JP2013210945A - Waveform analyzing device and waveform analyzing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To calculate a value of a normal/abnormal state of an object from sensor data.SOLUTION: A waveform analyzing device is provided which comprises a sensing section 30, a determination section 39 and a calculation section. The sensing section 30 observes a monitoring object by means of a sensor and acquires a plurality of time-series data items about a plurality of variations. The determination section 39 determines a state of the monitoring object. The calculation section calculates a normal or abnormal quantitative state value of the monitoring object. The determination section 39 computes likelihood of normality and likelihood of abnormality of the monitoring object on the basis of segment data and a determination model from the plurality of time-series data items, respectively, and determines a state of the monitoring object to greater likelihood between the likelihood of normality and of abnormality. The calculation section calculates a quantitative state value of normality or abnormality as a sum of adding a result of multiplying a known normal state value by a normality degree and a result of multiplying a known abnormal state value by an abnormality degree.

Description

  Embodiments described herein relate generally to a waveform analyzer and a method thereof.

  With the recent spread of sensor networks, sensor data analysis techniques are required in various scenes such as sensor fusion. However, it is difficult to improve the determination performance in a determination apparatus using multivariate time series data. Generally, discrimination accuracy is good, and there are SVMs and neural networks, etc., which are relatively frequently used. However, it may be difficult to accept in the field because it is difficult to construct or the basis of judgment is difficult to understand. . In sensor fusion technology, it is possible to determine the abnormality of information of multiple sensor data using a probabilistic model. In this case, it is the mainstream to convert sensor data from continuous time series values to discrete values and treat them as categorical data. .

Japanese Patent No. 3624546 Japanese Patent No. 3931879 JP 2005-165421 A JP 2007-64307 A

Hui Ding, Goce Trajcevski, Peter Scheuermann, Xiaoyue Wang and Eamonn Keogh (2008), Querying and Mining of Time Series Data: Experimental Comparison of Representations and Distance Measures.In Proceedings of the International Conference on Very Large Data Bases (VLDB2008), Vol .1,, Issue 2, pp. 1542-1552, 2008.

  It is desirable to be able to accurately determine the abnormality of the monitoring target using the sensing data of multiple sensors accumulated in the past and calculate and present the normal / abnormal state value of the target from such sensor data. Yes.

  According to the embodiment, a sensing unit that observes a monitoring target with a sensor and acquires a plurality of time-series data regarding a plurality of variables, a determination unit that determines a state of the monitoring target, and whether the monitoring target is normal or abnormal There is provided a waveform analysis device comprising a calculation unit for calculating a quantitative state value. The determination unit calculates the normal likelihood and the abnormal likelihood of the monitoring target based on the segment data and the determination model from the plurality of time series data, respectively, and the likelihood of the normal and abnormal is large Then, the state of the monitoring target is determined. The calculation unit calculates a normal or abnormal quantitative state value as a sum of a result obtained by multiplying a known normal state value by a normal degree and a result obtained by multiplying a known abnormal state value by an abnormal degree.

The figure which shows the structure of the abnormality determination system which concerns on embodiment Diagram showing the flow of training and learning process by server The figure which shows the example of the training data set in the training data storage part Diagram showing examples of waveform amplitude The figure which shows the example of the feature vector after converting to the power spectrum Diagram showing an example of waveform division Diagram showing another example of waveform division Diagram showing still another example of waveform division Diagram showing an example of division in the case of the power spectrum The figure which shows the detailed processing flow of pseudo judgment evaluation processing Diagram showing an example of division of training data The figure which shows the other example of division of training data The figure which shows the detail of step S205 of FIG. Diagram showing an example of conditional probability calculation Figure showing an example of finding nearest neighbor segment data Figure showing format example of frequency distribution table Figure showing an example of the score table Diagram showing how nearest neighbors are calculated Diagram showing how nearest neighbors are calculated The figure which shows the example of data (judgment model) in the best model storage part The figure which shows the process of a 1st modification. The figure which shows the process added by various modifications The figure which shows the process of a 2nd modification. The figure which shows the process of a 3rd modification. The figure which shows the process of a 4th modification The figure which shows an example of the determination model which concerns on a 5th modification The figure which shows the concept of the model type | formula which concerns on a 5th modification The figure which shows a mode that the top of judgment object data is scanned Diagram showing how data (waveform) is cut out The figure which shows the example of a display screen of the part of abnormality judgment among judgment object data Diagram showing the operation flow in the client Diagram showing the waveform of the car acceleration for normal and abnormal cases The figure which shows the procedure which calculates the abnormal degree of the basket swing quantitatively The figure which shows an example of the hardware constitutions for implement | achieving a server and a client Figure showing an example of segment template fitting The figure which shows an example of operation | movement of the client which concerns on a modification.

  Hereinafter, embodiments will be described with reference to the drawings.

  FIG. 1 shows a configuration of an abnormality determination system according to the embodiment. This abnormality determination system includes a server (monitoring center device) and a client (remote monitoring terminal). The server performs training learning using past sensor data (time-series data) obtained by observation of the monitoring target and a class (abnormal or normal) that identifies the status of the monitoring target at the time of acquisition of the sensor data. As a result, a determination model for determining new sensor data is generated. The client observes the monitoring target, acquires sensor data, and determines whether the monitoring target is normal or abnormal using the acquired sensor data and the determination model.

(server)
FIG. 2 is a flowchart showing the flow of the training learning process by the server.

  First, the server reads various parameters set by the user (S101). For example, parameters such as the maximum waveform division number Z_max (used in step S106) are read. Reading is performed from a recording medium such as a memory or a hard disk.

  Next, by performing initial setting, the parameter z of the number of waveform divisions is set to 0 (S102).

  Next, the training data input unit 12 reads the training data set from the training data storage unit 11 and inputs it to the waveform preprocessing unit 13 at the next stage (S103).

  FIG. 3 shows an example of a training data set in the training data storage unit 11.

  Each training data is composed of at least one type of time series data (sensor data) and a class. Here, the class is a determination result obtained when a maintenance person or the like determines the state of the target device (monitoring target) when the corresponding time-series data is acquired in the past. Classes are, for example, abnormal and normal. However, there may be multiple types of abnormal states such as abnormal type A and abnormal type B. Here, in order to make the explanation easy to understand, the case where there are two classes of normal and abnormal will be described. In the illustrated example, the training data includes time-series data of four variables (channels). The class of training data d1 to dN is normal, and the class of training data dN + 1 to dM is abnormal. The time series data of the four variables are obtained from the corresponding four sensors. Here, for simplification of explanation, it is assumed that the time series data has the same size (length in the time axis direction), but the size may be different for each variable (channel).

  Next, the waveform preprocessing unit 13 preprocesses each time series data included in the training data set (S104). As preprocessing, a feature vector such as an amplitude spectrum may be acquired by performing signal processing such as power spectrum conversion by FFT, short-time Fourier transform, and wavelet transform. Alternatively, waveform amplitude values at a plurality of predetermined times may be acquired. FIG. 4 shows examples of waveform amplitudes acquired at a plurality of predetermined times. FIG. 5 shows an example of the feature vector after conversion to the power spectrum. As preprocessing, the waveform may be further processed using a low-frequency pass filter (smoothing filter). This is effective when noise is added to the waveform amplitude or when it is desired to grasp the general characteristics of the waveform. Alternatively, only a waveform in a limited band may be extracted by a filter. Alternatively, various waveform approximation calculations such as line segment approximation, Chebyshev approximation, and APCA approximation may be performed as described in Non-Patent Document 1, for example. It is also possible to proceed to the next process without performing any pre-processing. In the following description, for ease of understanding, description will be made using the time-series data of FIG. 3 that has not undergone preprocessing.

  Next, in steps S105 to S110, the time-series data is divided into a plurality of sections by the waveform division number z while sequentially increasing the waveform division number z from 1 to the maximum waveform division number z_max, and the waveform division number z (1 ~ Z_max), an important interval for each variable is determined. Then, the important section of each variable at the time of the waveform division number at which the highest evaluation value is obtained is determined as the optimum section. In addition, the data portion (segment data) of each optimum section in each time series data of the training data set is stored in association with the corresponding class. Details of steps S105 to S110 will be described below.

  In step S105, the server increments the waveform division number z by one.

  In step S106, the server determines whether or not the waveform division number z exceeds the maximum division number z_max. If it exceeds, the process proceeds to step S111. If not, the process proceeds to step S107.

  In step S107, the waveform dividing unit 14 divides each time series data of the training data set on the time axis by the waveform division number z, and cuts out segment data. Here, the division method is simple and is divided so that the division width is equal. However, another division method may be used. The extracted segment data is stored in the segment storage unit 15.

  An example of waveform division is shown in FIG. 6 (when z = 1), FIG. 7 (when z = 2), and FIG. 8 (when z = 4). Note that no division is actually performed when z = 1. Each segment data (partial time series data) cut out in this way is stored in the segment storage unit 15 in association with the value of z, the training data ID, and the variable ID. If the pre-processed time-series data (feature vector) is a power spectrum, the data may be divided along the frequency axis direction as shown in FIG. Here, the case of z = 3 is shown, and the frequency band is divided into three. In the present embodiment, dividing time series data means dividing the time series data along the frequency axis direction when converting the time series data into a power spectrum.

  Next, in step S108, a pseudo determination evaluation process is performed by the pseudo determination evaluation unit 17, the probability likelihood calculation unit 16, and the best model selection unit 19.

  FIG. 10 is a flowchart showing a detailed process flow of the pseudo determination evaluation process (S108).

  First, the pseudo judgment evaluation unit 17 divides the training data set (segmented) into a plurality of divided sets, and labels the plurality of divided sets with 1 to Vmax (S201). One divided set may be composed of one training data, or may be composed of a plurality of training data. When one divided set is composed of one training data, the training data set is divided into the total number of training data, and thus Vmax matches the total number of training data. In the following, for the sake of simplicity, it is assumed that one divided set is composed of one training data unless otherwise specified.

  Next, by performing initialization, the divided set identifier v = 0 and the evaluation value q = 0.0 are set (S202).

  Next, v is incremented by 1 (S203).

  Next, the pseudo determination evaluation unit 17 selects, as the pseudo determination target data set Tv, the one indicated by the identifier v among the plurality of split sets divided in step S201. That is, a plurality of divided sets are divided into a pseudo determination target data set Tv and other divided sets. FIG. 11 shows an example when v = 1, and FIG. 12 shows an example when v = 2. As described above, since each of the plurality of divided sets is composed of one piece of training data, the pseudo determination target data set Tv includes one piece of training data. Therefore, hereinafter, the pseudo-determination target data Tv refers to a case where the pseudo-determination target data set Tv includes one piece of training data.

  Next, the pseudo determination evaluation unit 17 performs a modeling process using Leave Cross Validation by training learning, and acquires an evaluation value r (S205). In this process, the class of the pseudo determination target data Tv (determination result) is simulated and the remaining training data is used to estimate the class of the pseudo determination target data Tv. The evaluation value r is obtained by calculating whether the estimated result matches the actual class of the pseudo determination target data Tv. In particular, Leave-One-out Cross Validation (only one training data is included in one divided set) is effective when the training data is a small number. However, when there is only one training data to be included in the divided set, there is a problem that the pseudo judgment process takes too much time.If you want to avoid this problem, include multiple training data in one divided set and add one of the divided sets. It is only necessary to select the pseudo judgment target set and repeat the evaluation so that all the subsets become the pseudo judgment target set once. This is an evaluation method generally called Cross Validation. In this example, it is assumed that the training set includes one piece of training data as described above.

  In step S207, the pseudo determination evaluation unit 17 adds the evaluation value r to the evaluation value q.

  In step S208, the pseudo determination evaluation unit 17 determines whether or not the divided set identifier v exceeds Vmax. That is, it is determined whether each training data of the training data set is selected as the pseudo determination target data Tv (whether each of the plurality of divided sets is selected as the pseudo determination target data set Tv). When it does not exceed Vmax, the process returns to step S203, and when it exceeds, the process proceeds to step S209.

  In step S209, the pseudo determination evaluation unit 17 calculates the pseudo correct answer rate Gz (average evaluation value) by dividing the evaluation value q by v_max which is the number of evaluations (number of divided sets). Accordingly, one pseudo correct answer rate Gz (average evaluation value) is obtained corresponding to one waveform division number z.

  Step S210 (calculation of conditional probability) and step S211 (determination of important segment) will be described later.

  Returning to FIG. 2, in step S109, the pseudo judgment evaluation unit 17 determines whether the pseudo correct answer rate Gz calculated in step S209 is smaller than the pseudo correct answer rate Gz-1 when the previous waveform division number z-1 or not. Determine whether or not. When Gz is equal to or greater than Gz−1, it is determined that there is a possibility that a higher pseudo correct answer rate may be obtained, and the process returns to step S105, the waveform division number z is incremented by 1, and the same procedure is repeated. On the other hand, when the pseudo correct answer rate Gz is smaller than Gz-1, it is determined that a pseudo correct answer rate larger than this cannot be obtained, and the process proceeds to step S110.

  Prior to the description of steps S110 and S111, details of step S205 (modeling processing by training learning) in FIG. 10 will be described.

  FIG. 13 is a flowchart showing details of step S205 in FIG. Here, a case where the number of waveform divisions z = 4 will be described as an example.

First, in step S301, the probability / likelihood calculation unit 16 calculates the normal and abnormal occurrence probabilities based on the classes of training data in the training data set (segmented by z = 4) as prior probabilities p. Calculate as (Ci). For example, if the size of the training data set is 200, there are 140 normal classes and 60 abnormal classes, normal prior probability p (C ₁ = normal) = 0.7, abnormal prior probability p (C ₂ = Abnormal) = 0.3 (see the upper left of FIG. 14). This step S301 may be performed only once, and the processing of this step may be skipped from the next time.

  Next, in step S302, the pseudo determination evaluation unit 17 performs initialization and sets i indicating variable ID (channel ID) to 0 and j indicating section ID to 0.

  Next, the pseudo determination evaluation unit 17 increments the channel i by 1 in step S303a, and increments the variable (channel) j by 1 in step S303b.

  Next, in step S304, the pseudo determination evaluation unit 17 uses the k-nearest neighbor method proven in the time series data classification problem for the pseudo determination target data Tv to estimate the class of the pseudo determination target data Tv ( (Pseudo-judgment). In the k-nearest neighbor method, k cases closest to the pseudo judgment target are extracted in the feature space, and the class occupying the largest number among the classes of the k cases is selected as the pseudo judgment target. It is the determination method determined as an estimation class. This will be described in detail below.

  For each segment data of each variable (each channel) in the pseudo judgment target data Tv, the k nearest segment data is the same variable from the remaining training data (remaining divided sets) other than the pseudo judgment target data Tv Find in. Focusing on variable (channel) 1 and segment s1 of pseudo judgment target data dN + 1 in FIG. 15, the top k segment data s1 having the highest similarity (closest distance) to segment data s1 of pseudo judgment target data d1 Is found from the time series data of the variable (channel) 1 of the remaining training data. However, k = 5. In the illustrated example, the segment data s1 of the variable 1 in the training data d13, d14, d15, d17, d16 is specified. Here, for the calculation of the distance between the segment data, a scale such as Dynamic Time Warping (DTW) distance or Euclidean distance may be used. Here, the distances from the training data d13, d14, d15, d17, d16 to the variable 1 segment data s1 are calculated as 3.5, 9.3, 12.9, 13.2, and 14.1, respectively. In the figure, dist (x, y) indicates the distance between the segment data x and the segment data y.

When the top k (= 5) segment data are specified in this way, the class with the largest number among the classes related to these segment data is specified. When this is formulated, Equation 1-1 is obtained.

  In the example of FIG. 15, all classes of training data d13, d14, d15, d17, and d16 are normal. That is, the frequency of abnormality and the frequency of normality are freq (abnormality, normal) = (0, 5). Therefore, the estimation result is determined to be normal according to the above equation 1-1. Here, the actual class of the pseudo determination target data d1 is abnormal. Therefore, this estimation result is an incorrect answer (error).

  Next, in step S305, the pseudo determination evaluation unit 17 updates the normal and abnormal frequency distribution table for each variable and for each section based on the normal frequency and the abnormal frequency obtained in step S304. A format example of the frequency distribution table is shown in FIG. For example, the frequency distribution table is prepared for each waveform division number z. Initially, all items in the frequency distribution table are set to zero. In the above calculation example, 5 is added to the normal item of section s1 in the distribution table of channel 1 (upper left of FIG. 16), and nothing is added to the abnormal item.

  Next, in step S306, the pseudo evaluation determination unit 17 updates the score table according to whether the estimation in step S304 is correct or incorrect. The score table stores a score for each combination of all channels and segments (sections) selected in the process of proceeding with pseudo judgment evaluation (see the upper diagram of FIG. 17 described later). The initial value of each square in the score table is 0. If the answer is correct, a predetermined score (1 in this case) is added to the corresponding cell. For example, as shown in FIG. 18, if the estimation result in step S304 is correct for the segment s2 of the variable (channel) 1 in the pseudo determination target data d1, the cells corresponding to the variable (channel) 1 and the segment 2 are displayed. Add 1 to the score score (ch1, s2). That is, score (ch1, s2) = 0 + 1. A score table exists for each waveform division number z.

  In step S309, it is determined whether or not the segment j has reached jmax. If not, the process returns to step S303b to increment j and select the next segment. If reached, the process proceeds to the next step S310. FIG. 18 shows how the nearest neighbor calculation is performed in step S304 for the section s2 of the variable 1 (channel 1). Here, the estimation result is abnormal, and the pseudo determination target data dN + 1 is also abnormal, so it is a correct answer.

  In step S310, it is determined whether or not the variable (channel) i has reached imax. If not, the process returns to step S303a to select the next variable (channel). If reached, the next step S311 is performed. Proceed to FIG. 19 shows a state where the nearest neighbor calculation in step S304 is performed for section s2 of variable 3 (channel 3). Here, the estimation result is normal, and the pseudo-determination target data dN + 1 is incorrect, so it is incorrect.

  In step S311, the evaluation value of the pseudo determination target data Tv is calculated. If the judgment result is abnormal and correct for at least one of the total 16 judgments (S304) made for the segments s1 to s4 for each of the variables 1 to 4, the evaluation value r is 1.0, and the others When it is 0.0. Alternatively, 1.0 may be used when the number of correct answers is greater than the number of incorrect answers, and 0.0 may be set when the number of correct answers is less than or equal to the number of incorrect answers. Alternatively, the ratio of the number of correct answers to the number of determinations may be set as the evaluation value r. When step S311 is completed, this flow is ended, and the process returns to step S207 in FIG.

  When a plurality of training data are included in one divided set, the processing of steps S302 to S310 is performed for each training data, and thereafter, an evaluation value may be calculated according to the same criteria.

  In step S207 of FIG. 10, the evaluation value r is added to q to update q. Then, the process proceeds to selection of the next pseudo determination target data Tv, and the flow of FIG. 13 is performed in the same manner.

  When all the training data (all divided sets) are selected once as the pseudo judgment target data (pseudo judgment target data set) Tv, the process proceeds to step S209, and the pseudo correct answer rate (average evaluation value) Gz = q / Calculate v_max. That is, the average of evaluation values among all training data (all divided sets) is calculated.

  In the next step S210, the probability / likelihood calculation unit 16 performs normal and abnormal conditional probabilities p (X | C for each variable and each segment combination based on the frequency distribution table updated in step S305 of FIG. ). For example, for a set of variable 2 and segment s2, p (X2 = s2 | C) = (abnormal = 0.8, normal = 0.067) is calculated.

  Here, p (X | C) is the probability of normalization when the variable X takes each segment under the condition of C, that is, C = abnormal and C = normal. This shows the probability of an abnormality. That is, the probability of normalizing the normal / abnormal distribution in each segment with C = normal and C = abnormal is p (X | C). If only P (X) is normal, the probability of normality exceeds the probability of abnormality, but if the probability of normal occurrence in P (C) is five times higher than the probability of occurrence of abnormality, this difference in prior probabilities Must be corrected to 1/5 to consider the conditional probability. Therefore, here, the probability in each segment normalized by p (C) is calculated as p (X | C). For each C (ie normal and abnormal) value, the probability is calculated for each C so that the sum of the probabilities in all the segment types to be taken is 1.

  As an example, calculation examples of conditional probabilities for variable 2 and variable 3 are shown in the upper right and center of FIG. If the illustrated frequency distribution f (X2 | C) is obtained for the variable 2, the conditional probability p (X2 | C) is calculated as shown in the table based on this frequency distribution. That is, for each of normal and abnormal, the conditional probability is calculated by dividing the frequency of each variable by the total frequency. The same method is used for variable 3. Although not shown, conditional probabilities are similarly calculated for variables 1 and 4. If the abnormal conditional probability is 0.1, the normal conditional probability is 0.9, and so on, this is normal compared to the abnormal conditional probability of 0.55 and normal conditional probability of 0.45. The probability is very high, and such a probability is close to certainty.

  In step S211, an important segment (important section) is determined for each variable. That is, all the training data is selected as the pseudo judgment target data Tv, and the flow of FIG. 13 is performed for each, so that a score table is finally obtained as shown in the upper part of FIG. The pseudo judgment evaluation unit 17 selects each important segment (important section) of each variable (channel) based on this score table, and records the information in the segment storage unit 15. Specifically, in the score table, the segment (section) with the highest score is selected as an important segment for each variable. For example, in the variable (channel) 1, since the segment s1 has the highest score, the segment s1 is selected as the important segment. Similarly, for the variables 2 to 4, segments s2, s2, and s4 are selected as important segments. A selection method when there are a plurality of segments having the same score and other selection methods will be described later.

  When the important segment is determined in step S211, this flow is ended, and the process proceeds to step S109 in FIG.

  In step S109, as described a little earlier, it is checked whether the pseudo correct answer rate Gz calculated in step S209 is smaller than the pseudo correct answer rate Gz-1 when the waveform division number is z-1. If it is smaller, the best model selection unit 19 stores the important segment for each variable selected when Gz-1 is obtained in the best model storage unit 18 as the best segment (best section) in the next step S110. .

  Further, when it is determined in step S106 that the waveform division number z is larger than the maximum waveform division number z_max, the best segment (best section) in Gz-1 is specified and stored in the same manner. However, in this case, since z is incremented in step S105, it should be noted that in this case, Gz-1 matches Gz_max.

  In addition, once the best segment (best section) of each channel (variable) is obtained, if there is room to improve the pseudo judgment performance by merging other segments (other sections) in each channel, the merge approach It is good also considering the segment which introduce | transduced and merged as the best segment.

  Next, in step S111, the best model selection unit 19 obtains normal and abnormal prior probability information and conditional probabilities for each variable obtained in step S210 (corresponding to z from which the best model was obtained) as the best model storage unit. Store in 18. The conditional probability stored may be only the probability of the segment specified as the best segment in each variable.

  In addition, a model formula (described later) used for determination by the client is stored in the best model storage unit 18. The model formula can be automatically generated once the best segment for each variable is determined.

  The best model selection unit 19 reads the segment data of the best section of each variable and the corresponding class from the segment storage unit 15 and stores them in the best model storage unit 18. The segment to be read may be the entire training data, a predetermined number of training data for each of normal and abnormal, or may be determined by other criteria.

  The best model selection unit 19 also stores in the best model storage unit 18 detailed information (time length) of each section (segment) when divided by the selected waveform division number z. At least the detailed information of the section corresponding to the best segment is stored.

  These pieces of information data stored in the best model storage unit 18 form a determination model. An example of data stored in the best model storage unit 18 is shown in FIG.

  In the example of FIG. 20, the segment s2 is specified as the best segment in all the variables. In FIG. 20, the notation of p (X1 | C), p (X3 | C), p (X4 | C) is not shown in detail, and simply p (X1 | C), p ( Only X3 | C) and p (X4 | C) are shown. How to read the model formula in (2) will be described later.

  Next, the transmission unit 20 transmits the determination model stored in the best model storage unit 18 to the client. The model formula type data may be given to the client in advance, and the client may generate the model formula from the type data based on the best segment (best section) of each variable. In this case, the server may not include the model formula in the determination model sent to the client. When a plurality of clients are connected to one server, the determination model is transmitted to the plurality of clients. By receiving this determination model, the client is ready for abnormality determination.

(Supplement) Here is a supplementary explanation of the conditional probability. Usually, in order to estimate a determination result by a Bayes classifier, it is necessary to obtain a conditional probability p (Xi | C) in each variable (attribute) Xi. Here, the type of attribute value of attribute Xi becomes a problem. Normally, when calculating the conditional probability, it is calculated based on how often the discrete attribute value ai of the attribute Xi occurs, but in this embodiment, segment data obtained by cutting out the time series waveform Since there is no attribute value in itself, the probability cannot be calculated. Usually, time-series waveforms are clustered and divided into several category categories, and each category type is calculated as an attribute value in the domain of time-series clustering and time-series classification. It's not easy to decide how many types to choose. In addition, when the number of divisions increases, the number of clusters handled for each variable must be obtained when the types of variables handled at the same time increase, which is unrealistic. Although it is conceivable to handle the segments divided by each variable as all the attributes Xi, there is a problem that the probability calculation cost increases. Therefore, in this embodiment, in the attribute Xi of each variable, by setting the segment type with the highest pseudo judgment performance as the attribute value, the time series data is discretized within the framework of the Bayesian classification problem and the information is not dropped. Made it possible to handle. If it does in this way, it becomes possible to express the segment selected by each variable with certainty. For example, as described above, the conditional probability can be expressed as p (X2 = s2 | C) = (abnormal = 0.8, normal = 0.067).
(First modification of server)
As shown in the score table shown in FIG. 17, when there is only one segment (section) with the highest score in each variable, the important segment is uniquely determined. However, when there are a plurality of segments having the same point, this method cannot be uniquely determined. In such a case, an important segment (important section) is determined by the following method.

  FIG. 21 is a diagram for explaining an important segment determination method according to the first modification.

  In the example of FIG. 21, the scores of the segments (s1 and s4) of the variable (channel) 4 are the same (each is 9 points). For variables 1, 2, and 3, there is only one segment with the highest score, and segments s1, s2, and s2 are selected, respectively.

  In this case, a plurality of candidates are generated by taking all combinations of segments having the highest score among the variables 1 to 4. Then, as shown in the flowchart of FIG. 22, the likelihood of abnormality is calculated for each candidate (S222). Likelihood is a measure representing the plausibility of a decision and is defined as the total product of probabilities. The calculated likelihoods are compared, and the candidate with the highest likelihood is selected (S223).

  In the illustrated example, as candidate c1, (variable 1, variable 2, variable 3, variable 4) = (s1, s2, s2, s1), as candidate c2 (variable 1, variable 2, variable 3, variable 4) = (S1, s2, s2, s4) are generated, and the likelihood of abnormality is calculated for each candidate. Then, the candidate with the highest likelihood is selected (pseudo-judgment evaluation by likelihood calculation).

Here, the likelihood is calculated according to the following equation. Since a vector composed of normal likelihood and abnormal likelihood is obtained by the calculation of the following equation, the likelihood of the abnormal is selected, and the likelihood of abnormality is compared between the candidates. p (C) is a prior probability, and p (X _j = s _i | C) is a conditional probability. As the conditional probability, the value calculated in step S210 in FIG. 10 can be used.

  In the example of FIG. 21, when the likelihood of abnormality of each candidate is actually calculated, it becomes 0.087 for candidate c1 and 0.084 for candidate c2, and candidate c1 has a larger value, so candidate c1 is selected. That is, the important segment of the variable (channel) 4 is determined as s1. As shown in the example in Fig. 21, if there is only one variable with a tie segment, compare the conditional probabilities of abnormalities between these tie segments and select the segment with the largest value as the important segment. The same result can be obtained even if it is determined.

(Second modification of server)
(No. 1) In the present embodiment, the important segment is specified by the maximum value of the score, but as another method, it is possible to select a combination of segments between the highest variables in the likelihood calculation. This is because the best segment is selected for each variable, but it is not always the best determination accuracy when a pseudo determination is made for the entire variable. In the process of this modification, for example, according to the above equation 1-2, the likelihood of abnormality is calculated for all the combinations of segments between variables (S222 in FIG. 22), and the combination with the highest likelihood may be selected. (S223 in FIG. 22).

  (Part 2) Further, the following method is also possible as a second modification.

  That is, the lower limit threshold θ of the evaluation score is determined in advance, and all segments having a score higher than the lower limit threshold are selected as candidates, and the final important segment is determined from the candidates.

  FIG. 23 is a diagram for explaining an important segment determination method according to the second modification.

  The lower threshold θ is set to 5. Segments with scores greater than 5 in each of the variables (channels) 1-4 are selected as candidates. For variable 1, segments s1, s4, for variable 2, s2, for variable 3, s2, for variable 4, s1, s3, s4 are selected. When the selected segments are combined between the variables, the following six candidates c1 to c6 are obtained.

c1 = (s1, s2, s2, s1), c2 = (s4, s2, s2, s1), c3 = (s1, s2, s2, s3), c4 = (s4, s2, s2, s3), c5 = (S1, s2, s2, s4), c6 = (s4, s2, s2, s4)
The likelihood of abnormality is calculated for each candidate in the same manner as “No. 1” (S222 in FIG. 22). Then, the candidate with the highest likelihood is selected (S223 in FIG. 22). In this example, the likelihood L1 of the candidate c1 is 0.013, the likelihood L2 of the candidate c2 is 0.031, the likelihood L3 of the candidate c3 is 0.024, the likelihood L4 of the candidate c4 is 0.062, and the candidate c5 The likelihood L5 is calculated as 0.033, and the likelihood L6 of the candidate c6 is calculated as 0.093. Since the likelihood of the candidate c6 is the largest, the segment included in the candidate c6 is determined as the important segment. That is, segment s4 is determined as the important segment for variable 1, segment s2 is determined as variable 2, segment s2 is determined as variable 3, and segment s4 is determined as important segment in variable 4.

  According to such an important segment determination method, an important segment is not selected in a variable (channel) in which there is no segment with a score higher than the threshold θ. It can be said that such a variable is less necessary for abnormality detection, and there is an advantage that the data of the variable (channel) is not used for abnormality determination. When there are a plurality of candidates with the largest likelihood, a plurality of candidates may be selected.

(Modification 3 of the server)
Depending on the sensing target (monitoring target), it may not be necessary to consider the time difference of each sensor. In that case, a segment at the same position in each variable (channel) may be selected, and estimation may be performed by the k-nearest neighbor method in units of the same segment.

In this case, the conditional probability is calculated for each flow in FIG. 13 corresponding to step S205 in FIG. 10 (the conditional probability is calculated for each divided set identifier v). As shown in FIG. 24, normal and abnormal likelihoods are calculated by multiplying the normal and abnormal conditional probabilities of the same segment (for example, s2) between the variables, and further multiplying the prior probabilities, and the values are large. The state of the direction is adopted. This is shown by Equation 1-3 below.

  If the adopted state matches the state of the pseudo-determination target data, the score is incremented by 1, and if not, the score is not added. The same calculation is performed for other same segments (s1, s3, s4) between each variable, and normal or abnormal is selected, and if correct, the score is added. In this way, the score table is updated (the size of the score table is 1 × 4 compared to 4 × 4 in the present embodiment).

In the example of FIG.

(For ease of understanding, the values in the table of FIG. 14 are used here). Therefore, the likelihood of abnormality is higher than normal. Therefore, in this case, normal is the estimation result. Because pseudo-class determination target data d ₁ is normal, the estimation result is a correct answer, 1 is added to the score score (S2) for the segment s2.

In some cases, it is possible that the bias of the class prior distribution has already been corrected by the k-nearest neighbor method. Therefore, Formula 1-4, in which prior probability distribution p (C) is removed from the estimation formula shown in Formula 1-3, may be used instead of Formula 1-3.

(Fourth modification of server)
This modification shows another method for determining the best segment of each variable (channel).

  FIG. 25 is a diagram for explaining a method for determining the best segment according to the fifth modification.

  First, the segment length and provisional position of the best segment are determined in advance for each variable. Each segment (section) with a predetermined segment length is placed at a provisional position, and each segment is shifted back and forth along the time axis from the provisional position by the minimum movement interval Δ unit, and the position where the pseudo judgment evaluation value Gz is the best is obtained. The combination is determined, thereby determining the best segment. In the example of FIG. 25, the maximum shift width is the absolute value of the segment length. A combination with the highest likelihood of abnormality is selected for all the combinations of segments between the variables when moving in units of Δ. In this way, the combination of the position of each possible segment is searched, and the best segment for each variable is determined by performing pseudo judgment evaluation comparison.

(Fifth modification of server)
In this modification, an example will be described in which a stochastic dependency is known from the first variable to the second variable. Here, the first variable is described as variable (channel) X3, and the second variable is described as variable (channel) X2. The variable dependency relationship, that is, the dependency relationship between sensors, is specified in advance by the user to the server.

When there is a stochastic dependency from the variable X3 to the variable X2, the likelihood is calculated according to Equation 1-5, and the decision estimation equation (the equation that estimates the larger of normal and abnormal likelihoods) is Like -6.

When the prior probability related to the class distribution is not used as in the fourth modification, the determination may be made using Expression 1-7 instead of Expression 1-6.

As a difference from the first embodiment, in this modification, the likelihood is calculated by Expression 1-5. In addition, as a model formula stored in the best model storage unit 18, a formula shown in Formula 1-8 is used. The meaning of P (Xnew = s2 | C) will be described later. In this modification, the normal and abnormal conditional probabilities for the variable X2 to the variable X3 are also stored in the best model storage unit 18 and included in the determination model.

  An example of a determination model according to this modification is shown in FIG. FIG. 27 schematically shows the concept of the model formula of Formula 1-6.

  Here, p (X2 = s2 | X3 = s2, C) indicates that C2 and X3 = s2 have a probability dependency relationship between X2 = s2 depending on a combination of possible values. In other words, this means that there is an influence that should be a synergistic effect (positive or negative) between C and the combination of values that X3 = s2.

  A calculation example of the conditional probability p (X2 = s2 | X3 = s2, C) is shown below FIG. p (X2 = s2 | X3 = s2, C) is calculated from the frequency f (X2 = s2 | X3 = s2, C). Here, to obtain the frequency f (X2 = s2 | X3 = s2, C), the sum of the frequencies f (X2 = s2 | C) and p (X3 = s2 | C) is used as an approximation. This is one way to prevent the frequency of each cell in the conditional probability table from becoming too small when k is small. In the example shown in the figure, f (X2 = s2 | C = abnormal) = 4 and f (X3 = s2 | C = abnormal) = 2, so that f (X2 = s2 | X3 = s2, C = abnormal) = 4 + 2 = 6. Also, f (X2 = s2 | C = normal) = 1 and f (X3 = s2 | C = normal) = 2, so f (X2 = s2 | X3 = s2, C = normal) = 1 + 3 = 4 It becomes.

  As a calculation method other than this, if infrequent frequency is not an issue, 3 + 2 = 5 satisfying f (X3 = s2 | C = abnormal) and f (X3 = s2 | C = normal) Of these data, the frequency table of f (X2 = s2 | X3 = s2, C) may be created by counting the frequency of abnormalities and normality of f (X2 = s2).

  Note that the method for calculating the conditional probability from the frequency table is the same as p (X2 | C) and p (X3 | C) already described, and therefore, redundant description is omitted.

(client)
The client in FIG. 1 will be described.

  The client includes a sensing unit 30 that newly senses data using a plurality of sensors, and stores data sensed by the sensing unit 30 (this data will be referred to as determination target data) in the sensing data storage unit 31.

  The determination target data input unit 32 monitors whether or not the determination target data is stored in the sensing data storage unit 31. If new determination target data is input, the determination target data is read and input to the waveform preprocessing unit 33.

  The waveform preprocessing unit 33 performs waveform preprocessing in the same manner as described in the waveform preprocessing unit 13 of the server.

  The model receiving unit 34 receives the determination model sent from the server.

  The determination model storage unit 35 stores the determination model received by the model reception unit 34.

  As shown in FIG. 28, the segment selection unit 36 scans the determination target data at regular intervals based on the segment template composed of the best segments of each variable included in the determination model, and the scanned segment template Is read out and output to the abnormality determination unit 39.

  The abnormality determination unit 39 calculates the likelihood of abnormality and normality based on the cut-out data of each variable input from the segment selection unit 36 and the determination model. The abnormality determining unit 39 determines that the abnormality is abnormal if the likelihood of abnormality is larger than the normal likelihood, and otherwise determines that the abnormality is normal.

For example, when the data (waveform) is cut out as shown on the left in Fig. 29 for each variable, the distances to the corresponding segment data (Fig. 20 (4) or Fig. 26 (4)) are compared in the judgment model. Then, the determination result (normal or abnormal) of the top k ′ segments closest to each other is specified. However, 1 <k ′ <k, and k and k ′ are set in advance as system parameters. Then, the likelihood calculation is performed using the model formula of (2) in FIG. 20 or (2) in FIG. 26, and the likelihood that becomes normal and the likelihood that becomes abnormal are calculated. The way

Output as. When both likelihood values are the same, one of the predetermined values is taken as the determination result.

  More specifically, when following the model formula of FIG. 20 (2), first, the ratio between normal and abnormal (conditional probability) p (Xnew = s2 | C) is calculated for each variable. For example, for a variable, when k '= 5, normal count is 1, and abnormal count is 4, p (Xnew = s2 | C = normal) = 0.2, p (Xnew = s2 | C = abnormal) = 0.8. Then, p (Xnew = s2 | C) for each variable is multiplied by the normal and abnormal probabilities p (Xj = s2 | C) of the relevant section (section) of the conditional probability table included in the judgment model (this example) So in all variables the section is s2.) That is, p (X1 = s2 | C), p (X2 = s2 | C), p (X3 = s2 | C), p (X4 = s2 | C) for each variable p (Xnew = s2 | C) Multiply Further, normal and abnormal prior probabilities p (C) are multiplied, thereby obtaining normal and abnormal likelihoods. In the case of following the model formula of FIG. 26 (2), normal and abnormal probabilities of p (X2 | X3, C) are further read out and multiplied to obtain normal and abnormal likelihoods. Finally, the state of the higher likelihood of normal and abnormal is determined.

  In this way, the segment selection unit 36 and the abnormality determination unit 39 cut out the segment data by sliding the segment template from the determination target data in the same manner as the likelihood calculation by assigning the sliding window, and the determination model and the segment The likelihood is calculated using the data.

  By sliding the segment template in this way, the abnormality determination unit 39 outputs the determination results for the number of times scanned. For example, FIG. 29 shows a case where it is normal in a certain part of the first half but is determined to be abnormal in the second half.

  The notification display unit 38 notifies or displays the determination result by the abnormality determination unit 39. As an example, FIG. 30 shows a screen display in the case of highlighting only the abnormality determination portion of the determination target data. The determination result is notified, for example, to a remote monitoring terminal or a maintenance staff or a staff of the server using a display or a speaker.

  The device control unit 37 controls the operation to be monitored according to the determination result by the abnormality determination unit 39. For example, when it is determined that there is an abnormality, the monitoring target is stopped in an emergency.

  The determination result storage unit 40 has a predetermined time length (for example, the same time length as already stored training data) including the determination result in the abnormality determination unit 39 and the data of each variable extracted for determination from the determination target data. Accumulate time series data.

  The determination result transmission unit 41 transmits the time series data of each variable and the corresponding determination result to the server.

  The server determination result receiving unit 22 receives the time series data of each variable transmitted from the client and the corresponding determination result, and the notification unit 21 displays or notifies the monitoring member of these. After the monitoring person confirms that the determination result is correct, the notification unit 21 adds the time series data and the determination result to the training data storage unit 11 of the server in response to an instruction input from the monitoring person.

  If there is an error in the determination, the determination result is corrected according to the instruction input from the supervisor, and then the time series data and the corrected determination result are stored in the training data storage unit 11.

  The training data input unit 12 of the server may detect that the data in the training data storage unit 11 has been updated and recalculate the determination model.

  In this way, by preparing a mechanism that can always give new training data to the apparatus, the determination model can be refined, that is, the accuracy can be improved continuously. This means that there is a possibility that the abnormality determination accuracy can be improved in daily monitoring operations, and it is considered to be particularly effective in an area where high abnormality determination performance is required.

  FIG. 31 is a flowchart showing an operation flow from input of determination target data in the client until determination by the abnormality determination unit 39 is performed.

  The determination target data input unit 32 reads the determination target data in the sensing data storage unit 31 and inputs it to the waveform preprocessing unit 33 (S401).

  The waveform preprocessing unit 33 performs preprocessing on the determination target data (S402), and the segment selection unit 36 extracts data based on a segment template composed of the best segment of each variable included in the determination model (S403). ). Then, the cut out data is input to the abnormality determination unit 39 (S404).

  For each variable (S406), the abnormality determination unit 39 calculates k'-nearest neighbor (S407), and calculates a conditional probability (the ratio of normal and abnormal based on k'-nearest neighbor) (S408). When the processing of S407 and S408 is completed for each variable (YES in S409), the normal likelihood and the abnormal likelihood are calculated according to the above-described model formula (S410). The abnormality determination unit 39 compares the likelihood that the likelihood of abnormality is normal and gives the state having the larger value as the determination result (S411).

  Here, when performing remote monitoring, it may be insufficient to know whether the monitored elevator is in a discrete state of normal or abnormal. This is because in order to perform timely maintenance, it is necessary to prepare replacement parts and prepare maintenance personnel dispatch for parameter adjustment when it is found that the degree of abnormality is slightly high. In that case, it is necessary to present a quantitative value of how abnormal it is.

  In this embodiment, the elevator is remotely monitored by detecting an abnormality in the scheme of FIG. For example, by changing the parameters during the shipping test, it is possible to obtain in advance a car swing acceleration waveform during driving in a clearly normal state and a clearly abnormal state. In the following, a procedure for quantitatively calculating the degree of abnormality indicating how abnormal an inspection target waveform measured at the time of diagnosis is determined to be “abnormal” will be described.

  In the calculation of the k-nearest neighbor (S407) of the abnormality determination unit 39, the observed waveform obtained from the monitored elevator and the similar distance from the normal waveform or the abnormal waveform nearest to the k-nearest selected from the model waveform group Based on this, the degree of abnormality of the monitored elevator is calculated. When creating a judgment model, the normal state and abnormal state of the elevator are set in advance, and their observed waveforms are collected.

  As an example of an elevator abnormality, a description will be given by taking “car swinging” of a riding car as an example. Under normal conditions, the rotational torque of the hoisting machine is appropriately adjusted according to the load weight in the riding car, so that the car does not shake at the time of the initial movement of the car. However, such adjustment by load compensation may become inappropriate due to changes over time. For example, a basket shake can be adjusted in a regular maintenance operation every other month, but if the elevator becomes obsolete, the adjustment may suddenly go out during a regular maintenance operation every other month. . In this case, the load compensation at the time of the initial movement of the riding car becomes inappropriate, and the car shakes. The load compensation amount can be set in advance as an elevator control parameter, and when the load compensation amount deviates, the maintenance staff changes the compensation amount parameter to return to the normal state. In rare cases, the load sensor installed under the floor of the car may deteriorate or break down. In that case, replace the load sensor.

  In the normal state, such an elevator car swing is almost 0 mm immediately after the elevator door is closed, while the deviation between the floor of the passenger car and the floor of the passenger is almost 0 mm. It appears as a visible phenomenon such as a 10mm gap between the floor and the height of the boarding board. This can be clearly seen by observing the movement of the car with a security window, and appears as a misalignment of the security windows on the floor side and the basket side.

  When creating a normal / abnormal discrimination model, a car acceleration waveform in an abnormal state is collected in a state where the abnormal state is intentionally created. For example, the control parameter for determining the load compensation amount described above is intentionally adjusted so that a deviation of 10 mm is generated, and the car acceleration waveform at that time is collected.

  It is known in advance that the acceleration waveform in the abnormal state collected in this way is a quantitative state with a deviation of 10 mm. Therefore, for example, a normal state misalignment of 0 mm and an abnormal state misalignment of 10 mm, and using the similar distance of how much the respective car acceleration waveform is similar to the waveform to be inspected, normal / abnormal state A quantitative value representing the degree of abnormality can be calculated from the positional deviation ratio.

  Hereinafter, a specific calculation example will be described along the flowchart of FIG.

  First, the number of channels (variables) is 2, normal state value VN = 0mm, abnormal state value VA = 10mm, number of normal waveform group N in model n = 5, number of abnormal waveform group A waveform number m = 5 The normal waveform group of ch1 is (N11, N12, ..., N1n), the abnormal waveform group of ch1 is (A11, A12, ..., A1m), and the normal waveform group of ch2 is (N21, N22, ..., N2n) It is assumed that the abnormal waveform group of ch2 is (A21, A22, ..., A2m). In addition, the inspection target waveform is E, the ch1 inspection target waveform is E1, the ch2 inspection target waveform is E2, and the k-closest upper cut-off parameter K is K = 5.

  Next, for each channel, the top k waveforms (k-nearest neighbor waveforms) having a high similarity to the waveform to be examined (the distance between waveforms is small) are divided into a normal waveform group and an abnormal waveform group. In the case of ch1, for example, it is assumed that the k-nearest neighbor waveform (k-NN) vector is (A13, A12, N12, A15, N11). Here, the vector is an ordered list. Further, it is assumed that the waveform similarity vector of the nearest waveform of ch1 is (0.14, 0.21, 0.3, 0.43, 0.5). Similarly, it is assumed that the k-nearest neighbor waveform vector of ch2 is (A25, A21, A22, N21, N25) and the ch2 waveform similarity vector is (0.03, 0.12, 0.15, 0.21, 0.3).

  Here, the average value of the waveform similarity is obtained for each of the abnormal case and the normal case. That is, the waveform similarity average value SA1 of ch1 in the case of abnormality is SA1 = (0.14 + 0.21 + 0.43) /3≈0.26 (step S1), and the waveform similarity average value SN1 of ch1 in the normal case is SN1 = (0.3 + 0.5) /2=0.4 (step S2).

  Further, a ratio between the normal degree and the abnormal degree is obtained. Here, the normal weight and the abnormal weight are calculated so that the sum of the normal degree and the abnormal degree is 1. That is, the abnormal weight WA1 is WA1 = (1 / SA1) / (1 / SA1 + 1 / SN1) = (1 / 0.26) / (1 / 0.26 + 1 / 0.4) ≈0.61 (step S3), and the normal weight WN1 is WN1 = (1 / SN1) / (1 / SA1 + 1 / SN1) = (1 / 0.4) / (1 / 0.26 + 1 / 0.4) ≈0.39 (step S4). The normal weight and the abnormal weight are calculated after the waveform similarity average values are calculated by repeating the process for all k in the channel in the flowchart of FIG. 33 (step S5 = YES).

  Then, from the calculated channel ch1 (i = 1), the abnormal degree weight WAi, the normal degree weight WNi, the abnormal class state value VA, and the normal class state value VN are calculated and the quantitative estimate VE1 of the abnormal degree is calculated. (Step S6). Specifically, the estimated shift amount VE1 of channel i = 1 is VE1 = WA1 × VA1 + WN1 × VN1 = 0.61 × 10.0 mm + 0.39 × 0 mm≈6.1 mm.

  The above calculation is performed similarly for channel ch2 (i = 2). In case of abnormality, ch2 waveform similarity average value SA2 = (0.03 + 0.12 + 0.15) /3=0.1, and in normal case, ch2 waveform similarity average value SN2 = (0.21 + 0.3) / 2 ≒ 0.26, abnormal weight WA2 = (1 / SA2) / (1 / SA2 + 1 / SN2) = (1 / 0.1) / (1 / 0.1 + 1 / 0.26) ≒ 0.72, and normal weight WN2 = (1 /SN2)/(1/SN2+1/SN2=(1/0.26)/(1/0.1+1/0.26)≈0.28 Therefore, the estimated deviation VE2 = WA2 × VA2 + WN2 of channel i = 2 × VN2 = 0.72 × 10.0mm + 0.28 × 0mm ≒ 7.2mm.

  Finally, after the calculation for all the channels is completed (step S7 = YES), an estimated deviation amount VE (referred to as “quantitative state value”) is calculated by obtaining an average of the estimated deviation amounts VEi of each channel. (Step S8). That is, VE = (6.1 mm + 7.2 mm) /2≈6.7 mm. Therefore, the amount of deviation of the elevator car at the time of diagnosis is about 6.7 mm. In the case of this example in which the normal is 0 mm and the abnormality is 10 mm, the quantitative state value of 6.7 mm indicates that a considerable deviation has progressed.

  In this case, if parts need to be replaced, arrange them in time for maintenance. If the deviation seems to be further advanced, the maintenance inspection time can be adjusted in advance according to the state of the basket shaking.

  According to the embodiment described above, it is possible to identify the analysis target in the normal state / abnormal state from the data having an arbitrary waveform shape, and the continuous value of the target object is inconsistent with the discrete value. Can be estimated simultaneously. Therefore, it is possible to capture not only a discrete state change of the object but also a minute state change representing an abnormal sign.

  FIG. 34 shows an example of a hardware configuration for realizing a server and a client.

  The server includes a CPU 51, a RAM 52, a ROM 53, an HDD 54, an I / O controller 55, a display 56, a speaker 57, an I / O 58, and a network interface 59. The training data storage unit 11 and the segment storage unit 15 of the server are configured by the HDD 54, for example. The model transmitting unit 20 and the determination result receiving unit 22 can be configured by a network interface 59. The other elements 12, 13, 14, 16, 17, 19, and 21 can be configured by logic circuits as program modules that are executed by the CPU 51, for example. The program modules are stored in the ROM 53 or the HDD 54, read out by the CPU 51, developed in the RAM 52, and executed, whereby the operations of the corresponding logic circuits are realized.

  The client includes a CPU 61, a RAM 62, a ROM 63, an HDD 64, an I / O controller 65, a display 66, a speaker 67, an I / O 68, and a network interface 69. The sensing data storage unit 31, the determination model storage unit 35, and the determination result storage unit 40 of the client can be configured by the RAM 62 or the HDD 64, for example. The model receiving unit 34 and the determination result transmitting unit 41 can be configured by a network interface 69. The other elements 32, 33, 36, 37, and 38 can be configured by a logic circuit as a program module to be executed by the CPU 61, for example. The program modules are stored in the ROM 63 or the HDD 64, and the CPU 61 reads the program modules, develops them in the RAM 62, and executes them, thereby realizing the operations of the corresponding logic circuits.

  In this embodiment, the server and the client are separated. However, a part or all of the server functions may be performed by the client, or a part or all of the client functions may be performed by the server. In the present embodiment, the process executed by a computer includes a case where a single computer executes the process, and a case where the process is distributed and executed by a plurality of computers.

(Client variants)
In the explanation of the client so far, the likelihood is calculated by applying the segment template at regular intervals on the determination target data. However, this method sometimes exceeds the upper limit of the time required for the determination. In particular, information processing equipment such as a remote monitoring terminal installed in the field has severe restrictions on computing resources (memory amount, CPU, etc.) that can be used for judgment processing. If it is necessary to stop the monitoring target device (equipment) immediately in case of an abnormality, or if there is a communication performance limit on the communication path from the remote monitoring terminal in the field to the monitoring center server, scan at regular intervals. Sending all the results to the server is unrealistic.

  In such a case, an upper limit threshold is determined for each variable in advance, and the upper limit threshold for each variable is stored in the determination model storage unit 35. Then, when the value of the variable for any one variable or all of the variables exceeds the upper threshold, the device control unit 37 performs a method such as an emergency stop of the device. In the example of FIG. 35, an example is shown in which the segment template is applied so as to select a portion that exceeds the upper threshold in each variable.

  Thereafter, or in parallel with this, the segment template is applied so as to include a portion exceeding the upper limit threshold, and data is cut out as shown in FIG. 35, and determination by the abnormality determination unit 39 is performed based on the cut out data. When it is determined that the abnormality is normal, the emergency stop (control operation) is automatically canceled by the device control unit 37. The upper threshold may be given in advance by the user. Alternatively, some threshold candidates may be selected and compared using the above-described method of dividing and determining training data, and a threshold having the best pseudo-determination performance may be adopted from the compared candidates.

  FIG. 36 is a flowchart showing an example of the operation of the client according to the present modification.

  The determination target data input unit 32 monitors whether sensor data is input to the sensing data storage unit 31 (S501). When sensor data is not input, it is confirmed whether a stop instruction is input from a supervisor or the like. If it is input, this flow is terminated, and if it is not input, the process returns to S501.

  When sensor data is input in S502, the determination target data input unit 32 reads the sensor data from the sensing data storage unit 31 as determination target data, and outputs it to the abnormality determination unit 39 via the waveform preprocessing unit 33. The abnormality determination unit 39 determines whether all or any one or more of the variables exceed the respective upper thresholds while moving the segment template (S505). When it does not exceed, the process returns to step S501, and when it exceeds, emergency stop of the monitoring target device is performed via the device control unit 37 (S506).

  On the other hand, the segment selection unit 36 cuts out the data of each variable at the position of the segment template when it is determined that the upper limit threshold has been exceeded and sends it to the abnormality determination unit 39. The abnormality determination unit 39 The determination is made based on the determination model (S507).

  When it is determined to be normal, the abnormality determination unit 39 cancels the emergency stop or the like via the device control unit 37 (S509), and the determination result and data of a certain time length including the extracted data are determined result storage unit Store in 40 (S511). On the other hand, when it is determined that there is an abnormality, the abnormality determination unit 39 notifies or displays that fact via the notification display unit 38 (S510).

  According to the embodiment described above, it is possible to use a set of accumulated multi-channel sensor data and an abnormality determination result (class) corresponding to them to improve the determination accuracy while making it possible to specify the dependency relationship between the sensors. Can do. The determination can be performed in a form (positional relationship between sections for each variable) that can indicate the basis of which part of the waveform data contributes to the determination.

  The embodiment described above is a manufacturing system monitoring system, an elevator monitoring system, an air conditioning system monitoring system, a power system monitoring system, a vital sensing system monitoring system in health care and nursing care, and a health equipment monitoring system in semiconductor manufacturing equipment and product manufacturing lines. It can be used for various remote monitoring systems such as quality control, maintenance, and status monitoring.

  Although several embodiments of the present invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These novel embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the scope of the invention. These embodiments and modifications thereof are included in the scope and gist of the invention, and are included in the invention described in the claims and the equivalents thereof.

  11 ... Training data storage unit; 12 ... Training data input unit; 13 ... Waveform preprocessing unit; 14 ... Waveform division unit; 15 ... Segment storage unit; 16 ... Probability / likelihood calculation unit; 17 ... Pseudo judgment evaluation unit; 18 ... Best model storage unit; 19 ... Best model selection unit; 20 ... Model transmission unit; 21 ... Notification unit; 22 ... Determination result reception unit; 30 ... Sensing unit; 31 ... Sensing data storage unit; 32 ... Determination target data input unit 33 ... Waveform preprocessing unit; 34 ... Model reception unit; 35 ... Determination model storage unit; 36 ... Segment selection unit; 37 ... Device control unit; 38 ... Notification display unit; 39 ... Abnormality determination unit; Part; 41 ... judgment result transmission part

Claims (4)

Represents the prior probability of normality and abnormality of the monitoring target, and for each waveform of the plurality of variables to be monitored, the identification information of the best section, the segment data of the best section, the class related to the segment data, the class of the best section A storage unit for storing a determination model representing normal and abnormal conditional probabilities;
A sensing unit for observing the monitoring target with a plurality of sensors and acquiring a plurality of time-series data regarding a plurality of variables;
For each of the plurality of variables, a selection unit that selects segment data from the plurality of time-series data according to the identification information of the best section;
Based on the segment data from the plurality of time-series data and the determination model, the normal likelihood and the abnormal likelihood of the monitoring target are respectively calculated, and the monitoring is performed on the higher likelihood of the normal and abnormal A determination unit for determining the state of the target;
The normal weight and the abnormal weight based on the ratio of the normal degree and the abnormal degree are obtained from the average value of the waveform similarity degree in the normal case and the average value of the waveform similarity degree in the abnormal case by the nearest neighbor method, and the known normal state A waveform analyzer comprising: a calculation unit that calculates a normal or abnormal quantitative state value as a sum of a result obtained by multiplying a value by the normal degree and a result obtained by multiplying a known abnormal state value by the abnormal degree. The determination unit
Detecting the upper predetermined number of segment data by the nearest neighbor method from the segment data from the plurality of time series data and the segment data of the storage unit,
For each of the plurality of variables, the ratio of the normal class and the abnormal class in the upper predetermined number of segment data is multiplied by the normal and abnormal conditional probabilities, respectively, and a multiplication value is calculated between the plurality of variables. The apparatus according to claim 1, wherein the normal likelihood and the abnormal likelihood of the monitoring target are respectively calculated by multiplying and multiplying by the normal and abnormal prior probabilities. Data for storing a plurality of training data in which a plurality of time-series data relating to the plurality of variables and a normal class or an abnormal class representing the state of the monitoring target when the plurality of time-series data are acquired are combined. A storage unit;
A plurality of sections are designated for each of the plurality of variables, and for each of the plurality of variables, from the plurality of time series data included in the plurality of training data, a plurality of data of the plurality of sections A waveform dividing unit for extracting segment data;
For each of the plurality of variables, by using the plurality of segment data extracted by the waveform dividing unit to determine each of the plurality of sections by a nearest neighbor method, one of the plurality of sections is obtained. An evaluation unit for selecting the best interval,
For each of the plurality of variables, a conditional probability of normality and abnormality of the best interval is calculated based on the number of times determined to be normal and the number of times determined to be abnormal for each of the plurality of intervals. A calculation unit for calculating the normal and abnormal prior probabilities from the total number of normal classes and the total number of abnormal classes included in the training data;
The apparatus according to claim 1, further comprising: a transmission unit that generates the determination model and sends the determination model to the storage unit. Represents the prior probability of normality and abnormality of the monitoring target, and for each waveform of the plurality of variables to be monitored, the identification information of the best section, the segment data of the best section, the class related to the segment data, the class of the best section Storing a determination model representing normal and abnormal conditional probabilities in a storage unit;
Observing the monitoring target with a plurality of sensors to obtain a plurality of time-series data regarding a plurality of variables;
For each of the plurality of variables, selecting segment data from the plurality of time series data according to the identification information of the best interval;
Based on the segment data from the plurality of time-series data and the determination model, the normal likelihood and the abnormal likelihood of the monitoring target are respectively calculated, and the monitoring is performed on the higher likelihood of the normal and abnormal Determining a state of the object;
The normal weight and the abnormal weight based on the ratio of the normal degree and the abnormal degree are obtained from the average value of the waveform similarity degree in the normal case and the average value of the waveform similarity degree in the abnormal case by the nearest neighbor method, and the known normal state Calculating a normal or abnormal quantitative state value as a sum of a result obtained by multiplying the value by the normal degree and a result obtained by multiplying the known abnormal state value by the abnormality degree.

Images