Intelligent equipment management system based on large model

Abstract

The invention discloses an intelligent equipment management system based on a large model, and belongs to the technical field of computers. Through the technical scheme provided by the embodiment of the invention, the related information of the multiple devices in the target space can be utilized to judge the target event, determine the target device and determine the device control data of the target device, and finally, the device control data is utilized to control the target device to process the target event. The information related to multiple devices is fully utilized, the efficiency of processing the target event is greatly improved, and the safety and stability of the target space are improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an intelligent equipment management system based on a large model. Background Art

[0002] With the development of society, the number of equipment arranged in space is increasing to meet the needs of different situations. For example, in the field of fire protection, more and more fire protection equipment is arranged in space to meet the fire protection needs of different situations.

[0003] However, events that may affect space safety or stability may occur in space, and how to use equipment in space to handle such events is a hot topic of research. Summary of the invention

[0004] The embodiment of the present application provides a large model-based intelligent equipment management system, which can realize intelligent equipment management and control, so as to handle events that affect the safety or stability of the space. The technical solution is as follows:

[0005] On the one hand, a device intelligent management system based on a large model is provided, the system comprising:

[0006] A data acquisition module, used to acquire device operation data, device information and environmental data of the environment of multiple devices, wherein the device information includes device maintenance information, device operation log and device function information, and the multiple devices are deployed at different locations in the target space;

[0007] A target event identification module is used to determine whether there is a target event in the target space based on the device operation data of the multiple devices, the environmental data of the environment, the device maintenance information and the device operation log through a large model, wherein the target event is an event that affects the safety or stability of the target space;

[0008] a determination module, configured to determine, when a target event exists in the target space, at least one target device from the multiple devices through the large model based on the target event and device function information of the multiple devices, and device control data of each target device based on device operation data, device function information, environmental data of the environment in which the target device is located, and the target event, wherein the target device is a device associated with the target event;

[0009] The device control module is used to control the corresponding target device using the device control data to process the target event.

[0010] In one possible implementation, the target event identification module is used to determine, through the big model and based on the device maintenance information and device operation logs of the multiple devices, multiple candidate devices from the multiple devices, where the multiple candidate devices are devices whose credibility meets the credibility conditions; and determine, through the big model and based on the device operation data of the multiple candidate devices and the environmental data of the environment in which they are located, whether there is a target event in the target space.

[0011] In a possible implementation, the target event identification module is used to determine, for any one of the multiple devices, a first credibility of the device based on the device maintenance information of the device through the big model; determine a second credibility of the device based on the device operation log of the device through the big model; determine the credibility of the device based on the first credibility and the second credibility of the device through the big model; and determine the device among the multiple devices whose credibility is greater than or equal to the credibility threshold as the candidate device.

[0012] In a possible implementation, the target event identification module is used to input the equipment maintenance information of the device into the big model, and perform feature extraction on the equipment maintenance information through the big model to obtain the equipment maintenance feature of the device; map the equipment maintenance feature of the device through the big model to obtain the first credibility and first weight of the device; input the equipment operation log of the device into the big model, and perform feature extraction on the equipment operation log through the big model to obtain the operation log feature of the device; map the operation log feature of the device through the big model to obtain the second credibility and second weight of the device; and fuse the first credibility and the second credibility using the first weight and the second weight to obtain the credibility of the device.

[0013] In a possible implementation, the target event identification module is used to determine a plurality of first candidate events through the big model based on the device operation data of the plurality of candidate devices, wherein the first candidate events are events that may be caused by device factors; determine a plurality of second candidate events through the big model based on the environmental data of the environments in which the plurality of candidate devices are located, wherein the second candidate events are events that may be caused by environmental factors; and determine whether there is a target event in the target space based on the device operation data of the plurality of candidate devices, the environmental data of the environments in which they are located, the plurality of first candidate events, and the plurality of second candidate events.

[0014] In a possible implementation, the target event identification module is used to screen the multiple first candidate events and the multiple second candidate events to obtain multiple third candidate events, where the third candidate events are events that may be caused by device factors and environmental factors; based on the device operation data of the multiple candidate devices and the environmental data of the environment in which they are located, determine the event existence probability of each of the third candidate events; when the event existence probability of any third candidate event among the multiple third candidate events is greater than or equal to a probability threshold, determine that there is a target event in the target space and the target event is the third candidate event; when the event existence probabilities of the multiple third candidate events are all less than the probability threshold, determine that there is no target event in the target space.

[0015] In a possible implementation, the determination module is configured to determine, through the large model and based on device function information of the multiple devices, at least one reference event corresponding to each of the devices, where the reference event is an event that can be processed by the corresponding device; match the target event with at least one reference event corresponding to each of the devices to obtain the at least one target device, where the at least one reference event corresponding to the target device includes the target event;

[0016] The determination module is used to determine the target environment data of the target space through the big model based on the environmental data of the environment in which each of the target devices is located and the target event, wherein the target environment data is the predicted environment data after the target event in the target space is processed; and to determine the device control data of each of the target devices through the big model based on the target environment data, the environmental data, the device operation data of each of the target devices and the device function information.

[0017] In a possible implementation, the determination module is used to input the device function information of any one of the multiple devices into the big model, and perform feature extraction on the device function information of the device through the big model to obtain functional characteristics of the device; and perform multiple rounds of iterative decoding on the device function characteristics of the device through the big model to obtain at least one reference event corresponding to the device.

[0018] In a possible implementation, the determination module is used to determine the environmental impact parameters based on the target event through the large model; and to determine the target environmental data based on the environmental impact parameters and the environmental data through the large model.

[0019] In a possible implementation, the determination module is used to determine, for any target device among the at least one target device, the control data range of the target device based on the device operation data and device function information of the target device through the big model; and to determine, through the big model, the device control data of the target device from the control data range based on the target environment data and the environment data.

[0020] On the one hand, a method for intelligent equipment management based on a large model is provided, the method comprising:

[0021] Acquire device operation data, device information, and environmental data of the environment in which multiple devices are located, wherein the device information includes device maintenance information, device operation log, and device function information, and the multiple devices are deployed at different locations in the target space;

[0022] By means of a large model, based on the device operation data of the multiple devices, the environmental data of the environment in which they are located, the device maintenance information and the device operation log, it is determined whether there is a target event in the target space, wherein the target event is an event that affects the safety or stability of the target space;

[0023] In the case where a target event exists in the target space, at least one target device is determined from the multiple devices through the large model based on the target event and the device function information of the multiple devices, and device control data of each target device is determined based on the device operation data, device function information, environmental data of the environment in which the target device is located, and the target event, wherein the target device is a device associated with the target event;

[0024] The device control data is used to control a corresponding target device to process the target event.

[0025] In one possible implementation, through the big model, based on the device maintenance information and device operation logs of the multiple devices, multiple candidate devices are determined from the multiple devices, and the multiple candidate devices are devices whose credibility meets the credibility conditions; through the big model, based on the device operation data of the multiple candidate devices and the environmental data of the environment in which they are located, it is determined whether there is a target event in the target space.

[0026] In a possible implementation, for any device among the multiple devices, the first credibility of the device is determined based on the device maintenance information of the device through the big model; the second credibility of the device is determined based on the device operation log of the device through the big model; the credibility of the device is determined based on the first credibility and the second credibility of the device through the big model; and the device among the multiple devices whose credibility is greater than or equal to the credibility threshold is determined as the candidate device.

[0027] In a possible implementation, the equipment maintenance information of the device is input into the big model, and the big model is used to extract features from the equipment maintenance information to obtain equipment maintenance features of the device; the equipment maintenance features of the device are mapped through the big model to obtain a first credibility and a first weight of the device; the equipment operation log of the device is input into the big model, and the big model is used to extract features from the equipment operation log to obtain operation log features of the device; the operation log features of the device are mapped through the big model to obtain a second credibility and a second weight of the device; the first credibility and the second credibility are merged using the first weight and the second weight to obtain the credibility of the device.

[0028] In one possible implementation, through the big model, based on the device operation data of the multiple candidate devices, multiple first candidate events are determined, and the first candidate events are events that may be caused by device factors; through the big model, based on the environmental data of the environment in which the multiple candidate devices are located, multiple second candidate events are determined, and the second candidate events are events that may be caused by environmental factors; based on the device operation data of the multiple candidate devices, the environmental data of the environment in which they are located, the multiple first candidate events and the multiple second candidate events, it is determined whether there is a target event in the target space.

[0029] In a possible implementation, the multiple first candidate events and the multiple second candidate events are screened to obtain multiple third candidate events, wherein the third candidate events are events that may be caused by device factors and environmental factors; based on the device operation data of the multiple candidate devices and the environmental data of the environment in which they are located, the event existence probability of each of the third candidate events is determined; when the event existence probability of any third candidate event among the multiple third candidate events is greater than or equal to a probability threshold, it is determined that there is a target event in the target space and the target event is the third candidate event; when the event existence probabilities of the multiple third candidate events are all less than the probability threshold, it is determined that there is no target event in the target space.

[0030] In a possible implementation, at least one reference event corresponding to each of the devices is determined based on the device function information of the multiple devices through the large model, where the reference event is an event that can be processed by the corresponding device; the target event is matched with the at least one reference event corresponding to each of the devices to obtain the at least one target device, where the at least one reference event corresponding to the target device includes the target event;

[0031] Through the big model, based on the environmental data of the environment in which each of the target devices is located and the target event, the target environmental data of the target space is determined, and the target environmental data is the predicted environmental data after the target event in the target space is processed; through the big model, based on the target environmental data, the environmental data, the device operation data of each of the target devices and the device function information, the device control data of each of the target devices is determined.

[0032] In a possible implementation, for any one of the multiple devices, the device function information of the device is input into the big model, and the device function information of the device is feature extracted through the big model to obtain the functional characteristics of the device; the device function characteristics of the device are iteratively decoded for multiple rounds through the big model to obtain at least one reference event corresponding to the device.

[0033] In a possible implementation, the environmental impact parameters are determined based on the target event through the large model; and the target environmental data are determined based on the environmental impact parameters and the environmental data through the large model.

[0034] In a possible implementation, for any target device among the at least one target device, the control data range of the target device is determined through the big model based on the device operation data and device function information of the target device; and the device control data of the target device is determined from the control data range through the big model based on the target environment data and the environment data.

[0035] On the one hand, a computer device is provided, which includes one or more processors and one or more memories, wherein at least one computer program is stored in the one or more memories, and the computer program is loaded and executed by the one or more processors to implement the large model-based device intelligent management method.

[0036] On the one hand, a computer-readable storage medium is provided, in which at least one computer program is stored. The computer program is loaded and executed by a processor to implement the large model-based device intelligent management method.

[0037] Through the technical solution provided in the embodiment of the present application, it is possible to utilize the relevant information of multiple devices in the target space to judge the target event, determine the target device, and determine the device control data of the target device. Finally, the device control data can be used to control the target device to process the target event. By making full use of the information related to multiple devices, the efficiency of processing the target event is greatly improved, and the security and stability of the target space are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0039] Figure 1 It is a structural diagram of a device intelligent management system based on a large model provided in an embodiment of the present application;

[0040] Figure 2 This is a flowchart for determining whether a target event exists, provided in an embodiment of the present application;

[0041] Figure 3 is a flow chart for determining device control data provided by an embodiment of the present application;

[0042] Figure 4 It is a flow chart of a method for intelligent equipment management based on a large model provided in an embodiment of the present application. DETAILED DESCRIPTION

[0043] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0044] In this application, the terms "first", "second", etc. are used to distinguish identical or similar items with substantially the same effects and functions. It should be understood that there is no logical or temporal dependency between "first", "second", and "nth", nor is there any limitation on quantity and execution order.

[0045] Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a similar way to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines so that machines have the functions of perception, reasoning and decision-making.

[0046] Artificial intelligence technology is a comprehensive discipline that covers a wide range of fields, including both hardware-level and software-level technologies. The basic technologies of artificial intelligence generally include sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics and other technologies. Artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0047] Machine Learning (ML) is a multi-disciplinary subject that involves probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and other disciplines. It specializes in studying how computers simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge sub-models to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications are spread across all areas of artificial intelligence. Machine learning and deep learning usually include artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and teaching learning.

[0048] Normalization: Mapping sequences with different value ranges to the interval (0, 1) facilitates data processing. In some cases, the normalized values ​​can be directly implemented as probabilities.

[0049] Learning Rate: It is used to control the learning progress of the model. The learning rate can guide the model in how to use the gradient of the loss function to adjust the network weights in the gradient descent method. If the learning rate is too large, the loss function may directly cross the global optimal point, which is manifested as excessive loss; if the learning rate is too small, the loss function changes very slowly, which will greatly increase the convergence complexity of the network and it is easy to be trapped in the local minimum or saddle point.

[0050] Embedded Coding: Embedded coding mathematically represents a correspondence, that is, mapping data in X space to Y space through a function F, where the function F is an injective function, and the mapping result is structural preservation. The injective function indicates that the data after mapping is uniquely corresponding to the data before mapping, and structural preservation indicates that the size relationship of the data before mapping is the same as the size relationship of the data after mapping. For example, there are data X1 and X2 before mapping, and after mapping, Y1 corresponding to X1 and Y2 corresponding to X2 are obtained. If the data X1 before mapping is greater than X2, then correspondingly, the data Y1 after mapping is greater than Y2. For words, it means mapping the words to another space to facilitate subsequent machine learning and processing.

[0051] Attention weight: It can indicate the importance of a certain data in the training or prediction process. Importance indicates the influence of input data on output data. Data with high importance has a higher corresponding attention weight value, and data with low importance has a lower corresponding attention weight value. In different scenarios, the importance of data is different. The process of training the attention weight of the model is also the process of determining the importance of data.

[0052] The following is an explanation of the equipment intelligent management system based on the large model provided in the embodiment of the present application. Figure 1 The system includes a data acquisition module 101, a target event recognition module 102, a determination module 103 and a device control module 104.

[0053] The data acquisition module 101 is used to acquire device operation data, device information and environmental data of the environment in which multiple devices are located. The device information includes device maintenance information, device operation log and device function information. The multiple devices are deployed in different locations of the target space.

[0054] Among them, the target space is any selected space, for example, the target space is a space with fire protection needs, for example, the target space is a space such as a shopping mall, a school, and a hospital, and accordingly, the multiple devices refer to equipment related to fire protection. Alternatively, the target space is an agricultural space, for example, the target space is a smart greenhouse, and one or more crops are planted in the target space, and accordingly, the multiple devices refer to equipment related to agriculture. Alternatively, the target space is a space where people gather, for example, the target space is a scenic spot or an amusement park, and accordingly, the multiple devices refer to equipment related to crowd management or crowd guidance. Equipment operation data is used to describe the operation of the equipment, and the content contained in the equipment operation data is associated with the type of equipment. Equipment information refers to information directly related to the equipment, equipment maintenance information is used for the maintenance of the equipment, equipment operation log is used to record the historical operation of the equipment, and equipment function information is used to indicate the functions of the equipment. Environmental data of the environment in which the equipment is located is used to indicate the environmental conditions of the environment in which the equipment is located. Generally speaking, environmental data is obtained through environmental sensors on the equipment.

[0055] The target event identification module 102 is used to determine whether there is a target event in the target space through a large model based on the device operation data of the multiple devices, the environmental data of the environment, the device maintenance information and the device operation log. The target event is an event that affects the safety or stability of the target space.

[0056] Among them, the large model is a model obtained after pre-training and instruction fine-tuning. The large model refers to the large model parameters and the large amount of data used in training. In the embodiment of the present application, the large model has data processing capabilities. The target event has different meanings in different scenarios. When the target space is a space with fire protection needs, the target event is a fire. When the target space is an agricultural space, the target event is an event that affects the growth of crops, for example, the target event is water shortage, high temperature, low temperature or lack of fertilizer, etc., which is not limited in the embodiment of the present application. When the target space is a space where people gather, the target event is an event that may cause crowd congestion and exceed the carrying capacity of the facilities.

[0057] In order to explain the above implementation more clearly, the functions of the target event recognition module in the above implementation are introduced in the form of steps. Figure 2 , the target event recognition module is used to perform the following steps.

[0058] 201. Using the large model, based on the device maintenance information and device operation logs of the multiple devices, determine multiple candidate devices from the multiple devices, where the multiple candidate devices are devices whose credibility meets the credibility condition.

[0059] Among them, the credibility that meets the credibility condition means that the credibility is high, and correspondingly, the credibility that does not meet the credibility condition means that the credibility is low. The credibility condition is set by the technical personnel according to the actual situation, and the embodiment of the present application does not limit this.

[0060] In a possible implementation, for any device among the multiple devices, the first credibility of the device is determined based on the device maintenance information of the device through the large model. The second credibility of the device is determined based on the device operation log of the device through the large model. The credibility of the device is determined based on the first credibility and the second credibility of the device through the large model. The device whose credibility is greater than or equal to the credibility threshold among the multiple devices is determined as the candidate device.

[0061] Among them, the first credibility is also called the equipment maintenance credibility, which is used to evaluate the credibility of the equipment from the perspective of equipment maintenance. From the dimension of the number of equipment maintenance, if the number of equipment maintenance is in the maintenance number interval corresponding to the equipment, the first credibility of the equipment is high. If the number of equipment maintenance is greater than the maintenance number interval corresponding to the equipment or less than the maintenance number interval corresponding to the equipment, the first credibility of the equipment is low. This is because if the number of equipment maintenance is greater than the maintenance number interval corresponding to the equipment, it means that the equipment may have a fault and need high-frequency maintenance, so the first credibility of the equipment is low. If the number of equipment maintenance is less than the maintenance number interval corresponding to the equipment, it means that the equipment may lack maintenance, so the first credibility of the equipment is also low. In some embodiments, the maintenance number interval corresponding to the equipment refers to the range of the number of times the equipment needs to be maintained when the equipment is operating normally. The second credibility is also called the equipment operation credibility, which is used to evaluate the credibility of the equipment from the perspective of equipment operation. The equipment operation log can reflect the operation of the equipment, so the second credibility of the equipment can be determined using the equipment operation log.

[0062] In this implementation, the first credibility of the device is determined from the perspective of device maintenance, and the second credibility of the device is determined from the perspective of device operation. The credibility of the device is determined using the first credibility and the second credibility, and the accuracy of the credibility is high, so the accuracy of screening candidate devices is also high.

[0063] For example, for any device among the multiple devices, the device maintenance information of the device is input into the large model, and the device maintenance information is extracted through the large model to obtain the device maintenance feature of the device. The device maintenance feature of the device is mapped through the large model to obtain the first credibility and first weight of the device. The device operation log of the device is input into the large model, and the device operation log is extracted through the large model to obtain the operation log feature of the device. The operation log feature of the device is mapped through the large model to obtain the second credibility and second weight of the device. The first credibility and the second credibility are fused using the first weight and the second weight to obtain the credibility of the device.

[0064] The first weight is used to indicate the importance of the first credibility, and the second weight is used to indicate the importance of the second credibility.

[0065] For example, for any device among the multiple devices, the device maintenance information of the device is input into the large model. Through the large model, the device maintenance information is extracted and fully connected multiple times to obtain the first device maintenance sub-feature and the second device maintenance sub-feature of the device. The first device maintenance sub-feature and the second device maintenance sub-feature constitute the device maintenance feature of the device. The first device maintenance sub-feature includes the maintenance features of the device under multiple maintenance dimensions. Through the large model, the first device maintenance sub-feature is mapped to obtain the first credibility of the device. Through the large model, the second device maintenance sub-feature is fully connected and normalized to obtain the first weight. The operation log of the device is input into the large model. Through the large model, information is extracted to obtain the abnormal log in the operation log. Through the large model, the abnormal log is encoded to obtain the device log feature of the device. Through the large model, the device log feature is fully connected and normalized to obtain the second credibility of the device. Through the large model, the device log feature is normalized to obtain the second weight. The first credibility and the second credibility are merged by using the first weight and the second weight to obtain the credibility of the device.

[0066] In some embodiments, the first maintenance sub-feature includes maintenance features under the dimension of maintenance times, maintenance features under the dimension of maintenance reasons, maintenance features under the dimension of maintainers, and maintenance features under the dimension of maintenance results.

[0067] 202. Determine whether there is a target event in the target space based on the device operation data of the multiple candidate devices and the environmental data of the environment in which they are located, using the large model.

[0068] In a possible implementation, through the large model, based on the device operation data of the multiple candidate devices, multiple first candidate events are determined, and the first candidate events are events that may be caused by device factors. Through the large model, based on the environmental data of the environment in which the multiple candidate devices are located, multiple second candidate events are determined, and the second candidate events are events that may be caused by environmental factors. Based on the device operation data of the multiple candidate devices, the environmental data of the environment in which they are located, the multiple first candidate events, and the multiple second candidate events, it is determined whether there is a target event in the target space.

[0069] The “possible” in the above implementation means that the possibility is greater than 0.

[0070] In order to explain the above implementation more clearly, the above implementation is explained in several parts below.

[0071] Part 1: Determine multiple first candidate events based on the device operation data of the multiple candidate devices through the large model.

[0072] In a possible implementation, the large model is used to determine the device risk events of each candidate device based on the operation data of each candidate device. The large model is used to determine multiple first candidate events based on the device risk events of each candidate device.

[0073] Among them, the equipment risk event is an event that may occur and cause equipment failure. The first candidate event is an event that may be caused by equipment factors, and equipment failure may affect the safety and stability of the target space. Therefore, the first candidate event can be determined by using the equipment risk event.

[0074] For example, for any candidate device among multiple candidate devices, the device operation data of the candidate device is input into the large model, and the device operation data of the candidate device is extracted through the large model to obtain the device operation feature of the candidate device. Through the large model, the device operation feature of the candidate device is matched with the risk event features of multiple candidate device risk events to obtain the device risk event of the candidate device, and the device risk event is a candidate risk event whose similarity between the risk event features in the multiple candidate device risk events and the device operation features of the candidate device is greater than or equal to the similarity threshold. Through the large model, feature extraction is performed on the device risk events and device operation data of each candidate device to obtain the target risk event features of the device risk events of each candidate device. The target risk event features of the device risk events of each candidate device are fused to obtain the candidate event prediction features. Through the large model, the candidate event prediction features are matched with the reference event features of multiple first reference events to obtain the multiple first candidate events, and the first candidate event is the first reference event whose feature similarity between the reference event features and the candidate event prediction features is greater than or equal to the similarity threshold.

[0075] The similarity threshold is set by a technician according to actual conditions, and the present application embodiment does not limit this. The multiple first reference events are optional events that may be caused by device factors.

[0076] Part 2: Using the large model, based on the environmental data of the environments in which the multiple candidate devices are located, multiple second candidate events are determined, and the second candidate events are events that may be caused by environmental factors.

[0077] In a possible implementation, the large model is used to determine the spatial environment data of the target space based on the environmental data of the environments in which the multiple candidate devices are located. The large model is used to extract features from the spatial environment data of the target space to obtain spatial environment features of the target space. The large model is used to determine the multiple second candidate events based on the spatial environment features of the target space.

[0078] For example, the large model is used to splice the environmental data of the environments in which the multiple candidate devices are located to obtain the spatial environmental data of the target space. The large model is used to perform multiple convolutions on the spatial environmental data of the target space to obtain the spatial environmental features of the target space. The large model is used to match the spatial environmental features of the target space with the reference event features of multiple second reference events to obtain the multiple second candidate events, where the second candidate events are second reference events whose feature similarity between the reference event features and the spatial environmental features of the target space is greater than or equal to the similarity threshold.

[0079] The plurality of second reference events are optional events that may be caused by environmental factors.

[0080] The third part determines whether there is a target event in the target space based on the device operation data of the multiple candidate devices, the environmental data of the environment in which they are located, the multiple first candidate events, and the multiple second candidate events.

[0081] In a possible implementation, the multiple first candidate events and the multiple second candidate events are screened to obtain multiple third candidate events, which are events that may be caused by device factors and environmental factors. Based on the device operation data of the multiple candidate devices and the environmental data of the environment in which they are located, the event existence probability of each third candidate event is determined. When the event existence probability of any third candidate event among the multiple third candidate events is greater than or equal to the probability threshold, it is determined that there is a target event in the target space and the target event is the third candidate event. When the event existence probabilities of the multiple third candidate events are all less than the probability threshold, it is determined that there is no target event in the target space.

[0082] Among them, the probability threshold is set by technical personnel according to actual conditions, and the embodiments of the present application do not limit this.

[0083] For example, the multiple first candidate events are screened to obtain multiple fourth candidate events, and the fourth candidate events are events that may also be caused by environmental factors among the multiple first candidate events. The multiple second candidate events are screened to obtain multiple fifth candidate events, and the fifth candidate events are events that may also be caused by equipment factors among the multiple second candidate events. The multiple fourth candidate events and the multiple fifth candidate events are compared to obtain the multiple third candidate events, and the third candidate events are candidate events that exist in both the multiple fourth candidate events and the multiple fifth candidate events. The multiple third candidate events, the equipment operation data of each candidate device, and the environmental data of the environment in which they are located are input into the large model, and the multiple third candidate events, the equipment operation data of each candidate device, and the environmental data of the environment in which they are located are input into the large model through the large model for processing, and the event existence probability of each third candidate event is obtained. When the event existence probability of any third candidate event among the multiple third candidate events is greater than or equal to the probability threshold, it is determined that the target event exists in the target space and the target event is the third candidate event. When the event existence probabilities of the multiple third candidate events are all less than the probability threshold, it is determined that the target event does not exist in the target space.

[0084] In order to more clearly illustrate the technical solution provided in the above example, the method of calculating the event existence probability of each third candidate event based on the large model in the above example is explained below.

[0085] In some embodiments, for any third candidate event among multiple third candidate events, the third candidate event, the device operation data of each candidate device, and the environmental data of the environment in which it is located are input into the large model, and the large model is used to extract features of the third candidate event, the device operation data of each candidate device, and the environmental data of the environment in which it is located to obtain candidate event features of the third candidate event, device operation features of each candidate device, and environmental features of the environment in which it is located. Through the large model, the candidate event features of the third candidate event, the device operation features of each candidate device, and the environmental features of the environment in which it is located are fused to obtain the probability determination features of the third candidate event. Through the large model, the probability determination features of the third candidate event are fully connected and normalized to obtain the event existence probability of the third event.

[0086] The function of the determination module 103 is described below.

[0087] The determination module 103 is used to determine, when there is a target event in the target space, at least one target device from the multiple devices based on the target event and the device function information of the multiple devices through the large model, and the device control data of each target device based on the device operation data, device function information, environmental data of the environment in which each target device is located, and the target event, and the target device is a device associated with the target event.

[0088] The device associated with the target event refers to a device that can handle the target event. For example, when the target event is a fire, the target device refers to a device that can handle the fire. Device control data refers to data used to control a device, and the device control data can be regarded as a control instruction for the device.

[0089] In order to explain the above implementation more clearly, the functions of the determination module in the above implementation are introduced in the form of steps. Figure 3 , the determination module is used to perform the following steps.

[0090] 301. Determine at least one reference event corresponding to each device based on the device function information of the multiple devices through the large model, where the reference event is an event that can be processed by the corresponding device. Match the target event with the at least one reference event corresponding to each device to obtain the at least one target device, where the at least one reference event corresponding to the target device includes the target event.

[0091] In a possible implementation, for any device among the multiple devices, the device function information of the device is input into the large model, and the device function information of the device is feature extracted through the large model to obtain the functional characteristics of the device. The device function characteristics of the device are iteratively decoded for multiple rounds through the large model to obtain at least one reference event corresponding to the device. The target event is matched with at least one reference event corresponding to each device to obtain the at least one target device.

[0092] The functional feature is a semantic feature of the device functional information.

[0093] For example, for any device among the multiple devices, the device function information of the device is input into the large model, and the large model encodes the device function information of the device based on the attention mechanism to obtain the device function characteristics of the device. Through the large model, the device function characteristics of the device are iteratively decoded for multiple rounds based on the attention mechanism to obtain at least one reference event corresponding to the device. The target event is compared with at least one reference event corresponding to each device, and the device that has the target event in at least one reference event corresponding to the multiple devices is determined as the target device.

[0094] 302. Determine target environment data of the target space based on the environment data of the environment of each target device and the target event through the large model, where the target environment data is the predicted environment data after the target event in the target space is processed. Determine device control data of each target device based on the target environment data, the environment data, device operation data of each target device, and device function information through the large model.

[0095] In a possible implementation, the environmental impact parameter is determined based on the target event through the large model. The target environmental data is determined through the large model based on the environmental impact parameter and the environmental data. For any target device among the at least one target device, the control data range of the target device is determined through the large model based on the device operation data and device function information of the target device. The device control data of the target device is determined from the control data range based on the target environmental data and the environmental data through the large model.

[0096] For example, the target event is input into the large model, and the target event is extracted through the large model to obtain the event feature of the target event. Through the large model, the event feature is fully connected and normalized to obtain the environmental impact parameter. The environmental impact parameter and the environmental data are input into the large model, and the environmental impact parameter and the environmental data are encoded through the large model to obtain the environmental prediction feature. The environmental prediction feature is decoded through the large model to obtain the target environmental data. For any target device among the at least one target device, the device operation data and device function information of the target device are input into the large model, and the device operation data and device function information of the target device are encoded and iteratively decoded for multiple rounds to obtain the control data range of the target device. Through the large model, based on the target environmental data and the environmental data, the environmental data change range is determined. Through the large model, the environmental data influence range corresponding to multiple control data in the control data range is determined. The environmental data influence ranges corresponding to the multiple control data are matched with the environmental data change ranges to obtain the device control data of the target device, and the device control data of the target device is the device control data with the largest overlap between the environmental data influence range and the environmental data change range.

[0097] The functions of the device control module 104 are introduced below.

[0098] The device control module 104 is used to control the corresponding target device using the device control data to process the target event.

[0099] In a possible implementation, the device control module 104 is configured to send the device control data to a corresponding target device, so that the target device executes an action corresponding to the device control data to process the target event.

[0100] Among them, the target device is associated with the target event. When the target event is fire, the target device refers to the fire extinguishing device, and the action corresponding to the target device refers to the fire extinguishing device performing the fire extinguishing action. The device control data can control the working parameters of the fire extinguishing device, such as controlling the amount of fire extinguishing medium used by the fire extinguishing device. When the target event is water shortage for crops, the target device refers to the irrigation equipment for crops, and the action corresponding to the target device refers to the irrigation equipment performing the irrigation action. The device control data can control the working parameters of the irrigation equipment, such as controlling the irrigation amount of the irrigation equipment.

[0101] All the above optional technical solutions can be arbitrarily combined to form optional embodiments of the present application, which will not be described one by one here.

[0102] Through the technical solution provided in the embodiment of the present application, it is possible to utilize the relevant information of multiple devices in the target space to judge the target event, determine the target device, and determine the device control data of the target device. Finally, the device control data can be used to control the target device to process the target event. By making full use of the information related to multiple devices, the efficiency of processing the target event is greatly improved, and the security and stability of the target space are improved.

[0103] Figure 4 This is a flow chart of a device intelligent management system based on a large model provided in an embodiment of the present application, see Figure 4 , taking the execution subject as an electronic device as an example, the method includes:

[0104] 401. Obtain device operation data, device information, and environmental data of the environments in which multiple devices are located. The device information includes device maintenance information, device operation logs, and device function information. The multiple devices are deployed at different locations in the target space.

[0105] 402. Determine whether there is a target event in the target space through a large model based on the device operation data of the multiple devices, the environmental data of the environment, the device maintenance information, and the device operation log. The target event is an event that affects the safety or stability of the target space.

[0106] In a possible implementation, through the large model, based on the device maintenance information and device operation logs of the multiple devices, multiple candidate devices are determined from the multiple devices, and the multiple candidate devices are devices whose credibility meets the credibility conditions. Through the large model, based on the device operation data of the multiple candidate devices and the environmental data of the environment in which they are located, it is determined whether there is a target event in the target space.

[0107] In a possible implementation, for any device among the multiple devices, the first credibility of the device is determined based on the device maintenance information of the device through the large model. The second credibility of the device is determined based on the device operation log of the device through the large model. The credibility of the device is determined based on the first credibility and the second credibility of the device through the large model. The device whose credibility is greater than or equal to the credibility threshold among the multiple devices is determined as the candidate device.

[0108] In a possible implementation, the device maintenance information of the device is input into the large model, and the device maintenance information is feature extracted through the large model to obtain the device maintenance feature of the device. The device maintenance feature of the device is mapped through the large model to obtain the first credibility and first weight of the device. The device operation log of the device is input into the large model, and the device operation log is feature extracted through the large model to obtain the operation log feature of the device. The operation log feature of the device is mapped through the large model to obtain the second credibility and second weight of the device. The first credibility and the second credibility are fused using the first weight and the second weight to obtain the credibility of the device.

[0109] In a possible implementation, through the large model, based on the device operation data of the multiple candidate devices, multiple first candidate events are determined, and the first candidate events are events that may be caused by device factors. Through the large model, based on the environmental data of the environment in which the multiple candidate devices are located, multiple second candidate events are determined, and the second candidate events are events that may be caused by environmental factors. Based on the device operation data of the multiple candidate devices, the environmental data of the environment in which they are located, the multiple first candidate events, and the multiple second candidate events, it is determined whether there is a target event in the target space.

[0110] In a possible implementation, the multiple first candidate events and the multiple second candidate events are screened to obtain multiple third candidate events, which are events that may be caused by device factors and environmental factors. Based on the device operation data of the multiple candidate devices and the environmental data of the environment in which they are located, the event existence probability of each of the third candidate events is determined. When the event existence probability of any third candidate event among the multiple third candidate events is greater than or equal to the probability threshold, it is determined that there is a target event in the target space and the target event is the third candidate event. When the event existence probabilities of the multiple third candidate events are all less than the probability threshold, it is determined that there is no target event in the target space.

[0111] 403. When there is a target event in the target space, at least one target device is determined from the multiple devices through the large model based on the target event and the device function information of the multiple devices, and the device control data of each target device is determined based on the device operation data, device function information, environmental data of the environment in which each target device is located, and the target event. The target device is a device associated with the target event.

[0112] In one possible manner, at least one reference event corresponding to each of the multiple devices is determined based on the device function information of the multiple devices through the large model, and the reference event is an event that the corresponding device can handle. The target event is matched with at least one reference event corresponding to each of the devices to obtain the at least one target device, and the at least one reference event corresponding to the target device includes the target event. Based on the environmental data of the environment in which each of the target devices is located and the target event, the target environment data of the target space is determined through the large model, and the target environment data is the predicted environment data after the target event in the target space is processed. Based on the target environment data, the environmental data, the device operation data of each of the target devices, and the device function information, the device control data of each of the target devices is determined through the large model.

[0113] In a possible manner, for any device among the multiple devices, the device function information of the device is input into the macro model, and the device function information of the device is feature extracted through the macro model to obtain the functional features of the device. The device function features of the device are iteratively decoded for multiple rounds through the macro model to obtain at least one reference event corresponding to the device.

[0114] In a possible manner, the large model is used to determine the environmental impact parameter based on the target event, and the large model is used to determine the target environmental data based on the environmental impact parameter and the environmental data.

[0115] In one possible manner, for any target device among the at least one target device, the control data range of the target device is determined based on the device operation data and device function information of the target device through the macro model. The device control data of the target device is determined from the control data range based on the target environment data and the environment data through the macro model.

[0116] 404. Use the device control data to control a corresponding target device to process the target event.

[0117] Through the technical solution provided in the embodiment of the present application, it is possible to utilize the relevant information of multiple devices in the target space to judge the target event, determine the target device, and determine the device control data of the target device. Finally, the device control data can be used to control the target device to process the target event. By making full use of the information related to multiple devices, the efficiency of processing the target event is greatly improved, and the security and stability of the target space are improved.

[0118] In an exemplary embodiment, a computer-readable storage medium is also provided, such as a memory including a computer program, and the computer program can be executed by a processor to implement the large model-based device intelligent management method in the above embodiment. For example, the computer-readable storage medium can be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, etc.

[0119] In an exemplary embodiment, a computer program product or a computer program is also provided, which includes a program code, and the program code is stored in a computer-readable storage medium. A processor of a computer device reads the program code from the computer-readable storage medium, and the processor executes the program code, so that the computer device executes the above-mentioned large model-based device intelligent management method.

[0120] In some embodiments, the computer program involved in the embodiments of the present application may be deployed and executed on a computer device, or on multiple computer devices located at one location, or on multiple computer devices distributed at multiple locations and interconnected by a communication network. Multiple computer devices distributed at multiple locations and interconnected by a communication network may constitute a blockchain system.

[0121] A person skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware or by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a disk or an optical disk, etc.

[0122] The above are only optional embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A large model-based equipment intelligent management system, characterized in that: The system comprises: A data acquisition module, used to acquire device operation data, device information and environmental data of the environment of multiple devices, wherein the device information includes device maintenance information, device operation log and device function information, and the multiple devices are deployed at different locations in the target space; A target event identification module is used to determine whether there is a target event in the target space based on the device operation data of the multiple devices, the environmental data of the environment, the device maintenance information and the device operation log through a large model, wherein the target event is an event that affects the safety or stability of the target space; a determination module, configured to determine, when a target event exists in the target space, at least one target device from the multiple devices through the large model based on the target event and device function information of the multiple devices, and device control data of each target device based on device operation data, device function information, environmental data of the environment in which the target device is located, and the target event, wherein the target device is a device associated with the target event; The device control module is used to control the corresponding target device using the device control data to process the target event.

2. The system according to claim 1, characterized in that The target event identification module is used to determine a plurality of candidate devices from the plurality of devices through the large model based on the device maintenance information and device operation logs of the plurality of devices, wherein the plurality of candidate devices are devices whose credibility meets the credibility condition; Through the large model, based on the device operation data of the multiple candidate devices and the environmental data of the surrounding environment, it is determined whether there is a target event in the target space.

3. The system according to claim 2, characterized in that The target event identification module is used to determine, for any device among the multiple devices, a first credibility of the device based on device maintenance information of the device through the large model; Determining, by means of the large model and based on a device operation log of the device, a second credibility of the device; Determining the credibility of the device based on the first credibility and the second credibility of the device by the large model; A device whose credibility is greater than or equal to a credibility threshold among the multiple devices is determined as the candidate device.

4. The system according to claim 3, characterized in that The target event identification module is used to input the equipment maintenance information of the device into the big model, and extract the features of the equipment maintenance information through the big model to obtain the equipment maintenance features of the device; map the equipment maintenance features of the device through the big model to obtain the first credibility and first weight of the device; input the equipment operation log of the device into the big model, and extract the features of the equipment operation log through the big model to obtain the operation log features of the device; map the operation log features of the device through the big model to obtain the second credibility and second weight of the device; and fuse the first credibility and the second credibility using the first weight and the second weight to obtain the credibility of the device.

5. The system according to claim 2, characterized in that The target event identification module is used to determine a plurality of first candidate events based on the device operation data of the plurality of candidate devices through the large model, wherein the first candidate events are events that may be caused by device factors; Determine, by means of the large model, a plurality of second candidate events based on environmental data of the environments in which the plurality of candidate devices are located, wherein the second candidate events are events that may be caused by environmental factors; Based on the device operation data of the multiple candidate devices, the environmental data of the surrounding environment, the multiple first candidate events, and the multiple second candidate events, it is determined whether there is a target event in the target space.

6. The system according to claim 5, characterized in that The target event identification module is used to screen the multiple first candidate events and the multiple second candidate events to obtain multiple third candidate events, where the third candidate events are events that may be caused by device factors and environmental factors; Determining the event existence probability of each of the third candidate events based on the device operation data of the multiple candidate devices and the environmental data of the environment in which they are located; When the event existence probability of any third candidate event among the multiple third candidate events is greater than or equal to the probability threshold, determining that there is a target event in the target space and the target event is the third candidate event; When the event existence probabilities of the plurality of third candidate events are all less than the probability threshold, it is determined that there is no target event in the target space.

7. The system according to claim 1, characterized in that The determination module is used to determine at least one reference event corresponding to each of the devices based on the device function information of the multiple devices through the large model, where the reference event is an event that can be processed by the corresponding device; Matching the target event with at least one reference event corresponding to each of the devices to obtain the at least one target device, wherein the at least one reference event corresponding to the target device includes the target event; The determination module is used to determine the target environment data of the target space through the big model based on the environmental data of the environment in which each of the target devices is located and the target event, wherein the target environment data is the predicted environment data after the target event in the target space is processed; and to determine the device control data of each of the target devices through the big model based on the target environment data, the environmental data, the device operation data of each of the target devices and the device function information.

8. The system according to claim 7, characterized in that The determination module is used to input the device function information of any device among the multiple devices into the big model, and extract features of the device function information of the device through the big model to obtain functional features of the device; and perform multiple rounds of iterative decoding on the device function features of the device through the big model to obtain at least one reference event corresponding to the device.

9. The system according to claim 7, characterized in that The determination module is used to determine the environmental impact parameters based on the target event through the large model; and to determine the target environmental data based on the environmental impact parameters and the environmental data through the large model.

10. The system according to claim 7, characterized in that The determining module is used to determine, for any target device among the at least one target device, a control data range of the target device based on the device operation data and device function information of the target device through the large model; The device control data of the target device is determined from the control data range based on the target environment data and the environment data by means of the large model.

Images