Online fault diagnosis method for petrochemical equipment group

Abstract

The invention discloses a petrochemical equipment group-oriented fault online diagnosis method, which comprises the following steps of: decomposing a pre-constructed cloud large model according to a coupling relationship among unit devices in a petrochemical equipment group to obtain cloud sub-models; issuing the cloud sub-model to an edge node, and establishing an edge sub-model; inputting the obtained real-time data of each unit device into the corresponding edge sub-model, updating parameters of the edge sub-model and performing fault diagnosis on the real-time data of each unit device to obtain the updated edge sub-model and a fault diagnosis result of each unit device; uploading the updated parameters of the edge sub-model and the real-time data of each unit device to a cloud large model, and aggregating the updated parameters of the edge sub-model by the cloud large model to obtain an updated cloud large model; and obtaining a fault diagnosis result of the petrochemical equipment group based on the updated cloud large model according to the real-time data of each unit device.

Description

Technical Field

[0001] The present invention relates to the technical field of petrochemical equipment fault diagnosis, and in particular to an online fault diagnosis method for a petrochemical equipment group. Background Art

[0002] Fault diagnosis technology significantly impacts the diagnosis of faults and even accidents. Existing fault diagnosis methods for petrochemical production facilities primarily focus on localized units or equipment, with limited macro-global diagnosis. As petrochemical production facilities expand in scale, the interconnectedness and material flow between individual units become complex. This "single-brain" approach to localized fault diagnosis is no longer adequate for current fault diagnosis needs and will severely impact the diagnostic accuracy of complexly coupled equipment clusters. Furthermore, existing data-driven fault diagnosis methods are primarily based on historical data and lack the ability to incorporate real-time data in a timely manner, which can also impact fault diagnosis accuracy. Therefore, researching macro-global, real-time diagnostic methods for the entire equipment cluster is crucial for improving fault diagnosis accuracy.

[0003] Timeliness of fault diagnosis is also a crucial metric. Early fault detection can significantly reduce both life and economic losses. The large amount of data in a fleet of equipment places high demands on computing and storage. Therefore, improving the efficiency of macro-global diagnostic methods across the entire fleet can significantly enhance the real-time performance of fault diagnosis. The overcentralization of cloud computing resources creates insurmountable latency barriers. Edge computing resources, however, are distributed and geographically close to users, providing high-bandwidth and low-latency services. However, edge computing clearly lacks computing and storage capabilities, failing to provide sufficiently powerful computing or storage services. Summary of the Invention

[0004] The purpose of the present invention is to overcome the deficiencies in the prior art and provide an online fault diagnosis method for petrochemical equipment groups, which can meet the high-precision and real-time requirements of equipment group faults.

[0005] To achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0006] The present invention provides a method for online fault diagnosis of a petrochemical equipment group, comprising:

[0007] Based on the coupling relationship between each unit device in the petrochemical equipment group, the pre-built cloud-based large model is decomposed to obtain cloud-based sub-models;

[0008] Send the cloud sub-model to the edge node to establish the edge sub-model;

[0009] Inputting the acquired real-time data of each unit device into the corresponding edge sub-model, updating the parameters of the edge sub-model and performing fault diagnosis on the real-time data of each unit device, thereby obtaining an updated edge sub-model and fault diagnosis results for each unit device;

[0010] Uploading the updated parameters of the edge sub-model and the real-time data of each unit device to the cloud-based large model, and the cloud-based large model aggregating the updated parameters of the edge sub-model to obtain an updated cloud-based large model;

[0011] According to the real-time data of each unit device, based on the updated cloud-based big model, the fault diagnosis results of the petrochemical equipment group are obtained.

[0012] When applied, the present invention can perform real-time diagnosis of unit faults, while gathering multi-domain knowledge of the entire equipment group to perform real-time diagnosis of global complex faults, making full use of the advantages of edge-cloud collaborative technology.

[0013] Optionally, the real-time data of each unit device includes at least temperature, humidity, displacement, pressure, noise, liquid level and vibration frequency.

[0014] Optionally, the construction of the cloud-based large model includes:

[0015] A large cloud model is established based on the historical data of the petrochemical equipment group and the coupling relationship between each unit device in the petrochemical equipment group.

[0016] Optionally, the objective function of the cloud-based large model is expressed as:

[0017] ;

[0018] Where, Represents a large model in the cloud; represents the historical data of the i-th petrochemical equipment group corresponding to time t; express The corresponding target value.

[0019] Optionally, the updated edge sub-model is expressed as:

[0020] ;

[0021] Where, represents the jth updated edge sub-model; The jth marginal sub-model; represents the local learning rate; represents the gradient of the objective function of the cloud-based large model with respect to the j-th edge sub-model; represents the real-time data of the unit device input into the jth edge sub-model; represents the fault diagnosis result of the unit device output by the jth edge sub-model; Represents the objective function of the large model in the cloud.

[0022] Optionally, the updated cloud-based large model is represented as:

[0023] ;

[0024] Where, Indicates the updated cloud-based large model; represents the number of edge sub-models; represents the jth updated marginal sub-model.

[0025] Optionally, before inputting the acquired real-time data of each unit device into the corresponding edge sub-model, the real-time data of each unit device is pre-processed.

[0026] Optionally, the coupling relationship between each unit device in the petrochemical equipment group is a spatial domain coupling relationship.

[0027] Optionally, also include:

[0028] According to the coupling relationship between each unit device in the petrochemical equipment group, the updated cloud-based large model is decomposed to obtain the updated cloud-based sub-model;

[0029] Sending the updated cloud sub-model to the edge node to obtain a re-updated edge sub-model;

[0030] Repeat until the preset stop condition is reached.

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

[0032] The present invention utilizes the spatial coupling relationship between each unit device in the petrochemical equipment group, integrates historical and real-time data knowledge, and uses edge-cloud collaborative technology to perform edge-cloud model multi-domain real-time diagnosis, which will greatly improve the diagnostic accuracy of equipment group faults; the edge sub-model can perform real-time diagnosis of unit faults, and the cloud-based large model gathers multi-domain knowledge of the entire equipment group to perform real-time diagnosis of global complex faults, making full use of the advantages of edge-cloud collaborative technology, greatly improving the efficiency of the equipment group fault diagnosis algorithm, and ensuring the real-time performance of equipment group fault diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 FIG. 1 is a schematic diagram showing a framework of an online fault diagnosis method for a petrochemical equipment group according to an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations on the technical solution of the present invention. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0035] The term "and / or" simply describes a relationship between related objects, indicating that three possible relationships exist. For example, "A and / or B" can mean: A exists alone, A and B exist simultaneously, or B exists alone. Additionally, the character " / " generally indicates an "or" relationship between the related objects.

[0036] Example 1

[0037] like Figure 1 As shown, this embodiment introduces a method for online fault diagnosis of a petrochemical equipment group, which specifically includes the following steps:

[0038] Establish and decompose a cloud-based large model based on the spatial domain; based on the collected historical data samples and the spatial coupling relationship of the petrochemical equipment group, establish a global cloud-based large model in the cloud that reflects the spatial relationship of the entire equipment group, and decompose and pruned the cloud-based large model according to the coupling relationship between each unit device in the equipment group, and decompose it into cloud-based sub-models corresponding to each unit device.

[0039] The cloud-based big model will send the edge sub-model based on the spatial domain to each edge node (unit device) to establish the edge sub-model.

[0040] The edge nodes collect and pre-process the new data generated by the local unit devices in real time.

[0041] Based on the edge sub-model training diagnosis in the time domain, the edge sub-model is trained according to the real-time new data, the trained parameters are uploaded to the cloud-based large model, and the real-time diagnosis results of the local unit device are output.

[0042] The cloud-based big model aggregates the parameters uploaded by each edge node, and combines the summarized new parameters to perform global training diagnosis of the cloud-based big model, and outputs the composite fault diagnosis results of the diagnosed equipment group.

[0043] Edge computing can make up for the shortcomings of traditional centralized cloud computing. The powerful computing and storage capabilities of cloud computing can make up for the shortcomings of edge computing in this regard. The edge and cloud form a complementary end-to-end system, realizing a pattern of mutual coordination between centralized cloud and distributed edge computing, thereby providing more flexible and lower-latency services for fault diagnosis of petrochemical equipment groups.

[0044] Example 2

[0045] like Figure 1 As shown, based on Example 1, this embodiment also has the following designs:

[0046] Step 1: Build a cloud-based large model, decompose the petrochemical equipment group into each unit device, and according to the historical data of each unit device in the petrochemical equipment group, the time series is , the i-th data sample corresponding to time t is (To simplify the process, time t here corresponds to the i-th data sample), The sampled data after preprocessing the historical data of the petrochemical equipment group is used as the input of the cloud-based large model. The corresponding target value is , The output of the cloud-based big model at time t. The historical data is integrated with the spatial coupling relationship between each unit device to establish a cloud-based big model of the entire petrochemical equipment group system. , then the objective function of the cloud-based large model is expressed as:

[0047] .

[0048] Step 2: Based on the spatial domain coupling relationship between the unit devices in the petrochemical equipment group, the pre-built cloud-based large model is decomposed and pruned to obtain the cloud-based sub-model corresponding to each single unit device.

[0049] Step 3: Send the cloud sub-model to the edge node of the corresponding unit device to establish an edge sub-model. There are m edge sub-models, and the jth edge sub-model is represented as .

[0050] Step 4: Collect the real-time data of the local unit device at each edge node, pre-process and sample the real-time data of the unit device to obtain the pre-processed new data. The new data of the j-th edge node is expressed as , the new data includes at least temperature, humidity, displacement, pressure, noise, liquid level and vibration frequency.

[0051] Step 5: Inputting the preprocessed new data into the corresponding edge sub-model, updating the parameters of the edge sub-model, and performing fault diagnosis on the preprocessed new data to obtain the updated edge sub-model and the corresponding unit device fault diagnosis results;

[0052] In a specific embodiment, the new data of the j-th edge node after preprocessing is Input the jth edge sub-model, update the parameters of the jth edge sub-model, and perform the pre-processed new data of the jth edge node. Perform fault diagnosis and obtain the updated fault diagnosis results of the unit device corresponding to the j-th edge sub-model and the j-th edge node and output the fault diagnosis result to the corresponding unit device ; The updated j-th edge sub-model Expressed as:

[0053] ;

[0054] Where, represents the local learning rate; represents the gradient of the objective function of the cloud-based large model with respect to the j-th edge sub-model; Represents the objective function of the large model in the cloud.

[0055] In this way, new failure modes are continuously learned, and new data samples are combined with old model parameters to obtain new model parameters, thereby completing the training and updating between the edge sub-model and real-time data.

[0056] Step 6: Upload the updated edge sub-model parameters and the real-time data of each unit device to the cloud-based large model, and the cloud-based large model aggregates the updated edge sub-model parameters to obtain an updated cloud-based large model. Expressed as:

[0057] ;

[0058] Step 7: Based on the real-time data of each unit device, based on the updated cloud-based big model , get the fault diagnosis results of the petrochemical equipment group .

[0059] Step 7: Based on the spatial domain coupling relationship between each unit device in the petrochemical equipment group, the updated cloud-based large model Decompose to obtain an updated cloud sub-model; send the updated cloud sub-model to the edge node to obtain a re-updated edge sub-model, and perform the next round of real-time recursive update of the edge sub-model and the cloud large model.

[0060] In this way, each edge node has new data samples of the corresponding unit device edge sub-model for training, supporting efficient cloud-based large model updates with low parameter conflicts.

[0061] The edge sub-model on the edge node can continuously improve its accuracy through online recursive updates. At the same time, through two-way enhancement with the large cloud model, global optimization can be achieved to improve the reliability and robustness of the model.

[0062] In this way, through edge-cloud collaboration, the edge sub-model at the edge and the cloud large model at the cloud are continuously and iteratively updated synchronously according to a certain time period. The unit fault diagnosis model at the edge and the global fault diagnosis model at the cloud are gradually improved in the edge-cloud collaboration, so as to obtain the latest and optimal unit and system diagnosis models, and realize the two-way joint optimization of unit and system-level diagnosis models.

[0063] When applied, this embodiment integrates historical and real-time data knowledge based on the spatial coupling relationship between each unit device in the petrochemical production equipment cluster, and utilizes edge-cloud collaborative technology to perform multi-domain real-time diagnosis of the edge-cloud model. Specifically, a cloud-based large model is established based on the global macro knowledge of the entire equipment cluster aggregated on the cloud. The large model is then decomposed according to the corresponding unit device coupling relationship, and the decomposed edge sub-models are distributed to each corresponding unit device (i.e., edge node). The edge sub-models are fed with real-time new data for real-time training and diagnosis of unit device faults. The new data and new parameters are uploaded to the cloud, and the cloud-based large model aggregates the data of each edge sub-model for global real-time training and diagnosis of faults across the entire equipment cluster. This solves the latency and accuracy requirements for real-time diagnosis of equipment cluster faults.

[0064] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0065] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. 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.

[0066] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work 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 1a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0067] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0068] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the purpose of the present invention and the claims, which are all protected by the present invention.

Claims

1. A method for online fault diagnosis of petrochemical equipment groups, characterized in that: include: According to the coupling relationship between each unit device in the petrochemical equipment group, the pre-built cloud large model is decomposed to obtain the cloud sub-model; Send the cloud sub-model to the edge node to establish the edge sub-model; Inputting the acquired real-time data of each unit device into the corresponding edge sub-model, updating the parameters of the edge sub-model and performing fault diagnosis on the real-time data of each unit device to obtain the updated edge sub-model and the fault diagnosis results of each unit device; Uploading the updated parameters of the edge sub-model and the real-time data of each unit device to the cloud-based large model, and the cloud-based large model aggregates the updated parameters of the edge sub-model to obtain an updated cloud-based large model; According to the real-time data of each unit device, based on the updated cloud-based big model, the fault diagnosis results of the petrochemical equipment group are obtained.

2. The method for online fault diagnosis of petrochemical equipment group according to claim 1, characterized in that: The real-time data of each unit device at least includes temperature, humidity, displacement, pressure, noise, liquid level and vibration frequency.

3. The method for online fault diagnosis of petrochemical equipment groups according to claim 1, characterized in that: The construction of the cloud-based large model includes: A large cloud model is established based on the historical data of the petrochemical equipment group and the coupling relationship between each unit device in the petrochemical equipment group.

4. The method for online fault diagnosis of petrochemical equipment group according to claim 3, characterized in that: The objective function of the cloud-based large model is expressed as: ； In the formula, Represents a large model in the cloud; Represents the historical data of the i-th petrochemical equipment group corresponding to time t; express The corresponding target value.

5. The method for online fault diagnosis of petrochemical equipment group according to claim 1, characterized in that: The updated edge sub-model is expressed as: ； In the formula, represents the jth updated edge sub-model; The jth marginal sub-model; represents the local learning rate; represents the gradient of the objective function of the cloud-based large model with respect to the j-th edge sub-model; represents the real-time data of the unit device input into the jth edge sub-model; represents the fault diagnosis result of the unit device output by the jth edge sub-model; Represents the objective function of the large model in the cloud.

6. The method for online fault diagnosis of petrochemical equipment group according to claim 1, characterized in that: The updated cloud-based large model is represented as follows: ； In the formula, Indicates the updated cloud-based large model; represents the number of edge sub-models; represents the jth updated marginal sub-model.

7. The method for online fault diagnosis of petrochemical equipment groups according to claim 1 or 3, characterized in that: Before inputting the acquired real-time data of each unit device into the corresponding edge sub-model, the real-time data of each unit device is pre-processed.

8. The method for online fault diagnosis of petrochemical equipment groups according to claim 1, characterized in that: The coupling relationship between the unit devices in the petrochemical equipment group is a spatial domain coupling relationship.

9. The method for online fault diagnosis of petrochemical equipment groups according to claim 1, characterized in that: Also includes: According to the coupling relationship between each unit device in the petrochemical equipment group, the updated cloud-based large model is decomposed to obtain an updated cloud-based sub-model; Sending the updated cloud sub-model to the edge node to obtain a re-updated edge sub-model; Repeat until the preset stop condition is reached.

Images