (STEP 3).

  Here, measurement t1, t2, and t3 are performed at different times to obtain explanatory variables t1M1 and t1M2 for measurement t1, obtain explanatory variables t2M1 and t2M2 for measurement t2, and explain variable t3M1 for measurement t3. t3M2 is obtained (FIG. 7). As shown in FIG. 8, these are input to the same prediction model 30 to obtain prediction data (measured prediction data) t1F, t2F, t3F (right column in FIG. 7). In FIG. 8, three prediction models 30 are illustrated, but the prediction may be performed sequentially using one prediction model 30. Of course, three prediction models 30 may be provided. As shown in FIG. 7 and FIG. 8, the difference between the prediction data (measurement prediction data) t1F, t2F, t3F and the actually measured target variables t1MT, t2MT, t3MT of the measurement data M is an error (individual error) t1ER, t2ER, t3ER. Asking. Since the errors t1ER, t2ER, and t3ER are obtained by calculating as many times as the number of data measurements, the same quantity as the number of times of acquisition is obtained. Therefore, an average value of these is obtained, and this is calculated for each measurement t1, t2, and t3. It is held as an average error (t1AVER, t2AVER, t3AVER).

  Next, as shown in STEP 4 in FIG. 3, it is calculated how many times the error amount obtained from each measurement data is compared with the reference error obtained from the teacher data. It is concluded that the larger the magnification is, the more the state changes. In FIG. 9, the comparison subjects are “error (average) of measurement t1 (t1AVER)”, “error (average) of measurement t2 (t2AVER)”, and “error (average) of measurement t3 (t3AVER)”. It is assumed that there is a value corresponding to the area of the circle and there is a value corresponding to the area of the circle as shown in “standard error (average) (AVSTER)” as a comparison target. Assuming that the magnification is 1.2, 2.3, and 3.8, (error of measurement t1) <(error of measurement t2) <(error of measurement t3) holds. It can be concluded that it has occurred, then it can be concluded that a state change has occurred in measurement 2, and it can be concluded that measurement 1 has the lowest probability of state change.

  Alternatively, instead of obtaining an average of standard errors and errors, a distribution graph of standard error values is created, a distribution graph of error values is created, and state variations are detected using these distribution graphs. The distribution graph of the standard error value is as shown in FIG. 10A, the distribution graph of the error value at the measurement t1 is as shown in FIG. 10B, and the distribution graph of the error value at the measurement t2 is FIG. It is as shown in FIG. 10C, and the distribution graph of the error value of the measurement t3 is as shown in FIG.

  In the above case, the peak difference of the distribution graph is adopted as the degree of dissociation. The difference between the peak of the standard error value distribution graph and the peak of the error value distribution graph of measurement t1 is as shown by d1 in FIG. The difference between the peak of the distribution graph of the standard error value and the peak of the distribution graph of the error value of the measurement t2 is as shown by d2 in FIG. The difference between the peak of the distribution graph of the standard error value and the peak of the distribution graph of the error value of measurement t3 is as shown by d3 in FIG. Since the peak difference is t1 <t2 <t3, the determination of the state change is as described with reference to FIG.

  Further, as shown in STEP 4 of FIG. 3, an abnormality determination threshold is set for the magnification obtained above or a value σ obtained by standardizing the error for each number of measurements so as to detect the abnormality. Also good. For example, in the example of FIG. 9, if the threshold is “3.5”, it can be determined that the measurement 3 is abnormal.

An example of using the deviation value σ is shown in FIG. It is assumed that the above error distribution is as shown in the graph of FIG. An average value of the errors is obtained (S11). Using this average value, a deviation value σ is obtained (S12). Next, a state change is detected using the deviation value σ obtained above (S13). For example, since the standard error is also obtained every measurement, the standard error deviation value σ _ST is obtained from these distributions. On the other hand, deviation values σ _t1 , σ _t2 , and σ _t3 are also obtained for the measurements t1, t2, and t3. Of the deviation values σ _t1 , σ _t2 , and σ _t3 , it is assumed that the larger the difference or ratio or magnification from the standard error deviation value σ _ST , the higher the probability of a state change. A threshold value σ _SH is set for the deviation values σ _t1 , σ _t2 , and σ _t3 , and a state change (abnormality) can be made when the threshold value σ _SH is _exceeded .

  In the above description, one objective variable is obtained from two explanatory variables. However, the present invention can be applied to a prediction model that obtains one objective variable from one or more explanatory variables. Further, the state variation is not limited to abnormal variation, and can be applied to various state variation detections such as an excessive state, a shortage state, a low state, and a high state.

  In the above embodiment, when data is measured by the sensors A, B, and C, the sensor A prediction model in which the objective variable is data obtained by the sensor A and the explanatory variable is data obtained by the sensors B and C. Is configured only for one prediction model. However, in this way, in a system that measures data of three or more systems, a plurality of prediction systems can be constructed.

  For example, when data is measured by sensors A, B, and C, the target variable is data measured by sensor A, and the explanatory variable is data measured by sensors B and C. When the objective variable is data measured by the sensor B and the explanatory variable is data measured by the sensors A and C, the model is called a sensor B prediction model, and the objective variable is the data measured by the sensor C. When the explanatory variable is data measured by the sensors A and B, when referred to as a sensor C prediction model, the three prediction models are represented as a sensor A prediction model, a sensor B prediction model, and a sensor C prediction model. Can be created.

  It is assumed that the time series measurement values obtained by measuring the data of the sensors A, B, and C are t1, t2, and t3. Each of t1, t2, and t3 includes three types of data of sensors A, B, and C. In the sensor A prediction model, the measurement values t1, t2, and t3 of the sensors B and C are input, and three types of average errors, A_t1AVER, A_t2AVER, and A_t3AVER, of the sensor A prediction model can be obtained. In the sensor B prediction model, the measurement values t1, t2, and t3 of the sensors A and C are input, and three types of average errors B_t1AVER, B_t2AVER, and B_t3AVER of the sensor B prediction model can be obtained. In the sensor C prediction model, the measurement values t1, t2, and t3 of the sensors A and B are input, and three types of average errors C_t1AVER, C_t2AVER, and C_t3AVER of the sensor C prediction model can be obtained.

  When the reference error of each model obtained from the teacher data is A_AVSTER, B_AVSTER, and C_AVSTER, the respective magnifications and error distributions (deviations) of the pairs A_t1AVER, A_t2AVER, and A_t3AVER with reference to the reference error A_AVSTER can be obtained. The magnification and error distribution (deviation) of the pair B_t1AVER, B_t2AVER, B_t3AVER with respect to the reference error B_AVSTER can be obtained, and the magnification and error of the pair C_t1AVER, C_t2AVER, C_t3AVER with respect to the reference error C_AVSTER Distribution (deviation) can also be obtained.

  When the deviation is obtained as described above, for example, the deviation value σA_t1 of the measurement value t1 in the sensor A prediction model, the deviation value σB_t1 of the measurement value t1 in the sensor B prediction model, and the deviation value t1 of the measurement value t1 in the sensor C prediction model By obtaining the average of σC_t1 or the like, a new feature quantity of the measurement value t1 can be generated, and an abnormality can be detected by providing a threshold for this. The above is an example of the measurement value t1, but the measurement values t2 and t3 can be handled in the same manner.

10 CPU
DESCRIPTION OF SYMBOLS 11 Main memory 12 Bus 13 External storage interface 14 Input interface 15 Display interface 16 Data input interface 22 Mouse 23 External storage device 24 Input device 25 Display device 26 Sensor 30 Prediction model 31 Prediction model creation means 32 Teacher prediction data acquisition means 33 Standard error Acquisition means 34 Measurement prediction data acquisition means 35 Error acquisition means 36 State variation detection means

Claims (12)

A prediction model that predicts objective variables from explanatory variables by machine learning;
Using teacher data composed of a teacher objective variable that is an objective variable of a fixed value and a teacher explanatory variable that is an explanatory variable of a fixed value, the teacher explanatory variable is input to the prediction model and the teacher prediction data Teacher prediction data acquisition means for obtaining
A standard error acquisition means for acquiring a standard error which is a difference between the teacher prediction data and the teacher objective variable;
Measurement prediction that obtains measurement prediction data by inputting the measurement explanatory variable to the prediction model using measurement data composed of a measurement objective variable that is a measured objective variable and a measurement explanatory variable that is a measured explanatory variable Data acquisition means;
Error acquisition means for acquiring an error that is a difference between the measurement prediction data and the measurement objective variable;
A state fluctuation detecting device comprising: a state fluctuation detecting means for obtaining a degree of dissociation between the standard error and the error and detecting a state fluctuation based on the degree of dissociation.   The state variation detection apparatus according to claim 1, further comprising a prediction model creating unit that creates the prediction model using the teacher data.   The state fluctuation detecting means detects a state fluctuation based on a ratio of the standard error to the error, a magnification of the error with respect to the standard error, and a dissociation degree which is a value obtained by standardizing the error at each measurement time. The state variation detection apparatus according to claim 1 or 2, wherein   The state change detection means has a dissociation degree threshold value, and determines that there is a state change when the obtained dissociation degree exceeds the dissociation degree threshold value. State change detection device of statement.   5. The state change detection means has a dissociation degree abnormality threshold value to be determined as abnormal, and determines that an abnormality has occurred when the obtained dissociation degree exceeds the dissociation degree abnormality threshold value. The state variation detection device according to any one of the above. The teacher data and the measurement data are composed of a plurality of data corresponding to measurement times,
2. The standard error obtaining unit obtains an average of the standard errors of the measurement times to obtain a standard error, and the error obtaining unit obtains an average of errors of the measurement times to obtain an error. 6. The state variation detection device according to any one of 5 above. Computer
A prediction model that predicts objective variables from explanatory variables by machine learning,
Using teacher data composed of a teacher objective variable that is an objective variable of a fixed value and a teacher explanatory variable that is an explanatory variable of a fixed value, the teacher explanatory variable is input to the prediction model and the teacher prediction data Teacher prediction data acquisition means to obtain,
A standard error acquisition means for acquiring a standard error which is a difference between the teacher prediction data and the teacher objective variable;
Measurement prediction that obtains measurement prediction data by inputting the measurement explanatory variable to the prediction model using measurement data composed of a measurement objective variable that is a measured objective variable and a measurement explanatory variable that is a measured explanatory variable Data acquisition means,
Error acquisition means for acquiring an error that is a difference between the measurement prediction data and the measurement objective variable;
A program for detecting state fluctuations, characterized by obtaining a degree of dissociation between the standard error and the error and functioning as a state fluctuation detecting means for detecting a state fluctuation based on the degree of dissociation. Said computer further
The state change detection program according to claim 7, wherein the state change detection program functions as a prediction model creation unit that creates the prediction model using the teacher data.   Using the computer as the state variation detection means, the state variation is detected based on the ratio of the standard error to the error, the magnification of the error with respect to the standard error, and the degree of dissociation obtained by standardizing the ratio or magnification. The state change detection program according to claim 7 or 8, wherein the program is made to function as described above.   8. The computer according to claim 7, wherein the state change detecting means has a dissociation degree threshold and functions to determine that there is a state change when the obtained dissociation degree exceeds the dissociation degree threshold. The state change detection program according to any one of items 9 to 9.   The computer has the dissociation degree abnormality threshold value to be determined as abnormal as the state variation detection means, and functions to determine that an abnormality has occurred when the obtained dissociation degree exceeds the dissociation degree abnormality threshold value. The state change detection program according to any one of claims 7 to 10. The teacher data and the measurement data are composed of a plurality of data,
The computer is used as the standard error acquisition means, and the average error of the measurement batch is calculated to be a standard error, and the computer is used as the error acquisition means to calculate an error of the measurement batch average. The state change detection program according to any one of claims 7 to 11, wherein the program is made to function as follows.

Images