Adaptive design method and system for alarm delayer based on change detection

Abstract

The invention relates to the technical field of industrial alarm, and provides an alarm delayer adaptive design method and system based on change detection, and the method comprises the steps: obtaining a process variable, and generating an alarm duration time sequence and an alarm interval time sequence in combination with an alarm threshold value; searching a mean value change point of the alarm duration time sequence and the alarm interval time sequence by using a change detection method; on the basis of the mean value change point, utilizing piecewise linear representation and hypothesis testing methods to confirm whether the process variable is in a normal condition or an abnormal condition; and determining whether to update the delay factor based on the mean value change point and the state of the process variable, and determining the time of updating the delay factor by using Bayesian estimation after determining that the delay factor is updated. Therefore, the number of false alarms and missed alarms is controlled.

Description

Technical Field

[0001] The invention belongs to the technical field of industrial alarms, and in particular relates to a method and system for adaptively designing an alarm delayer based on change detection. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] The alarm system is a component of the automated monitoring system. When the process variable exceeds the normal allowable range, the alarm mechanism generates an alarm to remind the operator to deal with the alarm. However, most existing industrial alarm systems do not achieve good alarm performance and it is difficult to ensure system stability and data reliability. Therefore, designing an alarm system with good performance is of great significance in the field of industrial alarm.

[0004] Alarm systems are often affected by false alarms and missed alarms, which degrade the performance of the alarm system. False alarms are alarms that occur even when the process variable is in normal operating conditions, and missed alarms are alarms that do not occur when the process variable is in abnormal conditions. The presence of too many false alarms and missed alarms is a major cause of alarm overload. These noise alarms drown out the real alarms, making it difficult for operators to extract valid information from the large amount of noise information. This will not ensure the safe and efficient operation of industrial plants and may cause accidents. Therefore, in industrial production processes, reducing the occurrence of false alarms and missed alarms is crucial to improving the performance of alarm systems.

[0005] Alarm delayers are widely used in the field of industrial alarms to reduce the occurrence of false alarms and missed alarms in alarm systems. The false alarm rate (FAR) and missed alarm rate (MAR) are reduced by designing the parameter delay factor of the alarm delayer. However, most of the existing alarm delayer design methods assume that the process variables are stationary and design fixed parameters for the alarm delayer based on the historical data of the process variables. For non-stationary process variables, these alarm delayer design methods with fixed parameters cannot maintain the desired false alarm rate and missed alarm rate. Therefore, it is necessary to design an alarm design method that can update the alarm delayer parameters online, so as to more effectively improve the performance of the alarm system. Summary of the invention

[0006] In order to solve the technical problems existing in the above-mentioned background technology, the present invention provides an alarm delayer adaptive design method and system based on change detection, which determines the state of the process variable after the change based on the piecewise linear representation and mean hypothesis test method, and uses the confidence interval width designed by Bayesian estimation technology to determine the update point to update the delay factor. It can update the parameter delay factor of the alarm delayer when the statistical characteristics of the process variable change, so as to control the number of false alarms and missed alarms.

[0007] In order to achieve the above object, the present invention adopts the following technical solution:

[0008] A first aspect of the present invention provides an adaptive design method for an alarm delayer based on change detection, comprising:

[0009] Obtain process variables and combine them with alarm thresholds to generate alarm duration series and alarm interval time series;

[0010] Using the change detection method, the mean change points of the alarm duration series and the alarm interval time series are searched;

[0011] Based on the mean change point, piecewise linear representation and hypothesis testing methods are used to confirm whether the process variable is in a normal or abnormal state;

[0012] Based on the mean change point and the state of the process variable, it is determined whether to update the delay factor, and after confirming the update of the delay factor, the time to update the delay factor is determined by using Bayesian estimation.

[0013] Further, the process variable is compared with an alarm threshold to generate an alarm signal;

[0014] The alarm duration is the duration from the occurrence of the alarm to the clearing of the alarm in the alarm signal;

[0015] The alarm interval time is the interval time from the clearing of an alarm to the occurrence of the next alarm in the alarm signal.

[0016] Further, the steps of the change detection method include:

[0017] For the alarm duration series or the alarm interval time series, calculate the statistics;

[0018] By maximizing the statistic, detect the mean change point and calculate the P value corresponding to the mean change point;

[0019] For each mean change point, the P value is compared with the selected value to determine whether to retain the mean change point.

[0020] Furthermore, the piecewise linear representation is used to segment the process variable into several short time series.

[0021] Furthermore, the hypothesis testing method is used to determine whether the short time series is in a normal state or an abnormal state.

[0022] Further, the updating of the delay factor includes three situations: updating the delay factor based on the alarm duration, updating the delay factor based on the alarm interval time, and no need to update the delay factor.

[0023] Furthermore, the Bayesian estimation determines the reliability of the delay factor by estimating the confidence interval of the false alarm rate or the missed alarm rate, and updates the delay factor when the delay factor is reliable.

[0024] A second aspect of the present invention provides an alarm delay adaptive design system based on change detection, comprising:

[0025] A data acquisition module is configured to: acquire process variables and generate an alarm duration sequence and an alarm interval time sequence in combination with an alarm threshold;

[0026] A change detection module is configured to: search for mean change points of the alarm duration series and the alarm interval time series using a change detection method;

[0027] The abnormality judgment module is configured to: confirm whether the process variable is in a normal state or an abnormal state based on the mean change point by using a piecewise linear representation and a hypothesis testing method;

[0028] The delay factor updating module is configured to: determine whether to update the delay factor based on the mean change point and the state of the process variable, and after confirming to update the delay factor, determine the time to update the delay factor by using Bayesian estimation.

[0029] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the above-mentioned method for adaptively designing an alarm delayer based on change detection.

[0030] The fourth aspect of the present invention provides a computer device, including a computer-readable storage medium, a processor, and a computer program stored on the computer-readable storage medium and executable on the processor, wherein when the processor executes the program, the steps in the adaptive design method of an alarm delayer based on change detection as described above are implemented.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] The present invention determines the state of the process variable after the change based on the piecewise linear representation and mean hypothesis test method, and uses the confidence interval width designed by the Bayesian estimation technology to determine the update point to update the delay factor. It can update the parameter delay factor of the alarm delayer when the statistical characteristics of the process variable change, so as to control the number of false alarms and missed alarms.

[0033] The present invention takes into account the design of the delay timer of the process variable under normal and abnormal conditions and the two performance indicators of false alarm and missed alarm. By estimating the confidence interval of the false alarm rate or the missed alarm rate, the delay factor is updated to maintain the expected performance of false alarm and missed alarm. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0035] Figure 1 is a flow chart of an adaptive design method of an alarm delayer based on change detection according to a first embodiment of the present invention;

[0036] Figure 2 It is a piecewise schematic diagram of the piecewise linear representation technology of the first embodiment of the present invention;

[0037] Figure 3 is a diagram of the update and application of alarm delay parameters of the first embodiment of the present invention;

[0038] Figure 4 This is a time series diagram of the alarm duration and alarm interval collected in the first embodiment of the present invention;

[0039] Figure 5 A two-dimensional schematic diagram of delay factor design according to Embodiment 1 of the present invention;

[0040] Figure 6 A three-dimensional schematic diagram of a delay factor design according to Embodiment 1 of the present invention;

[0041] Figure 7 It is a structural diagram of a computer device according to a fourth embodiment of the present invention. DETAILED DESCRIPTION

[0042] To make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0043] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0044] Embodiment 1

[0045] To address the problem that the alarm delayer with fixed parameters described in the background art cannot maintain the desired FAR and MAR, this embodiment provides an alarm delayer adaptive design method based on change detection.

[0046] The present embodiment provides an adaptive design method for an alarm delayer based on change detection, which can update the parameter delay factor of the alarm delayer when the statistical characteristics of the process variable change, thereby controlling the number of false alarms and missed alarms.

[0047] The present embodiment provides an alarm delayer adaptive design method based on change detection, which considers whether the alarm delayer parameters need to be updated when the process variable is in a non-stationary state, how to determine the update time after determining that the alarm delayer parameters need to be updated, and whether the designed alarm delayer parameters are reliable. Through this scheme, we can update the parameter delay factor of the alarm delayer when the process variable is a non-stationary signal, thereby maintaining the performance expected by FAR and MAR.

[0048] This adaptive method can maintain the desired performance by continuously updating the parameter delay factor of the delay timer when the statistical characteristics of the alarm duration or alarm interval corresponding to the alarm variable change. This embodiment considers the design of the delay timer of the process variable under normal and abnormal conditions and the two performance indicators of false alarm and missed alarm. Therefore, this embodiment mainly involves three issues:

[0049] First, how to detect the characteristic change when the distribution of alarm duration or alarm interval time changes; this embodiment uses a change detection algorithm to detect its obvious change points to solve this problem;

[0050] Second, after detecting the change in the mean value of the alarm duration or the alarm interval, how to determine whether the corresponding process variable is in a normal state or an abnormal state; this embodiment determines the state of the process variable after the change based on the piecewise linear representation and the mean hypothesis test method; then consider updating the delay factor to maintain the performance of the alarm;

[0051] Third, how to determine the update point of the delay factor m after detecting a significant mean change; since the number of samples of the new distribution of the process variable may not be sufficient to design a reliable delay factor, this embodiment uses the confidence interval width designed using the Bayesian estimation technology to determine the update point to update the delay factor.

[0052] This embodiment provides an adaptive design method for an alarm delayer based on change detection. The alarm delayer is designed through mean change and Bayesian estimation, and the alarm delayer parameters can be updated in an online manner. The alarm delayer designed by measuring the design uncertainty of FAR and MAR can maintain the expected performance of false alarms and missed alarms.

[0053] The present embodiment provides an adaptive design method for an alarm delayer based on change detection. First, a change detection method is used to detect whether a mean change has occurred in the alarm duration and the alarm interval. A piecewise linear representation and hypothesis test are used to determine whether the corresponding process variable is in a normal state or an abnormal state after the mean change, thereby determining whether the delay factor needs to be updated. Secondly, a delay factor is designed through a Bayesian estimation technique to ensure that the designed delay factor can reliably achieve the expected values ​​of FAR and MAR. Finally, the delay factor is updated when a reliable delay factor is designed.

[0054] The present embodiment provides an alarm delayer adaptive design method based on change detection, comprising the following steps:

[0055] Step 1: Collect the alarm duration and alarm interval time available at the current moment to form a sample data set and

[0056] Step 2: Detect alarm duration using change detection method and alarm interval The mean change point k d,cp and k i,cp , in order to detect whether their data sets have undergone significant mean changes.

[0057] Step 3: Based on the mean changes of alarm duration and alarm interval, the mean change points are identified by using piecewise linear representation and mean hypothesis test method ( or ) corresponding to the process variable historical data In normal or abnormal condition.

[0058] Step 4: Determine whether the delay factor m needs to be updated based on the mean change of the alarm duration and the alarm interval and the state of the corresponding process variable after the change. s ; if m s Need to be updated, perform the following steps.

[0059] Step 5: Design the alarm delay parameter m using Bayesian estimation techniques s , and determine its reliability; if m s is unreliable and requires Td or T i More data until m is obtained s Become reliable.

[0060] Step 6: If the m obtained is s is reliable, then the alarm delay parameter is given by m s-1 Update to m s .

[0061] Step 7: Collect T d (T i ) and repeat the above steps.

[0062] This embodiment provides an adaptive design method for an alarm delayer based on change detection, such as Figure 1 As shown, the specific steps include:

[0063] Step 1: Collect historical data of process variables at the current moment and with the high alarm threshold x th Generate corresponding alarm variable sample data (alarm signal) To obtain the alarm duration sequence available up to the current moment and alarm interval time series

[0064] The alarm duration is the duration from the occurrence of an alarm to the clearing of the alarm in each alarm event. o and t c are the sampling indices of alarm occurrence and alarm clearing respectively, and the duration of the kth alarm is the sampling index length from its alarm occurrence to its alarm clearing:

[0065] T d (k)=(t c (k)-t o (k)+1) (1)

[0066] Among them, the alarm occurrence is the alarm variable x a (t) Event of switching from non-alarm state "0" to alarm state "1":

[0067]

[0068] Correspondingly, the alarm clear is x a (t) An event that switches from “1” to “0”.

[0069] The alarm interval is the interval between the clearing of an alarm and the occurrence of the next alarm. The kth alarm interval is the length of the sampling index between the clearing of the kth alarm and the occurrence of the next alarm:

[0070] Ti (k)=(t o (k+1)-t c (k)+1) (3)

[0071] According to formula (1) and formula (3), the data set of alarm duration and alarm interval can be obtained: and

[0072] Step 2: Use change detection methods to search for alarm duration sequences in the dataset and alarm interval time series Find the mean change point that meets the conditions in the alarm duration and alarm interval time series, indicating that the sequence has undergone a significant mean change.

[0073] Step 201, The statistic U k,K It can be calculated as:

[0074]

[0075] in,

[0076]

[0077] Where sgn(·) indicates that the operand takes the value "-1", "0" or "1" when it is a negative number, zero or a positive number.

[0078] Step 202, by maximizing |U k,K |, detect mean change point k d,cp :

[0079]

[0080] Step 203, the corresponding P value is calculated as:

[0081]

[0082] Step 204, similarly, for Perform steps 201-203 to obtain the mean change point k i,cp and the corresponding P-values.

[0083] Step 205, select α as the probability of type I error, for example, α = 0.01; if the P value is less than the selected value α, it means or There is a significant mean change, k d,cp or k i,cp If it is an obvious mean change point, then the mean change point k is retained. d,cp or k i,cp Therefore, considering the or Update the delay factor m, and the corresponding collected data samples of x are or Among them, t d,cp ,t i,cp It is k d,cp , k i,cp The starting time when the corresponding alarm duration and alarm interval in the historical data of process variable x change in the mean.

[0084] Step 3: Due to the non-stationary nature of the process variable x, it is necessary to determine the data segment or In normal or abnormal state. Based on the alarm duration or alarm interval time series after the mean value changes, the piecewise linear representation and hypothesis testing method are used to confirm whether the corresponding process variable historical data is in normal or abnormal state.

[0085] Among them, the piecewise linear representation is used to segment the historical data of the process variables to detect whether its distribution trend has changed, and the hypothesis test is used to determine whether the segmented historical data of the process variables is in a normal state or an abnormal state.

[0086] For non-stationary process variables, the statistical characteristics of the process variables may be changing. The long data segments of the process variables can be divided into short data segments with the same adjacent sample trends through the piecewise linear representation method, and then the mean of the data segment samples is calculated. By using hypothesis testing, it can be determined whether these data segments are in a normal state or an abnormal state.

[0087] Step 301: Use piecewise linear representation technology to represent long time series (t cp =min{t d,cp ,t i,cp},N=max{N d ,N i}) is divided into short time series Each short time series It is formed by clustering and fitting adjacent data samples with the same trend.

[0088] Example: Generate a process variable:

[0089]

[0090] Where e(t) is a function with a mean of 0 and a variance of σ 2 is a Gaussian white noise of 0.25.

[0091] Applying piecewise linear representation technology, we get the following results: Figure 2 As shown, the piecewise linear representation converts the long time series of process variables into Short time series and

[0092] Step 302: Determine the short time series using hypothesis testing method In normal or abnormal condition:

[0093] (1) Calculation The mean of is:

[0094]

[0095] (2) According to the central limit theorem, μ r is Gaussian distributed, so μ r The 95% confidence interval for

[0096] [μ r,- ,μ r,+ ]=[μ r -1.96σ r ,μ r +1.96σ r ] (9)

[0097] in,

[0098]

[0099] (3) Considering the mean change of process variables, the hypothesis testing method is adopted:

[0100] H0:μ r =x th ,H1:μ r ≠x th (11)

[0101] (4) If the high alarm threshold x th Greater than (less than) μ r,+ (μ r，- ), the null hypothesis H0 is rejected, indicating that the sample mean μ r Statistically smaller (greater) than x th , then the data segment In normal (abnormal) condition; otherwise, H1 is rejected. Therefore, the indicator variable I(r) is used to represent the values ​​"1", "0" and "-1" of the normal, unclassified and abnormal state data segments:

[0102]

[0103] Step 4: According to the alarm duration T d Or alarm interval time T iThe mean change of and the normal or abnormal state of the process variable historical data x determine whether to update the delay factor, forming three situations of whether to update the alarm delay parameter m, as shown in Table 1, including:

[0104] (1) In T d The change point k is detected in d,cp , and the corresponding data segment In normal state or from abnormal to normal state, based on Update m;

[0105] (2) In T i The change point k is detected in i,cp , and the corresponding data segment In an abnormal state or from normal to abnormal state, based on Update m;

[0106] (3) In T d The change point k is detected in d,cp And the corresponding data segment In abnormal state or from normal to abnormal state, or in T i The change point k is detected in i,cp And the corresponding data segment In the normal state or from abnormal to normal state, there is no need to update m.

[0107] In summary, based on the mean changes of the alarm duration and the alarm interval and the corresponding state of the process variable after the change, it can be sorted into three situations: updating the delay factor based on the alarm duration, updating the delay factor based on the alarm interval, and no need to update the delay factor. If the alarm duration (alarm interval) changes in mean and the corresponding process variable historical data is in a normal state (abnormal state) after the change, the delay factor is updated based on the alarm duration (alarm interval).

[0108] Table 1. Update of alarm delay parameters

[0109] Normal state Abnormal status From normal to abnormal state From abnormal to normal state <![CDATA[T d Changes]]> <![CDATA[Based on T d Update m]]> No need to update m No need to update m <![CDATA[Based on T d Update m]]> <![CDATA[T i Changes]]> No need to update m <![CDATA[Based on T i Update m]]> <![CDATA[Based on T i Update m]]> No need to update m <![CDATA[T d &T i Changes]]> <![CDATA[Based on T d Update m]]> <![CDATA[Based on T i Update m]]> <![CDATA[Based on T i Update m]]> <![CDATA[Based on T d Update m]]>

[0110] After using the alarm delay parameter delay factor m, the false alarms and missed alarms less than m in the alarm duration series and the alarm interval time series will be removed. Therefore, the relationship between the alarm duration and the alarm interval time and the delay factor shows that the optimal value of the alarm delay parameter can be determined by the statistical characteristics of the alarm duration and interval time.

[0111] If determined based on Design delay factor m, FAR is equal to where T d Cumulative probability greater than (or equal to) m:

[0112]

[0113] The optimal value of the delay factor m is calculated by making F(m) closest to the expected value F0:

[0114]

[0115] Wherein, F0 is the expected value of F(m), for example, F0=0.05.

[0116] If determined based on The delay factor m can be designed similarly based on Get the best value of m:

[0117]

[0118] Where M(m) is MAR, which means The cumulative probability of being greater than (or equal to) m; M0 = 0.05 is the expected value of MAR.

[0119] In order to record the change of m, take m0 as the initial value and change m s =[m0,m1,m2,...,m S ] as the update sequence of m.

[0120] Step 5: If the delay factor needs to be updated, use Bayesian estimation technology to design a reliable alarm delay parameter m, and determine when to update m to maintain the expected value F0 or M0 of FAR or MAR. That is, use Bayesian estimation technology to design the delay factor and determine its reliability. If the designed delay factor is unreliable, further alarm duration and alarm interval time data need to be obtained until a reliable delay factor is obtained.

[0121] After determining that the delay factor needs to be updated, the time to update the delay factor is determined using Bayesian estimation. The reliability of the designed delay factor is determined by estimating the confidence intervals of FAR and MAR, and the design uncertainty of FAR and MAR is measured.

[0122] Among them, the delay factor is designed and its reliability is determined based on the Bayesian estimate and its confidence interval. It is considered reliable when the upper limit of the confidence interval is less than the set value.

[0123] Here we take FAR as an example, and MAR can be obtained similarly.

[0124] In the Bayesian estimation method, FAR is considered as a random variable with a value of [0,1]. Considered as Without any additional information, it is reasonable to assume that The prior PDF of is a continuous uniform distribution:

[0125]

[0126] exist Greater than (or equal to) m s The number of samples is given by Indicates; because each alarm duration is independent, Following a binomial distribution, for , so given get The conditional probability is:

[0127]

[0128] The Bayesian estimate of is obtained through its posterior probability:

[0129]

[0130] in, is the estimated m s FAR; The (1-α) confidence interval is the narrowest interval Satisfies the equation:

[0131]

[0132] Here, α is a certain confidence level, for example, α=0.05.

[0133] Let F u is the upper bound of F0, indicating the worst allowed case of FAR. Then, if Upper limit of Less than (or equal to) F u , then it means m s The design value is reliable:

[0134]

[0135] Among them, F u Defined by the user, for example, F u =0.1.

[0136] Step 6: When designing the m s When equation (20) is satisfied, the alarm delay parameter is m s-1 Update to m s That is, a reliable delay factor is obtained, and the parameters of the alarm delayer are updated to the delay factor.

[0137] Update the delay factor m at the moment when the design obtains a reliable delay factor s, that is, the latest collection time during the alarm duration or alarm interval.

[0138] Step 7: Collect new data of alarm duration and alarm interval, and repeat the above steps. The online design of alarm delayer is realized by collecting new data.

[0139] The following is an application of the alarm delay adaptive design method provided in this embodiment in a specific example.

[0140] Taking the industrial data of a thermal power plant as an example, the process variable x is the pressure difference between the upper and middle layers of the catalyst in the reaction zone of furnace #4, and is configured with a high alarm threshold x th =150Pa. Figure 3 (a) shows a 2-hour real-time data sample of x(t) with a sampling period of 1 second. The initial value of the alarm delay parameter is m0=3.

[0141] Step 1: Based on the real-time data x(t) of the process variable and the alarm threshold x th Get the alarm duration T d And the alarm interval time T i Data sets such as Figure 4 As shown in (a) and (c) in .

[0142] Step 2: Use the change detection method to detect T d and T i When k = 231 (line B) is collected, T is detected at t = 3043 and t = 3016. d The mean change point k d,cp =217 and T i The mean change point and k i,cp =209 (line A), and the corresponding P values ​​are 4.9111×10 -37 and 2.9744×10 -20 All less than 0.01, T d and T i The corresponding statistic U(k) is Figure 4 As shown in (b) and (d) in .

[0143] Step 3: Dataset after mean change and The corresponding data segment is Using piecewise linear representation Divided into two data segments and According to the mean hypothesis test, the indicator variable value I(r) = {-1,-1} is obtained, so the data segment In abnormal condition.

[0144] Step 4:i If a change point is detected in the data segment and the corresponding data segment is in an abnormal state after the change, it is necessary to Update delay factor m.

[0145] Step 5, using Bayesian estimation techniques, based on The alarm delay parameter m1 is designed to be 4, and the estimated MAR value is 0.0652, with a confidence interval of [0, 0.121]. The upper limit of MAR is 0.121, which is significantly higher than F u = 0.1, so m1 = 4 is unreliable. Further collect T i Samples to (Line C), the corresponding data segment In an abnormal state. The design reliable parameter m1 = 6, the estimated MAR value is 0.0521, and the confidence interval is [0, 0.094]. Figure 5 and Figure 6 shown.

[0146] Step 6, at time t = 3297, the alarm delay parameter is updated from m0 = 3 to m1 = 6, such as Figure 3 The dashed line in (c) is shown in FIG. Calculate the data segment using the delay factor m1=6 The MAR is 0.0435, which is close to F0 = 0.05. Apply the updated delay factor and the fixed delay factor to the process variable data to generate alarm variables respectively. and like Figure 3 As shown in (d) and (e) in the figure, the FAR and MAR are shown in Table 2. It can be seen that the use of updated delay factors after the statistical characteristics of the process variables change can more effectively reduce the number of missed alarms.

[0147] Table 2. FAR and MAR results

[0148]

[0149]

[0150] The present embodiment provides an adaptive design method for an alarm delayer based on change detection. First, it is determined whether the parameter delay factor of the alarm delayer needs to be updated. According to the mean change of the alarm duration and the alarm interval time and the normal and abnormal states of the corresponding process variables, three situations of whether to update the delay factor are formed; secondly, the time to update the alarm delayer parameters is determined based on the confidence intervals of FAR and MAR, and the design uncertainty of FAR and MAR is measured using Bayesian estimation technology; finally, the alarm delayer parameters are updated and applied to the process variable data to maintain the desired number of false alarms and missed alarms.

[0151] Embodiment 2

[0152] This embodiment provides an alarm delay adaptive design system based on change detection, which specifically includes:

[0153] A data acquisition module is configured to: acquire process variables and generate an alarm duration sequence and an alarm interval time sequence in combination with an alarm threshold;

[0154] A change detection module is configured to: search for mean change points of the alarm duration series and the alarm interval time series using a change detection method;

[0155] The abnormality judgment module is configured to: confirm whether the process variable is in a normal state or an abnormal state based on the mean change point by using a piecewise linear representation and a hypothesis testing method;

[0156] The delay factor updating module is configured to determine whether to update the delay factor based on the mean change point and the state of the process variable, and after confirming to update the delay factor, determine the time to update the delay factor by using Bayesian estimation.

[0157] It should be noted here that each module in this embodiment corresponds to each step in Example 1 one by one, and the specific implementation process is the same, which will not be repeated here.

[0158] Embodiment 3

[0159] This embodiment provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps in the adaptive design method of an alarm delayer based on change detection as described in the first embodiment above are implemented.

[0160] Embodiment 4

[0161] This embodiment provides a computer device, such as Figure 7 As shown, it includes a display device, an input device, a computer-readable storage medium (volatile memory and non-volatile storage medium), a processor, a communication interface (i.e., a network interface), and a computer program stored on the computer-readable storage medium and executable on the processor, wherein the processor, the communication interface, and the computer-readable storage medium can be connected via a bus or other means. The communication interface is used to receive and send data, and when the processor executes the program, the steps in the adaptive design method of an alarm delayer based on change detection as described in the first embodiment above are implemented.

[0162] Among them, any reference to memory, storage, database or other media provided by the present application and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0163] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0164] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0165] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0166] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for adaptively designing an alarm delayer based on change detection, characterized in that: include: Obtain process variables and combine them with alarm thresholds to generate alarm duration series and alarm interval time series; Using the change detection method, the mean change points of the alarm duration series and the alarm interval time series are searched; Based on the mean change point, piecewise linear representation and hypothesis testing methods are used to confirm whether the process variable is in a normal or abnormal state; Based on the mean change point and the state of the process variable, it is determined whether to update the delay factor, and after confirming the update of the delay factor, the time to update the delay factor is determined by using Bayesian estimation.

2. The adaptive design method of alarm delayer based on change detection as claimed in claim 1, characterized in that: The process variable is compared with an alarm threshold to generate an alarm signal; The alarm duration is the duration from the occurrence of the alarm to the clearing of the alarm in the alarm signal; The alarm interval time is the interval time from the clearing of an alarm to the occurrence of the next alarm in the alarm signal.

3. The adaptive design method of alarm delayer based on change detection as claimed in claim 1, characterized in that: The steps of the change detection method include: For the alarm duration series or the alarm interval time series, calculate the statistics; By maximizing the statistic, detect the mean change point and calculate the P value corresponding to the mean change point; For each mean change point, the P value is compared with the selected value to determine whether to retain the mean change point.

4. The adaptive design method of alarm delayer based on change detection as claimed in claim 1, characterized in that: The piecewise linear representation is used to segment the process variable into several short time series.

5. The adaptive design method of alarm delayer based on change detection as claimed in claim 4, characterized in that: The hypothesis testing method is used to determine whether the short time series is in a normal state or an abnormal state.

6. The adaptive design method of alarm delayer based on change detection as claimed in claim 1, characterized in that: The updating of the delay factor includes three cases: updating the delay factor based on the alarm duration, updating the delay factor based on the alarm interval time, and no need to update the delay factor.

7. The adaptive design method of alarm delayer based on change detection as claimed in claim 1, characterized in that: The Bayesian estimation determines the reliability of the delay factor by estimating the confidence interval of the false alarm rate or the missed alarm rate, and updates the delay factor when the delay factor is reliable.

8. An adaptive design system for alarm delay based on change detection, characterized in that: include: A data acquisition module is configured to: acquire process variables and generate an alarm duration sequence and an alarm interval time sequence in combination with an alarm threshold; A change detection module is configured to: search for mean change points of the alarm duration series and the alarm interval time series using a change detection method; The abnormality judgment module is configured to: confirm whether the process variable is in a normal state or an abnormal state based on the mean change point by using a piecewise linear representation and a hypothesis testing method; The delay factor updating module is configured to: determine whether to update the delay factor based on the mean change point and the state of the process variable, and after confirming to update the delay factor, determine the time to update the delay factor by using Bayesian estimation.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps in the adaptive design method of an alarm delayer based on change detection as described in any one of claims 1 to 7 are implemented.

10. A computer device comprising a computer-readable storage medium, a processor, and a computer program stored in the computer-readable storage medium and executable on the processor, characterized in that: When the processor executes the program, the steps in the adaptive design method of an alarm delayer based on change detection as described in any one of claims 1-7 are implemented.

Images