JP2006106870A - Process management device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To efficiently discriminate whether or not a process is abnormal from the change of spectrums obtained by carrying out the wavelet-analysis of main component scores corresponding to process result data outputted from a process side. <P>SOLUTION: When the monitor control of an air conditioning device at each object building side is started, the main component analysis of process result data stored in a process result data storing means 7 is executed by a statistical analysis means 11, and the wavelet-analysis processing and contour line patterning processing of main component scores "Z<SB>i</SB>" obtained by the main component analysis is carried out by a process status deciding means 9 so that the area of a land can be calculated, and when this exceeds a preliminarily set threshold, it is decided that the process side is abnormal. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a process management apparatus for predicting abnormalities and inoperability in various industrial and consumer facilities, and estimating the cause of a failure. In particular, among many functions of a BEMS apparatus, failure detection, failure The present invention relates to a process management apparatus that performs prediction and failure diagnosis.

  In recent years, with the advancement of computer processing technology, image data obtained by photographing various industrial and consumer facilities is processed to determine the presence or absence of abnormalities in these various facilities and to determine the soundness of the facilities. A method has been developed (see Patent Document 1).

  However, in this method, the presence / absence of an abnormality to be monitored is determined by image processing. Therefore, when applied to building monitoring, etc., humidity data and temperature at each part in the building can be obtained unless the abnormality of each part of the building can be grasped through the image. Even if process data such as data and valve opening data is abnormal, there is a problem that the occurrence of abnormality cannot be detected.

  For this reason, in a building with a large-scale building in recent years, the discovery and treatment of abnormalities occurring in building equipment such as an air conditioning system is slowed down, and the possibility of causing deterioration of the indoor environment and energy waste is increasing.

  Therefore, BEMS (Bui1ding and Energy Management System: Japanese Building, Environment, Energy Management System) devices are now attracting attention as a system that solves these problems and is expected to have a high energy saving effect.

  BEMS equipment manages all building management items or multiple building management items from building management to energy management in a lump, and is important as an energy-saving measure for commercial buildings such as offices and stores. It has become.

At this time, as a method of determining the presence or absence of abnormality of the building equipment provided in the building to be monitored, a method of analyzing the spectrum by performing Fourier transform on the process data output from the building equipment, and determining the presence or absence of failure, or A method that estimates the characteristic model of a building facility using a system identification method, etc., compares the process data output from this characteristic model with the process data output from the building facility, and determines the presence or absence of a failure is used. The
JP 2001-31723 A

  By the way, in various building facilities, the air conditioning system that performs heating and cooling is indispensable for the building, and the optimum temperature is monitored by monitoring the temperature and humidity of each part of the building and the power consumption of each air conditioner. It is extremely important to get into the driving situation.

  However, in large-scale buildings, the temperature and humidity of each part of the building may change suddenly depending on the opening and closing status of the entrance door, etc. The spectrum that changes from moment to moment due to Fourier transformation, or by system identification techniques, There is a problem that a characteristic model that fluctuates cannot be obtained.

  In such large-scale buildings, several air conditioning systems equipped with a large number of air conditioners must be installed on each floor. If any of the air conditioners malfunctions, the air conditioning system will not operate normally, and a comfortable air conditioning environment cannot be provided to the occupants. Further, even if energy saving control or the like is introduced, usually, only sensors necessary for the control are usually installed, and those for detecting a failure and its factor are generally not installed.

  For this reason, even if a malfunction occurs in the equipment, it is not discovered and it is operated as it is, and the state in which the equipment efficiency is greatly reduced is continued and energy is often wasted.

  Therefore, in recent years, in order to quickly find defects in each facility installed in a building, many buildings and the center are connected by a communication line, and the facilities in each building are concentrated by a supervisor placed in the center. In many cases, a monitoring system is installed that can monitor the presence or absence of abnormality of each facility provided on the building side on the center side even when a management device is not installed on each building side.

  However, in such a monitoring system, a huge amount of data is concentrated on the center side, so there is a limit to the detection of malfunctions due to manual monitoring, and a huge amount of data is automatically analyzed to improve efficiency. There has been a strong demand for the development of a device that can determine the presence or absence of a failure.

  In view of the above circumstances, the present invention simplifies the configuration of the entire apparatus, shortens the processing time, and determines whether the process is normal from the change in spectrum obtained by wavelet analysis of process performance data output from the process side. The purpose is to provide a process management device that can efficiently estimate and estimate system performance, perform system performance verification, failure detection, failure prediction, and failure diagnosis, reduce energy consumption, and reduce maintenance costs. Yes.

  In order to achieve the above object, the present invention provides a process result data storage means for storing a plurality of process result data obtained by measuring a process in time series, and each of the process result data storage means stored in the process result data storage means. It is characterized in that it includes a process state determination means for performing wavelet analysis of the process performance data and determining the presence or absence of abnormality by converting the analysis result into a three-dimensional contour pattern.

  According to the present invention, it is possible to efficiently determine whether a process is normal from a change in spectrum obtained by wavelet analysis of process performance data output from the process side while simplifying the configuration of the entire apparatus and reducing processing time. Discrimination and estimation can be performed, system performance verification, failure detection, failure prediction, and failure diagnosis can be performed with high accuracy, energy consumption can be reduced, and maintenance costs can be reduced.

<< Description of Wavelet Analysis Method Used in the Present Invention >>
First, prior to specific description of the process management apparatus according to the present invention, a wavelet analysis method used in the process management apparatus according to the present invention will be described.

In the conventional analysis method, frequency characteristics and the like are analyzed by using the following Fourier transform formula for signal data “x (t)” such as process data output from the monitoring target.

On the other hand, in the present invention, frequency characteristics are analyzed using the following (continuous) wavelet transform equation which has been attracting attention as a technique for analyzing unsteady signals in recent years.

  Here, “φ (•)” is a wavelet analysis and is a basis function called a mother wavelet, and “φ ((t−b) / a)” is a mother wavelet “φ (t)” that is only “b”. Translated and scaled (enlarged / reduced) by “a”.

  At this time, the width of the mother wavelet “φ (t)” is “a” times corresponding to the scale parameter “a”, and “1 / a” corresponds to the frequency. When it is large, the basis function is expanded, and when the scale parameter “a” is small, the basis function is narrowed.

The commonly used Fourier transform is the basis function of the (continuous) wavelet transform,
φ (t) = e− ^jωt (3)
a = ω−1 (4)
b = 0 (5)
The basis function has an infinite extent in the time domain.

  For this reason, the spectrum obtained by the Fourier transform becomes a linear function only on the frequency axis, the time information of the signal is lost, and the time dependency of which part of the observation data is characteristic is determined. I can't.

In contrast, in the wavelet transform, for example, a temporally local function called a Gabor function shown in the following equation is used.

  As a result, the spectrum obtained by the wavelet transform becomes a quadratic function with respect to the frequency axis and the time axis, and has temporal information of the signal, so that the time dependence of which part of the observation data is characteristic is determined. be able to.

  Details on the wavelet transform theory are introduced in, for example, Kohei Arai “Basic Theory of Wavelet Analysis”, Morikita Publishing Co., Ltd. (2000).

  However, if all of the process data output from each sensor provided in the building is analyzed using such a wavelet analysis method, the analysis result becomes too complex, making it difficult to detect, predict, diagnose, etc. Therefore, there is a risk that the discovery of the failure will be delayed.

  Therefore, in the present invention, when the number of sensors provided in the building is large, the principal component of the process data output from each sensor is extracted using the principal component analysis method, and the main component obtained by this extraction processing is extracted. By analyzing the components using the wavelet analysis method, early detection, prediction and diagnosis of failures are realized.

<< First Embodiment >>
<Description of Configuration of Embodiment>
FIG. 1 is a block diagram showing the overall configuration of a first embodiment of a process management apparatus according to the present invention.

  The process management apparatus 1 shown in this figure is connected to each measurement monitoring control means 5 installed on the target building side (department building 2, office building 3, hospital building 4, etc.) via the network 6, and each measurement monitoring. The process result data storage means 7 that captures and accumulates the detection results (process result data) collected by the control means 5 and the principal component analysis (PCA: Principa1 Components anlysis) are stored in the process result data storage means 7. Statistical analysis means 8 for generating the principal component by analyzing the actual process performance data, and each installed on the target building side that is subject to monitoring control by wavelet analysis of the principal component obtained by the statistical analysis means 8 The process state determination means 9 for determining whether the air conditioner is normal or abnormal, and the analysis data such as the contour line pattern obtained by the process state determination means 9 are accumulated. Analysis data storage means 10.

  The process result data storage means 7 includes the energy consumption of the air conditioner and other process values from each measurement monitoring control means 5 installed on the target building side such as each department store building 2, each office building 3, each hospital building 4, etc. When process performance data including (outside temperature, room temperature, indoor humidity, etc.) is output, it is fetched via the network 6, and based on the determination result of the process state determination means 9, the normal process performance data The accumulated process result data is supplied to the statistical analysis means 8 while being divided into process result data at the time of failure.

  The statistical analysis means 8 is obtained by the process simulation means 11 and the process simulation means 11 which perform simulation when the process is normal and simulation when various failures occur, using various parameters such as various buildings and load conditions. The simulation data storage means 12 for storing the simulation results and the simulation results stored in the simulation data storage means 12 or the process result data stored in the process result data storage means 7 using principal component analysis. Principal component analysis means 13 is provided that performs statistical analysis and performs a pre-stage process necessary for early and easy extraction of the difference between the characteristic value in the normal state and the characteristic value in the abnormal state.

  Before starting monitoring and control of the air conditioners on the target building side, using various parameters such as various buildings and load conditions, simulations when the process is in a normal state and various failure states The simulation results are accumulated and the simulation results are accumulated, and the simulation results are analyzed by principal component analysis, which is necessary to quickly and easily extract the difference between the normal state characteristic values and the abnormal state characteristic values. In addition to obtaining process variable connection weight parameters, using this process variable connection weight parameter, principal component analysis of simulation results when the process is in a normal state and simulation results when various fault conditions occur, and each main analysis The score is supplied to the process state determination means 9. After this, when performing monitoring control of the air conditioners and the like on each target building side, the process result data supplied via the process result data storage means 7 is principally analyzed using the process variable coupling weight parameter, The main analysis score is supplied to the process state determination means 9. Then, after the operation period of the process management device 1 becomes long and the process result data storage means 7 has sufficiently stored the process result data at the normal time and the process result data at the time of failure, these process result data are used. Update the process variable binding weight parameter.

  The process state determination means 9 uses the (continuous) wavelet transform equation shown in the above equation (2) to perform wavelet analysis on each main analysis score obtained by the principal component analysis means 13 of the statistical analysis means 8. The contour of the analysis result obtained by the wavelet analysis means 14 is converted into a contour pattern and stored in the analysis data storage means 10, and on the sea surface based on the contour line pattern stored in the analysis data storage means 15. A land area calculating means for calculating the corresponding amplitude height and dividing the contour line pattern obtained by the contour line patterning means 15 into a sea surface portion and a land portion based on the amplitude height and calculating the area of the land portion. 16 and the contour line pattern stored in the analysis data storage means 10, etc. When the land area obtained by the land area calculating means 16 exceeds a preset threshold value while displaying a three-dimensional contour line pattern having an axis, the process being monitored and controlled is abnormal. In addition to supplying the determination result to the process result data storage means 7, the analysis data storage means 10 and the like, a normal / abnormality determination means 17 for displaying a process abnormality on the display device is provided.

  And before starting monitoring control of the air conditioner etc. of each target building side, each main analysis score obtained by the principal component analysis means 13 of the statistical analysis means 8 is subjected to wavelet analysis, and the analysis result is converted into a contour pattern, After accumulating in the analysis data accumulating means 10, the height of the amplitude corresponding to the sea surface is obtained from the normal contour line pattern and the abnormal contour line pattern accumulated in the analysis data accumulating means 10. After that, when performing monitoring control such as an air conditioner provided on each target building side, wavelet analysis of the principal component score of the process result data output from the principal component analysis means 13 of the statistical analysis means 8 is performed, A contour line pattern is obtained and stored in the analysis data storage means 10. In parallel with this operation, the height of the amplitude corresponding to the sea surface is used to determine the area of the land portion (the portion having a value larger than the amplitude height corresponding to the sea surface) included in the contour pattern. When a preset threshold value is exceeded, it is determined that an abnormality has occurred in the process, the determination result is supplied to the process result data storage means 7, the analysis data storage means 10, etc., and the process abnormality is displayed on the display device. Is displayed.

  The analysis data storage means 10 captures each time a new contour line pattern is output from the contour line patterning means 15 of the process state determination means 9 and stores it as time series data in units of one day while determining the process state. When a read request is issued from the land area calculation means 16 of the means 9, the accumulated contour line pattern is read and supplied to the land area calculation means 16 of the process state determination means 9. Whether the contour line pattern output from the contour line patterning unit 15 of the process state determining unit 9 is normal or abnormal is determined based on the normal / abnormal determination result output from the normal / abnormal determining unit 17 of the process state determining unit 9. Classify and accumulate.

<Description of Operation of Embodiment>
Next, referring to the block diagram shown in FIG. 1 and the flowchart shown in FIG. 2, the operation before starting the monitoring control of the air conditioner provided in each target building, the air conditioner on the target building side, etc. The operation of the process management apparatus 1 will be described separately from the operation when monitoring control is performed.

[Operation before starting monitoring and control of air conditioners, etc. at each target building]
First, before starting monitoring control of an air conditioner or the like provided on each target building side, the simulation process shown in the flowchart of FIG. 2 is started, and each of the processes input to the process simulation means 11 of the statistical analysis means 8 is started. Each data related to the target building, each load condition data, and the like are arranged for each time.

  After that, the process simulation means 11 reads information on the target building side that has been input in advance, and data such as the size of the room, the heat capacity, the wall area, and the thermal conductivity of the wall provided on the building side. Is set (step S1), and based on each load condition data and the like, the outside air temperature, the outside air humidity, and the amount of solar radiation, which are meteorological elements related to the air conditioning load, are set in advance (step S2), For the simulation time corresponding to the change, the amount of heat obtained from the window glass surface using the amount of solar radiation, the amount of heat generated by humans, electrical equipment such as OA, the amount of heat generated by lighting, etc. are calculated in advance, and heat load data is set. (Step S3).

  Thereafter, the process simulation means 11 checks whether or not the sampling control period is an integer multiple of the simulator calculation period, and when it is the sampling period (step S4), the room temperature is measured by the DDC (Direct Digita1 Contro11er) control algorithm method. Based on the deviation between the set value and room temperature, the difference between the humidity set value and room humidity, etc., the PID parameter (speed type) is calculated (step S5).

  Next, the process simulation means 11 uses the theoretical calculation formula of the heat exchanger and the calculation formula of the air diagram in the order of the cooling coil, the heating coil, and the humidifier, and each state point (temperature) of the air in the air conditioner , Humidity, enthalpy, etc.) (step S6), the wall of the room in the building is divided into the outer wall (with windows), the floor, the inner wall, and the ceiling, and the heat transfer equation (heat transfer model) The heat balance is described (step S7), the dynamic characteristic change between the room temperature and the wall temperature is calculated by the differential decomposition method, and the air state point at the air conditioner inlet is obtained by the mixing calculation of the recirculated air and the introduced outside air ( Step S8).

  Hereinafter, the above-described calculation is executed by the process simulation means 11 until the simulation end time is reached while the time data is being updated, and the simulation when the process is in a normal state and the simulation when the various fault states occur. Is performed, and the process result data when the process is in a normal state and the process result data when various failure states are stored in the simulation data storage unit 12 (steps S4 to S10).

  Further, in parallel with this operation, the process results when the process accumulated in the simulation data accumulating unit 12 by the principal component analyzing unit 13 is normal before the monitoring control of the air conditioner etc. of each target building is started. Data and process performance data when various types of failures occur are analyzed, and process variable connection weights necessary to quickly and easily extract the difference between the normal state characteristic value and the abnormal state characteristic value A parameter is required.

For example, the process performance data is multivariate data with the number of variables “X ₁ , X ₂ ,..., X _p ” and the number of observation objects is n, and using this multivariate data, p When creating fewer m new variables “Z ₁ , Z ₂ ,..., Z _m ”, the process variable coupling weight parameters “a ₁₁ , a _{12 ,.} Calculated.

Var (Z ₁ ) ≧ Var (Z ₂ ) ≧… ≧ Var (Z _m ) (7)
a _i1 ² + a _i2 ² + ... 10 a _ip ² = 1 (8)
Where i: i = 1, 2,..., M
Var (Z _i ): Variance At this time, the new variable “Z ₁ ” is made independent of the new variable “Z ₂ ”. The new variable “Z ₃ ” is made independent of the new variable “Z ₁ ” and the new variable “Z ₂ ”. Hereinafter, the same process is performed from the new variable “Z ₄ ” to the new variable “Z _m ”.

Under such conditions, the process variable connection weight parameter _{"a 11, a 12, ...} , a mp" to seek the original variables _{"X 1, X 2, ...} , X p" covariance matrix of The process variable coupling weight parameters “a ₁₁ , a ₁₂ ,..., A _mp ” are none other than finding the eigenvectors.

However, the original variable _{"X 1, X 2, ...} , X p" and the units of the new variable _{"Z 1, Z 2, ......} , Z m" from that and units of different, in front of the principal component analysis , The original variables “X ₁ , X ₂ ,..., X _p ” are normalized (standardized).

That is, when the average value of all target data of the kth variable is “M _k ” and the standard deviation is “S _k ”, each variable “X _kj ” is normalized as follows.

X _kj → (X _kj −M _k ) / S _k (9)
However, j: j = 1 to n (n is the number of data)
Next, the process variable connection weight parameters “a ₁₁ , a ₁₂ ,..., A _mp ” are used by the principal component analysis means, and the operation result data (or The original variables “X ₁ , X ₂ ,..., X _p ” constituting the process result data at the time of various fault conditions are combined to form new variables “Z ₁ , Z ₂ ,. ..., _{Z m} "is created.

_{Z 1 (t) = a 11} · X 1 (t) + a 12 · X 2 (t) + ... Ten _{a 1p · X p (t)} ... (10)
_{Z 2 (t) = a 21} · X 1 (t) + a 22 · X 2 (t) + ... Ten _{a 2p · X p (t)} ... (11)
...
...
Z _m (t) = a _m1 · X ₁ (t) + a _m2 · X ₂ (t) + ... 10 _amp · X _p (t) (12)
However, _amp : process variable connection weight parameter t: time In this case, it is assumed that new variables (principal component scores) “Z ₁ , Z ₂ ,..., Z _m ” have the following properties.

The new variable “Z ₁ ” allows the information of the original variables “X _{1 to} X _p ” to be aggregated to the maximum extent. Further, the new variable “Z ₂ ” is configured so that the information of the original variables “X _{1 to} X _p ” is aggregated as much as possible after the new variable “Z ₁ ”. Hereinafter, the same applies from the new variable “Z ₃ ” to the new variable “Z _m ”.

Then, the principal component analysis means 13 performs principal component analysis processing on all process result data when the process accumulated in the simulation data accumulation means 12 is in a normal state, and applies to all process result data when various failure states occur. when the principal component analysis process and is completed, the main component score "Z _m" obtained at the time of normal in the principal component analysis, principal component scores "Z _m" at the time of failure and is supplied to the process state judging means 9 .

In parallel with the above-described operation, the normal component score “Z _m ” output from the principal component analysis unit 13 by the wavelet analysis unit 14 and the contour line patterning unit 15 of the process state determination unit 9 and the failure time are displayed. The principal component score “Z _m ” is subjected to wavelet analysis, and is also formed into a contour line pattern to generate a normal contour line pattern and an abnormal contour line pattern, which are stored in the analysis data storage means 10. When the wavelet analysis unit 14 and the contour line patterning unit 15 finish the wavelet analysis process and the contour line patterning process for all the principal component scores “Z _m ” output from the principal component analysis unit 13, the land area calculation unit 16, the height of the amplitude corresponding to the sea level is obtained and stored based on the normal contour line pattern stored in the analysis data storage means 10 and the abnormal contour line pattern.

[Operation when monitoring control of air conditioners etc. of each target building]
Thereafter, when the process control device 1 starts monitoring and controlling the air conditioners and the like of each target building, the process result data storage means 7 sets the target buildings such as the department store building 2, the office building 3, the hospital building 4, and the like. Each process result data output from each measurement monitoring control means 5 installed on the side, for example, process result data including energy consumption of each air conditioning process, other process values (outside temperature, room temperature, indoor humidity, etc.), etc. Captured and stored sequentially.

Further, in parallel with this operation, the process variable combination weight parameters “a ₁₁ , a ₁₂ obtained by the principal component analysis unit 13 of the statistical analysis unit 8 before starting the monitoring control of the air conditioners and the like on each target building side are started. ,..., A _mp ″ are used, and the operations shown in the above equations (10) to (12) are performed on the process result data stored in the process result data storage means 7 to obtain the principal component score “ Z _m ″ is obtained and supplied to the process state determination means 9.

Then, the wavelet analysis unit 14 of the process state determination unit 9 uses the (continuous) wavelet transform equation shown in the above equation (2) to perform wavelet analysis of the principal component score “Z _m ”, and then performs contour line patterning unit 15, the wavelet analysis result is contoured and accumulated in the analysis data accumulating unit 10, and obtained by the land area calculating unit 16 before the monitoring control of the air conditioner etc. of each target building is started. The height indicating the amplitude of the sea surface is used, and the contour line pattern obtained by the contour line patterning means 15 is subjected to level discrimination to obtain the land area (land area).

  In parallel with this operation, the normal / abnormal determination means 17 of the process state determination means 9 reads the contour line pattern stored in the analysis data storage means 10 and sets the frequency axis and time axis on the display device. A three-dimensional contour line pattern is displayed, and the land area output from the land area calculating means 16 is compared with a preset threshold value, and the land area output from the land area calculating means 16 is calculated. When the threshold value is smaller than a preset threshold value, it is determined that the process is normal, the determination result is supplied to the analysis data storage means 10 and the current contour line pattern output from the contour line patterning means 15 is normal. A time contour pattern is accumulated and a message indicating that the process is normal is displayed on the display device.

  When the land area output from the land area calculation means 16 exceeds a preset threshold value, the normal / abnormality determination means 17 determines that the process is abnormal, and the determination result is analyzed. The current contour line pattern supplied to the data storage means 10 and the like and output from the contour line patterning means 15 is stored as a contour line pattern at the time of abnormality, and a message indicating that the process is abnormal is displayed on the display device. Is done.

When the operation period of the process management apparatus 1 becomes long and the process result data storage unit 7 has enough process result data at normal time and process result data at failure, the process result data is stored by the principal component analysis unit 13. The normal process performance data and the failure process performance data stored in the data storage means 7 are analyzed, and new process variable coupling weight parameters “a ₁₁ , a ₁₂ ,..., A _mp ” are obtained. The process variable connection weight parameter “a ₁₁ , a ₁₂ ,..., A _mp ” is updated.

  In parallel with this operation, when the operation period of the process management apparatus 1 becomes long, and the contour line pattern at the normal time and the contour line pattern at the abnormal time are sufficiently stored in the analysis data storage means 10, the land area calculating means 16, the height of the amplitude corresponding to the sea level is obtained again based on the normal contour line pattern stored in the analysis data storage means 10 and the abnormal contour line pattern, and stored as a new height. Is done.

<Simulation result of embodiment>
Next, simulation contents when the normality / abnormality determination capability of the process management apparatus 1 is confirmed will be described using the explanatory diagrams shown in FIGS. Since audio analysis software is used as wavelet conversion software for convenience, in the explanatory diagrams shown in FIGS. 3 to 6, the horizontal axis is in seconds, but the actual unit is “× 100 hours”. is there.

  First, the air conditioning process provided in the department store building 2 is used as the target process, and the following five variables (P = 5) that can be normally measured as the variables of the process performance data, that is, the actual operation currently in progress Variables that can collect process performance data without adding new sensors are used for air conditioning processes.

Room temperature (with measurement noise) (13)
Cold water operation valve MV (14)
Indoor humidity (15)
Outside temperature (16)
Outside air humidity (17)
Then, under the condition that the operation time of the air conditioning process is from “8:30” to “20:00 (calculation cycle: 1 minute)”, while changing the air conditioning load conditions such as the outside air temperature and the indoor heat load for 6 days, 1 Of the process performance data collected during the cooling period, the process variable combination weight parameter is used using 4,140 (6 days × 11.5 hours × 60) process performance data corresponding to 6 days. “A ₁₁ , a ₁₂ ,..., A _mp ” was obtained.

Next, among the process result data at the normal time, the process result data for one day was selected, and the principal component analysis was performed using the process variable coupling weight parameters “a ₁₁ , a ₁₂ ,..., A _mp ” described above. When the principal component score “Z _i ” that changes is obtained as shown in the trend graph of FIG. 3 and this principal component score “Z _i ” is subjected to wavelet analysis, as shown in the explanatory diagram of FIG. A three-dimensional contour line pattern 18 having an axis and a time axis and a two-dimensional contour line pattern 19 were obtained.

Further, the process performance data for one day including the process performance data when the capacity of the air supply fan is reduced is selected, and the above-described process variable combination weight parameters “a ₁₁ , a ₁₂ ,..., A _mp ” are used. When the principal component analysis is performed and the principal component score “Z _i ” obtained by the principal component analysis is wavelet analyzed, a three-dimensional contour pattern having a frequency axis and a time axis as shown in FIG. 20 and a two-dimensional contour line pattern 21 were obtained.

Further, process performance data for one day including process performance data when the capacity of the cooling coil is reduced is selected, and the above-described process variable coupling weight parameters “a ₁₁ , a ₁₂ ,..., A _mp ” are used. When the principal component analysis is performed and the principal component score “Z _i ” obtained by the principal component analysis is wavelet analyzed, as shown in the explanatory diagram of FIG. 6, a three-dimensional contour pattern 22 having a frequency axis and a time axis is obtained. As a result, a two-dimensional contour line pattern 23 was obtained.

  As can be seen from these simulation results, the land area of the contour line pattern 18 shown in FIG. 4, the land area of the contour line pattern 20 shown in FIG. 5, and the land area of the contour line pattern 22 shown in FIG. By using a value corresponding to the land area of the contour line pattern 18 shown in FIG. 4 as a threshold value, a contour line pattern different from the three-dimensional contour line pattern 18 for the normal process performance data shown in FIG. 4, for example, the three-dimensional figure shown in FIG. When the contour line pattern 20 is obtained, or when the three-dimensional contour line pattern 22 shown in FIG. 6 is obtained, it can be determined that an abnormality has occurred in the air conditioning process.

  Further, when the three-dimensional contour line pattern 20 shown in FIG. 5 is displayed together with a message indicating the occurrence of an abnormality on the display device, it is recognized from the waveform shape that the ability of the air supply fan has decreased, and 3 shown in FIG. When the dimensional contour line pattern 22 is displayed, it can be recognized from the waveform shape that the capacity of the cooling coil has decreased.

<Effect of embodiment>
As described above, in the first embodiment, when the monitoring control of the air conditioner or the like on each target building side is started, the process variable combination weight parameters “a ₁₁ , a _{12 ,.} The process result data stored in the result data storage means 7 is subjected to principal component analysis, and the principal component score “Z _i ” obtained by the principal component analysis is subjected to wavelet analysis processing, contour line pattern processing, and When the area is obtained and this exceeds a preset threshold value, the process side is judged to be abnormal, so the process performance data output from the process side is converted to the principal component score. This reduces the number of data types subject to wavelet analysis, shortens the time required for wavelet analysis processing and abnormality determination, and reduces wavelet analysis. It is possible to efficiently discriminate and estimate whether the process is normal from the change in spectrum obtained by analysis, and to perform system performance verification, failure detection, failure prediction, and failure diagnosis with high accuracy, while reducing energy consumption and maintenance costs. Can be reduced.

Moreover, in 1st Embodiment, before starting monitoring control of the air conditioner etc. of each object building side, the process performance data at the time of normal obtained by the simulation process of the statistical analysis means 8, and the process performance data at the time of abnormality , And a process variable coupling weight parameter “a ₁₁ , a ₁₂ ,..., A _mp ”, a normal contour line pattern, an abnormal contour line pattern, and a height indicating the amplitude of the sea surface are obtained. Even when there are many types of process result data output from the side, the number of data types subject to wavelet analysis can be reduced from the time the process starts to shorten the time required for wavelet analysis processing and abnormality determination. However, it is possible to efficiently discriminate and estimate whether the process is normal from the change in spectrum obtained by wavelet analysis, Performance verification and fault detection of the stem, failure prediction, along with a failure diagnosis can be performed with high accuracy, it is possible to reduce energy consumption, reduce maintenance costs.

  In the first embodiment described above, the statistical analysis unit 8 performs principal component analysis on the process result data stored in the process result data storage unit 7 to reduce the number of types of data while determining the process state. The data is supplied to the means 9, and wavelet analysis processing, contour line patterning processing, land area calculation processing, normality / abnormality determination processing, and the like are performed. The process result data stored in the process result data storage unit 7 is directly supplied to the process state determination unit 9, and wavelet analysis processing, contour line patterning processing, land area calculation processing, normal / abnormality determination processing, etc. May be performed.

  This simplifies the configuration of the entire process management device 1 and shortens the processing time, while efficiently determining whether the process is normal from the change in spectrum obtained by wavelet analysis of the process performance data output from the process side. It is possible to estimate, perform system performance verification, failure detection, failure prediction, and failure diagnosis with high accuracy, reduce energy consumption, and reduce maintenance costs.

<< Second Embodiment >>
FIG. 7 is a block diagram showing an example of the overall configuration of the second embodiment of the process management apparatus according to the present invention. In this figure, the same parts as those in FIG. 1 are denoted by the same reference numerals.

  The process management apparatus 31 shown in this figure is different from the process management apparatus 1 shown in FIG. 1 in that the contour line patterns output from the process state determining means 9 are distinguished and stored as normal contour lines and abnormal contour lines. Instead of the analysis data accumulating means 10, the contour line pattern output from the process state determination means 9 is stored separately in a normal contour line pattern and a contour line pattern for each failure content, and the failure output from the failure factor determination means 33. The fault knowledge data storage means 32 for storing the land patterns according to the contents is provided, and when the process state determination means 9 determines that the plant is abnormal, the contour line pattern output from the process state determination means 9 and the failure Based on the land pattern for each failure content stored in the knowledge data storage means 32, System system is by providing the failure factor determination means 33 determines have become abnormal.

  The failure knowledge data storage unit 32 is based on the normal / abnormal determination result output from the normal / abnormal determination unit 17 of the process state determination unit 9, the failure content output from the failure factor determination unit 33, and the like. The new contour line pattern output from the contour line patterning means 15 is classified into a normal contour line pattern and a contour line pattern for each failure content, and accumulated as time series data in units of one day, while the process state determining means When a read request is issued from the land area calculation means 16 of 9, the stored normal contour line patterns and the contour line patterns classified by failure contents are read and supplied to the land area calculation means 16. Further, the land pattern for each new failure content from the failure factor determination means 33, for example, the contour line pattern obtained at the time of system A failure is level-discriminated as shown in FIG. As shown in FIG. 9, the land pattern 34 at the time of system A failure with the grid for “0” is level discriminated as shown in FIG. 9, and the grid corresponding to the land is “1” for the sea The land pattern 35 at the time of failure of the B system with the grid set to “0”, and the contour line pattern obtained at the time of failure of the C system as shown in FIG. Each time the land pattern 36 at the time of C system failure is output with the value of “0” being stored, it is captured and stored, and when a read request is issued from the failure factor determination means 33, Land of It reads the pattern, and supplies the failure factor determination means 33.

The failure factor determination means 33 performs the simulation operation of the statistical analysis means 8 before starting the monitoring control of the air conditioner etc. on each target building side, and the principal component score “Z _i corresponding to the process result data for each failure content”. ”Is generated, the contour pattern for each failure content obtained when the process state determination means 9 performs wavelet analysis and contour patterning, the height indicating the amplitude of the sea surface is taken in, and the failure is detected at the height indicating the amplitude of the sea surface. The contour patterns according to contents are subjected to level discrimination to generate land patterns 34 to 36 according to fault contents as shown in FIGS. 8 to 10 and stored in the fault knowledge data storage means 32.

  Then, when the monitoring control of the air conditioners and the like on each target building side is performed and the normal / abnormal determination unit 17 of the process state determination unit 9 determines that the process is abnormal, as shown in FIG. 11 and FIG. After assigning “1” to the grid corresponding to the land portion of the contour line pattern 37 obtained by the converting means 15 and creating the land pattern 38 assigned “0” to the sea, the land pattern 38 After obtaining the sum of products of each of the land patterns 34 to 36 for each failure content stored in the failure knowledge data storage means 32 and selecting the land pattern for each failure content having the maximum product sum value. Then, it is estimated that the same failure as the failure content of this land pattern has occurred, the failure content is displayed on the display device, and the current failure content is supplied to the failure knowledge data storage means 32, Contours pattern 37 as a fault condition by the contour pattern output from the high-line patterning means 15, accumulates.

  Thereafter, when the operation period of the process management device 31 is extended and the contour line patterns for each failure content are sufficiently accumulated in the failure knowledge data accumulation means 32, the height of the sea used by the land area calculation means 16 is calculated. At the same time, the contour line pattern for each failure content stored in the failure knowledge data storage means 32 is discriminated by the level of the sea, and the grid corresponding to the land is “1”, and the grid corresponding to the sea is “0”. Is created, and the land pattern for each failure content stored in the failure knowledge data storage means 32 is updated.

  As described above, in the second embodiment, a land pattern for each failure content corresponding to the contour line pattern for each failure content is created and stored in the failure knowledge data storage means 32, and the process state determination means 9 makes the process abnormal. Is determined, the failure factor determination means 33 takes in the contour line pattern output from the process state determination means 9 to create a land pattern and stores it in the land pattern and failure knowledge data storage means 32. Since the similarity with the land pattern for each failure content is determined and the failure content is estimated based on this determination result, it can be obtained by wavelet analysis of process performance data output from the process side Efficiently discriminate and estimate whether the process is normal from the change in the spectrum, and estimate the cause when a failure occurs, Performance verification and fault detection of the stem, failure prediction, along with a failure diagnosis can be performed with high accuracy, it is possible to reduce energy consumption, reduce maintenance costs.

  Further, in the second embodiment, the contour line pattern at the time of failure is divided into a grid, and “1” is assigned to the grid portion corresponding to the land, and “0” is assigned to the grid corresponding to the ocean. A land pattern is created and stored in the failure knowledge data storage unit 32. When the process state determination unit 9 determines that the process is abnormal, the failure state determination unit 33 outputs the process pattern determination unit 9. Divide the contour line pattern into a grid, assign "1" to the grid corresponding to the land, and create a land pattern that assigns "0" to the sea corresponding to the sea. The product sum of each land pattern for each failure content stored in the data storage means 32 is summed, and the failure content is estimated based on the failure content of the land pattern with the maximum product sum value. Therefore, it is possible to simplify the creation of the failure content determination data and the failure content determination processing, and facilitate the manufacture of the process management device 31, while the process result data output from the process side is subjected to wavelet analysis. Efficiently discriminates and estimates whether the process is normal from changes, quickly estimates the cause of a failure, and enables accurate system performance verification, failure detection, failure prediction, and failure diagnosis, as well as energy consumption And maintenance costs can be reduced.

In the second embodiment, the principal component score “Z _i ” for each failure content is generated by the simulation operation of the statistical analysis means 8 before starting the monitoring control of the air conditioner etc. of each target building. Then, the component state score “Z _i ” is subjected to wavelet analysis processing and contour line patterning processing by the process state determination means 9 to create a contour line pattern for each failure content and a height indicating the amplitude of the sea surface. The contour pattern for each failure content is discriminated at the indicated height, and a land pattern for each failure content is created and stored in the failure knowledge data storage means 32. Based on the data, the land pattern for each failure content used in the failure content estimation process is calculated, and the process side Efficiently discriminate and estimate whether the process is normal from the changes in the spectrum obtained by wavelet analysis of the process performance data output from the system, and in the event of a failure, the cause is estimated to verify system performance, failure detection, and failure Prediction and failure diagnosis can be performed with high accuracy, energy consumption can be reduced, and maintenance costs can be reduced.

  Further, in the second embodiment, when the operation period of the process management device 31 becomes long and the contour line pattern for each failure content is sufficiently accumulated in the failure knowledge data accumulation unit 32, it is used by the land area calculation unit 16. The level of the contour line for each failure content stored in the failure knowledge data storage means 32 is discriminated based on the level of the ocean, and the grid corresponding to the land is set to “1”, corresponding to the ocean. Since the land pattern assigned with “0” is created and the land pattern according to the failure content stored in the failure knowledge data storage means 32 is updated, the error pattern output from the process side is displayed. Based on the process performance data, the land pattern for each failure content used in the failure content estimation process is updated, and the land pattern for each failure content is changed according to the state change on the plant side. The process performance data output from the process side is optimized while efficiently determining whether or not the process is normal from the changes in the spectrum obtained by wavelet analysis. Performance verification, failure detection, failure prediction, and failure diagnosis can be performed with high accuracy, energy consumption can be reduced, and maintenance costs can be reduced.

<< Other embodiments >>
In addition, this invention is not limited to the said embodiment, It can implement variously in the range which does not deviate from the summary.

  Further, in the above-described embodiment, the failure detection, failure prediction, and failure diagnosis functions of the BEMS device that controls the air conditioning process of each building are used as specific examples. However, it is obvious that the present invention can be applied to general processes. is there.

  In addition, as a normal / abnormal determination algorithm, an algorithm that is determined based on whether or not a preset threshold value is deviated is used. An algorithm for determining normal / abnormal using an instantaneous value is used. It may be used. However, if such a judgment algorithm is used, there is a possibility of misjudgment due to the influence of data fluctuation due to “Noise Factor”. The degree of misjudgment caused by noise may be reduced in combination with an algorithm for evaluating the degree of normality and abnormality.

  In the above-described embodiment, the present invention has been described by taking as an example a BEMS device that remotely monitors and controls each air-conditioning facility installed in a plurality of buildings. You may carry out with. In this case, the process management apparatus according to the present invention is installed in the building, not remote.

It is a block diagram which shows the example of whole structure of 1st Embodiment of the process management apparatus by this invention. It is a flowchart which shows the simulation operation example of the process simulation means shown in FIG. It is a trend graph which shows an example of the principal component score obtained when the statistical analysis means shown in FIG. A normal component score is supplied to the process state determination means shown in FIG. 1 to obtain a wavelet analysis, a contour line pattern, a three-dimensional contour pattern having a frequency axis and a time axis, and a two-dimensional contour line. It is explanatory drawing which shows a pattern. The principal component score when the capacity of the air supply fan decreases is supplied to the process state judging means shown in FIG. 1, and the three-dimensional contour line having the frequency axis and the time axis obtained when the wavelet analysis and the contour line pattern are created. It is explanatory drawing which shows a pattern and a two-dimensional contour line pattern. A three-dimensional contour pattern having a frequency axis and a time axis obtained by supplying a principal component score when the capacity of the cooling coil is reduced to the process state judging means shown in FIG. It is explanatory drawing which shows a two-dimensional contour line pattern. It is a block diagram which shows the example of whole structure of 2nd Embodiment of the process management apparatus by this invention. It is explanatory drawing which shows the land pattern example at the time of the A system failure output from a failure factor determination means by the failure factor determination means shown in FIG. It is explanatory drawing which shows the land pattern example at the time of the B system failure output from a failure factor determination means by the failure factor determination means shown in FIG. It is explanatory drawing which shows the land pattern example at the time of C system failure output from a failure factor determination means by the failure factor determination means shown in FIG. It is explanatory drawing which shows the example of the contour line pattern output from the process state determination means shown in FIG. It is explanatory drawing which shows the land pattern example produced by the failure factor determination means shown in FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1,31: Process management apparatus 2: Department store building 3: Office building 4: Hospital building 5: Measurement monitoring control means 6: Network 7: Process performance data storage means 8: Statistical analysis means 9: Process state determination means 10: Analysis Data storage means 11: Process simulation means 12: Simulation data storage means 13: Principal component analysis means 14: Wavelet analysis means 15: Contour line patterning means 16: Land area calculation means 17: Normal / abnormality determination means 18, 20, 22: Three-dimensional contour line pattern 19, 21, 23: Two-dimensional contour line pattern 32: Failure knowledge data storage means 33: Failure factor determination means 34: Land pattern when system A fails 35: Land pattern when system B fails 36: C Land pattern at the time of system failure 37: Contour pattern 38: Land pattern N

Claims (7)

Process result data storage means for storing a plurality of process result data obtained by measuring processes in time series,
Wavelet analysis of each process performance data stored in the process performance data storage means, and a process state determination means for determining the presence or absence of abnormality by converting this analysis result into a three-dimensional contour line pattern;
A process management apparatus comprising: The process management apparatus according to claim 1, wherein
The process result data stored in the process result data storage means is subjected to principal component analysis using a process variable combination weight parameter designated in advance, and the principal component score obtained by the principal component analysis is used as the process state. Statistical analysis means to be supplied to the judgment means;
A process management apparatus comprising: The process management apparatus according to claim 2, wherein
The statistical analysis means includes a process simulation means for simulating the normal state of the process and various failure states,
After calculating the process variable coupling weight parameter used in the principal component analysis based on the process performance data obtained by the simulation process, each process variable data accumulated means is stored in the process performance data accumulation means using the process variable coupling weight parameter. The process result data is subjected to principal component analysis, and the principal component score obtained by the principal component analysis is supplied to the process state determination means.
A process management apparatus characterized by that. In the process management device according to any one of claims 1, 2, and 3,
Failure knowledge data storage means for storing a land pattern by failure content corresponding to a contour line pattern by failure content;
When the process state determination means determines that the process is abnormal, a land pattern is created from the contour line pattern output from the process state determination means, and is stored in the land pattern and the failure knowledge data storage means. A failure factor determination means for determining similarity with a land pattern for each failure content and estimating the failure content based on the determination result;
A process management apparatus comprising: The process management apparatus according to claim 4, wherein
The land pattern for each failure content stored in the failure knowledge data storage unit and the land pattern generated by the failure factor determination unit are obtained by dividing the contour line pattern into a grid pattern and setting “1” in the grid portion corresponding to the land. Assigned, created by assigning "0" to the cell corresponding to the sea,
When the failure factor determination means determines the similarity of each land pattern, each failure pattern of the land pattern obtained based on the contour line pattern output from the process state determination means is stored in the failure knowledge data storage means. Take the product sum with each square of the land pattern for each failure content, and estimate the failure content based on the failure content of the land pattern with the largest product sum value,
A process management apparatus characterized by that. In the process management device according to any one of claims 4 and 5,
Using the process result data at the time of abnormality obtained by simulation, create contour lines for each failure content, create a land pattern for each failure content based on these contour patterns for each failure content, and store the failure knowledge data Accumulate in the means,
A process management apparatus characterized by that. In the process management device according to any one of claims 4, 5, and 6,
The failure factor determination means determines the content of the failure when the process state determination means determines that the process is abnormal, and is determined to be process abnormality among the contour line patterns stored in the failure knowledge data storage means. Contour lines to which the fault contents stored in the fault knowledge data accumulating means are added when the fault contents are added to the contour pattern pattern and the contour line patterns with the fault contents are sufficiently accumulated in the fault knowledge data accumulating means. Create a land pattern for each failure content from the pattern, and update the land pattern for each failure content stored in the failure knowledge data storage means,
A process management apparatus characterized by that.