Dynamic equipment fault early warning method based on GRU algorithm and SPC

Abstract

The invention discloses a dynamic equipment fault early warning method based on a GRU algorithm and an SPC. The dynamic equipment fault early warning method specifically comprises the following steps of S1, collecting original data of dynamic equipment; s2, performing data processing on the training set, and storing a standardized model of the training set; s3, constructing a GRU model; s4, before the model is applied, processing measuring point parameter values; s5, estimating by applying a GRU model; s6, calculating deviation of the parameters of the main power equipment; s7, judging the health condition of the mobile equipment according to the deviation degree em; and S8, defining an SPC rule capable of changing the health degree of the mobile equipment in real time by referring to eight abnormal judgment criteria of the SPC standard. Considering deep learning model structure analysis, the invention provides a model based on deep learning, the model can quantify the operation state of the dynamic equipment, namely, health degree estimation, and the change trend of the health degree is judged in combination with an SPC rule, so as to carry out early warning on the health condition of the dynamic equipment.

Description

Technical Field

[0001] The present invention relates to a method for early warning of a moving equipment failure, and in particular to a method for early warning of a moving equipment failure based on a GRU algorithm and SPC. Background Art

[0002] As a core component of industrial production, the operating status of dynamic equipment directly affects the stability and safety of the production line. Therefore, accurate and timely evaluation and prediction of the status of dynamic equipment is of great significance for preventing equipment failures and ensuring production continuity.

[0003] Dynamic equipment is one of the common machines and equipment in industrial production. It refers to rotating equipment driven by a driver (that is, equipment that consumes energy), such as pumps, generators, compressors, fans, etc. Its energy can be electric power, pneumatic power, steam power, etc. The reliability research of dynamic equipment is of great significance to improving its service life and preventing accidents. With the improvement of the integration and informatization of large-scale electromechanical dynamic equipment systems, the difficulty of fault diagnosis and logistics support has increased. In order to ensure the continuous and stable operation of these systems, not affect the normal execution of tasks, reduce resource waste, and improve the efficiency of equipment maintenance and repair, it is necessary to keep track of the health of the equipment at all times, and make appropriate repair and maintenance decisions based on the system health to improve work efficiency.

[0004] The core of the health status evaluation of dynamic equipment is to quantitatively evaluate the comprehensive performance of dynamic equipment and classify the safety level, so as to intuitively show the health level of dynamic equipment. The health status evaluation is based on the state operation data, and the system performance degradation state is modeled to construct a one-dimensional health curve to characterize the degree of performance degradation or deviation.

[0005] As an important branch of artificial intelligence, deep learning has made breakthrough progress in image recognition, natural language processing, speech recognition and other fields in recent years. Its powerful model matching ability and adaptive characteristics make deep learning have significant advantages in dealing with complex and nonlinear problems. Especially when dealing with large-scale, high-dimensional data, deep learning can automatically extract features and improve the accuracy and generalization ability of prediction.

[0006] Dynamic equipment status assessment refers to the evaluation and prediction of the current status and possible future development trends of equipment by monitoring the equipment's operating data combined with the equipment's historical information and expert knowledge. This process not only helps to timely discover the precursors of equipment failure, but also provides a scientific basis for equipment maintenance, thereby avoiding production line interruptions and safety accidents caused by equipment failure.

[0007] Method based on professional knowledge: This method is based on human experience and knowledge to infer fault characteristics. That is, after a device fails, experts dissect the device to find the problem, and combine the changes in historical operating parameters before the failure to qualitatively or quantitatively describe the connection relationship between components, fault propagation mode, etc., summarize the qualitative or quantitative characteristics of the equipment failure, and use these characteristics to complete equipment failure warning and monitoring; Among them, the advantages are: when the monitored object is relatively simple and the process knowledge and production experience are sufficient, the use effect is better; Disadvantages: The accuracy of the warning is highly dependent on the richness of expert knowledge and the level of expert knowledge; Many experiences are difficult to describe in a reasonable formal expression.

[0008] As deep learning technology continues to mature and its application scenarios expand, more and more researchers are beginning to explore the application of deep learning in dynamic equipment status assessment. By building a status assessment model based on deep learning, the automatic analysis and processing of equipment operation data can be realized, and the key features reflecting the equipment status can be extracted, and then the equipment operation status can be evaluated and predicted. Compared with traditional manual assessment methods, this method has higher accuracy and efficiency. Summary of the invention

[0009] In order to solve the shortcomings of the above technology, the present invention provides a dynamic equipment fault early warning method based on GRU algorithm and SPC.

[0010] In order to solve the above technical problems, the technical solution adopted by the present invention is: a dynamic equipment fault early warning method based on GRU algorithm and SPC, which specifically includes the following steps:

[0011] Step S1: Collect the original data of the moving equipment, obtain the monitoring parameters that can reflect the performance of the equipment and save them;

[0012] Step S2: extracting data from the historical operation data of the equipment as a training set, performing data processing on the training set, and saving a standardized model of the training set;

[0013] Step S3: construct a GRU model;

[0014] Step S4: before applying the model, the key measuring points of the moving equipment to be evaluated are selected and the parameter values ​​of the measuring points are processed;

[0015] Step S5: Apply the GRU model to estimate the operating point parameter values ​​of the dynamic equipment to be evaluated;

[0016] Step S6: Calculate the overall deviation of the moving equipment and summarize the deviation of the moving equipment parameters;

[0017] Step S7: By deviation e m Size to judge the health of the moving equipment;

[0018] Step S8: Referring to the eight abnormality judgment criteria of the SPC standard, define SPC rules that can run the health changes of dynamic equipment in real time.

[0019] Preferably, in step S1, the collected data include: equipment operation data, equipment factory test data, all maintenance records within the equipment operation data time period, and the ranges corresponding to all measuring points of the equipment.

[0020] Preferably, in step S2, the intercepted data is historical operation data of 1 year or more and the time interval is 1 second;

[0021] Data processing includes parameter screening, data cleaning, and parameter standardization;

[0022] During the parameter screening process, the selection of operating parameters is obtained from the monitoring parameters saved in step S1;

[0023] During the data cleaning process, only the data during the normal operation period of the equipment is retained, and the data during the failure period needs to be eliminated;

[0024] Parameter standardization uses the Z-score method, that is, the formula X = (x-μ) / σ is applied to make the variance of the new X data set 1 and the mean 0, so that the processed data conforms to the standard normal distribution.

[0025] Preferably, in step S3, each operating parameter value of the dynamic equipment in the training set is used as the target of the model, and other operating parameter values ​​except the target parameter are used as input of the model to perform supervised training, and the training result model is saved.

[0026] Preferably, the specific process of step S4 is: obtaining the operating data of the current dynamic equipment to be evaluated, screening the operating parameter items of the equipment, and standardizing the screening result set, and the processing basis comes from the parameter list and standardized model saved in step S1 and step S2.

[0027] Preferably, in step S6, the deviation formula is:

[0028]

[0029] In the formula, x i , They represent the actual value and estimated value of the i-th parameter among the m key parameters to be estimated, respectively, and w i represents the weight of the deviation of the i-th parameter in the overall deviation, e m Represents the overall deviation of m parameters to be estimated.

[0030] Preferably, in step S7, the deviation is mapped to a threshold value within a range of 0-100 by a mathematical method, and this mapping value is called healthiness. The mapping formula is:

[0031]

[0032] In the formula, h x It indicates the health value of the mapping when the overall deviation is x, and p indicates the deviation cutoff value. That is, when the overall deviation is greater than p, the mapped health is less than 60, otherwise it is greater than or equal to 60. When the health is less than 60 points, it indicates that the equipment is abnormal and needs maintenance.

[0033] Preferably, in step S8, for the case in step S7 where the health level is greater than or equal to 60, when monitoring the fluctuation of the health level score between 60 and 100, refer to the eight abnormality criteria of the SPC standard to define SPC rules that can monitor the health level changes of the moving equipment in real time.

[0034] Preferably, the specific contents of the SPC rules are as follows: input the window data into the SPC extended rule model to calculate the rate of change of health; intercept a section of data from the very end of the data, namely the trend diagnosis area, to calculate the mean and standard deviation used in the trend diagnosis area, and verify whether this section of data meets the calculation conditions before calculation. When there is downtime or equipment replacement in the window data, the data stability zone must meet 35 minutes; the health score that does not meet 35 minutes will be discarded, this section of data will not be calculated, and the trend diagnosis area will not be able to determine whether a trend alarm will occur.

[0035] Preferably, the window data is the health of the mobile device for 35 minutes, which is divided into three sections, specifically including:

[0036] Data stability area: used to cache a period of unstable data during the device shutdown and restart process;

[0037] Data accumulation area: used to calculate the floating mean and standard deviation of health;

[0038] Trend diagnosis area: It is the current value diagnosed by threshold value.

[0039] The present invention takes into account the structural analysis of deep learning models, so a deep learning model is proposed. The model can quantify the operating status of the dynamic equipment, that is, estimate the health, and judge the changing trend of the health in combination with the SPC rule, and then give an early warning of the health status of the dynamic equipment. The advantages are as follows:

[0040] 1. Modern industrial production places higher demands on the real-time operation of equipment status. Deep learning models, combined with efficient computing platforms and optimization algorithms, can complete data analysis and status judgment in a very short time, and realize real-time or near real-time operation of dynamic equipment status, which helps to detect potential faults in a timely manner, avoid unplanned downtime, and ensure production safety and continuity;

[0041] 2. Deep learning gives the equipment status assessment system a high degree of intelligence. Through continuous learning and optimization, these systems can adapt to environmental changes, equipment aging and other factors to improve prediction accuracy and robustness. At the same time, deep learning can also be combined with other intelligent technologies to achieve more comprehensive equipment status assessment and fault diagnosis;

[0042] 3. The dynamic equipment status assessment system based on deep learning can predict equipment failures in advance and realize predictive maintenance. Compared with traditional planned maintenance and post-maintenance, predictive maintenance can significantly reduce the number of maintenance times, spare parts inventory and labor costs, while avoiding production interruptions and losses caused by sudden equipment failures. Therefore, the application of deep learning helps enterprises reduce overall maintenance costs and improve economic benefits.

[0043] 4. Through real-time operation (step 7) and accurate prediction (step 8), the deep learning model can promptly detect and solve potential equipment problems and effectively avoid equipment failures. This preventive maintenance strategy can significantly improve the reliability and stability of dynamic equipment, extend the service life of equipment, and reduce production risks and safety hazards caused by equipment failures. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a flow chart of the present invention.

[0045] Figure 2 It is a schematic diagram of the GRU memory unit structure of the present invention.

[0046] Figure 3 It is a schematic diagram of window data of the present invention. DETAILED DESCRIPTION

[0047] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0048] like Figure 1 The specific steps of the dynamic equipment failure early warning method based on the GRU algorithm and SPC are as follows:

[0049] Step 1: Collect the original data of the equipment: the equipment's operating data, the equipment's factory test data, all maintenance records within the equipment's operating data period, and the corresponding ranges of all measuring points of the equipment;

[0050] Dynamic equipment status assessment relies on a large amount of operating data, and the quality of the data directly affects the accuracy of the assessment results. However, in practical applications, due to factors such as the accuracy of data acquisition equipment, sampling frequency, and interference during data transmission, the collected data often has problems such as noise and missing values. These problems will seriously affect the training effect of the deep learning model and reduce the accuracy of the assessment results. It is necessary to analyze the operating mechanism of the dynamic equipment and the data missing situation, obtain the operating parameters that can reflect the performance of the equipment, and save the names of these parameter items.

[0051] Step 2: Extract 1 year (or more than 1 year) of historical operation data from the equipment at a time interval of 1 second, use the extracted data as a training set, process the training set, and save the standardized model of the training set;

[0052] The specific process is as follows: perform parameter screening, data cleaning, parameter standardization and other operations for the training set, and save the standardized model of the training set. In the parameter screening process, the selection of operating parameters is obtained from the parameter items saved in step 1 (the screening criteria are to analyze the operating mechanism of the dynamic equipment and the data missing situation, as well as the key measurement points of the equipment operation); in the data cleaning process, only the data within the normal operating period is retained, and the data within the fault period needs to be eliminated; the standardization process adopts the method Z-score, that is, the formula X = (x-μ) / σ is applied to make the variance of the new X data set 1 and the mean 0, so that the processed data will conform to the standard normal distribution.

[0053] Step 3: Build a GRU model, take each operating parameter value of the dynamic device in the training set as the target of the model, and take other operating parameter values ​​except the target parameter as the input of the model, perform supervised training, and save the training result model.

[0054] The target parameter refers to the key parameter that is used as the value to be evaluated when training the prediction model for the key parameter, and the remaining parameters are input as evaluation values, and the output result is the model trained for the key parameter.

[0055] The essential reason for the vanishing gradient problem of RNN is that when data propagates from the previous moment to the next moment, it is affected by the next moment, resulting in the loss of information from the previous moment. Therefore, the solution to the vanishing gradient problem is to avoid the influence of the input of the next moment on the output of the hidden layer of the previous moment. When the time step is large or the time step is small, the gradient of RNN is more likely to decay or explode. Although clipping the gradient can deal with gradient explosion, it cannot solve the problem of gradient decay. Therefore, in practice, it is difficult for RNN to capture dependencies with large time step distances in time series. The Gated Recurrent Neural Network (GRNN) was proposed to better capture dependencies with large time step distances in time series. It controls the flow of information through learnable gates. The structure of GRU is as follows: Figure 2 shown.

[0056] Among them, X t represents the input data at the current time t, h t-1 Represents the implicit state output at the historical time t-1. σ and tanh represent the sigmoid function and the hyperbolic tangent function respectively. Sigmoid and tanh are activation functions, and their mathematical formulas are as follows:

[0057]

[0058] In order to simplify the internal structure and improve the computing efficiency, the GRU network structure integrates the input gate, forget gate, and output gate structure in LSTM into an update gate and a reset gate. Using an update gate can realize the forgetting and selective memory of the neural network, which greatly reduces the parameters. The GRU principle is as follows:

[0059] 1. The update gate determines how much historical information and current information to use to update the current implicit state. The update gate at time t is:

[0060] z t =σ(W z ·[h t-1 ,x t ]) (Formula 3)

[0061] Among them, z t is the gate update signal, z t The size of determines the degree of memory of the candidate hidden state, h t-1 is the historical implicit state, x t represents the input data at time t, W z is the weight matrix, and σ is the sigmoid function.

[0062] 2. The reset gate determines how much historical information is retained. The reset gate at time t is:

[0063] r t =σ(W r ·[h t-1 , x t ]) (Formula 4)

[0064] Among them, r t is the reset signal. The larger the reset signal value is, the more historical information needs to be remembered. r is the weight matrix.

[0065] 3. In the update gate z t and reset gate r t Under the action of , the candidate hidden state at the current moment, the implicit output state h t Can be updated to:

[0066]

[0067] Among them, the candidate hidden state is:

[0068]

[0069] In the above formula, the candidate hidden state is responsible for fusing the information features of the input data and historical data. This operation is combined with the reset signal r obtained by the reset gate. t W corresponds to the training parameters, and represents the final unit state at the current moment, which includes two processes: forgetting and remembering. (1-z t ) and the implicit state h at the previous moment t-1 The product of represents the forgetting process, z t The closer it is to 1, the more information from the previous moment will be forgotten. t The product of the candidate hidden state represents the memory process, z t The size determines the degree of memory of the candidate hidden state, that is, how much of the previous hidden state is retained.

[0070] Step 4: Before applying the model, select the key measuring points of the dynamic equipment to be evaluated (select the measuring points with the greatest impact on the equipment as the key measuring points) and process the parameter values ​​of the measuring points;

[0071] The specific process is: obtain the operating data of the current dynamic equipment to be evaluated, filter the operating parameter items of the equipment, and standardize the filter result set. The processing basis of these two steps comes from the parameter items and standardized models saved in steps 1 and 2.

[0072] Step 5: Apply the GRU model to estimate the operating point parameter values ​​of the dynamic equipment to be evaluated.

[0073] The specific content is: according to step 4, apply the GRU training model to estimate the operating parameter values ​​of the current moving equipment.

[0074] Step 6: The calculated results of the parameters of each operating point of the moving equipment will deviate from the actual values. The deviation of each parameter reflects the degree to which the parameter deviates from the optimal value. In order to reflect the overall deviation of the moving equipment, the deviation of each parameter needs to be summarized. The formula is:

[0075]

[0076] In the formula, x i , They represent the actual value and estimated value of the i-th parameter among the m key parameters to be estimated, respectively, and w i represents the weight of the deviation of the i-th parameter in the overall deviation, e m Represents the overall deviation of m parameters to be estimated.

[0077] Step 7: The health of the equipment can be measured by the deviation (e m ) size. The greater the deviation, the worse the health. However, the deviation range is greater than 0, which is not convenient for qualitatively judging the health. Therefore, the deviation is mapped to a threshold range of 0-100 through mathematical methods, and this mapping value is called health. The mapping formula is:

[0078]

[0079] In the formula, h x It indicates the health value of the mapping when the overall deviation is x, and p indicates the deviation cutoff value. That is, when the overall deviation is greater than p, the mapped health is less than 60, otherwise it is greater than or equal to 60. When the health is less than 60 points, it indicates that the equipment is abnormal and needs maintenance.

[0080] Step 8: For the case where the health level is greater than or equal to 60 in step 7, monitor the fluctuation of the health level between 60 and 100, refer to the eight abnormality criteria of the SPC standard, and define the SPC rules that can monitor the health level changes of the moving equipment in real time. The specific contents are as follows:

[0081] Input the window data into the SPC extended rule model (this model converts the SPC criteria into a model for judging the trend of the health score) to calculate the rate of change of the health; intercept a period of data from the very end of the data (before the trend diagnosis area) to calculate the mean and standard deviation used in the trend diagnosis area, and verify whether this period of data meets the calculation conditions before calculation (the calculation conditions are to detect whether the window data meets the requirements of trend diagnosis): when there is downtime or equipment replacement in the window data, there is enough cache period to stabilize the data, that is, the data stability zone must meet 35 minutes at this time; if the 35-minute health score is not met, the window data will be discarded, this period of data will not be calculated, and the trend diagnosis area will not be able to determine whether a trend alarm will occur.

[0082] The window data is the health of the dynamic equipment every 35 minutes (equal intervals / minute), which is divided into three sections, as follows:

[0083] Data stability area: used to cache the period of unstable data during the device shutdown and restart process;

[0084] Data accumulation area: used to calculate the floating mean and standard deviation of health;

[0085] Trend diagnosis area: the current value diagnosed by threshold value;

[0086] SPC mainly refers to the application of statistical analysis technology to monitor the production process in real time, scientifically distinguish between random fluctuations and abnormal fluctuations in product quality during the production process, and thus issue early warnings for abnormal trends in the production process so that production managers can take timely measures to eliminate abnormalities and restore process stability, thereby achieving the purpose of improving and controlling quality. The eight criteria for judging abnormalities in the SPC standard are as follows:

[0087] Rule 1: 1 point falls outside three standard deviations from the center line;

[0088] Rule 2: 9 consecutive points fall on the same side of the center line;

[0089] Rule 3: 6 consecutive points increasing or decreasing;

[0090] Rule 4: Adjacent points in a row of 14 points alternate up and down;

[0091] Rule 5: 2 out of 3 consecutive points fall outside of 2 standard deviations of the same measure of the center line;

[0092] Rule 6: 4 out of 5 consecutive points fall outside one standard deviation on the same side of the center line;

[0093] Rule 7: 15 consecutive points fall within one standard deviation on either side of the center line;

[0094] Rule 8: 8 consecutive points fall on both sides of the center line and none of them is within one standard deviation;

[0095] Step 9: Apply the SPC rules defined in step 8 to monitor the health sequence value of the submersible pump calculated in step 7 in real time, determine whether the operation of the submersible pump is abnormal, and thus trigger the alarm mechanism.

[0096] The above implementation modes are not limitations of the present invention, and the present invention is not limited to the above examples. Changes, modifications, additions or substitutions made by technicians in this technical field within the scope of the technical solution of the present invention also belong to the protection scope of the present invention.

Claims

1. A dynamic equipment failure early warning method based on GRU algorithm and SPC, characterized by: The specific steps include: Step S1: Collect the original data of the moving equipment, obtain the monitoring parameters that can reflect the performance of the equipment and save them; Step S2: extracting data from the historical operation data of the equipment as a training set, performing data processing on the training set, and saving a standardized model of the training set; Step S3: construct a GRU model; Step S4: before applying the model, the key measuring points of the moving equipment to be evaluated are selected and the parameter values ​​of the measuring points are processed; Step S5: Apply the GRU model to estimate the operating point parameter values ​​of the dynamic equipment to be evaluated; Step S6: Calculate the overall deviation of the moving equipment and summarize the deviation of the moving equipment parameters; Step S7: By deviation e m Size to judge the health of the moving equipment; Step S8: Referring to the eight abnormality judgment criteria of the SPC standard, define SPC rules that can run the health changes of dynamic equipment in real time.

2. The method for early warning of dynamic equipment failure based on the GRU algorithm and SPC according to claim 1 is characterized in that: In the step S1, the collected data include: the operation data of the equipment, the factory test data of the equipment, all the maintenance records within the time period of the equipment operation data, and the ranges corresponding to all the measuring points of the equipment.

3. The method for early warning of dynamic equipment failure based on GRU algorithm and SPC according to claim 1 is characterized in that: In step S2, the intercepted data is historical operation data of 1 year or more and the time interval is 1 second; Data processing includes parameter screening, data cleaning, and parameter standardization; During the parameter screening process, the selection of operating parameters is obtained from the monitoring parameters saved in step S1; During the data cleaning process, only the data during the normal operation period of the equipment is retained, and the data during the failure period needs to be eliminated; Parameter standardization uses the Z-score method, that is, the formula X = (x-μ) / σ is applied to make the variance of the new X data set 1 and the mean 0, so that the processed data conforms to the standard normal distribution.

4. The method for early warning of dynamic equipment failure based on the GRU algorithm and SPC according to claim 1 is characterized in that: In step S3, each operating parameter value of the equipment in the training set is used as the target of the model, and other operating parameter values ​​except the target parameter are used as the input of the model to perform supervised training, and the training result model is saved.

5. The method for early warning of dynamic equipment failure based on GRU algorithm and SPC according to claim 1 is characterized in that: The specific process of step S4 is: obtaining the operating data of the current dynamic equipment to be evaluated, screening the operating parameter items of the equipment, and standardizing the screening result set, and the processing basis comes from the parameter list and standardized model saved in steps S1 and S2.

6. The method for early warning of dynamic equipment failure based on GRU algorithm and SPC according to claim 1 is characterized in that: In step S6, the deviation formula is: In the formula, x i , They represent the actual value and estimated value of the i-th parameter among the m key parameters to be estimated, respectively, and w i represents the weight of the deviation of the i-th parameter in the overall deviation, e m Represents the overall deviation of m parameters to be estimated.

7. The method for early warning of moving equipment failure based on GRU algorithm and SPC according to claim 1 is characterized in that: In step S7, the deviation is mapped to a threshold value within a range of 0-100 by mathematical methods, and this mapping value is called health. The mapping formula is: In the formula, h x It indicates the health value of the mapping when the overall deviation is x, and p indicates the deviation cutoff value. That is, when the overall deviation is greater than p, the mapped health is less than 60, otherwise it is greater than or equal to 60. When the health is less than 60 points, it indicates that the equipment is abnormal and needs maintenance.

8. The method for early warning of dynamic equipment failure based on GRU algorithm and SPC according to claim 1 is characterized in that: In step S8, for the case in step S7 where the health level is greater than or equal to 60, when monitoring the fluctuation of the health level score between 60 and 100, the SPC rules that can monitor the health level changes of the moving equipment in real time are defined with reference to the eight abnormality criteria of the SPC standard.

9. The method for early warning of moving equipment failure based on GRU algorithm and SPC according to claim 7 is characterized in that: The specific contents of the SPC rules are as follows: input the window data into the SPC extended rule model to calculate the rate of change of health; intercept data from the very end of the data, namely the trend diagnosis area, to calculate the mean and standard deviation used in the trend diagnosis area, and verify whether the data meets the calculation conditions before calculation. When there is downtime or equipment replacement in the window data, the data stability zone must meet 35 minutes. The health score that does not meet 35 minutes will be discarded, the intercepted data will not be calculated, and the trend diagnosis area will not be able to determine whether a trend alarm will occur.

10. The method for early warning of moving equipment failure based on GRU algorithm and SPC according to claim 9, characterized in that: The window data is the health of the mobile device for 35 minutes, which is divided into three sections, including: Data stability area: used to cache a period of unstable data during the device shutdown and restart process; Data accumulation area: used to calculate the floating mean and standard deviation of health; Trend diagnosis area: It is the current value diagnosed by threshold value.

Images