JP2005149137A - Remote supervisory system, method and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a remote supervisory system or the like for remotely supervising a plurality of facilities such as power facilities and capable of discovering the symptoms of failures of the facilities to be supervised at early stages and correctly locating failures, etc. <P>SOLUTION: The remote supervisory system 1 for remotely supervising a plurality of facilities 2 each having a plurality of sensors 21 includes a sensor value acquisition means 11 for constantly acquiring sensor measurements detected by the plurality of sensors 21 and sent from the facilities 2; a predictive model construction means 12 by which the relationship between the plurality of sensor measurements detected by the plurality of sensors 21 is determined on the basis of the sensor measurements detected by the plurality of sensors 21 and acquired by the sensor value acquisition means 11 while the facilities 2 are operating normally, to store the relationship as a predictive model; and a failure symptom detecting means 13 for determining the predictive sensor values of the plurality of sensors 21 on the basis of the sensor measurements acquired and the predictive model stored by the predictive model construction means 12 when the sensor value acquisition means 11 acquires the sensor measurements of the plurality of sensors 21 and for detecting the symptoms of failures of the facilities 2 according to the differences between the sensor measurements and the predictive sensor values. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a system for remotely monitoring a plurality of facilities such as an air conditioner, a generator, and the like, and in particular, a remote monitoring system that can detect a failure sign of a monitored facility at an early stage and correctly identify a failure site. About.

  Conventionally, methods based on the concepts of post-maintenance and time-based maintenance have been used as maintenance methods for power equipment and the like. However, the post-maintenance that repairs after the occurrence of a failure has a problem that the operation of the facility must be stopped for a long time for the repair. Further, in the time-based maintenance in which the maintenance work is periodically performed, the maintenance work is performed even when the target equipment is in a healthy state, and thus there is a problem that time and cost are wasted.

  Therefore, there is a concept of state-based maintenance that performs appropriate maintenance based on the state of the target equipment. In such a method, the values of various sensors provided in the equipment and the status of each part of the equipment are inspected, and the result Maintenance work is determined based on the above. However, if there are multiple equipment to be maintained and they are scattered in remote locations, it is difficult for the workers to check the status of each equipment at the site. A system for remotely monitoring the facilities has been proposed.

In such a remote monitoring system (for example, the device described in Patent Literature 1, 2, or 3 below), the value of the sensor provided in each facility is received by the monitoring system provided in a place away from the facility, The system determines whether the equipment is normal or abnormal based on the value. In general, the determination method is that the system notifies each sensor of normality / abnormality based on whether or not the value of each received sensor exceeds a preset threshold value, and monitoring is performed based on the notification. A person such as a staff member judges the occurrence of a failure of a facility or a failure part. There is also a method in which the system determines that a facility failure occurs when some predetermined sensors have abnormal values.
JP 2001-41615 A JP 2003-4281 A JP 2001-21192 A

  However, in the above conventional methods, in many cases, the failure of the target equipment is determined based on the fact that the value of each sensor provided in the monitored equipment has become an abnormal value. It was difficult to detect signs of failure at an early stage because it was detected that a failure occurred or that the degree of abnormality was considerably advanced. Similarly, the failure part, the failure type, and the like are mainly determined based only on the fact that the value of each sensor has become an abnormal value. Therefore, it may be difficult to accurately identify the failure part. Further, as described above, when a person determines the occurrence of a failure, a failure part, etc., skill is required, and there may be individual differences in the determination.

  Therefore, an object of the present invention is a system for remotely monitoring a plurality of facilities such as power facilities, and can remotely detect a failure sign of a monitoring target facility at an early stage and correctly identify a failure site or the like. Is to provide.

In order to achieve the above object, one aspect of the present invention is that a remote monitoring system that remotely monitors a plurality of facilities each including a plurality of sensors is detected by the plurality of sensors transmitted from the facilities. Based on the measured sensor values acquired by the sensor values acquiring means, which are detected by the plurality of sensors when the equipment is operating normally, A prediction model construction unit that obtains a correlation between a plurality of actual sensor values detected by the plurality of sensors and stores the correlation as a prediction model; and the sensor value acquisition unit calculates the actual sensor values of the plurality of sensors. At the time of acquisition, predicted sensor values of the plurality of sensors are obtained from the acquired actual sensor values and the prediction model stored by the prediction model construction means, and the actual sensor values and the prediction sensors are obtained. Based on the difference between the service value is that it has a fault indication detecting means for detecting a failure indication of the equipment. Therefore, according to the present invention, failure signs are detected by paying attention to the correlation between a plurality of sensors, so that early detection of failures is possible. In addition, since a plurality of monitoring target facilities that are scattered at different locations can be monitored at one location, labor saving can be achieved in facility monitoring work and maintenance work. Further, since the failure diagnosis is automatically performed by the system, no skill is required as in the case where a person makes a judgment, and there is no individual difference.

  Furthermore, in the above-described invention, a preferable aspect thereof is a failure model in which a failure model that is a tendency of a difference between the measured sensor value and the predicted sensor value at the time of failure of the facility is stored in advance for each type of failure. When a failure sign is detected by the storage means and the failure sign detection means, a tendency of a difference between the measured sensor value and the predicted sensor value used for the detection, and a failure model stored by the failure model storage means And failure type determining means for determining the type of failure based on the above. As a result, the failure type can be specified more accurately.

  In the above invention, a preferable aspect is that the correlation obtained by the prediction model construction means is obtained based on principal component analysis, and the difference between the measured sensor value and the predicted sensor value is the measured sensor value of each sensor. It is characterized by being expressed by a residual vector having a difference between a predicted sensor value and a component as a component.

  In the above invention, a preferable aspect is that the correlation obtained by the prediction model construction means is obtained based on principal component analysis, and the difference between the measured sensor value and the predicted sensor value is the measured sensor value of each sensor. The failure sign detection means detects the failure sign based on the length of the residual vector, and the failure type determination means The failure type is determined based on the direction of the residual vector.

  In order to achieve the above object, another aspect of the present invention provides a remote monitoring method in a remote monitoring system for remotely monitoring a plurality of facilities each having a plurality of sensors, wherein the remote monitoring system transmits from the facilities. The first step of acquiring measured sensor values detected by the plurality of sensors as needed and the remote monitoring system are detected by the plurality of sensors when the facility is operating normally. A second step of obtaining a correlation between a plurality of actual sensor values detected by the plurality of sensors based on the acquired actual sensor values, and storing the correlation as a prediction model; and the remote monitoring system, When the actual sensor values of the plurality of sensors are acquired, predicted sensor values of the plurality of sensors are obtained from the acquired actual sensor values and the stored prediction model, and the actual sensor Based on the difference between the predicted sensor values and is to have a third step of detecting a failure indication of the equipment.

  Furthermore, in the above-mentioned invention, a preferable aspect thereof is that the remote monitoring system predetermines a failure model that has a tendency of a difference between the actually measured sensor value and the predicted sensor value at the time of failure of the facility for each type of failure. Further, when the remote monitoring system detects a failure sign in the third step, the tendency of the difference between the measured sensor value and the predicted sensor value used for the detection, and the storage And a step of determining a type of failure based on the failure model.

  In order to achieve the above object, according to still another aspect of the present invention, a remote monitoring program that causes a remote monitoring system to execute a process of remotely monitoring a plurality of facilities each having a plurality of sensors is transmitted from the facility. A first step of acquiring measured sensor values detected by the plurality of sensors as needed, and an actual sensor value detected and acquired by the plurality of sensors when the facility is operating normally. A second step of obtaining a correlation between a plurality of actual sensor values detected by the plurality of sensors, and storing the correlation as a prediction model, and when acquiring actual sensor values of the plurality of sensors In addition, a predicted sensor value of the plurality of sensors is obtained from the acquired measured sensor value and the stored predicted model, and a failure of the facility is determined based on a difference between the measured sensor value and the predicted sensor value. And a third step of detecting the weather is to be executed by the remote monitoring system.

Furthermore, in the above-mentioned invention, a preferable aspect thereof is that the remote monitoring system predetermines a failure model that has a tendency of a difference between the actually measured sensor value and the predicted sensor value at the time of failure of the facility for each type of failure. Further, when a failure sign is detected in the third step, based on the tendency of the difference between the measured sensor value and the predicted sensor value used for the detection and the stored failure model And causing the remote monitoring system to execute a step of determining the type of failure.

  Further objects and features of the present invention will become apparent from the embodiments of the invention described below.

  Embodiments of the present invention will be described below with reference to the drawings. However, such an embodiment does not limit the technical scope of the present invention. In the drawings, the same or similar elements are denoted by the same reference numerals or reference symbols.

  FIG. 1 is a configuration diagram according to an embodiment of a remote monitoring system to which the present invention is applied. A remote monitoring system 1 shown in FIG. 1 is a remote monitoring system according to this embodiment, and remotely monitors a plurality of power equipments 2 installed at a plurality of locations. The remote monitoring system 1 detects early signs of failure in the power equipment 2 based on the correlation of the values of the plurality of sensors 21 (21a, 21b, 21c,...) Attached to each power equipment 2, The failure type is to be specified accurately.

  As described above, the power facility 2 shown in FIG. 1 is a facility to be monitored by the remote monitoring system 1, and each power facility 2 includes a plurality of sensors 21. The power equipment 2 is, for example, a generator or a gas engine heat pump type air conditioner (GHP), and the sensor 21 is, for example, a thermometer, a pressure gauge, or the like attached to each place of the equipment. A communication line 3 shown in FIG. 1 is a line for transmitting values (for example, temperature, pressure, etc.) detected by the respective sensors 21 to the remote monitoring system 1, and each power equipment 2 and the remote monitoring system. 1 is connected.

  As shown in the figure, the remote monitoring system 1 includes a sensor information acquisition unit 11 (sensor value acquisition unit), a prediction model construction unit 12 (prediction model construction unit), a failure sign detection unit 13 (failure sign detection means), a failure It has a type identification unit 14 (failure type determination unit) and a failure model storage unit 15 (failure model storage unit), and can be constructed by a so-called computer system including a CPU, a memory, an OS, an application program, and the like. it can. First, the sensor information acquisition unit 11 is a part that interfaces with the power equipment 2, and the values of the sensors 21 that are periodically detected and transmitted by the plurality of sensors 21 are transmitted via the communication line 3. Receive and store sequentially.

  Next, the prediction model construction unit 12 constructs and holds correlations between the values of the plurality of sensors 21 from the values of the plurality of sensors 21 acquired when the power equipment 2 is operating normally. It is a part to do. The constructed correlation is referred to as a prediction model. A specific method for constructing the prediction model will be described later. For example, a correlation between a plurality of predetermined sensors 21 is constructed as a mathematical expression by multivariate analysis such as principal component analysis. For example, the correlation of the values detected by the three sensors 21a, 21b, and 21c of the power facility 2 shown in FIG. 1 during normal operation is modeled by a mathematical formula, and the three sensors 21a, 21b, and 21c are modeled by this prediction model. Can be predicted. The prediction model construction unit 12 may be composed of a program that describes the procedure of construction processing of the prediction model, a control device that executes processing according to the program, a data storage device that stores the constructed prediction model, and the like. it can.

  Further, the failure sign detection unit 13 includes the sensor value of the power equipment 2 acquired by the sensor information acquisition unit 11 during the actual operation of the power equipment 2 and the prediction model that the prediction model construction unit 12 has constructed for the power equipment 2. This is a part for detecting (judging) a failure sign of the power equipment 2. Although details of the failure sign detection method will be described later, the difference between the predicted sensor value based on the prediction model and the actually measured sensor value acquired by the sensor 21 is obtained, and the target power equipment 2 is based on the difference. It is determined whether or not is in an abnormal state. The failure sign detection unit 13 can be configured by a program describing a procedure of failure sign detection processing, a control device that executes processing according to the program, and the like.

  Next, the failure type identification unit 14 is a part that identifies the type of failure when the failure sign detection unit 13 detects a failure sign. Specifically, the failure type is identified by comparing the difference between the failure model stored in the failure model storage unit 15 to be described later and the predicted sensor value obtained by the failure sign detection unit 13 and the measured sensor value. To do. Although the details of the specifying method will be described later, the failure type is information for identifying the content of the failure, such as a failure part and a failure phenomenon. The failure type specifying unit 14 includes a program describing a procedure for specifying a failure type, a control device that executes processing according to the program, and an output device for notifying a user of the remote monitoring system 1 of a specific result. (A monitor, a printer, etc.).

  The failure model storage unit 15 is a part that stores a failure model used in the failure type identification unit 14 described above. The failure model means a failure pattern in which a difference pattern (trend) between the predicted sensor value and the measured sensor value at the time of the failure such as a failure experienced in the past is known, and is remotely monitored in advance. Stored in the system 1. Although details of such a failure model will be described later, this failure model is generally determined for each failure type of the power equipment 2. The failure model storage unit 15 can be composed of a storage device such as a hard disk.

  FIG. 2 is a flowchart illustrating the processing contents in the remote monitoring system 1. Hereinafter, based on FIG. 2 etc., the processing content after the monitoring target power equipment 2 is newly installed or after the repair of the power equipment 2 is completed will be described. For convenience of explanation, the description will be given focusing on one power equipment 2 having the sensors 21a, 21b, and 21c (for example, the power equipment 2 shown in the upper left of FIG. 1).

  First, the power equipment 2 to be monitored is activated and a trial operation is performed to confirm that the operation is normal. Thereafter, detection is performed periodically (for example, every hour) by a plurality of sensors 21 attached to the power equipment 2, and the detected sensor values are sequentially transmitted to the remote monitoring system 1 via the communication line 3.

  In the remote monitoring system 1, the sensor information acquisition unit 11 sequentially receives and stores the transmitted sensor values (step S11 in FIG. 2). FIG. 3 is a diagram for explaining sensor values and prediction models acquired and stored by the sensor information acquisition unit 11. (A) of FIG. 3 has illustrated the sensor value which the sensor information acquisition part 11 acquired in time series, Here, sensor value a of three sensors 21a, 21b, 21c with which one power equipment 2 is provided, b and c are shown. “Data acquisition No.” indicates how many times the sensor value is detected after the power equipment 2 is started. For example, the values “60, 20, 35” on the right side of “1” are respectively The sensor values of the sensors 21a, 21b, and 21c detected at the first time (after one hour) after activation.

  The sensor information acquisition unit 11 continuously receives and stores sensor values. At the time of acquiring a predetermined number (number of times) of sensor values, the prediction model construction unit 12 described above performs the construction process of the prediction model. Start (steps S12 and S13 in FIG. 2). Note that the predetermined number is a number necessary for constructing a prediction model, that is, a number sufficient to perform statistical processing described later and perform so-called learning, for example, 50. Determined by value.

  As described above, the construction of the prediction model is to determine the correlation between the values in the plurality of sensors 21, and the prediction model construction unit 12 uses the sensor values for the predetermined number of times acquired and stored. Thus, a mathematical formula or the like for determining the correlation between sensor values is obtained by a so-called statistical analysis method. For example, in the example shown in FIG. 3A, the sensor values a, b, and c (data of the portion shown in A in the figure) for 50 times are used, for example, by the principal component analysis technique. The relationship between b and c is expressed by the following equation (1).

t = ap ₁ + bp ₂ + cp ₃ (1)
Here, t is a three-dimensional space composed of sensor values a, b, and c, and is an axial value representing the principal component direction of the space in which the points composed of the used (acquired) sensor values are distributed. Further, p ₁ , p ₂ , and p ₃ are constants that define the axis representing the principal component direction, and the model construction process performed by the prediction model construction unit 12 is, for example, p ₁ , p ₂ , p ₃ Is a process for determining the value of. FIG. 3B is a diagram illustrating a prediction model based on the above equation (1). A black point such as a point represented by x in the figure is a point formed of the acquired sensor value, and has a component (a, b, c). For example, the data acquisition number in FIG. The sensor value “1” is represented as a point (60, 20, 35) in the three-dimensional space in FIG. In this way, when points are plotted in a three-dimensional space consisting of a, b, and c for all data used for constructing a prediction model, there is usually a tendency in the distribution of those points. For example, FIG. A space (s) is formed as shown in FIG. The axis indicating the longitudinal direction of the space s is the axis t indicating the principal component direction described above. The modeling based on the equation (1) means that a point x consisting of sensor values a, b, c detected by the sensors 21a, 21b, 21c is a principal component direction in a three-dimensional space consisting of a, b, c. To be plotted on the axis t representing

  In this way, the prediction model construction unit 12 constructs a prediction model using the sensor value acquired during normal operation of the power equipment 2, and stores the constructed prediction model. In the above-described example, in the principal component analysis adopted for the prediction model construction, modeling is performed with a one-dimensional axis t. However, modeling may be performed with a two-dimensional plane. In the above example, the values of the three sensors 21 are used. However, the number of the sensors 21 is not limited to three, and the modeling order in the principal component analysis is changed according to the number. May be. Furthermore, as long as the correlation between the plurality of sensors 21 can be obtained, other methods such as regression analysis may be used instead of the principal component analysis.

  The process described above is the learning phase (S10 in FIG. 2) in the remote monitoring system 1, and then enters the failure sign detection phase (S20 in FIG. 2) which is the actual monitoring process. As described above, the sensor information acquisition unit 11 sequentially receives and stores sensor values transmitted from the power equipment 2 (step S21 in FIG. 2). In the example shown in FIG. 3A, the data of the portion shown in B is sequentially acquired. Here, every time the sensor information acquisition unit 11 receives a new sensor value, the failure sign detection unit 13 determines whether or not the received sensor value is an abnormal value. Specifically, a predicted value of a sensor value is obtained from the predicted model constructed by the predicted model construction unit 12 and the received sensor value, and the difference between the predicted value and the received sensor value (ie, measured value) is calculated. Based on the magnitude of this difference, it is determined whether or not it is abnormal (step S22 in FIG. 2).

FIG. 4 is a diagram for explaining a difference between the predicted value and the actually measured value. FIG. 4 shows the case of the example described based on the equation (1) and FIG. 3, and the three-dimensional space composed of sensor values a, b, and c and the axis t in the principal component direction are shown in FIG. The same as (b). For example, when the sensor value received by the sensor information acquisition unit 11 is located at x (a1, b1, c1) in FIG. 4, the failure sign detection unit 13 sets the value (a1, b1, c1) of x. Substituting into the equation (1), t1 is obtained. This t1 is a value on the axis t in the principal component direction, and represents the position indicated by _xt in FIG. 4, that is, the position projected from x to the axis t. The x _t a well point indicated by, b, a point in the three-dimensional space consisting of c, and a component that _{(a t 1, b t 1} , c t 1). The _{(a t 1, b t 1} , c t 1) is the predicted value of the sensor value which the fault indications detection unit 13 obtains. Then, the difference of the measured and predicted values of the failure sign detection unit 13 described above finding is, this predicted value _{x t (a t 1, b} t 1, c t 1) and the measured value x (a1, b1, c1) This is a residual vector indicated by e in FIG. In this example, the failure sign detection unit 13 determines whether or not the actual measurement value x (a1, b1, c1) is abnormal based on the magnitude of the residual vector e. For example, if the square of the length of the residual vector e (hereinafter referred to as SPE (square prediction error) value) exceeds a 95% confidence limit value, it is determined that the sensor value (actual measurement value x) is abnormal. To do. The 95% confidence limit value is a value usually used in statistical analysis, and means a value that does not exceed this value with a probability of 95%. In the case of this embodiment, this value is obtained by the learning face described above.

  As described above, when it is determined whether or not the acquired sensor value is abnormal (step S22 in FIG. 2), the failure sign detection unit 13 acquires the determination result for the sensor value acquired this time and the past. Based on the determination result of whether or not the sensor value is an abnormal value, it is determined whether or not there is a sign of failure in the target power equipment 2 (step S23 in FIG. 2). Specifically, when it is determined that the acquired sensor value is abnormal continuously for a predetermined number of times, it is determined that the power equipment 2 has a sign of failure. This is because there may be a case where the sensor value shows an abnormal value on a one-off basis even during normal operation without any failure in the power equipment 2. For example, in the above-described example, if the SPE value exceeds the 95% confidence limit value 10 times continuously in the process in step S22 of FIG. 2, it is determined that a failure sign has been detected.

  FIG. 5 is a diagram illustrating the SPE values in time series. The position where the “observation days” in the figure is “0” is the time when the power equipment 2 is started, and thereafter, the SPE values obtained each time the value is detected by the sensor 21 are sequentially plotted on the right side. Yes. Further, the period C shown in the figure is a period corresponding to the learning phase described above, and the above-described failure signs are detected in the subsequent period. In addition, the line shown by E in the figure is the line showing the above-mentioned 95% confidence limit value, and when the SPE value exceeding this line is obtained, the sensor value at that time is abnormal as described above. To be judged.

  In the example shown in FIG. 5, the SPE value often exceeds the E line after the learning period, but the observation days exceed the predetermined number of times (for example, 10 times) continuously until around 50 days. It is not determined that there is a sign of failure. However, when the period shown in D of FIG. 5 is reached, the SPE value exceeds the E line for a predetermined number of times, and at this point, the failure sign detection unit 13 determines that the power equipment 2 has a failure sign. to decide.

  Here, when it is determined that the acquired sensor value is abnormal continuously several times, it is determined that the power equipment 2 has a failure sign. However, depending on the equipment and sensors to be monitored, a plurality of sensors A failure sign of the power equipment 2 may be determined based on a single acquired value of 21 (for example, the sensors 21a, 21b, and 21c).

  If no failure sign is detected as a result of the failure sign detection in the failure sign detection unit 13 (No in step S23 in FIG. 2), the remote power facility 2 to be monitored is stopped. As long as there is no reason to end the monitoring by the monitoring system 1 (No in step S24 in FIG. 2), the processing from step S21 in FIG. 2 described above is repeatedly executed. If there is a reason for terminating the monitoring (Yes in step S24 in FIG. 2), the monitoring is terminated.

  On the other hand, if a failure sign is detected as a result of the detection of the failure sign (Yes in step S23 in FIG. 2), the process proceeds to the type specific processing for the failure in which the sign is detected (S20 → S30 in FIG. 2). ).

  By the processing as described above, the failure sign detection phase (S20) is completed. As described above, in the remote monitoring system 1, the values of the plurality of sensors 21 attached to the same power facility 2 are mutually different. Focusing on the fact that there will be some relationship, failure signs are detected based on those relationships during normal operation, that is, based on the prediction model. Therefore, it is possible to detect a failure sign more accurately and earlier than the conventional method of determining a failure based only on the value of each sensor 21.

  FIG. 6 is a diagram for explaining the effect of the remote monitoring system 1. The two graphs shown in FIG. 6 are obtained by plotting values a and b detected by the sensor 21a and the sensor 21b attached to the same power equipment 2 in time series. In this graph, the solid line is the value when the power equipment 2 is operating normally, and as described above, there is a correlation between a and b. For example, the value of b is indicated by F in the figure. If it moves like a part, this correlation will be broken. Usually, in such a case, it can be said that there is some problem in the power equipment 2. Since the remote monitoring system 1 determines the detection of a failure sign based on this correlation as described above, when a value such as the portion indicated by F is detected, there is a failure sign in the power equipment 2. It can be judged that there is. On the other hand, in the conventional method, it is usually determined that the power equipment 2 is faulty only when each of a and b exceeds a predetermined value (for example, the threshold value in FIG. 6). Therefore, even if values such as the portion indicated by F are detected, these values do not exceed the threshold value, so that it is not determined that there is a failure. Therefore, it is not possible to find a defect at an early stage.

  Next, returning to FIG. 2, when a sign is detected in the failure sign detection phase (S20 in FIG. 2), as described above, the remote monitoring system 1 identifies the type of the failure (see FIG. 2). 2 S30). FIG. 7 is a diagram for explaining the failure type identification process. First, the failure type identification unit 14 of the remote monitoring system 1 analyzes the difference between the predicted value of the sensor value obtained by the failure sign detection unit 13 and the actual measurement value (step S31 in FIG. 2). Specifically, for example, in the case of the example described with reference to FIGS. 3 and 4, the component (ea, eb, ec) value of the residual vector e obtained by the failure sign detection unit 13 is extracted. The component (ea, eb, ec) value of the extracted residual vector e is, for example, shown as “actual value pattern” in FIG. 7B.

  Next, the failure type identification unit 14 compares the difference between the predicted value and the actual measurement value with the failure model stored in the failure model storage unit 15 described above, and identifies the type of failure detected this time ( Step S32 in FIG. Specifically, from the failure models stored for each failure type, a failure model in which the difference between the predicted value obtained this time and the actual measurement value and the pattern (trend) match or is very similar is selected, The failure type of the selected failure model is specified as the type of failure detected this time. This is because if the failure type is the same, the difference pattern (trend) between the predicted value and the actual measurement value is almost the same, and if the failure type is different, the pattern (trend) is different. This is because it is known experimentally.

  FIG. 7A illustrates three residual vectors in the example described with reference to FIGS. In the figure, x1, x2, and x3 are three actually measured values of sensor values, and e1, e2, and e3 are residual vectors corresponding to them. As described above, the pattern (trend) of the difference between the predicted value and the actually measured value, in this case, the pattern of each component of the residual vector, that is, the direction of the residual vector is substantially the same for each failure type. Become. Accordingly, in FIG. 7A, since the directions of the residual vectors e1 and e3 are substantially the same, it can be determined that the failure types relating to the actual measurement values x1 and x3 are the same, and the actual measurement values having different residual vector directions. It can be determined that the failure type related to x2 is different from these.

  FIG. 8 is a diagram showing each component of the residual vector obtained by the experiment. The example of FIG. 8 shows the components of the residual vector obtained from the values detected by the 12 sensors provided in the electric heat pump, the horizontal axis indicates 12 components, and the vertical axis indicates those component values. ing. (A) and (b) in the figure are data at different dates and times when the electric heat pump has a failure "clogged heat exchanger", and (c) and (d) in the figure are "refrigerant" Data of different date and time when the failure "leakage" occurs. When the correlation coefficients of these four residual vectors are calculated, (a) and (b), (c) and (d), which are the same failure types, have high values, and low values between different failure types. It becomes. Therefore, it can be seen from this experiment that the failure type can be specified by the component pattern of the residual vector.

  FIG. 7B is a diagram illustrating more specifically the specification of the failure type. This is also due to the same conditions as the example described with reference to FIGS. What is shown on the left side of the figure is an example of a failure model stored in the failure model storage unit 15, and corresponding residual vectors for a plurality of failure types (type A, type B, type C, etc.). The component (pattern) of is stored. The failure type includes, for example, “heat exchanger clogged”, “refrigerant leak”, and a failure part in the electric heat pump. Further, for these failure models, residual vectors obtained from sensor values actually obtained in the case of failures experienced in the past are used.

  As described above, the “measured value pattern” shown on the right side of the figure indicates the residual vector component obtained this time, and this pattern is collated with the pattern of each failure model. In the example shown in the figure, since the pattern of the actual measurement value is similar to the failure model of type B, in this case, the failure indicating the sign is specified as type B. In the determination of similarity, for example, the inner product of the unit vector of the residual vector related to the actual measurement value and the unit vector of the residual vector related to each failure model is calculated, and if the result is close to 1, it is determined to be similar . That is, as described above, if the directions of the residual vectors are almost the same, it is determined that the failure types are the same.

  As described above, when the identification of the failure type is completed, the failure type identification unit 14 notifies the failure information to the user of the remote monitoring system 1 (step S33 in FIG. 2). The content of the failure information is, for example, the power equipment 2 that has detected the failure sign and its failure type. Further, the notification means includes display on a monitor and printing on paper by a printer. Thereafter, the remote monitoring system 1 ends the monitoring if there is a reason to end the monitoring process such as when the power facility 2 should be stopped due to the detected failure sign (Yes in step S24 in FIG. 2). Then, inspection, repair, and the like of the power equipment 2 in which the failure sign is detected are performed as necessary. On the other hand, if there is no reason to end the monitoring process (No in step S24 in FIG. 2), the monitoring process is continued, and the process from step S21 in FIG. 2 described above is repeatedly executed.

  Although the processing of the failure type identification phase (S30) is completed by the processing described above, in the remote monitoring system 1, the tendency of the difference between the above-described predicted value of the sensor value and the actual measurement value that can accurately identify the failure. Therefore, the failure type can be specified based on the above, so that the failure type can be specified and diagnosed more accurately than in the past. In the specific description of the failure type identification phase (S30), the description is based on the content based on the principal component analysis. However, other multivariate analysis methods such as regression analysis are used in the processing from the construction of the prediction model described above. It can also be used.

  As described above, according to the remote monitoring system 1 according to the present embodiment, failure detection and failure type focusing on the correlation between the plurality of sensors 21 by multivariate analysis on the values detected by the plurality of sensors 21. Thus, it is possible to detect failures early and accurately diagnose failures. In addition, since a plurality of monitoring objects (power equipment 2) scattered at different locations can be monitored at one place, labor saving in equipment monitoring work and maintenance work can be achieved. Further, since the failure diagnosis is automatically performed by the system, no skill is required as in the case where a person makes a judgment, and there is no individual difference.

  In the present embodiment, the power facility 2 is the monitoring target of the remote monitoring system 1, but the monitoring target is not limited to the power facility, and any other device may be used as long as it can acquire a sensor value.

  The protection scope of the present invention is not limited to the above-described embodiment, but covers the invention described in the claims and equivalents thereof.

1 is a configuration diagram according to an embodiment of a remote monitoring system to which the present invention is applied. 3 is a flowchart illustrating processing contents in the remote monitoring system 1; It is a figure for demonstrating the sensor value and prediction model which the sensor information acquisition part 11 acquired and stored. It is a figure for demonstrating the difference of the predicted value and actual value of a sensor value. It is the figure which illustrated SPE value in time series. It is a figure for demonstrating the effect of this remote monitoring system. It is a figure for demonstrating failure classification specific processing. It is the figure which showed each component of the residual vector obtained by experiment.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Remote monitoring system 2 Power equipment 3 Communication line 11 Sensor information acquisition part (sensor value acquisition means)
12 Prediction model construction part (prediction model construction means)
13 Failure sign detection unit (failure sign detection means)
14 Failure type identification part (failure type determination means)
15 Failure model storage unit (failure model storage means)
21, 21a, 21b, 21c sensor

Claims (8)

A remote monitoring system for remotely monitoring a plurality of facilities each having a plurality of sensors,
Sensor value acquisition means for acquiring, as needed, measured sensor values detected by the plurality of sensors transmitted from the facility;
When the equipment is in normal operation, based on the actual sensor values detected by the plurality of sensors and acquired by the sensor value acquisition means, between the actual sensor values detected by the plurality of sensors. A prediction model building means for obtaining a correlation and storing the correlation as a prediction model;
When the sensor value acquisition unit acquires the measured sensor values of the plurality of sensors, the predicted sensor values of the plurality of sensors are obtained from the acquired actual sensor values and the prediction model stored by the prediction model construction unit. And a failure sign detection means for detecting a failure sign of the equipment based on the difference between the measured sensor value and the predicted sensor value. In claim 1, further comprising:
A failure model storage means for storing in advance a failure model that is a tendency of a difference between the measured sensor value and the predicted sensor value at the time of failure of the facility for each type of failure;
When the failure sign detection means detects a failure sign, based on the tendency of the difference between the measured sensor value and the predicted sensor value used for the detection and the failure model stored in the failure model storage means. And a failure type determination means for determining the type of failure. In claim 1 or 2,
Correlation obtained by the prediction model construction means is obtained based on principal component analysis,
The remote monitoring system, wherein the difference between the actual sensor value and the predicted sensor value is represented by a residual vector whose component is the difference between the actual sensor value and the predicted sensor value of each sensor. In claim 2,
Correlation obtained by the prediction model construction means is obtained based on principal component analysis,
The difference between the measured sensor value and the predicted sensor value is represented by a residual vector whose component is the difference between the measured sensor value and the predicted sensor value of each sensor,
The failure sign detection means detects the failure sign based on the length of the residual vector;
The remote monitoring system, wherein the failure type determination means determines the failure type based on a direction of the residual vector. A remote monitoring method in a remote monitoring system for remotely monitoring a plurality of facilities each having a plurality of sensors,
A first step in which the remote monitoring system acquires, as needed, measured sensor values detected by the plurality of sensors, transmitted from the facility;
Correlation between the plurality of actual sensor values detected by the plurality of sensors based on the actual sensor values detected and acquired by the plurality of sensors when the remote monitoring system is operating normally. A second step of obtaining a relationship and storing the correlation as a prediction model;
When the remote monitoring system acquires measured sensor values of the plurality of sensors, the remote monitoring system obtains predicted sensor values of the plurality of sensors from the acquired measured sensor values and the stored predicted model, and And a third step of detecting a failure sign of the equipment based on a difference between the predicted sensor value and the predicted sensor value. In claim 5,
The remote monitoring system stores in advance a failure model that is a tendency of a difference between the measured sensor value and the predicted sensor value at the time of failure of the facility for each type of failure,
Furthermore, when the remote monitoring system detects a failure sign in the third step, the tendency of the difference between the actual sensor value and the predicted sensor value used for the detection, the stored failure model, and A remote monitoring method comprising a step of determining a type of failure based on the method. A remote monitoring program for causing a remote monitoring system to execute a process of remotely monitoring a plurality of facilities each having a plurality of sensors,
A first step of acquiring, as needed, measured sensor values detected by the plurality of sensors, transmitted from the facility;
When the equipment is in normal operation, a correlation between a plurality of measured sensor values detected by the plurality of sensors is obtained based on measured sensor values detected and acquired by the plurality of sensors. A second step of storing the correlation as a prediction model;
When the measured sensor values of the plurality of sensors are acquired, a predicted sensor value of the plurality of sensors is obtained from the acquired measured sensor value and the stored prediction model, and a difference between the measured sensor value and the predicted sensor value is obtained. A remote monitoring program for causing the remote monitoring system to execute a third step of detecting a failure sign of the equipment based on the above. In claim 7,
The remote monitoring system stores in advance a failure model that is a tendency of a difference between the measured sensor value and the predicted sensor value at the time of failure of the facility for each type of failure,
Further, when a failure sign is detected in the third step, the failure is determined based on the tendency of the difference between the measured sensor value and the predicted sensor value used for the detection and the stored failure model. A remote monitoring program that causes the remote monitoring system to execute a step of determining a type.

Images