Management method and device for operation of fire pump house

Abstract

The invention provides a management method for operation of a fire pump house. The method comprises the following steps: at least acquiring regional image information of a fire pump house; processing the fire-fighting water pump room area image information by adopting a fire-fighting water pump monitoring model to generate fire-fighting water pump room area monitoring information; and judging whether the fire-fighting water pump room area monitoring information accords with a preset condition, and generating a corresponding management strategy when the fire-fighting water pump room area monitoring information accords with the preset condition. Therefore, the management method ensures that the large model can correctly provide information through the professional knowledge base and the knowledge graph in the field of maintenance operation of the fire pump house. In addition, the detection rule is defined by using a natural language, so that the utilization rate of physical defense and civil defense can be improved to the greatest extent. And finally, multi-modal identification of a risk scene is realized by using computer vision and a large model, so that known risks can be effectively prevented, left risks can be found through anomaly identification, and a management system is continuously improved.

Description

Technical Field

[0001] The present invention relates to the field of fire protection technology, and in particular to a management method and device for fire water pump room operations. Background Art

[0002] Fire pump room maintenance is a crucial component of fire protection facility safety. To ensure the safety and accuracy of pump room maintenance operations, advanced technologies are needed to monitor and control various risk factors during operations.

[0003] In the process of implementing the prior art, the inventors found that:

[0004] Existing smart fire warning systems require predefined anomaly scenarios. Identifying anomalies requires a large amount of labeled anomaly data. The enormous workload of data collection and annotation often makes it impractical in the short term, making implementation difficult. An even more formidable obstacle is the sheer number of states in which anomalies can occur, making them difficult to exhaustively enumerate and accurately simulate. This makes it difficult for existing systems to identify new safety anomalies and results in inaccurate alerts for known risks.

[0005] Therefore, it is necessary to provide a management solution for fire water pump room maintenance operations with high risk identification accuracy and small training set requirements to solve the problems of the existing system's difficulty in identifying new safety anomalies and inaccurate warnings for known risks. Summary of the Invention

[0006] The embodiment of the present application provides a management solution for fire pump room maintenance operations with high risk identification accuracy and low training set requirements to solve the problems of existing systems in identifying new safety anomalies and inaccurate warnings for known risks.

[0007] Specifically, a method for managing fire pump room operations includes the following steps:

[0008] At least obtain image information of the fire pump room area;

[0009] The fire pump monitoring model is used to process the image information of the fire pump room area to generate the fire pump room area monitoring information;

[0010] Determine whether the fire water pump room area monitoring information meets the preset conditions, and generate a corresponding management strategy when it meets the preset conditions.

[0011] Furthermore, the fire pump detection model is generated by pre-training, specifically including:

[0012] Build an initial fire pump monitoring model based on the fire pump room maintenance operation knowledge graph and template library;

[0013] The initial fire pump detection model is trained to generate a final fire pump monitoring model.

[0014] Furthermore, the fire water pump room maintenance operation knowledge graph is pre-set, specifically including: obtaining a keyword list of knowledge terminal nodes, extracting associated knowledge point segment slices based on the keyword list, and generating a fire water pump room maintenance operation knowledge graph based on the keyword list and the associated knowledge point segment slices.

[0015] Furthermore, the template library is pre-set and specifically includes:

[0016] Get the initial template library;

[0017] Based on the fire water pump room maintenance operation knowledge graph, the initial template library is processed to generate a final template library.

[0018] Furthermore, the process of processing the initial template library to generate the final template library further includes:

[0019] The initial template library is processed using an accuracy assessment model to generate a final template library.

[0020] Furthermore, the fire pump detection model includes at least several target recognition sub-models and anomaly detection sub-models.

[0021] Furthermore, the setting of the preset conditions also includes verification using similarity identification technology.

[0022] Furthermore, the preset conditions are tested, including:

[0023] Using a prefabricated test set to test the accuracy of the preset conditions;

[0024] The accuracy of the preset conditions is evaluated through the accuracy rate.

[0025] Furthermore, a prefabricated test set is used to test the accuracy of the preset conditions, specifically including:

[0026] Testing the preset conditions through an automated test model, comparing the generated test results with a preset test set, and generating an accuracy rate for the preset conditions;

[0027] The preset test set includes a number of scenario data, wherein the scenario data contains test cases and expected outputs.

[0028] The present application also provides a fire pump room operation management device, including:

[0029] An acquisition module, configured to acquire at least image information of a fire pump room area;

[0030] The processing module is used to process the image information of the fire pump room area using the fire pump monitoring model to generate fire pump room area monitoring information;

[0031] The judgment execution module is used to judge whether the fire water pump room area monitoring information meets the preset conditions, and generate a corresponding management strategy when the preset conditions are met.

[0032] A computer-readable storage medium stores a computer program, wherein the computer program is executed by a processor to implement steps of a method for managing fire water pump room operations.

[0033] The technical solutions provided in the embodiments of the present application have at least the following beneficial effects:

[0034] This management approach leverages a specialized knowledge base and knowledge graph in the field of fire pump room maintenance to ensure that the large model accurately provides information. Furthermore, by defining detection rules using natural language, it maximizes the utilization of both physical and human defenses. Finally, by using computer vision and large models to achieve multimodal recognition of risk scenarios, it effectively mitigates known risks, identifies residual risks through anomaly recognition, and continuously improves management systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0036] Figure 1 A flowchart of a method for managing fire water pump room operations provided in an embodiment of the present application.

[0037] Figure 2 A schematic diagram of the structure of a management device for fire water pump room maintenance operations provided in an embodiment of the present application. DETAILED DESCRIPTION

[0038] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0039] Please refer to Figure 1 , a fire pump room operation management method provided in this application includes the following steps:

[0040] S110: Obtain at least the image information of the fire pump room area.

[0041] S120: Processing the fire pump room area image information using a fire pump monitoring model to generate fire pump room area monitoring information.

[0042] Furthermore, the fire water pump detection model is generated by pre-training, specifically including: constructing an initial fire water pump monitoring model based on the fire water pump room maintenance operation knowledge graph and template library; training the initial fire water pump detection model to generate a final fire water pump monitoring model.

[0043] Furthermore, the fire water pump room maintenance operation knowledge graph is pre-set, specifically including: obtaining a keyword list of knowledge terminal nodes, extracting associated knowledge point segment slices based on the keyword list, and generating a fire water pump room maintenance operation knowledge graph based on the keyword list and the associated knowledge point segment slices.

[0044] Specifically, the keyword list for knowledge edge nodes is pre-established. First, domain experts are engaged to collect and maintain a keyword list that covers all important concepts and terms in the domain. Next, natural language processing tools, such as part-of-speech tagging and entity recognition, are used to automatically update and expand the keyword list. Finally, formal definitions and context for the keywords are determined to ensure accuracy and consistency.

[0045] In this embodiment, the natural language processing tool adopts NLTK (Natural Language Toolkit). It should be noted that the protection scope of the present invention is not limited to this. Other natural language processing tools, such as Spacy, Gensim, Stanford CoreNLP, WordNet, TextBlob, BERT, AllenNLP, etc. are all within the protection scope of the present invention.

[0046] Extracting related knowledge point segment slices based on the keyword list specifically includes: analyzing the text corpus and extracting combinations of knowledge points that frequently appear together. Next, using topic modeling techniques (such as LDA) to identify and separate segments and establish the associations between them. Finally, creating a segment index and marking strongly associated knowledge points for rapid retrieval and integration. It can be understood that the text corpus can be understood as a list of knowledge points.

[0047] Based on the keyword list and associated knowledge point segment slices, a knowledge graph for fire pump room maintenance operations is generated. Specifically, once the keyword list and associated knowledge point segment slices are generated, the knowledge graph for fire pump room maintenance operations can be constructed. It should be noted that in this embodiment, when constructing the knowledge graph, graph database technology is used to store and manage the relationships between different knowledge points. This allows for capturing long-span, complex relationships, enabling efficient query and reasoning mechanisms across multiple knowledge points to solve complex problems.

[0048] The knowledge graph construction process described in this embodiment uses natural language processing tools to process the keyword list, which can fully expand the keyword list and greatly reduce the content required for knowledge graph construction. This can compensate for the lack of knowledge materials in the field of fire pump room maintenance operations and greatly reduce data processing time. In addition, by using topic modeling technology to establish strongly associated knowledge point and paragraph slices, the knowledge graph has high accuracy.

[0049] Furthermore, after the fire pump room maintenance knowledge graph is completed, the template library is also improved. The template library is pre-set and specifically includes: obtaining an initial template library; based on the fire pump room maintenance knowledge graph, processing the initial template library to generate a final template library.

[0050] Specifically, the initial template library collects common questions and answer patterns in the industry to form a standardized template library, and then uses natural language generation (NLG) technology to automatically generate answers that conform to the template to ensure the accuracy and consistency of the content.

[0051] On the basis of the above, the knowledge graph of fire pump room maintenance operations and the initial template library are combined to dynamically select and fill in the most suitable template according to the input question. It should be noted that the template library recorded in this embodiment is pre-defined according to the usage scenario, and the appropriate judgment criteria are determined by the "usage scenario". For example, if the user asks to "write a network security report", the system will extract the network report template from the template library. This extraction process mainly involves feature annotation of user questions and comparing them with manually annotated feature questions to standardize user questions. When the question is classified as a security report, the system will take out the template library stored in the knowledge graph. The two are combined, rather than looking for one and then the other.

[0052] Furthermore, the process of processing the initial template library to generate the final template library also includes: using an accuracy evaluation model to process the initial template library to generate the final template library.

[0053] Specifically, the accuracy assessment model is built based on statistical analysis and machine learning techniques to determine the effectiveness and weight of prompt words. This accuracy assessment model identifies and flags uncertain and ambiguous keywords and phrases. Users can also set thresholds and logic rules to provide reasonable rejection or improvement feedback for issues identified as uncertain. By using accuracy assessment and feedback to project prompt words, uncertainty can be detected and addressed, ensuring the accuracy and credibility of the output information.

[0054] Specifically, after completing the generation of the fire water pump room maintenance operation knowledge graph and template library, an initial fire water pump monitoring model is constructed based on this. Specifically, the fire water pump detection model includes at least several target recognition sub-models and anomaly detection sub-models.

[0055] It should be noted that in specific industrial scenarios, there isn't a single target, but rather a series of time-sequenced action recognitions. For example, to identify the action of connecting an oil pipeline, we need to identify the worker extracting the pipeline, climbing a ladder, connecting the pipeline, and then climbing down the ladder. The recognition of different actions (targets) requires separate training; it forms a collection. The fire pump monitoring model is the business agent that runs the entire inspection action. It is responsible for connecting the entire inspection process, performing a workflow loop inspection, and completing multimodal state recognition based on the output of the target recognition algorithm and other system inputs to ensure that people and objects are in a safe state. Multiple target recognition sub-models and anomaly detection sub-models are applied separately to achieve different detection tasks.

[0056] It should also be noted that the fire pump detection model in this embodiment can use a convolutional neural network (CNN), a recurrent neural network (RNN), etc., to achieve the technical effects to be achieved by this embodiment. In this embodiment, the anomaly detection sub-model uses a clustering algorithm and a time series data processing model. The clustering algorithm includes but is not limited to k-means, DBSCAN, etc., and the time series data processing model includes but is not limited to a long short-term memory network (LSTM).

[0057] It should also be noted that after the initial fire water pump monitoring model is constructed, the initial fire water pump detection model is trained to generate a final fire water pump monitoring model.

[0058] The initial fire pump detection model is trained using a training set that includes one or more of images, videos, and sounds. Specifically, training data is collected and prepared, including multiple modal data such as images, videos, and sounds, and annotated.

[0059] Use the large-scale datasets collected above to train sub-models for various functions, and optimize the models to adapt to specific scenarios through transfer learning or adaptive learning. In addition, it is necessary to use historical data to train anomaly detection models so that they can identify situations that are inconsistent with known normal behavior.

[0060] S130: Determine whether the fire water pump room area monitoring information meets the preset conditions, and generate a corresponding management strategy when the preset conditions are met.

[0061] Specifically, the preset conditions are pre-set and can also be understood as detection rules for determining risk scenarios.

[0062] Specifically, we first design and establish a rule language or paradigm that allows natural language input to define detection rules. Next, we identify key prompt words and phrases as triggers for rule creation and modification. Finally, we use natural language processing (NLP) techniques, including grammatical parsing and semantic analysis, to parse the user-provided natural language rules. Of course, the preconditions can be dynamically generated from natural language input.

[0063] The following example shows an alarm judgment scenario:

[0064] The Large Language Model (LLM) uses the regular expression a&&b&&c to output judgments based on the alarm content.

[0065] If it is false, please answer: The alarm exists, please go to the Alarm tab to check it immediately.

[0066] If true, please answer: The current alarm does not exist, please pay attention.

[0067] Give the alarm content: a: No alarm in the control cabinet = false; b: No alarm in the power cabinet = true; c: No alarm in the water pump = false;

[0068] or:

[0069] The hall alarm status is a set, where each value consists of "sensor number: Boolean value". For example, sensor number 3 is expressed as 3:true. When there are four sensors in total, the overall expression is {1:true, 2:false; 3:true; 4:true}.

[0070] If all the values in the set are true, the answer is: No alarm in the control cabinet = true.

[0071] If false appears in the set, the answer is: no alarm in the control cabinet = false.

[0072] Give an answer based on the following control cabinet alarm situation: {1: true, 2: false; 3: true; 4: true}.

[0073] Furthermore, the setting of the preset conditions also includes verification using similarity identification technology, and performing a secondary confirmation on the detection rule or paradigm using similarity identification technology.

[0074] Specifically, a similarity calculation algorithm (such as the cosine similarity algorithm) is first used to reconfirm and verify the natural language rules. Next, the natural language rules are converted into a standardized representation in the rule base for comparison and matching. Finally, a semantic similarity model, such as BERT or WordEmbeddings, is used to evaluate the similarity between the natural language rules and existing rules.

[0075] Furthermore, the test of the preset conditions specifically includes: using a prefabricated test set to test the accuracy of the preset conditions; and evaluating the accuracy performance of the preset conditions through the accuracy.

[0076] Specifically, a prefabricated test set is used to test the accuracy of the preset conditions, which specifically includes: testing the preset conditions through an automated test model, comparing the generated test results with the preset test set, and generating the accuracy of the preset conditions; the preset test set includes several scenario data, and the several scenario data are written with test cases and expected outputs.

[0077] A pre-built test suite is a set of pre-built tests that covers various scenarios and contexts, with test cases and expected outputs covering different types of detection rules or patterns. The accuracy and effectiveness of these detection rules or patterns are evaluated by executing them using an automated testing framework and comparing the results with the pre-built test suite.

[0078] The acquired fire water pump room area image information is processed through the fire water pump monitoring model to generate fire water pump room area monitoring information. Through judgment, it is determined whether the fire water pump room area monitoring information meets the preset conditions. If it meets the preset conditions, the corresponding management strategy is generated.

[0079] It should be noted that the output of the fire pump monitoring model is an information fusion of the detection results of each sub-model. The decision algorithm comprehensively judges the risk situation and takes corresponding preventive measures. In this embodiment, the decision algorithm uses a voting system or weighted average method to obtain the final similarity score.

[0080] The fire pump room maintenance management method provided in this embodiment leverages a specialized knowledge base and knowledge graph within the field to ensure that the large model accurately provides information. Furthermore, by defining detection rules using natural language, the utilization of both physical and human defenses can be maximized. Finally, by using computer vision and large models to achieve multimodal recognition of risk scenarios, known risks can be effectively prevented, and residual risks can be identified through anomaly recognition, enabling continuous improvement of management systems.

[0081] This embodiment also provides a management device for fire pump room maintenance operations, including:

[0082] The acquisition module 10 is used to acquire at least the image information of the fire pump room area. Since it has been described in detail in step S110 of the above-mentioned method for managing fire pump room operations, it will not be repeated here.

[0083] The processing module 20 is used to process the fire pump room area image information using the fire pump monitoring model to generate fire pump room area monitoring information. Since it has been described in detail in step S120 of the above-mentioned fire pump room operation management method, it will not be repeated here.

[0084] The judgment execution module 30 is used to judge whether the fire water pump room area monitoring information meets the preset conditions, and when the preset conditions are met, generate a corresponding management strategy. Since step S130 of the fire water pump room operation management method has been described in detail, it will not be repeated here.

[0085] In addition, an embodiment of the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium may store a program, which, when executed, includes part or all of the steps of any one of the fire water pump room maintenance management methods described in the above method embodiments.

[0086] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0087] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a memory and includes several instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned memory includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk, and other media that can store program code.

[0088] Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program. The program can be stored in a computer-readable memory, and the memory can include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0089] The above description, with reference to the accompanying drawings, illustrates an exemplary flowchart for managing fire pump room maintenance operations according to an embodiment of the present invention. It should be noted that the numerous details included in the above description are merely illustrative of the present invention and are not intended to limit the present invention. In other embodiments of the present invention, the method may include more, fewer, or different steps, and the order, inclusion, functionality, and other relationships between the steps may differ from those described and illustrated.

Claims

1. A method for managing fire pump room operations, characterized in that: The following steps are involved: At least obtain image information of the fire pump room area; The fire pump monitoring model is used to process the image information of the fire pump room area to generate the fire pump room area monitoring information; Determine whether the fire water pump room area monitoring information meets the preset conditions, and generate a corresponding management strategy when it meets the preset conditions.

2. The fire pump room operation management method according to claim 1, characterized in that: The fire pump detection model is generated by pre-training and specifically includes: Build an initial fire pump monitoring model based on the fire pump room maintenance operation knowledge graph and template library; The initial fire pump detection model is trained to generate a final fire pump monitoring model.

3. The fire pump room operation management method according to claim 2, characterized in that: The fire water pump room maintenance operation knowledge graph is pre-set, specifically including: obtaining a keyword list of knowledge terminal nodes, extracting associated knowledge point segment slices based on the keyword list, and generating a fire water pump room maintenance operation knowledge graph based on the keyword list and the associated knowledge point segment slices.

4. The method for managing fire pump room operations according to claim 3, wherein: The template library is pre-set and specifically includes: Get the initial template library; Based on the fire water pump room maintenance operation knowledge graph, the initial template library is processed to generate a final template library.

5. The method for managing fire water pump room operations according to claim 3, characterized in that: The process of processing the initial template library to generate the final template library also includes: The initial template library is processed using an accuracy assessment model to generate a final template library.

6. The fire pump room operation management method according to claim 1, characterized in that: The fire pump detection model includes at least several target recognition sub-models and anomaly detection sub-models.

7. The fire pump room operation management method according to claim 1, characterized in that: The setting of the preset conditions also includes verification using similarity identification technology.

8. The fire pump room operation management method according to claim 1, characterized in that: It also includes a test of the preset conditions, specifically including: Using a prefabricated test set to test the accuracy of the preset conditions; The accuracy of the preset conditions is evaluated through the accuracy rate.

9. The fire pump room operation management method according to claim 1, characterized in that: The accuracy of the preset conditions is tested using a pre-made test set, specifically including: Testing the preset conditions through an automated test model, comparing the generated test results with a preset test set, and generating an accuracy rate for the preset conditions; The preset test set includes a number of scenario data, wherein the scenario data contains test cases and expected outputs.

10. A management device for fire pump room operations, characterized in that: Specifically include: An acquisition module, configured to acquire at least image information of a fire pump room area; A processing module is used to process the image information of the fire pump room area using a fire pump monitoring model to generate fire pump room area monitoring information; The judgment execution module is used to judge whether the fire water pump room area monitoring information meets the preset conditions, and generate a corresponding management strategy when the preset conditions are met.

Images