Signal optimization method based on diagnosability evaluation result
By constructing a fault detectability and isolability model and calculating the joint mean-variance difference index, sensor signals are optimized, solving the problem of unbalanced signal optimization in existing technologies and improving the comprehensiveness and accuracy of fault diagnosis.
Patent Information
- Application Number
- CN202511699691.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-19
- Publication Date
- 2026-02-13
AI Technical Summary
Existing methods, when optimizing sensor signals, cannot comprehensively quantify the signal's ability to detect and isolate faults within a unified framework, leading to unbalanced diagnostic results and failing to guarantee the comprehensiveness of the diagnostic results.
Construct fault detectability and fault isolateability models, calculate maximum mean difference (MMD) and maximum variance difference (MVD) values, merge them into a joint mean-variance difference (JMVD) value, define detectability and isolateability assessment indices, and select the best signal through a diagnostic threshold.
It enables a comprehensive quantitative assessment of signal fault detection and isolation capabilities, improving the comprehensiveness and accuracy of fault diagnosis, and enhancing the ability to identify complex fault characteristics and resist interference.
Smart Images
Figure CN121524569A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of sensors, in particular to a signal optimization method based on diagnosability evaluation results. BACKGROUND
[0002] The signal optimization method based on diagnosability evaluation results is a key link of complex equipment fault diagnosis, aiming to select the most sensitive and most discriminant monitoring signals to the fault through quantitative evaluation of multi-source sensor signals to improve the accuracy of diagnosis, which is the basis for building an efficient and reliable diagnosis model. The sensors are placed at the key parts of the equipment to collect physical signals reflecting the running state of the equipment in real time. High-quality monitoring signals are the premise of accurate fault diagnosis. In order to comprehensively evaluate the diagnosability of signals, a scientific diagnosability evaluation system needs to be established to ensure the final effect of fault diagnosis.
[0003] At present, due to the complexity of equipment running state and the diversity of fault modes, when optimizing sensor signals, the existing methods usually focus on the time-frequency domain features of signals and simple separability metrics, which cannot comprehensively quantify the detection ability and isolation ability of signals to faults in a unified framework. When only considering a single dimension index, the selected signals will be unbalanced in fault detection and isolation performance, which cannot guarantee the comprehensiveness of the diagnosis results.
[0004] Therefore, the present application provides a signal optimization method based on diagnosability evaluation results to solve the above problems. SUMMARY
[0005] In view of the deficiencies of the prior art, the present application provides a signal optimization method based on diagnosability evaluation results, which solves the problem of unbalanced fault detection and isolation performance of the selected signals in the background art, and cannot guarantee the comprehensiveness of the diagnosis results.
[0006] To achieve the above purpose, the present application provides the following technical scheme: A signal optimization method based on diagnosability evaluation results, the method comprising the following steps: S1, acquiring monitoring signal data collected by multiple candidate sensors under normal state and multiple fault states of equipment; S2, constructing a fault detectability model based on the monitoring signal data, the fault detectability model being used to quantify the difference between fault states and normal states in sensor signals; S3, constructing a fault isolability model based on the monitoring signal data, the fault isolability model being used to quantify the difference between different fault states in sensor signals; S4, defining a diagnosability threshold, the diagnosability threshold being calculated based on the difference between normal state signals; S5, calculating a maximum mean difference MMD value, which is used to evaluate the mean difference of distribution; S6, calculating a maximum variance difference MVD value, which is used to evaluate the variance difference of distribution; S7, based on the MMD value and MVD value, calculating a joint mean-variance difference JMVD value, which integrates the mean difference and variance difference; S8, according to the JMVD value, calculating a detectability evaluation index Det and an isolability evaluation index Iso; S9, based on the detectability evaluation index and isolability evaluation index, calculating a diagnosability evaluation index Diag, and selecting a signal from candidate sensors according to the Diag value.
[0007] Preferably, the monitoring signal data in S1 includes the following steps: S11, collecting monitoring signal data of equipment in normal state and fault state through vibration sensors, temperature sensors and current sensors, the fault state including multiple fault modes, and the fault mode set being represented as: ; wherein represents the total number of fault modes, is the fault mode set, is the th fault mode; S12, pre-processing the collected monitoring signal data, including denoising and normalization processing, to remove environmental noise and dimension influence, to generate standardized signal data.
[0008] Preferably, the construction of the fault detectability model in S2 includes: The fault detectability model is defined as: for fault mode , its detectability: ; wherein is a difference evaluation operator, represents the sensor signal under fault state , and represents the sensor signal under normal state; The greater the value of the detectability , the stronger the detectability of the fault mode on the sensor signal.
[0009] Preferably, the construction of the fault isolability model in S3 includes: The fault isolability model is defined as: for fault mode and Its separability: ; in For the difference evaluation operator, and These represent the fault states respectively. and Sensor signals below; The isolation The larger the value, the more likely it is to indicate a fault mode. and The stronger the isolation of the sensor signal.
[0010] Preferably, the diagnostic threshold defined in S4 includes: Diagnostic threshold: ; in and These are two different normal state signals; when At that time, the failure mode It can be detected when At that time, the failure mode and It can be isolated.
[0011] Preferably, the calculation of the MMD value in S5 includes: The empirical estimation of MMD values is based on the definition of the regenerating kernel Hilbert space RKHS, and the formula is: ; in and There are two sample sets, representing sensor signal data under different states. The mapping function projects the data onto a high-dimensional feature space. RKHS is indicated. Represents the norm, and For sample set and Size, and For sample set and The first in The and the first One data point; By employing kernel function techniques, the Gaussian kernel function is used to avoid explicit mapping. The specific calculation is as follows: ; in For the kernel function, a Gaussian kernel is preferred. , is a kernel parameter.
[0012] Preferably, the calculation of MVD value in S6 includes: MVD value is defined as the difference of distribution variance quantified in RKHS, which is formulated as: ; where and denote two distributions, and denote the variance embeddings of distributions and , which are calculated as , and are the variance embedding vectors, is the supremum, and are data points and under the function ; The simplified calculation of MVD is: ; where the variance embeddings and are the variance embedding vectors of distributions and .
[0013] Preferably, the calculation of JMVD value in S7 includes: JMVD value is a fusion of MMD and MVD, which is defined as: ; where and are the mean embeddings of distributions and , and are the variance embedding vectors of distributions and , is a tuning coefficient, used to balance the weight of mean difference and variance difference; JMVD value is used as a difference evaluation operator to replace the difference calculation in fault detectability model and fault isolability model.
[0014] Preferably, the calculation of detectability evaluation index and isolability evaluation index in S8 includes: The detectability evaluation index Det is defined as: ; wherein Det is the detectability value of the fault mode, Diag is the diagnosability threshold, N is the total number of fault modes; The isolability evaluation index Iso is defined as: ; wherein , and respectively represent the standard deviation and mean value of to evaluate the volatility of isolation difficulty.
[0015] Preferably, the S9 calculates the diagnosability evaluation index and preferably the signal includes: The diagnosability evaluation index Diag is defined as Diag = Det + Iso, wherein Det is the detectability evaluation index, and Iso is the isolability evaluation index; According to the Diag value, all candidate sensor signals are sorted, and the sensor with the largest Diag value is selected as the optimal signal for fault diagnosis application; The preferred process includes: calculating the Diag value of each sensor, generating a diagnosability evaluation report, and automatically selecting the signal based on the threshold to improve the accuracy of fault detection and isolation.
[0016] Compared with the prior art, the present application provides a signal optimization method based on the diagnosability evaluation result, which has the following beneficial effects: 1. In the present application, when performing sensor signal optimization, a unified fault detectability model and fault isolability model are constructed, and the detectability evaluation index and the isolability evaluation index are calculated based on the joint mean-variance difference value, which realizes comprehensive quantitative evaluation of the signal fault detection ability and fault isolation ability, ensures that the performance requirements of different dimensions in the signal optimization process are balanced, solves the problem that the existing method cannot comprehensively quantify the signal diagnosis ability in a unified framework, and improves the comprehensiveness and accuracy of fault diagnosis.
[0017] 2. In the present application, when performing diagnosability evaluation, the maximum variance difference value is introduced to quantify the high-order statistical characteristics of signal distribution, and it is combined with the maximum mean difference value to form a joint mean-variance difference value, so that the evaluation method not only pays attention to the distribution mean difference, but also captures the distribution variance change sensitive to faults, solves the problem that the existing method is not sensitive to certain fault modes due to ignoring the variance difference, enhances the recognition ability of complex fault characteristics, and improves the precision and reliability of signal optimization.
[0018] 3. In the signal optimization process, a diagnosability threshold based on normal state signals is defined, and the activation condition of signal contribution is set accordingly, so that the calculation of detectability evaluation index and isolability evaluation index can adaptively filter out minor differences caused by normal fluctuations and noise, ensuring that only fault feature differences are included in the final diagnosability evaluation index, solving the problem of deviation in the existing method due to lack of adaptive threshold mechanism, and improving the anti-interference and robustness of the signal optimization result. BRIEF DESCRIPTION OF DRAWINGS
[0019] Figure 1 A flowchart of a signal optimization method based on diagnosability evaluation results according to the present application. DETAILED DESCRIPTION
[0020] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.
[0021] Please refer to Figure 1 The signal optimization method based on diagnosability evaluation results comprises the following steps: S1, acquiring monitoring signal data collected by a plurality of candidate sensors under normal state and a plurality of fault states; S2, constructing a fault detectability model based on the monitoring signal data, the fault detectability model being used to quantify the difference between the fault state and the normal state in the sensor signal; S3, constructing a fault isolability model based on the monitoring signal data, the fault isolability model being used to quantify the difference between different fault states in the sensor signal; S4, defining a diagnosability threshold, the diagnosability threshold being calculated based on the difference between the normal state signals; S5, calculating a maximum mean difference MMD value, the MMD value being used to evaluate the distribution mean difference; S6, calculating a maximum variance difference MVD value, the MVD value being used to evaluate the distribution variance difference; S7, calculating a joint mean-variance difference JMVD value based on the MMD value and the MVD value, the JMVD value fusing the mean difference and the variance difference; S8, calculating a detectability evaluation index Det and an isolability evaluation index Iso according to the JMVD value; S9, based on the detectability evaluation index and the isolability evaluation index, calculate the diagnosability evaluation index Diag, and select the signal from the candidate sensors according to the Diag value; The monitoring signal data in S1 includes the following steps: S11, collect monitoring signal data of the equipment in normal state and fault state through vibration sensor, temperature sensor and current sensor, the fault state includes multiple fault modes, and a fault mode set is represented as: ; Wherein represents the total number of fault modes, is a fault mode set, is the i-th fault mode; S12, pre-process the collected monitoring signal data, including denoising and normalization processing, to remove environmental noise and dimension influence, and generate standardized signal data; S2 includes constructing a fault detectability model: The fault detectability model is defined as: for the fault mode , its detectability: ; Wherein is a difference evaluation operator, represents the sensor signal under the fault state , and represents the sensor signal under the normal state; The greater the value of the detectability , the stronger the detectability of the fault mode on the sensor signal; S3 includes constructing a fault isolability model: The fault isolability model is defined as: for the fault mode and , its isolability: ; Wherein is a difference evaluation operator, and represent the sensor signals under the fault states and , respectively; The greater the value of the isolability , the stronger the isolability of the fault mode and on the sensor signal; S4 includes defining a diagnosability threshold: Diagnosability threshold: ; wherein and are two different normal state signals; when , the fault mode can be detected, when , the fault mode and can be isolated; The calculation of MMD value in S5 includes: The empirical estimation of MMD value is based on the definition of Reproducing Kernel Hilbert Space (RKHS), which is given by: ; wherein and are two sample sets representing sensor signal data under different states, is a mapping function to project data into a high-dimensional feature space, denotes RKHS, denotes norm, and are the sizes of sample sets and , and are the th and th data points in sample sets and ; By kernel trick, Gaussian kernel function is used to avoid explicit mapping, which is calculated as: ; wherein is the kernel function, preferably Gaussian kernel , is the kernel parameter; The calculation of MVD value in S6 includes: MVD value is defined as the difference of distribution variance quantified in RKHS, which is given by: ; wherein and represent two distributions, and represent the variance embeddings of distributions and , which are calculated as , and are the variance embedding vectors, is the supremum, and For data points and under function ; The simplified calculation is: ; where variance embeddings and are the variance embedding vectors of distributions and ; The calculation of JMVD values in S7 includes: The JMVD value fuses MMD and MVD, defined as: ; where and are the mean embedding of distributions and , and are the variance embedding vectors of distributions and , is a tuning coefficient, used to balance the weights of mean difference and variance difference; The JMVD value is used as a difference evaluation operator to replace the difference calculation in the fault detectability model and the fault isolability model; The calculation of detectability evaluation index and isolability evaluation index in S8 includes: The detectability evaluation index Det is defined as: ; where is the detectability value of the fault mode , is the diagnosability threshold, is the total number of fault modes; The isolability evaluation index Iso is defined as: ; where , and respectively represent the standard deviation and mean of , used to evaluate the volatility of isolation difficulty; The calculation of diagnosability evaluation index and the selection of signals in S9 includes: The diagnosability evaluation index Diag is defined as Diag = Det + Iso, where Det is the detectability evaluation index and Iso is the isolability evaluation index; All candidate sensor signals are ranked according to Diag values, and the sensor with the largest Diag value is selected as the optimal signal for fault diagnosis application; The preferred process includes calculating Diag values for each sensor, generating a diagnosability assessment report, and automatically selecting signals based on thresholds to improve the accuracy of fault detection and isolation; The calculation of MVD values in S6 further includes: The calculation of MVD values is based on variance embedding vectors in reproducing kernel Hilbert space (RKHS), where the variance embedding vector and are defined as the mapping of variance statistics of distributions and in RKHS, specifically: ; where is the mean embedding of distribution , denotes the tensor product, denotes the expectation operator; The MVD value is calculated by kernel function technique, avoiding explicit mapping, and when using Gaussian kernel function, the empirical estimation formula of MVD is: ; where and are sample sets, is a mapping function, and H is RKHS; This MVD value is used to quantify the variance difference of the distribution, enhancing the ability to capture the characteristics of the fault fluctuation; In the calculation of JMVD value, the setting of adjustment coefficient is based on the requirements of fault diagnosis scene, optimized by cross-validation and grid search, and the preferred range is ∈[0.2, 0.8], to balance the contribution of mean difference and variance difference, and JMVD value is used as a difference evaluation operator , when applied to fault detectability model and fault isolability model, its calculation ensures the non-negativity, symmetry and triangle inequality of distance measure, thus ensuring the mathematical rigor of the evaluation result; In the calculation of detectability evaluation index Det, when , the term represents the normalized actual detection difficulty, and the summation process considers the contribution of all fault modes, eliminating the dimensional influence, and in the calculation of isolability evaluation index Iso, the vector is defined as , where std( ) and mean( respectively represent the standard deviation and mean value of , the coefficient of variation of isolation difficulty is evaluated, and the value is large, which indicates that the fault isolation performance is poor. These index calculations combine the JMVD value to improve the adaptability and robustness of signal optimization; A maximum variance difference MVD calculation method is applied to fault diagnosability evaluation, characterized in that it comprises the following steps: Two sample sets and are obtained, representing sensor signal data in different states; The data is mapped to the reproducing kernel Hilbert space RKHS through the kernel function, and the variance embedding vector and are calculated, wherein , is the mean embedding of the distribution ; The MVD value is defined as , wherein represents the RKHS norm; The MVD value is used to quantify the variance difference of the distribution, and is used as an auxiliary index for fault diagnosability evaluation, combined with the MMD value.
[0022] A signal optimization method based on diagnosability evaluation results operates as follows: Step 1: Obtain monitoring signal data of the equipment in normal state and multiple fault modes through vibration sensor, temperature sensor and current sensor candidate sensors, and the fault mode set is defined as , wherein represents the total number of faults, and the original signal collected is processed by noise removal and normalization to eliminate environmental noise and dimension influence, generating standardized signal data to provide reliable input for subsequent evaluation. This step guarantees data quality and is the basis of the entire method.
[0023] Step 2: The method constructs a fault detectability model and a fault isolability model to quantify the difference of signals in different states. The fault detectability model is defined as , wherein represents the signal in the fault state , represents the normal state signal, is a difference evaluation operator. This model quantifies the difference between fault and normal states, and the larger the value, the stronger the detectability. The fault isolability model is defined as , which is used to quantify the difference between different fault modes and . The larger the value, the better the isolation. At the same time, the diagnosability threshold , based on two normal signal calculations, as a basis for determining whether the fault can be detected and isolated, only when and h , it is considered that the fault can be diagnosed.
[0024] Step three, the method enters the difference measurement calculation phase, focusing on the fusion of mean difference and variance difference. First, the maximum mean difference MMD value is calculated, based on the RKHS definition, the formula is ; where and are sample sets, is a mapping function, and the MMD value is used to evaluate the mean difference of the distribution. At the same time, the maximum variance difference MVD value is calculated, defined as , where and are variance embedding vectors of distributions and , and the MVD value quantifies the variance difference of the distribution and captures the signal fluctuation characteristics. Based on MMD and MVD, the joint mean-variance difference JMVD value is calculated, the formula is JMVD , where is a regulation coefficient used to balance the mean and variance contributions, and the JMVD value is used as a difference evaluation operator fun to replace the original model operator, ensuring a more comprehensive evaluation.
[0025] Step four, the method calculates the detectability evaluation index Det and the isolability evaluation index Iso based on the JMVD value. The Det index is defined as the sum of the actual detection difficulty after normalization when , the formula is Det , otherwise it is 0, and the larger the value, the stronger the signal detectability. The Iso index is defined as the coefficient of variation when , the formula is Iso , where , otherwise it is 0, and the larger the value, the better the isolation performance. These indices combined with the JMVD value enhance the adaptability and robustness of the evaluation.
[0026] Step five, the method completes signal optimization through the diagnosability evaluation index Diag. The Diag index is defined as Diag = Det + Iso, which integrates the detection and isolation capabilities. According to the Diag value, all candidate sensor signals are sorted, and the sensor with the largest Diag value is selected as the optimal signal. The optimization process includes calculating the Diag value of each sensor, generating a diagnosability evaluation report, and automatically selecting signals based on thresholds, thereby improving the accuracy of fault detection and isolation. The entire principle step forms a closed-loop process from data acquisition to optimization decision, ensuring the scientificity and reliability of signal optimization through mathematical modeling and quantitative indices.
[0027] It has to be noted that, in the present document, the terms "first", "second", etc. merely serve to identify a subject or action, without necessarily requiring or implying any such actual relationship or order between such subjects or actions. Moreover, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without more constraints, exclude the presence of additional identical elements in the process, method, article, or apparatus that comprises the element.
[0028] While embodiments of the application have been shown and described, it is to be understood that the application is not limited to the details of the embodiments described, since numerous further modifications and changes can be apparent to one skilled in the art without departing from the spirit and scope of the application as defined by the appended claims and their equivalents.
Claims
1. A signal optimization method based on diagnosticability assessment results, characterized in that, The method includes the following steps: S1. Acquire monitoring signal data collected by multiple candidate sensors under normal and various fault conditions; S2. Based on the monitoring signal data, a fault detectability model is constructed. The fault detectability model is used to quantify the difference between the fault state and the normal state in the sensor signal. S3. Based on the monitoring signal data, construct a fault isolation model, which is used to quantify the differences in sensor signals under different fault states. S4. Define a diagnostic threshold, which is calculated based on the difference between normal state signals; S5. Calculate the maximum mean difference (MMD) value, which is used to assess the difference in distribution means; S6. Calculate the maximum variance difference (MVD) value, which is used to assess the variance difference of the distribution. S7. Based on the MMD and MVD values, calculate the joint mean-variance difference (JMVD) value, which integrates the mean difference and variance difference. S8. Calculate the detectability assessment index Det and the isolability assessment index Iso based on the JMVD value; S9. Based on the detectability assessment index and the isolability assessment index, calculate the diagnostic assessment index Diag, and select the best signal from the candidate sensors according to the Diag value.
2. The signal optimization method based on diagnosticability assessment results according to claim 1, characterized in that, The monitoring signal data in S1 includes the following steps: S11. Monitoring signal data of the equipment under normal and fault conditions are collected through vibration sensors, temperature sensors, and current sensors. The fault conditions include multiple fault modes, and the set of fault modes is represented as follows: ; in Indicates the total number of failure modes. For a set of failure modes, For the first One failure mode; S12. Preprocess the collected monitoring signal data, including denoising and normalization, to remove environmental noise and dimensional influences and generate standardized signal data.
3. The signal optimization method based on diagnosability assessment results according to claim 1, characterized in that, The fault detectability model constructed in S2 includes: The fault detectability model is defined as: for fault modes Its detectability: ; in For the difference evaluation operator, Indicates fault status The sensor signal below, This represents the sensor signal under normal conditions. The detectability The larger the value, the more likely it is to indicate a fault mode. The stronger the detectability of the sensor signal.
4. The signal optimization method based on diagnosticability assessment results according to claim 1, characterized in that, The fault isolation model constructed in S3 includes: The fault isolation model is defined as follows: for each fault mode and Its separability: ; in For the difference evaluation operator, and These represent the fault states respectively. and Sensor signals below; The isolation The larger the value, the more likely it is to indicate a fault mode. and The stronger the isolation of the sensor signal.
5. The signal optimization method based on diagnosability assessment results according to claim 1, characterized in that, The diagnostic threshold defined in S4 includes: Diagnostic threshold: ; in and These are two different normal state signals; when At that time, the failure mode It can be detected when At that time, the failure mode and It can be isolated.
6. The signal optimization method based on diagnosticability assessment results according to claim 1, characterized in that, The calculation of the MMD value in S5 includes: The empirical estimation of MMD values is based on the definition of the regenerating kernel Hilbert space RKHS, and the formula is: ; in and There are two sample sets, representing sensor signal data under different states. The mapping function projects the data onto a high-dimensional feature space. RKHS is indicated. Represents the norm, and For sample set and Size, and For sample set and The first in The and the first One data point; By employing kernel function techniques, the Gaussian kernel function is used to avoid explicit mapping. The specific calculation is as follows: ; in For the kernel function, a Gaussian kernel is preferred. , For kernel parameters.
7. The signal optimization method based on diagnosticability assessment results according to claim 1, characterized in that, The calculation of the MVD value in S6 includes: The MVD value is defined as the quantification of the variance of the distribution in the RKHS, and the formula is: ; in and Representing two distributions respectively, and Represents distribution and The variance embedding is calculated as , and For variance embedding vectors, For the upper bound, and For data points and In function The mapping value below; The simplified calculation is as follows: ; Where variance embedding and For distribution and The variance embedding vector.
8. The signal optimization method based on diagnosticability assessment results according to claim 1, characterized in that, The calculation of the JMVD value in S7 includes: JMVD value combines MMD and MVD, and is defined as: ; in and For distribution and Mean embedding, and For distribution and The variance embedding vector, For adjustment coefficients, The weights used to balance the differences in mean and variance; JMVD value as a difference evaluation operator It is used to replace the differential calculations in the fault detectability model and the fault isolation model.
9. The signal optimization method based on diagnosticability assessment results according to claim 1, characterized in that, The calculation of the detectability assessment index and the isolability assessment index in S8 includes: The detectability assessment index Det is defined as follows: ; in Fault mode Detectability value, The diagnostic threshold, This represents the total number of failure modes. The isolability assessment index Iso is defined as: ; in , and They represent The standard deviation and mean are used to assess the volatility of isolation difficulty.
10. The signal optimization method based on diagnosticability assessment results according to claim 1, characterized in that, The diagnosticability assessment index and preferred signals calculated in S9 include: The diagnosticability assessment index Diag is defined as Diag = Det + Iso, where Det is the detectability assessment index and Iso is the isolation assessment index. All candidate sensor signals are sorted according to their Diag values, and the sensor with the largest Diag value is selected as the optimal signal for fault diagnosis applications. The optimization process includes: calculating the Diag value for each sensor, generating a diagnostic assessment report, and automatically selecting signals based on thresholds to improve the accuracy of fault detection and isolation.