A multi-theme intelligent monitoring method for time series data in an industrial complex environment

Abstract

The application discloses a kind of time series data multi-theme intelligent monitoring methods for industrial complex environment, it is related to power plant monitoring alarm technical field, including constructing unit monitoring item precondition;Through unit monitoring item theme definition alarm state details;Through comprehensive alarm model, data source is recalculated.The time series data multi-theme intelligent monitoring method for industrial complex environment provided in the application reduces the false alarm rate and workload of monitoring system by constructing unit monitoring item precondition, improves the stability and reliability of system, realizes the fine management of equipment operating state by constructing unit monitoring item theme, accurately monitors equipment operating state, discovers potential problems in real time, improves the safety and efficiency of power plant operation, recalculates data source by comprehensive alarm model, improves the flexibility and adaptability of alarm systemThe application achieves better effect in alarm accuracy, adaptability and efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power plant monitoring and alarm, in particular to a time series data multi-theme intelligent monitoring method for industrial complex environment. BACKGROUND

[0002] In recent years, with the continuous improvement of industrial automation and intelligence level, the role of power plant monitoring system in power production and management has become increasingly prominent. Traditional power plant monitoring systems mostly use relational database-based design, which collects and stores a large amount of sensor data to monitor the equipment running state in real time. However, with the explosive growth of data volume and the diversification of monitoring needs, the shortcomings of traditional relational databases in handling high-frequency data, real-time performance and scalability have gradually emerged. In order to solve these problems, time series databases have emerged. Time series databases are optimized for time series data, can efficiently process and store large-scale time series data, provide fast read-write performance and efficient query capability, and become an important technical foundation for modern power plant monitoring systems.

[0003] Currently, the application of time series databases in power plant monitoring mainly focuses on real-time data collection, storage and query. By using time series databases, power plants can realize real-time monitoring of equipment running state and efficient query of historical data, thereby improving the safety and stability of power plant operation. However, with the continuous improvement of power plant monitoring needs, relying only on real-time data collection and storage cannot meet the complex and variable monitoring needs. How to build a more intelligent and comprehensive monitoring and alarm system based on time series databases has become one of the current research hotspots.

[0004] Although the application of time series databases in power plant monitoring systems has made significant progress, the existing monitoring and alarm methods still have many shortcomings. First, traditional monitoring and alarm methods usually use simple threshold setting and logical judgment, which is difficult to adapt to complex and variable equipment running environment. This method not only easily produces false alarms and missed alarms, but also is difficult to fine-tune the management of equipment running state. Second, the existing monitoring and alarm systems lack effective methods in handling multi-theme monitoring items and comprehensive alarm models, and cannot realize comprehensive analysis and processing of multi-dimensional data. Especially when facing data source recalculation and complex alarm conditions, the existing systems often appear to be inadequate, and it is difficult to provide accurate and reliable alarm information. SUMMARY

[0005] In view of the above problems, the present application is proposed.

[0006] Therefore, the technical problem solved by the present application is that the existing monitoring and alarm method has low accuracy, is prone to false alarms and missed alarms, is difficult to adapt to complex and variable equipment running environment, and is difficult to fine-tune the management of equipment running state.

[0007] To address the aforementioned technical problems, this invention provides the following technical solution: a multi-topic intelligent monitoring method for time-series data in complex industrial environments, comprising: constructing preconditions for unit monitoring items; defining alarm status details through unit monitoring item themes; and recalculating the data source based on a comprehensive alarm model.

[0008] As a preferred embodiment of the time-series data multi-topic intelligent monitoring method for complex industrial environments described in this invention, the construction of the preconditions for unit monitoring items includes collecting precondition data for unit monitoring items, including equipment status, environmental parameters, maintenance status, historical alarm status, and equipment start / stop status.

[0009] For equipment status-related prerequisites, alarm monitoring will not be activated if the boiler temperature is not between 150℃ and 250℃; alarm monitoring will not be activated if the generator speed is below 1800 rpm.

[0010] For environmental parameter-related prerequisites, if the ambient temperature is not between -10℃ and 40℃, alarm monitoring will not be activated; if the ambient humidity is not between 20% and 80%, alarm monitoring will not be activated.

[0011] For maintenance status prerequisites, alarm monitoring will not be activated if the equipment is in maintenance status.

[0012] For historical alarm status-related preconditions, if there are unprocessed historical alarms, new alarm monitoring will not be initiated.

[0013] For equipment start / stop status prerequisites, alarm monitoring will not be activated if the equipment is not started or has been running for less than 30 minutes.

[0014] As a preferred embodiment of the multi-topic intelligent monitoring method for time-series data in complex industrial environments described in this invention, the preconditions for constructing unit monitoring items include defining preconditions using Boolean logic, with each precondition represented by a Boolean matrix P. i It means that P i This indicates the alarm status of the i-th device, where 1 represents an alarm and 0 represents normal. The overall preconditions are expressed as follows:

[0015]

[0016] Where ⊙ represents element-wise multiplication, and n represents the number of prerequisite conditions;

[0017] If P total If all elements in the table are 1, the precondition is met and the monitoring and alarm work continues; otherwise, it does not.

[0018] As a preferred embodiment of the multi-topic intelligent monitoring method for time-series data in complex industrial environments described in this invention, the step of defining alarm status details through unit monitoring item topics includes constructing unit monitoring item topics, which provide monitoring alarm rule details for defining alarm status details, including equipment over-limit topics, assessment condition topics, input / output status, equipment start / stop topics, interface monitoring topics, feature value topics, packaging point topics, and wall temperature monitoring topics. The data source is bound to the monitoring item topic rules to generate alarm information.

[0019] A nonlinear transformation is performed on the data source vector D, where the data source vector is D = [D1, D2, ..., D]. m ], can undergo nonlinear transformation, represented as

[0020] σ(D)=[σ(D1),σ(D2),…,σ(D m )]

[0021]

[0022] Where D represents the data source vector, which comes from various sensors or monitoring devices, and the vector dimension is m; σ(D) represents the nonlinear transformation of the data source vector. i ) represents the transformation result of the i-th data source value;

[0023] Using multidimensional tensors and rule tensor Applying rules and performing tensor product operations on the data source can be represented as follows:

[0024]

[0025] in, Let M be a multidimensional tensor, representing the rule matrix of the i-th monitored item topic, with dimensions (p×q×r). i,juv Let represent the element of the i-th tensor at position (j, u, v). Let R be the rule tensor, which is the matrix of the i-th rule. i,uvk Let A1 represent the element at position (u,v,k) of the i-th rule tensor, and let A1 represent the intermediate alarm state vector obtained through tensor product operation and rule application. 1,jk This represents the element at position (j,k) after the tensor product operation and rule application.

[0026] Pooling operations on the data source are represented as:

[0027]

[0028] Where, pool(D) ijThis represents the max pooling result at position (i,j), where (u,v)∈N(i,j)'

[0029] Combining convolution tensors and weighting coefficient α j The convolution operation is represented as:

[0030]

[0031] Here, A2 represents the intermediate alarm state vector obtained through pooling and convolution operations.

[0032] As a preferred embodiment of the multi-topic intelligent monitoring method for time-series data in complex industrial environments described in this invention, the step of defining alarm status details through unit monitoring item topics includes performing a Fourier transform on the data source and outputting frequency domain information, represented as:

[0033]

[0034] in, This represents the Fourier transform of the data source vector, outputting frequency domain information. D(t) represents the value of the data source signal at time t, and f represents a specific frequency component in the Fourier transform result.

[0035] Combined with weighting coefficient β k The output is represented as:

[0036]

[0037] Where A3 represents the weight coefficient, which represents the weight of the k-th Fourier transform;

[0038] Applying a Laplace transform to the data source, the transformation from the time domain to the complex frequency domain is expressed as follows:

[0039]

[0040] in, The Laplace transform of the data source vector yields a transformation from the time domain to the complex frequency domain, where s represents a complex variable;

[0041] Combined with weighting coefficient γ l , is represented as:

[0042]

[0043] Where, γ l represents the weight coefficient, represents the weight of the l-th Laplace transform, and A4 represents the intermediate alarm state vector obtained through the Laplace transform and the weight coefficient.

[0044] The results of all parts are superimposed, and the final alarm state vector A is output through a nonlinear activation function. The comprehensive alarm model is expressed as follows:

[0045] A = f(A1 + A2 + A3 + A4)

[0046]

[0047] Where, f(z) j Let z represent the final nonlinear activation function, and z represent the combination result of the intermediate alarm state vectors. j Let represent the j-th element in vector z.

[0048] As a preferred embodiment of the time-series data multi-topic intelligent monitoring method for complex industrial environments described in this invention, wherein: the alarm status details defined by the unit monitoring item topic include the monitoring items and associated unit monitoring item topics, if the alarm conditions in the monitoring item topic details are met, the real-time alarm status and historical alarm status of the unit will be generated, and the alarm status will include the event start time, data source value, alarm duration, maximum value within the duration, and minimum value information;

[0049] When the alarm status changes, the previous alarm will be canceled, a new alarm will be generated, and the real-time alarm status and historical alarm status of the unit will be recorded.

[0050] As a preferred embodiment of the multi-topic intelligent monitoring method for time-series data in complex industrial environments described in this invention, the step of recalculating the data source through the comprehensive alarm model includes, for units that need to adjust historical alarm theme rules, re-executing the comprehensive alarm model judgment through the monitoring alarm recalculation function, recalculating the bound data source according to the new theme rules, specifying the recalculation time period, and automatically generating the latest real-time monitoring item alarm status and unit historical alarm status.

[0051] Another objective of this invention is to provide a multi-topic intelligent monitoring system for time-series data in complex industrial environments. This system can define alarm status details through unit monitoring item themes, thus solving the problem that current monitoring and alarm methods are difficult to manage in a refined manner regarding equipment operating status.

[0052] As a preferred embodiment of the multi-topic intelligent monitoring system for time-series data in complex industrial environments described in this invention, it includes: a precondition construction module for unit monitoring items, a comprehensive alarm model construction module, and a data source recalculation module; the precondition construction module for unit monitoring items is used to construct preconditions for unit monitoring items; the comprehensive alarm model construction module is used to construct a comprehensive alarm model by defining alarm status details through unit monitoring item themes; and the data source recalculation module is used to re-execute the comprehensive alarm model judgment through the monitoring alarm recalculation function.

[0053] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement a method for intelligent monitoring of time-series data on multiple themes in complex industrial environments.

[0054] A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of a method for intelligent monitoring of multiple themes of time-series data in complex industrial environments.

[0055] The beneficial effects of this invention are as follows: The multi-topic intelligent monitoring method for time-series data in complex industrial environments provided by this invention effectively filters and controls monitoring alarms by constructing preconditions for unit monitoring items, significantly reducing the false alarm rate and workload of the monitoring system, and improving the stability and reliability of the system. By constructing unit monitoring item themes and defining detailed alarm status, it achieves refined management of equipment operating status, accurately monitors equipment operating status, detects potential problems in real time, and improves the safety and efficiency of power plant operation. By recalculating the data source through a comprehensive alarm model, it improves the flexibility and adaptability of the alarm system, can reflect equipment operating status more promptly, and optimizes the overall performance of the monitoring system. This invention achieves better results in terms of alarm accuracy, adaptability, and efficiency. Attached Figure Description

[0056] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 The first embodiment of the present invention provides an overall flowchart of a multi-topic intelligent monitoring method for time-series data in complex industrial environments.

[0058] Figure 2 This is an architecture diagram of a multi-topic intelligent monitoring method for time-series data in complex industrial environments, provided in the first embodiment of the present invention.

[0059] Figure 3 This is a business architecture diagram of a multi-topic intelligent monitoring method for time-series data in complex industrial environments, provided in the first embodiment of the present invention.

[0060] Figure 4 The following is an overall flowchart of a multi-topic intelligent monitoring system for time-series data in complex industrial environments, provided as a third embodiment of the present invention. Detailed Implementation

[0061] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0062] Example 1

[0063] Reference Figures 1-3 As an embodiment of the present invention, a method for intelligent monitoring of multiple themes of time-series data in complex industrial environments is provided, comprising:

[0064] S1: Prerequisites for building unit monitoring items.

[0065] Furthermore, Figure 2 This is a system architecture diagram of an embodiment of the present invention. The diagram illustrates the relationship between the DCS data source, the time-series database, the monitoring subject, and the power plant monitoring and alarm system.

[0066] It should be noted that the DCS data source is forwarded to the time series database for storage through the gateway. The time series database stores analog quantities, switch quantities, characteristic values, and packaged points in the DCS according to the time series. Depending on the data source type, it implements compression modes such as dynamic threshold, jump storage, timed storage, lossy storage, and lossless storage.

[0067] The time-series database pushes high-concurrency, rapidly changing data to the power plant monitoring and alarm system via MQTT data subscription, and provides an RPC batch query interface to collect statistical information such as time, day, month, and year from the data source.

[0068] It should be noted that the power plant monitoring and alarm system provides topics for equipment exceeding limits, assessment conditions, operational status, equipment start-up and shutdown, interface monitoring, characteristic values, packaging points, and wall temperature monitoring. Each topic is configured with a corresponding data source list to monitor the operational alarm status of each unit. Figure 3This is the business architecture diagram of a multi-topic alarm system. The diagram illustrates the relationships between unit monitoring items, unit monitoring item prerequisites, unit monitoring item topics, unit alarm status, unit monitoring item recalculation, and the power plant monitoring and alarm system.

[0069] Furthermore, regarding the preconditions for monitoring new generating units, power plant generating units have numerous operating components, many of which are relatively problematic and interconnected. Therefore, providing a unit alarm monitoring method with configurable preconditions would effectively solve these problems. If the preconditions are not met, the monitoring and alarm work will not proceed; if the preconditions are met, the monitoring and alarm work will continue.

[0070] It should be noted that the precondition data for the unit monitoring items includes equipment status, environmental parameters, maintenance status, historical alarm status, and equipment start / stop status.

[0071] For equipment status-related prerequisites, alarm monitoring will not be activated if the boiler temperature is not between 150℃ and 250℃; alarm monitoring will not be activated if the generator speed is below 1800 rpm.

[0072] For environmental parameter-related prerequisites, if the ambient temperature is not between -10℃ and 40℃, alarm monitoring will not be activated; if the ambient humidity is not between 20% and 80%, alarm monitoring will not be activated.

[0073] For maintenance status prerequisites, alarm monitoring will not be activated if the equipment is in maintenance status.

[0074] For historical alarm status-related preconditions, if there are unprocessed historical alarms, new alarm monitoring will not be initiated.

[0075] For equipment start / stop status prerequisites, alarm monitoring will not be activated if the equipment is not started or has been running for less than 30 minutes.

[0076] It should also be noted that Boolean logic is used to define preconditions, and each precondition is represented by a Boolean matrix P. i It means that P i This indicates the alarm status of the i-th device, where 1 represents an alarm and 0 represents normal. The overall preconditions are expressed as follows:

[0077]

[0078] Where ⊙ represents element-wise multiplication, and n represents the number of prerequisite conditions;

[0079] If P total If all elements in the table are 1, the precondition is met and the monitoring and alarm work continues; otherwise, it does not.

[0080] S2: Define alarm status details through unit monitoring item topics.

[0081] Furthermore, new unit monitoring topics are created. These topics primarily provide detailed monitoring and alarm rules to define alarm status details, including topics such as equipment exceeding limits, assessment conditions, input / output status, equipment start / stop, interface monitoring, characteristic value, package point, and wall temperature monitoring. By binding the data source with the monitoring topic rules, it is easy to assign specific meanings to each time block of the data source, thereby generating alarm information.

[0082] It should be noted that the unit monitoring item topic is constructed, and the monitoring item topic provides detailed monitoring alarm rules to define alarm status details, including equipment over-limit topic, assessment condition topic, input / output status, equipment start / stop topic, interface monitoring topic, characteristic value topic, package point topic, wall temperature monitoring topic, data source is bound to the monitoring item topic rules to generate alarm information;

[0083] A nonlinear transformation is performed on the data source vector D, where the data source vector is D = [D1, D2, ..., D]. m ], can undergo nonlinear transformation, represented as

[0084] σ(D)=[σ(D1),σ(D2),…,σ(D m )]

[0085]

[0086] Where D represents the data source vector, which comes from various sensors or monitoring devices, and the vector dimension is m; σ(D) represents the nonlinear transformation of the data source vector. i ) represents the transformation result of the i-th data source value;

[0087] Using multidimensional tensors and rule tensor Applying rules and performing tensor product operations on the data source can be represented as follows:

[0088]

[0089] in, Let M be a multidimensional tensor, representing the rule matrix of the i-th monitored item topic, with dimensions (p×q×r). i,juv Let represent the element of the i-th tensor at position (j, u, v). Let R be the rule tensor, which is the matrix of the i-th rule. i,uvk Let A1 represent the element at position (u,v,k) of the i-th rule tensor, and let A1 represent the intermediate alarm state vector obtained through tensor product operation and rule application. 1,jkThis represents the element at position (j,k) after the tensor product operation and rule application.

[0090] Pooling operations on the data source are represented as:

[0091]

[0092] Where, pool(D) ij This represents the max pooling result at position (i,j), where (u,v)∈N(i,j)'

[0093] Combining convolution tensors and weighting coefficient α j The convolution operation is represented as:

[0094]

[0095] Here, A2 represents the intermediate alarm state vector obtained through pooling and convolution operations.

[0096] It should also be noted that performing a Fourier transform on the data source and outputting frequency domain information is represented as follows:

[0097]

[0098] in, This represents the Fourier transform of the data source vector, outputting frequency domain information. D(t) represents the value of the data source signal at time t, and f represents a specific frequency component in the Fourier transform result.

[0099] Combined with weighting coefficient β k The output is represented as:

[0100]

[0101] Where A3 represents the weight coefficient, which represents the weight of the k-th Fourier transform;

[0102] Applying a Laplace transform to the data source, the transformation from the time domain to the complex frequency domain is expressed as follows:

[0103]

[0104] in, The Laplace transform of the data source vector yields a transformation from the time domain to the complex frequency domain, where s represents a complex variable;

[0105] Combined with weighting coefficient γ l , is represented as:

[0106]

[0107] Where, γ lrepresents the weight coefficient, represents the weight of the l-th Laplace transform, and A4 represents the intermediate alarm state vector obtained through the Laplace transform and the weight coefficient.

[0108] The results of all parts are superimposed, and the final alarm state vector A is output through a nonlinear activation function. The comprehensive alarm model is expressed as follows:

[0109] A = f(A1 + A2 + A3 + A4)

[0110]

[0111] Where, f(z) j Let z represent the final nonlinear activation function, and z represent the combination result of the intermediate alarm state vectors. j Let represent the j-th element in vector z.

[0112] S3: Recalculate the data source based on the comprehensive alarm model.

[0113] Furthermore, a new unit monitoring item is created and associated with the unit monitoring item topic in S2. When the alarm conditions in the monitoring item topic details are met, the unit's real-time alarm status and historical alarm status will be generated. The alarm status will include information such as the event start time, data source value, alarm duration, maximum value within the duration, and minimum value. When the alarm status changes, the previous alarm will be canceled, a new alarm will be generated, and the unit's real-time alarm status and historical alarm status will be recorded.

[0114] It should be noted that, in addition to the task of recalculating monitoring items for the unit, this invention also provides a function for recalculating unit monitoring alarms. For units that need to adjust the historical alarm theme rules, the monitoring alarm recalculation function can be used to recalculate the bound data source according to the new theme rules. It can support recalculating a specified time period and quickly generate the latest real-time monitoring item alarm status and unit historical alarm status.

[0115] It should also be noted that the real-time alarm status will be pushed in real time via EventHubs and WebSocket. The start time and end time of alarm termination and the generation of new alarms, as well as the maximum and minimum values ​​of the data source during the alarm period, are stored by PostgreSQL.

[0116] Example 2

[0117] One embodiment of the present invention provides a multi-topic intelligent monitoring method for time-series data in complex industrial environments. To verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculations and simulation experiments.

[0118] First, the test subjects were five generating units of a power plant, numbered Unit A, Unit B, Unit C, Unit D, and Unit E. A comparative experiment was conducted on the five generating units, applying both the monitoring and alarm method of this invention and existing monitoring and alarm methods.

[0119] Each unit is equipped with necessary sensors, including temperature sensors, speed sensors, ambient temperature sensors, and ambient humidity sensors. Ensure that the sensors are connected to a time-series database and can acquire and store data in real time.

[0120] Equipment status prerequisites: Boiler temperature between 150℃ and 250℃, generator speed not less than 1800 rpm.

[0121] Environmental parameters prerequisites: ambient temperature between -10℃ and 40℃, and ambient humidity between 20% and 80%.

[0122] Prerequisite for maintenance status: The equipment is not in maintenance status.

[0123] Prerequisite for historical alarm status: No unprocessed historical alarms.

[0124] Prerequisites for equipment start-up / shutdown status: Equipment running time exceeds 30 minutes.

[0125] Each unit collects real-time data on temperature, speed, ambient temperature, and humidity to ensure accuracy and timeliness. The collected data is compared with predefined conditions; if the conditions are met, the process proceeds to the next stage; otherwise, alarm monitoring is stopped. Based on the monitoring item's theme rules, the data source undergoes nonlinear transformation and multidimensional tensor operations to generate a preliminary alarm status. Pooling, Fourier transform, and Laplace transform operations are then performed on the preliminary alarm status, combined with weighting coefficients, to finally generate the alarm status vector.

[0126] For situations requiring adjustments to historical alarm rules, the comprehensive alarm model is re-executed to generate the latest alarm status. The alarm status and related parameters of each unit are recorded and compared with existing alarm systems.

[0127] Table 1 Experimental Data

[0128]

[0129] Regarding the number of alarms and false alarms, the table shows that the alarm counts of the units applying this invention (Unit A, Unit B, and Unit E) are all controlled within a reasonable range, and the number of false alarms is significantly lower than that of the units using the prior art (Unit C and Unit D). Unit A had 15 alarms and only 1 false alarm, with an alarm accuracy rate as high as 93.33%; in contrast, Unit C had 20 alarms and 5 false alarms, with an alarm accuracy rate of only 75.00%.

[0130] Regarding alarm response time, the table data shows that the units using this invention perform excellently in this aspect. Unit B's alarm response time is 4 seconds, significantly lower than Unit C's 8 seconds and Unit D's 9 seconds.

[0131] Regarding data recalculation time, the data recalculation time for Unit A is 10 seconds, for Unit B it is 12 seconds, while the existing technology for Units C and D requires 20 seconds and 25 seconds respectively.

[0132] Regarding alarm accuracy, the table shows that the alarm accuracy of the units using this invention is significantly higher than that of units using existing technologies. Unit E has an alarm accuracy of 93.75%, while the alarm accuracy of Unit D is only 72.73%.

[0133] In summary, the present invention achieves better results in terms of alarm accuracy, adaptability, and efficiency.

[0134] Example 3

[0135] Reference Figure 4 As an embodiment of the present invention, a multi-topic intelligent monitoring system for time-series data in complex industrial environments is provided, including a precondition construction module for unit monitoring items, a comprehensive alarm model construction module, and a data source recalculation module.

[0136] The module for constructing preconditions for unit monitoring items is used to build preconditions for unit monitoring items. The module for constructing a comprehensive alarm model is used to construct a comprehensive alarm model by defining alarm status details based on the unit monitoring item theme. The module for recalculating the data source is used to re-execute the comprehensive alarm model judgment through the monitoring alarm recalculation function.

[0137] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0138] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0139] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0140] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc. It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0141] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for multi-topic intelligent monitoring of time-series data in an industrial complex environment, characterized in that, The method comprises the following steps: constructing a unit monitoring item precondition; defining an alarm state detail through a unit monitoring item theme; judging a data source for recalculation through a comprehensive alarm model; the step of constructing a unit monitoring item precondition comprises collecting unit monitoring item precondition data, including device state, environmental parameters, maintenance state, historical alarm state, and device start-stop state; for the device state precondition, if the boiler temperature is not between 150-250℃, the alarm monitoring is not started; if the generator speed is lower than 1800rpm, the alarm monitoring is not started; for the environmental parameter precondition, if the environmental temperature is not between -10-40℃, the alarm monitoring is not started; if the environmental humidity is not between 20-80%, the alarm monitoring is not started; for the maintenance state precondition, if the device is in maintenance state, the alarm monitoring is not started; for the historical alarm state precondition, if there is an unprocessed historical alarm, the new alarm monitoring is not started; for the device start-stop state precondition, if the device is not started or the running time is less than 30 minutes, the alarm monitoring is not started; The construction machine group monitoring item precondition includes defining preconditions using Boolean logic, each precondition being represented by a Boolean matrix P i represents the alarm state of the i-th device, 1 indicating an alarm and 0 indicating normal, the overall precondition being represented by: i represents the alarm state of the i-th device, 1 indicating an alarm and 0 indicating normal, the overall precondition being represented by: wherein, ⊙ represents element-wise multiplication, and n represents the number of preconditions; If P total All elements are 1, the precondition is met, the monitoring alarm is working continuously, otherwise not; the step of defining an alarm state detail through a unit monitoring item theme comprises constructing a unit monitoring item theme, the monitoring item theme provides monitoring alarm rule details, which are used to define alarm state details, including device limit theme, assessment condition theme, input-output state, device start-stop theme, interface monitoring theme, characteristic value theme, packaging point theme, wall temperature monitoring theme, binding the data source and the monitoring item theme rule, and generating alarm information; The data source vector D is non-linearly transformed, where D = [D1, D2,..., D m ,] and the non-linear transformation is represented as σ(D) = [σ(D1), σ(D2),..., σ(D m )] where D represents a data source vector, data from various sensors or monitoring devices, with a vector dimension of m, σ(D) represents a nonlinear transformation of the data source vector, σ(D i ) represents the transformation result of the i-th data source value; Using a multi-dimensional tensor M i and a regular tensor R i The regular application and tensor product operation on the data source is represented as: wherein M i represents a multi-dimensional tensor, is a regular matrix of the ith monitoring item subject, with dimensions (p x q x r), M i,juv represents an element of the ith tensor at position (j, u, v), Ri represents a regular tensor, is a matrix of the ith regular, R i,uvk represents an element of the ith regular tensor at position (u, v, k), Ai represents an intermediate alarm state vector obtained by tensor product operation and regular application, A 1,jk represents an element at position (j, k) after tensor product operation and regular application; pooling operation is performed on the data source, which is represented as: where pool(D) ij denotes the max-pooling result at (i, j) position, (u, v) ∈ N(i, j) Convolutional tensor and a weight coefficient α j a convolution operation, denoted as: wherein, A2 represents an intermediate alarm state vector obtained through the pooling operation and the convolution operation; the step of defining an alarm state detail through a unit monitoring item theme comprises performing Fourier transform on the data source, and outputting frequency domain information, which is represented as: wherein D(f) represents the Fourier transform of the data source vector, outputting frequency domain information, D(t) represents the value of the data source signal at time t, and f represents a specific frequency component in the Fourier transform result Combination weight coefficient β k Output, denoted as: wherein, A3 represents a weight coefficient, and represents the weight of the kth Fourier transform; performing Laplace transform on the data source, and outputting the transform from the time domain to the complex frequency domain, which is represented as: wherein denotes the Laplace transform of the data source vector, resulting in a transformation from the time domain to the complex frequency domain, s denotes a complex variable; Combination weight coefficient γ l is expressed as: wherein, γl represents a weight coefficient, and represents the weight of the lth Laplace transform; A4 represents an intermediate alarm state vector obtained through the Laplace transform and the weight coefficient; the results of all parts are superimposed, and the final alarm state vector A is output through a nonlinear activation function, i.e. the comprehensive alarm model is represented as: A=f(A1+A2+A3+A4) where f(z) = max(0, z) j represents the final non-linear activation function, z represents the combination result of the intermediate alarm state vector, z j represents the jth element in the vector z; the step of defining an alarm state detail through a unit monitoring item theme comprises monitoring items and associated unit monitoring item themes, if the alarm conditions in the monitoring item theme details are met, the unit real-time alarm state and the unit historical alarm state are generated, and the alarm state comprises event start time, data source value, alarm duration, maximum value, and minimum value information within the duration; when the alarm state changes, the previous alarm is cancelled, a new alarm is generated, and the unit real-time alarm state and the unit historical alarm state are recorded; The re-calculation of the data source by the integrated alarm model judgment comprises: for the unit needing to adjust the historical alarm theme rule, through the monitoring alarm re-calculation function, re-executing the integrated alarm model judgment, re-calculating the bound data source according to the new theme rule, re-specifying the re-calculation time period, and automatically generating the latest real-time monitoring item alarm state and the unit historical alarm state.

2. A system employing the multi-theme intelligent monitoring method for time series data in an industrial complex environment according to claim 1, characterized in that: The method comprises a unit monitoring item precondition construction module, an integrated alarm model construction module and a data source re-calculation module. The unit monitoring item precondition construction module is used for constructing the unit monitoring item precondition. The integrated alarm model construction module is used for constructing the integrated alarm model through the unit monitoring item theme definition alarm state details. The data source re-calculation module is used for re-executing the integrated alarm model judgment through the monitoring alarm re-calculation function.

3. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that, The processor executes the computer program to realize the steps of the method for the multi-theme intelligent monitoring of the time series data in the industrial complex environment.

4. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the steps of the method for the multi-theme intelligent monitoring of the time series data in the industrial complex environment.

Images