Gas concentration distribution data determination method and device and electronic equipment

Abstract

The invention discloses a gas concentration distribution data determination method and device and electronic equipment. The method comprises the following steps: acquiring gas concentration data corresponding to a plurality of local point locations in a predetermined space; according to the multiple pieces of gas concentration data, the incidence relation of the gas concentrations among the multiple preset point locations is determined, the preset space comprises the multiple preset point locations, and the multiple preset point locations comprise multiple local point locations; and determining gas concentration distribution data in the predetermined space according to the space parameters of the predetermined space and the association relationship. According to the invention, the technical problem that the gas distribution concentration in the space is difficult to accurately determine due to less sensor distribution in the prior art is solved.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to a method, device and electronic equipment for determining gas concentration distribution data. Background Art

[0002] In the related art, most of the existing methods for determining gas concentration distribution data are based on real-time measurement data of gas sensors. However, due to the small number of sensors distributed, there is a technical problem that it is difficult to accurately determine the gas distribution concentration in the space.

[0003] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0004] The embodiments of the present invention provide a method, device and electronic device for determining gas concentration distribution data, so as to at least solve the technical problem in the related art that sensors are sparsely distributed and it is difficult to accurately determine the gas distribution concentration in a space.

[0005] According to one aspect of an embodiment of the present invention, a method for determining gas concentration distribution data is provided, comprising: acquiring gas concentration data corresponding to multiple local points in a predetermined space; determining a correlation relationship between gas concentrations between multiple predetermined points based on the multiple gas concentration data, wherein the predetermined space includes the multiple predetermined points, and the multiple predetermined points include the multiple local points; determining the gas concentration distribution data in the predetermined space based on spatial parameters of the predetermined space and the correlation relationship.

[0006] Optionally, determining the correlation relationship between gas concentrations of multiple predetermined points based on multiple gas concentration data includes: determining one-dimensional concentration vectors corresponding to the multiple gas concentration data respectively; obtaining a high-dimensional concentration vector based on the multiple one-dimensional concentration vectors, wherein the dimension of the high-dimensional concentration vector is higher than the dimension corresponding to the multiple one-dimensional concentration vectors, and the high-dimensional concentration vector includes point feature vectors corresponding to the multiple predetermined points respectively; determining the weight corresponding to each point feature vector in the high-dimensional concentration vector based on the spatial position relationship between the multiple predetermined points and the gas concentration correlation relationship between the multiple local points; obtaining an attention-weighted feature vector based on the weight corresponding to each point feature vector in the high-dimensional concentration vector, wherein the attention-weighted feature vector represents the correlation relationship between the gas concentrations of the multiple predetermined points.

[0007] Optionally, determining the gas concentration distribution data in the predetermined space based on the spatial parameters of the predetermined space and the association relationship includes: determining the spatial feature vector corresponding to the spatial parameters; inputting the spatial feature vector corresponding to the spatial parameters and the attention-weighted feature vector into a linear layer to obtain a linear transformation vector, wherein the linear layer includes a predetermined weight matrix and a predetermined bias vector; inputting the linear transformation vector into an activation function to obtain a nonlinear vector, wherein the nonlinear vector weights the key feature vector and downgrades the non-key feature vector, the key feature vector is a feature vector whose corresponding importance index is greater than a predetermined threshold, and the non-key feature vector is a feature vector whose corresponding importance index is less than or equal to the predetermined threshold; combining the nonlinear vector with the attention-weighted feature vector to obtain a combined vector; and determining the gas concentration distribution data in the predetermined space based on the combined vector.

[0008] Optionally, determining the correlation relationship between the gas concentrations of multiple predetermined points based on multiple gas concentration data includes: acquiring flow field data in the predetermined space; and determining the correlation relationship between the gas concentrations of multiple predetermined points based on the flow field data and the multiple gas concentration data.

[0009] Optionally, based on multiple gas concentration data, the correlation relationship between the gas concentrations of multiple predetermined points is determined, including: determining first credible indexes corresponding to the multiple gas concentration data respectively based on a particle image velocity method, and determining second credible indexes corresponding to the multiple gas concentration data respectively based on a concentration experiment method; determining target credible indexes corresponding to the multiple gas concentration data respectively based on the multiple first credible indexes and the multiple second credible indexes; when multiple target credible indexes are all greater than a predetermined credible threshold, determining the correlation relationship between the gas concentrations of multiple predetermined points based on the multiple gas concentration data.

[0010] Optionally, after determining the gas concentration distribution data in the predetermined space based on the spatial parameters of the predetermined space and the association relationship, it also includes: when the gas concentration distribution data is greater than a predetermined concentration distribution threshold, determining the setting parameters corresponding to the safety equipment in the predetermined space; and controlling the safety equipment to be set to the setting parameters.

[0011] Optionally, after determining the gas concentration distribution data in the predetermined space based on the spatial parameters of the predetermined space and the association relationship, it also includes: determining a toxicity result of whether the gas is a toxic gas; when the toxicity result is that the gas is a toxic gas, determining a toxicity distribution threshold of the toxic gas; when the gas concentration distribution data is greater than the toxicity distribution threshold, determining a planned route corresponding to the predetermined space based on the gas concentration distribution data.

[0012] According to one aspect of an embodiment of the present invention, a device for determining gas concentration distribution data is provided, comprising: an acquisition module for acquiring gas concentration data corresponding to multiple local points in a predetermined space; a first determination module for determining a correlation relationship between gas concentrations between multiple predetermined points based on the multiple gas concentration data, wherein the predetermined space includes the multiple predetermined points, and the multiple predetermined points include the multiple local points; a second determination module for determining the gas concentration distribution data in the predetermined space based on the spatial parameters of the predetermined space and the correlation relationship.

[0013] According to one aspect of an embodiment of the present invention, there is provided an electronic device, comprising: a processor; and a memory for storing instructions executable by the processor; wherein the processor is configured to execute the instructions to implement any of the above-described methods for determining gas concentration distribution data.

[0014] According to one aspect of an embodiment of the present invention, a computer-readable storage medium is provided. When instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device can execute the above-mentioned method for determining gas concentration distribution data.

[0015] In an embodiment of the present invention, gas concentration data corresponding to multiple local points in a predetermined space are obtained; based on the multiple gas concentration data, the correlation relationship between the gas concentrations of the multiple predetermined points is determined, wherein the predetermined space includes multiple predetermined points, and the multiple predetermined points include multiple local points; based on the spatial parameters and the correlation relationship of the predetermined space, the gas concentration distribution data in the predetermined space is determined. By selecting multiple local points for collecting gas concentration data in the predetermined space, key gas concentration information can be effectively collected while the cost is controllable, thereby helping to provide an accurate data basis for the subsequent determination of gas concentration distribution data. By determining the correlation relationship between the gas concentrations of the multiple predetermined points based on the multiple gas concentration data, wherein the predetermined space includes multiple predetermined points, and the multiple predetermined points include multiple local points; by determining the correlation relationship between the gas concentrations of the multiple predetermined points based on the multiple gas concentration data, the diffusion and flow pattern of the gas in the space, as well as the mutual influence between the gas concentrations of different predetermined points, can be accurately reflected. Furthermore, the predetermined space includes multiple predetermined points, and the multiple predetermined points include multiple local points, so that there is no need to deploy gas concentration measurement devices at all locations, and global perception of gas concentration can be achieved under conditions of limited resources, thereby improving resource utilization efficiency. Based on the spatial parameters and correlation relationships of the predetermined space, the gas concentration distribution data in the predetermined space is determined, and the geometric characteristics of the predetermined space and the possible flow path of the gas are fully considered, and the influence on the gas concentration distribution is fully considered. By analyzing the combined correlation relationship, the gas distribution of the entire predetermined space can be inferred, even in areas that are not directly measured, thereby helping to accurately determine the global distribution of gas in the space, thereby solving the technical problem in the related technology that there are few sensors distributed and it is difficult to accurately determine the gas distribution concentration in the space. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0017] Figure 1 is a flow chart of a method for determining gas concentration distribution data according to an embodiment of the present invention;

[0018] Figure 2 is a diagram of a method for determining gas concentration distribution data provided by an optional embodiment of the present invention;

[0019] Figure 3 4 is a structural block diagram of a device for determining gas concentration distribution data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0021] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0022] First, some nouns or terms that appear in the description of the embodiments of the present application are subject to the following explanations:

[0023] Neural network: Neural network is an algorithmic structure designed to imitate the structure of the human brain. It consists of a large number of nodes (or "neurons") connected to each other. Each neuron can receive input, process it, and then output it to other neurons. The work of neural network can be divided into two main stages: forward propagation and back propagation. It occupies a core position in the field of machine learning and artificial intelligence, and can be applied to image recognition, speech processing, natural language processing and other fields.

[0024] MLP: MLP, or Multi-Layer Perceptron, is a common artificial neural network model and one of the earliest deep learning networks. Its structure includes multiple fully connected layers (that is, each neuron is connected to each neuron in the previous layer) and activation functions. MLP receives the input of raw data, performs weighted summation through neurons in the hidden layer, and then performs nonlinear transformation through the activation function, and finally generates the final prediction result of the model in the output layer. MLP can capture all feature combinations of input data and learn complex nonlinear functional relationships through multiple hidden layers.

[0025] Attention mechanism module: Attention mechanism is an important technology in the field of deep learning, especially in sequence models such as natural language processing (NLP) and computer vision. It enables the model to focus on important parts of the input data, thereby improving overall performance and efficiency. The attention mechanism enables the model to identify the most important information and focus on processing it while ignoring less relevant parts by assigning different weights (or attention scores) to different parts of the input data. This mechanism mimics the human attention process, improves the accuracy and efficiency of the model, and increases the transparency of the model.

[0026] Transformer module: Transformer module, or converter module, is a revolutionary architecture in the field of natural language processing (NLP). It is built entirely on the attention mechanism and abandons the traditional loop and convolution structure. It can comprehensively consider the degree of correlation between other elements in the input sequence and the current element, so as to more accurately understand the meaning of the input data. Transformer consists of multiple encoders and decoders, each of which contains components such as multi-head self-attention layers, feedforward neural network layers, residual connections, and layer normalization.

[0027] Particle Image Velocimetry: Particle Image Velocimetry (PIV) is a method that uses multiple cameras to record the position of particles in the flow field and analyzes the images to measure the flow velocity. The basic principle is to spread tracer particles in the flow field and use a pulsed laser light source to irradiate the measured flow field area. Through two or more consecutive exposures, the image of the particles is recorded on the film or CCD camera. Then, an optical method is used to process the PIV film or CCD recorded image point by point to obtain the flow field velocity distribution.

[0028] Concentration experiment method: Concentration experiment method is a method used to accurately measure and analyze the concentration of chemical substances in a laboratory or field environment. This method usually involves placing one or more chemical sensors under specific environmental conditions, which can monitor the concentration changes of target chemicals in real time. The steps of concentration experiment method include experimental design, which is to determine the layout of sensors, the parameters of the experimental environment and the monitoring time points; followed by data acquisition, in which the sensors record the concentration data of chemical substances at different time and spatial locations; followed by data processing, which cleans, converts and standardizes the collected concentration data in preparation for subsequent analysis and modeling; finally, the processed concentration data is used for model training or verification to evaluate the diffusion pattern of chemical substances, the effectiveness of decontamination strategies or the feasibility of emergency response measures.

[0029] Example 1

[0030] According to an embodiment of the present invention, an embodiment of a method for determining gas concentration distribution data is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0031] Figure 1 is a flow chart of a method for determining gas concentration distribution data according to an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:

[0032] S102, obtaining gas concentration data corresponding to a plurality of local points in a predetermined space;

[0033] In step S102 provided in the present application, gas concentration data corresponding to multiple local points in a predetermined space are obtained.

[0034] The predetermined space is a predetermined three-dimensional space with specific boundaries and structures, for example, a public indoor space where gas concentration needs to be measured.

[0035] Among them, local points are involved, which are specific locations selected in a predetermined space for installing gas concentration collection equipment to collect gas concentration data. The local points are sampling points in the entire space, through which gas concentration information in a local area of ​​the space can be obtained.

[0036] Among them, gas concentration data is involved, and the gas concentration data is data related to the concentration value of the gas collected at a local point.

[0037] By selecting multiple local points in a predetermined space for collecting gas concentration data, key gas concentration information can be effectively collected while keeping costs under control, thereby helping to provide an accurate data basis for subsequent determination of gas concentration distribution data, and further helping to accurately determine the gas distribution concentration in the predetermined space.

[0038] S104, determining a correlation relationship between gas concentrations of a plurality of predetermined points based on the plurality of gas concentration data, wherein the predetermined space includes the plurality of predetermined points, and the plurality of predetermined points include a plurality of local points;

[0039] In step S104 provided in the present application, the correlation relationship between the gas concentrations of a plurality of predetermined points is determined.

[0040] Among them, the correlation relationship is involved, which is the relationship between the gas concentrations collected at multiple predetermined points, which is used to reflect the correlation of gas concentration changes. For example, if a certain gas is released at one end of the predetermined space, by analyzing the changes in the concentration of a certain gas at the measured predetermined points, it can be inferred how the certain gas diffuses in the predetermined space.

[0041] Among them, predetermined points are involved, which are specific locations pre-selected in a predetermined space for collecting gas concentrations, and the predetermined points include multiple local points. For example, in a predetermined space, the predetermined points may include the four corners of the predetermined space, the center position, several points near the wall, and any possible vents or entrances. The selection of predetermined points can take into account the geometric characteristics of the predetermined space and the possible flow path of the gas to ensure the representativeness and comprehensiveness of the data.

[0042] By determining the correlation between gas concentrations of multiple pre-selected predetermined points for collecting gas concentrations based on multiple gas concentration data, the diffusion and flow patterns of gases in space, as well as the mutual influence between gas concentrations at different predetermined points, can be accurately reflected through this correlation. And by analyzing this correlation, the gas distribution of the entire predetermined space can be inferred, even in areas that are not directly measured, which helps to accurately determine the global distribution of gas in the space. And the predetermined space includes multiple predetermined points, and the multiple predetermined points include multiple local points, so that there is no need to deploy gas concentration measurement devices at all locations, so that global perception of gas concentration can be achieved under conditions of limited resources, thereby improving resource utilization efficiency.

[0043] S106, determining gas concentration distribution data in the predetermined space according to the spatial parameters and the correlation relationship of the predetermined space.

[0044] In step S106 provided in the present application, gas concentration distribution data in a predetermined space is determined.

[0045] Among them, space parameters are involved, which are parameters used to describe the geometric characteristics and physical properties of a predetermined space. The space parameters may include the size (length, width, height), shape, internal structure, location and size of ventilation holes, location and size of obstacles, etc. of the space.

[0046] Among them, gas concentration distribution data is involved, and the gas concentration distribution data is data used to reflect the distribution of gas concentration in the entire predetermined space.

[0047] By taking the spatial parameters of the predetermined space as the basis for analysis, the geometric characteristics of the predetermined space and the possible flow path of the gas are fully considered, and the impact on the gas concentration distribution is helpful for the comprehensiveness and accuracy of the subsequent gas concentration distribution. By combining the spatial parameters of the predetermined space and the correlation between the gas concentrations collected at multiple predetermined points, the gas distribution of the entire predetermined space can be inferred, even in areas that are not directly measured, which helps to accurately determine the global distribution of the gas in the space, thereby effectively solving the technical problem of less distributed sensors and difficulty in accurately determining the gas distribution concentration in the space.

[0048] Through the above steps S102-S106, the gas concentration data corresponding to multiple local points in the predetermined space are obtained; based on the multiple gas concentration data, the correlation relationship between the gas concentrations of the multiple predetermined points is determined, wherein the predetermined space includes multiple predetermined points, and the multiple predetermined points include multiple local points; based on the spatial parameters and the correlation relationship of the predetermined space, the gas concentration distribution data in the predetermined space is determined. By selecting multiple local points for collecting gas concentration data in the predetermined space, key gas concentration information can be effectively collected while the cost is controllable, thereby helping to provide an accurate data basis for the subsequent determination of gas concentration distribution data. By determining the correlation relationship between the gas concentrations of the multiple predetermined points based on the multiple gas concentration data, wherein the predetermined space includes multiple predetermined points, and the multiple predetermined points include multiple local points; by determining the correlation relationship between the gas concentrations of the multiple predetermined points based on the multiple gas concentration data, the diffusion and flow pattern of the gas in the space, as well as the mutual influence between the gas concentrations of different predetermined points, can be accurately reflected. Furthermore, the predetermined space includes multiple predetermined points, and the multiple predetermined points include multiple local points, so that there is no need to deploy gas concentration measurement devices at all locations, and global perception of gas concentration can be achieved under conditions of limited resources, thereby improving resource utilization efficiency. Based on the spatial parameters and correlation relationships of the predetermined space, the gas concentration distribution data in the predetermined space is determined, and the geometric characteristics of the predetermined space and the possible flow path of the gas are fully considered, and the influence on the gas concentration distribution is fully considered. By analyzing the combined correlation relationship, the gas distribution of the entire predetermined space can be inferred, even in areas that are not directly measured, thereby helping to accurately determine the global distribution of gas in the space, thereby solving the technical problem in the related technology that there are few sensors distributed and it is difficult to accurately determine the gas distribution concentration in the space.

[0049] As an optional embodiment, based on multiple gas concentration data, the correlation relationship between gas concentrations of multiple predetermined points is determined, including: determining one-dimensional concentration vectors corresponding to the multiple gas concentration data respectively; obtaining a high-dimensional concentration vector based on the multiple one-dimensional concentration vectors, wherein the dimension of the high-dimensional concentration vector is higher than the dimension corresponding to the multiple one-dimensional concentration vectors, and the high-dimensional concentration vector includes point feature vectors corresponding to the multiple predetermined points respectively; determining the weight corresponding to each point feature vector in the high-dimensional concentration vector based on the spatial position relationship between the multiple predetermined points and the gas concentration correlation relationship between the multiple local points; obtaining an attention-weighted feature vector based on the weight corresponding to each point feature vector in the high-dimensional concentration vector, wherein the attention-weighted feature vector represents the correlation relationship between the gas concentrations of multiple predetermined points.

[0050] In this embodiment, specific steps of determining the correlation relationship between gas concentrations at a plurality of predetermined points based on a plurality of gas concentration data are described.

[0051] Among them, a one-dimensional concentration vector is involved, which is obtained based on gas concentration data corresponding to multiple local points, and each gas concentration data corresponds to a one-dimensional vector.

[0052] Among them, a high-dimensional concentration vector is involved, which is a high-dimensional concentration vector obtained by mapping multiple one-dimensional concentration vectors. The high-dimensional concentration vector includes point feature vectors corresponding to multiple predetermined points. Obtaining the high-dimensional concentration vector can be achieved through a multi-layer perceptron (MLP) module.

[0053] Among them, a point feature vector is involved, and the point feature vector is a feature vector used to reflect the spatial position of each predetermined point in the high-dimensional concentration vector. For example, the point feature vector may include the spatial position information of the predetermined point.

[0054] Among them, the spatial position relationship is involved, which is a position relationship used to represent the relative position information of the predetermined point in the three-dimensional space, and the spatial position may include the distance, direction and possible obstacles between multiple predetermined points. For example, in the predetermined space, the spatial position relationship may include the straight-line distance between the gas concentration detection sensors.

[0055] Among them, an attention weighted feature vector is involved. The attention weighted feature vector is a feature vector obtained according to the weight corresponding to the feature vector of each point in the high-dimensional concentration vector, which is used to represent the correlation relationship between the gas concentrations of multiple predetermined points. The attention weighted feature vector can be generated by adding an attention mechanism to the neural network to obtain the corresponding attention weighted feature, thereby enhancing the information of the predetermined points that play a key role in the perception of the global concentration distribution, and suppressing those relatively unimportant information, so that when determining the global gas concentration distribution, the data of the predetermined points will be valued.

[0056] In the steps involved in this embodiment, first, multiple gas concentration data collected from multiple local points are represented as one-dimensional concentration vectors, and then these one-dimensional concentration vectors are converted into high-dimensional concentration vectors, by increasing the dimension of the vector, so as to be able to capture more information about the correlation between gas concentrations and spatial distribution. After obtaining the high-dimensional concentration vector, the weight corresponding to each point feature vector is determined based on the spatial position relationship between the predetermined points and the correlation relationship between the gas concentrations previously determined between the local points. These weights reflect the relative importance of the points in the global perception. Finally, based on the weight corresponding to each point feature vector in the high-dimensional concentration vector, an attention-weighted feature vector is obtained to represent the correlation relationship between the gas concentrations of multiple predetermined points.

[0057] By converting the one-dimensional concentration vector into a high-dimensional concentration vector, the accurate capture of more features and information of multiple predetermined points and the corresponding gas concentration is achieved, and the high-dimensional concentration vector includes the point feature vectors corresponding to the multiple predetermined points, which can accurately reflect the spatial position information of each predetermined point. By determining the weight corresponding to each point feature vector in the high-dimensional concentration vector according to the spatial position relationship between the multiple predetermined points and the gas concentration correlation relationship between the multiple local points, not only the gas concentration of the predetermined point itself but also the mutual influence between the predetermined points are considered. According to the weight corresponding to each point feature vector in the high-dimensional concentration vector, the attention weighted feature vector representing the correlation relationship between the gas concentrations of the multiple predetermined points is obtained, which helps to accurately analyze which gas concentration data of the predetermined points is the most critical for the gas concentration distribution in the predetermined space, ensuring that the subsequent analysis can accurately focus on the truly critical features, thereby helping to infer the gas concentration distribution of the entire predetermined space as accurately as possible under the condition of limited gas concentration detection sensor layout, helping to improve the accuracy and efficiency of determining the gas distribution concentration in the space, and thus helping to solve the technical problem that the sensor distribution is small and it is difficult to accurately determine the gas distribution concentration in the space.

[0058] Determine the one-dimensional concentration vectors corresponding to the multiple gas concentration data, and obtain a high-dimensional concentration vector based on the multiple one-dimensional concentration vectors, wherein the dimension of the high-dimensional concentration vector is higher than the dimension corresponding to the multiple one-dimensional concentration vectors, and the high-dimensional concentration vector includes point feature vectors corresponding to the multiple predetermined points. The specific process can be realized by a trained multi-layer perceptron (MLP) module. According to the spatial position relationship between the multiple predetermined points and the gas concentration correlation relationship between the multiple local points, determine the weight corresponding to each point feature vector in the high-dimensional concentration vector, and obtain the attention weighted feature vector based on the weight corresponding to each point feature vector in the high-dimensional concentration vector, wherein the attention weighted feature vector represents the correlation relationship between the gas concentrations of the multiple predetermined points. The process can be realized by a trained single-layer attention mechanism module.

[0059] As an optional embodiment, the gas concentration distribution data in the predetermined space is determined based on the spatial parameters and the correlation relationship of the predetermined space, including: determining the spatial feature vector corresponding to the spatial parameters; inputting the spatial feature vector corresponding to the spatial parameters and the attention weighted feature vector into the linear layer to obtain a linear transformation vector, wherein the linear layer includes a predetermined weight matrix and a predetermined bias vector; inputting the linear transformation vector into the activation function to obtain a nonlinear vector, wherein the nonlinear vector weights the key feature vector and reduces the weight of the non-key feature vector, the key feature vector is a feature vector whose corresponding importance index is greater than a predetermined threshold, and the non-key feature vector is a feature vector whose corresponding importance index is less than or equal to the predetermined threshold; combining the nonlinear vector with the attention weighted feature vector to obtain a combined vector; and determining the gas concentration distribution data in the predetermined space based on the combined vector.

[0060] In this embodiment, specific steps of determining gas concentration distribution data in a predetermined space based on the spatial parameters and the correlation relationship of the predetermined space are described.

[0061] Among them, a spatial feature vector is involved, which is a feature vector obtained according to corresponding spatial parameters and used to reflect the geometric and physical characteristics of a predetermined space. For example, in a predetermined space, the spatial feature vector may include data such as the length, width, height, location coordinates of vents, and material type of the wall of the predetermined space.

[0062] Among them, a linear layer is involved, which is a layer contained in a neural network for processing input data (such as spatial feature vectors and attention-weighted feature vectors) through linear transformation, which can be matrix multiplication and vector addition. The weight matrix and bias vector in the linear layer are learned through the training process and are used to adjust the linear combination of the spatial feature vector and the attention-weighted feature vector. For example, in a neural network, the linear layer can multiply the input vector with the weight matrix, and then add the bias vector to generate a new vector.

[0063] Among them, a linear transformation vector is involved, which is a vector obtained by linearly transforming the input vector (such as the spatial feature vector and the attention weighted feature vector) through the weight matrix and the bias vector in the linear layer. For example, in a neural network, in the process of obtaining the linear transformation vector, the specific value of the linear transformation (i.e., scale and offset adjustment) is mainly determined by the learning dynamics during the training process, rather than a pre-set fixed value.

[0064] Among them, a predetermined weight matrix is ​​involved, which is a matrix used to adjust the linear combination of input vectors (such as spatial feature vectors and attention weighted feature vectors) in the linear layer of the neural network. The value of the weight matrix is ​​learned through the training process, which reflects the contribution of the input features to the output results.

[0065] Among them, a predetermined bias vector is involved, which is a parameter vector used to adjust the output in the linear layer of the neural network. The value of the bias vector is learned during the training process and is used to better fit the data and ensure that the vector after linear transformation can more accurately reflect the input features.

[0066] Among them, the activation function is involved, which is a function used to introduce nonlinear characteristics in neurons in the neural network. For example, the ReLU function sets all negative values ​​to 0 and keeps the positive values ​​unchanged, so that the output nonlinear vector has nonlinear characteristics, which helps the neural network to better analyze the nonlinear characteristics of the gas concentration distribution in the predetermined space.

[0067] Among them, a nonlinear vector is involved, which is a vector with nonlinear characteristics obtained after processing based on the activation function. The nonlinear vector contains the features after nonlinear transformation and can better capture the complex relationship between data.

[0068] Among them, a key feature vector is involved, which is a feature vector in a nonlinear vector whose corresponding importance index is greater than a predetermined threshold. This key feature plays a key role in analyzing the gas concentration distribution in a predetermined space.

[0069] Among them, weighted processing is involved, which is the processing of increasing the weight corresponding to the key feature vector for the key feature vector. In the neural network, the influence of the key feature vector can be enhanced by increasing the weight to reflect the importance of the key feature vector in analyzing the gas concentration distribution in the predetermined space.

[0070] Among them, a non-critical feature vector is involved, and the non-critical feature vector is a feature vector whose corresponding importance index is less than or equal to a predetermined threshold in a non-linear vector. The importance of the non-critical feature vector in analyzing the gas concentration distribution in the predetermined space is relatively less than the importance corresponding to the critical feature vector. For example, the non-critical feature vector can be a color feature of the predetermined space.

[0071] Among them, it involves weight reduction processing, which is the processing of reducing the weights corresponding to non-critical feature vectors. In the neural network, the influence of non-critical feature vectors can be weakened by reducing the weights to ensure that the analysis process focuses on the features directly related to the gas concentration distribution, such as the features described by the key feature vectors.

[0072] Among them, an important index is involved, which is an index used to measure the importance of a feature vector in analyzing the gas concentration distribution in a predetermined space. For example, the higher the importance index corresponding to a feature vector, the more important the feature vector is in analyzing the gas concentration distribution in the predetermined space.

[0073] Among them, a predetermined threshold is involved, which is used to define the importance of the feature vector for analyzing the gas concentration distribution in the predetermined space, so as to distinguish the key feature vector from the non-key feature vector. For example, if the predetermined threshold is set to 0.5, then the feature vector with an importance index greater than 0.5 will be regarded as a key feature vector, and the feature vector with an importance index less than or equal to 0.5 will be regarded as a non-key feature vector.

[0074] Among them, a combined vector is involved, which is a vector obtained by combining a nonlinear vector with an attention weighted feature vector for determining gas concentration distribution data in a predetermined space. The combined vector integrates key information of spatial features and gas concentration distribution.

[0075] In the steps involved in this embodiment, the corresponding spatial feature vector is determined according to the spatial parameters for subsequent processing and analysis. The spatial feature vector and the attention weighted feature vector are input into the linear layer, and a linear transformation is performed through a predetermined weight matrix and a bias vector to obtain a linear transformation vector. The linear transformation vector is input into the activation function to introduce nonlinear characteristics and obtain a nonlinear vector to accurately capture complex feature relationships. In this process, the nonlinear vector weights the key feature vectors (i.e., the feature vectors whose corresponding importance index is greater than a predetermined threshold), thereby enhancing their importance in subsequent analysis; while the nonlinear vector downgrades the non-key feature vectors (i.e., the feature vectors whose corresponding importance index is less than or equal to a predetermined threshold), thereby reducing their influence. This process can be implemented through the feedforward application network of the Transformer module. The nonlinear vector is combined with the attention weighted feature vector to obtain a combined vector. Finally, according to the obtained combined vector, the gas concentration distribution data in the predetermined space is determined. This process can be implemented through the residual connection module of the Transformer module.

[0076] The spatial feature vector and the attention weighted feature vector are input into the linear layer to obtain the linear transformation vector, which helps to reduce the redundancy of the input features and improve the efficiency of subsequent analysis. The linear transformation vector is input into the activation function to introduce nonlinear characteristics and obtain a nonlinear vector to accurately capture the complex feature relationship, so that the gas concentration distribution in the predetermined space can be accurately determined. In this process, the nonlinear vector weights the key feature vectors (i.e., the feature vectors whose corresponding importance index is greater than the predetermined threshold), which enhances their importance in subsequent analysis; while the nonlinear vector downweights the non-key feature vectors (i.e., the feature vectors whose corresponding importance index is less than or equal to the predetermined threshold), reducing their influence. Through weighting and downweighting, it helps to focus on those feature vectors that are closely related to the gas concentration distribution, while reducing the influence of irrelevant or less contributing feature vectors, thereby helping to optimize the allocation of data analysis resources and improve the efficiency and accuracy of processing complex feature relationships. The nonlinear vector is combined with the attention-weighted eigenvector to obtain a combined vector, which provides more comprehensive information for the subsequent determination of the gas concentration distribution data in the predetermined space based on the combined vector, helps to improve the accuracy and efficiency of determining the gas distribution concentration in the space, and further helps to solve the technical problem that there are few sensors distributed and it is difficult to accurately determine the gas distribution concentration in the space.

[0077] As an optional embodiment, determining the correlation between gas concentrations between multiple predetermined points based on multiple gas concentration data includes: acquiring flow field data in a predetermined space; and determining the correlation between gas concentrations between multiple predetermined points based on the flow field data and multiple gas concentration data.

[0078] In this embodiment, specific steps of determining the correlation relationship between gas concentrations at a plurality of predetermined points based on a plurality of gas concentration data are described.

[0079] Among them, flow field data is involved, and the flow field data is data used to reflect the flow of gas in a predetermined space, and may include the speed and direction of the gas, etc. According to the flow field data, how the gas diffuses and distributes in the space can be analyzed.

[0080] In the steps involved in this embodiment, flow field data reflecting the flow of gas in a predetermined space is obtained. After the flow field data is obtained, the correlation relationship between the gas concentrations of multiple predetermined points is determined based on the flow field data and multiple gas concentration data.

[0081] By acquiring the flow field data in the predetermined space, it is helpful to understand the physical environment information of how the gas flows and diffuses in the predetermined space, and the gas concentration data at multiple local points can directly reflect the real-time distribution of the gas at different locations. Through the flow field data and multiple gas concentration data, the physical information of the gas flow in the space and the chemical information of the actual concentration can be integrated to form a more comprehensive data set, so that the correlation between the gas concentrations of multiple predetermined points determined based on the flow field data and multiple gas concentration data can better reflect the changes in gas concentration and the actual situation of diffusion, so that the distribution of gas concentration in the space can be accurately determined, which helps to solve the technical problem that there are few sensors and it is difficult to accurately determine the gas distribution concentration in the space.

[0082] As an optional embodiment, the correlation relationship between the gas concentrations of multiple predetermined points is determined based on multiple gas concentration data, including: determining the first credible indexes corresponding to the multiple gas concentration data respectively based on the particle image velocity method, and determining the second credible indexes corresponding to the multiple gas concentration data respectively based on the concentration experiment method; determining the target credible indexes corresponding to the multiple gas concentration data respectively based on the multiple first credible indexes and the multiple second credible indexes; when the multiple target credible indexes are all greater than the predetermined credible threshold value, determining the correlation relationship between the gas concentrations of multiple predetermined points based on the multiple gas concentration data.

[0083] In this embodiment, specific steps of determining the correlation relationship between gas concentrations at a plurality of predetermined points based on a plurality of gas concentration data are described.

[0084] Among them, the particle image velocimetry is involved, which is an experimental technique for measuring the velocity field of fluid flow. The particle image velocimetry sprays tiny particles in the fluid and irradiates it with a laser, then uses a high-speed camera to take an image of the particles, and calculates the velocity and direction of the fluid by analyzing the position changes of the particles in the image. The particle image velocimetry can provide two-dimensional or three-dimensional flow field data, which can be used to analyze the diffusion path and velocity of gas in space.

[0085] Among them, a first credibility index is involved, which is an index determined by particle image velocity method and reflects the reliability of data related to gas flow. The first credibility index can be used to evaluate the credibility of data related to gas flow, such as flow field data.

[0086] Among them, the concentration experiment method is involved, which is a method for measuring the gas concentration distribution in space. It can include using a chemical sensor network to measure the gas concentration at different points in real time, or obtaining data through gas diffusion simulation experiments in the laboratory.

[0087] Among them, a second credibility index is involved, which is an index determined by a concentration experiment method and reflects the reliability of data related to gas concentration. The second credibility index can be used to evaluate data related to gas concentration, such as the credibility of gas concentration data.

[0088] Among them, the target credibility index is involved, which is an indicator obtained by combining the first credibility index (which can be the reliability of flow field data) and the second credibility index (which can be the reliability of gas concentration data) to evaluate the comprehensive credibility of data obtained at a specific predetermined point.

[0089] Among them, a predetermined credible threshold is involved, which is a pre-set threshold used to determine whether the target credible index is high enough to ensure that the gas concentration data can be used to determine the correlation between the gas concentrations at predetermined points. For example, when the target credible index is lower than the predetermined credible threshold, the data can be considered unreliable and needs to be remeasured or corrected.

[0090] In the steps involved in this embodiment, first, the particle image velocity method is used to determine the first credible indexes corresponding to the multiple gas concentration data to evaluate the credibility of the gas flow data (which can be flow field data). Then, the concentration experiment method is used to determine the second credible indexes corresponding to the multiple gas concentration data to evaluate the credibility of the gas concentration data. Combining the multiple first credible indexes and the multiple second credible indexes, the target credible indexes corresponding to the multiple gas concentration data are obtained. When the multiple target credible indexes are all greater than the predetermined credible threshold, the correlation relationship between the gas concentrations of the multiple predetermined points is determined according to the corresponding multiple gas concentration data.

[0091] By using the particle image velocity method to determine the first credible index corresponding to multiple gas concentration data, the data quality is quantified, thereby realizing an accurate assessment of the credibility of the gas flow data (which can be flow field data), thereby helping to determine whether the data is suitable for subsequent analysis. By using the concentration experiment method to determine the second credible index corresponding to multiple gas concentration data, the credibility of the gas concentration data is accurately assessed, thereby helping to determine whether the data is suitable for subsequent analysis. By combining multiple first credible indexes and multiple second credible indexes, target credible indexes corresponding to multiple gas concentration data are obtained, and the gas flow and gas concentration aspects are comprehensively considered, thereby realizing a dual reliability measurement of the credibility of the gas concentration data at a specific point. On this basis, when multiple target credible indexes are all greater than a predetermined credible threshold, the correlation relationship between the gas concentrations of multiple predetermined points is determined according to the corresponding multiple gas concentration data, thereby reducing the erroneous analysis caused by data quality problems, and providing a comprehensive and reliable data basis for the accuracy and reliability of the subsequent analysis of the gas distribution concentration in the predetermined space, thereby helping to accurately determine the gas distribution concentration in the space, thereby helping to solve the technical problem that the sensors are less distributed and it is difficult to accurately determine the gas distribution concentration in the space.

[0092] As an optional embodiment, after determining the gas concentration distribution data in the predetermined space based on the spatial parameters and the association relationship of the predetermined space, it also includes: when the gas concentration distribution data is greater than a predetermined concentration distribution threshold, determining the setting parameters corresponding to the safety equipment in the predetermined space; and controlling the safety equipment to be set to the setting parameters.

[0093] In this embodiment, the specific steps after determining the gas concentration distribution data in the predetermined space according to the spatial parameters and the correlation relationship of the predetermined space are described.

[0094] Among them, a predetermined concentration distribution threshold is involved, which is a threshold pre-set in a predetermined space for determining whether the gas concentration has reached a defined level (such as a level requiring emergency action).

[0095] Among them, safety equipment is involved, which is a pre-installed device in a predetermined space for handling changes in gas concentration. The safety equipment may include air purifiers, exhaust fans, chemical reactors, decontamination spray systems, etc.

[0096] Among them, setting parameters are involved, which are parameters used to control the operation of safety equipment to adapt to the current gas concentration distribution.

[0097] In the steps involved in this embodiment, after determining the gas concentration distribution data in the predetermined space according to the spatial parameters and the association relationship of the predetermined space, it is necessary to further determine whether the gas concentration distribution data is greater than a predetermined concentration distribution threshold. When the gas concentration distribution data is greater than the predetermined concentration distribution threshold, the setting parameters corresponding to the safety device in the predetermined space are determined, and the safety device is controlled to be set to the setting parameters.

[0098] After determining the gas concentration distribution data in the predetermined space according to the spatial parameters and the correlation relationship of the predetermined space, further judging whether the gas concentration distribution data is greater than the predetermined concentration distribution threshold value helps to timely and accurately identify whether the gas concentration distribution in the current space will pose a threat. In the case where the gas concentration distribution data is greater than the predetermined concentration distribution threshold value, the setting parameters corresponding to the safety equipment in the predetermined space are determined, and the safety equipment is controlled to be set to the setting parameters, ensuring that the safety equipment is set only when action is really needed, avoiding overreaction or invalid operation, thereby ensuring personnel safety and saving resources.

[0099] As an optional embodiment, after determining the gas concentration distribution data in the predetermined space based on the spatial parameters and the correlation relationship of the predetermined space, it also includes: determining the toxicity result of whether the gas is a toxic gas; when the toxicity result is that the gas is a toxic gas, determining the toxicity distribution threshold of the toxic gas; when the gas concentration distribution data is greater than the toxicity distribution threshold, determining a planned route corresponding to the predetermined space based on the gas concentration distribution data.

[0100] In this embodiment, the specific steps after determining the gas concentration distribution data in the predetermined space according to the spatial parameters and the correlation relationship of the predetermined space are described.

[0101] Among them, toxic gases are involved, which are gases that cause serious harm or fatal effects on human health under certain conditions. The toxic gas can enter the human body through inhalation, skin contact or ingestion, causing acute or chronic poisoning. For example, the toxic gas may include aerosols of chemical warfare agents, toxic industrial chemicals, and biological toxins, which can cause harm to the human body or even endanger life safety when they reach a certain concentration in the air.

[0102] Among them, a toxicity result is involved, and the toxicity result is a judgment result used to describe whether the current gas is a toxic gas.

[0103] Among them, a toxicity distribution threshold is involved, which is whether the concentration of toxic gases in the current predetermined space has reached a threshold that causes harm to human health.

[0104] Among them, it involves a planned route, which is a route determined based on gas concentration distribution data when the gas concentration distribution data is greater than the toxicity distribution threshold, and is used for safe evacuation of personnel or equipment operation planning. For example, in an emergency such as a toxic gas leak, the planned route can guide personnel to avoid high-concentration dangerous areas, find the safest evacuation route, or instruct safety equipment (such as a decontamination spray system) to operate where it is most needed.

[0105] In the steps involved in this embodiment, after determining the gas concentration distribution data in the predetermined space according to the spatial parameters and the association relationship of the predetermined space, a toxicity result is further determined as to whether the current gas in the predetermined space is a toxic gas. And when the toxicity result is that the gas is a toxic gas, a toxicity distribution threshold of the toxic gas is determined to further determine whether the gas concentration distribution data is greater than the toxicity distribution threshold. When the gas concentration distribution data is greater than the toxicity distribution threshold, a planned route corresponding to the predetermined space is determined according to the gas concentration distribution data.

[0106] After determining the gas concentration distribution data in the predetermined space based on the spatial parameters and correlation of the predetermined space, the toxicity result is determined to determine whether the current gas in the predetermined space is a toxic gas, thereby achieving accurate distinction between harmless gas and toxic gas so that targeted safety measures can be taken. In the case where the toxicity result is that the gas is toxic, the toxicity distribution threshold of the toxic gas is determined, and it is further determined whether the gas concentration distribution data is greater than the toxicity distribution threshold, thereby achieving accurate determination of whether the current gas concentration distribution is safe or dangerous, thereby helping to decide whether it is necessary to subsequently determine the planned route corresponding to the predetermined space, reducing unnecessary evacuation and decontamination operations. In the case where the gas concentration distribution data is greater than the toxicity distribution threshold, the planned route corresponding to the predetermined space is determined based on the gas concentration distribution data, ensuring that personnel can evacuate quickly and safely to low-concentration or non-toxic areas, or instructing safety equipment (such as decontamination spray systems) to operate where it is most needed.

[0107] Based on the above embodiments and optional embodiments, an optional implementation is provided, which is described in detail below.

[0108] In the related art, most of the existing methods for determining gas concentration distribution data are based on real-time measurement data of gas sensors. However, due to the small number of sensors distributed, there is a technical problem that it is difficult to accurately determine the gas distribution concentration in space.

[0109] To address the above-mentioned problems, no effective solution has been proposed yet.

[0110] In view of this, a method for determining gas concentration distribution data is provided in an optional embodiment of the present invention, which can also be called a sparse sampling perception and intelligent disinfection method, which can effectively solve the technical problem in related technologies that there are few sensors distributed and it is difficult to accurately determine the gas distribution concentration in the space.

[0111] Figure 2 is a diagram of a method for determining gas concentration distribution data provided by an optional embodiment of the present invention, such as Figure 2 As shown, the details will be introduced below.

[0112] S1. Obtaining gas concentration data corresponding to multiple local points in a predetermined space;

[0113] S2. Determine, based on the plurality of gas concentration data, a correlation relationship between gas concentrations at a plurality of predetermined points, wherein the predetermined space includes a plurality of predetermined points, and the plurality of predetermined points include a plurality of local points;

[0114] Specifically, S2 also includes:

[0115] S211, determining one-dimensional concentration vectors corresponding to the plurality of gas concentration data respectively;

[0116] S212, obtaining a high-dimensional concentration vector based on the multiple one-dimensional concentration vectors, wherein the dimension of the high-dimensional concentration vector is higher than the dimension corresponding to the multiple one-dimensional concentration vectors, and the high-dimensional concentration vector includes point feature vectors corresponding to the multiple predetermined points respectively;

[0117] S213, determining a weight corresponding to each point feature vector in the high-dimensional concentration vector according to a spatial position relationship between a plurality of predetermined points and a gas concentration correlation relationship between a plurality of local points;

[0118] S214. Obtain an attention weighted feature vector according to the weight corresponding to each point feature vector in the high-dimensional concentration vector, wherein the attention weighted feature vector represents the correlation relationship between the gas concentrations of multiple predetermined points.

[0119] Specifically, S2 also includes:

[0120] S221, acquiring flow field data in a predetermined space;

[0121] S222, determining a correlation relationship between gas concentrations at a plurality of predetermined points based on the flow field data and the plurality of gas concentration data;

[0122] S223, determining first credibility indexes corresponding to the plurality of gas concentration data respectively according to the particle image velocity method, and determining second credibility indexes corresponding to the plurality of gas concentration data respectively according to the concentration experiment method;

[0123] S224, determining target credibility indexes corresponding to the plurality of gas concentration data respectively according to the plurality of first credibility indexes and the plurality of second credibility indexes;

[0124] S225. When the credibility indexes of multiple targets are all greater than a predetermined credibility threshold, determine the correlation between the gas concentrations of multiple predetermined points based on the multiple gas concentration data.

[0125] For example, this process can be implemented by a multi-layer perceptron (MLP) module, a single-layer attention mechanism module, and a Transformer module. Among them, the MLP module consists of many neurons, each of which is connected to other neurons. Each connection is evaluated by a weight coefficient that reflects the importance of a given link in the neural network. The multi-layer perceptron can arbitrarily map a low-dimensional vector (the same as the one-dimensional concentration vector mentioned above) space to another high-dimensional vector (the same as the high-dimensional concentration vector mentioned above) space, and is also advanced in adjusting the weight coefficient to capture unknown information hidden in the data.

[0126] The multi-layer perceptron formula is as follows:

[0127]

[0128] Among them, y j represents the final output value of the jth neuron, f j () represents the activation network (same as the activation function above), x i represents the i-th eigenvalue of the input layer, b j Represented as a j-column bias weight matrix, w ij Represented as a weight matrix with i rows and j columns, S j Represents an intermediate state of the network, and n represents the dimension of the input vector, that is, how many features are input into the neural network.

[0129] However, the multi-layer perceptron has the disadvantage of not being able to obtain global information, and the trainable parameters are limited, so it is difficult to learn deep-level associated features. Therefore, a single-layer attention mechanism module is added after the MLP module. The attention mechanism module can aggregate limited local distribution information to the global distribution of the entire space. The module is used to calculate the query, key, value (Q, K, V) of the attention, and apply a linear transformation of the input to obtain all triple values. Among them, Q is the element to be queried, K represents the position code of each element, and V is the value of each element in the source sequence. The calculation formula of the attention mechanism is as follows:

[0130]

[0131] Among them, Attention(Q,K,V) represents the output of the attention mechanism, Softmax() represents the activation function, Softmax is applied row by row, and W Q , W K , W V denote the linear transformation matrices that transform the input feature I into query (Q), key (K), and value (V), respectively. k The dimensions of the columns of the linear transformation matrix representing the query (Q) and key (K), is the input feature, d n is the number of input elements, d e is the dimension of the rows of the matrices of K and Q values, and is the linear transformation matrix of the input, d e is the dimension of the matrix column of V values, d v is the dimension of the rows of the matrix V. In addition, the dot product of the query and the key is divided by To alleviate the vanishing gradient problem.

[0132] However, the single-layer attention mechanism module is shallow because there are fewer adjustable parameters, which limits the model's ability to learn deep-level flow field and concentration field correlation features. To solve this problem, the Transformer architecture is connected to the multi-layer perceptron and the single-layer attention mechanism module to learn deep-level correlation features. Figure 2 As shown in part (c) of the figure, it consists of several stacked attention mechanism blocks. Each Transformer module consists of a multi-head attention mechanism module and a feedforward neural network. In addition, a residual connection is used, followed by a normalization module. Multi-head attention allows the model to jointly focus on the information of subspaces represented by different positions. Its architecture is shown in Figure 2 As shown in part (d), it includes three steps: attention calculation, multi-head concatenation and linear transformation, which are expressed as follows:

[0133] MultiHead(I)=Contact(head1,head2,…,head z )W O

[0134]

[0135] Among them, MultiHead() represents the output of the multi-head attention mechanism, head i represents the output of the ith attention head (i=1, 2, 3, ..., z), z is the number of attention heads, W i Q , W i K , W i V denote the linear transformation matrices of the query (Q), key (K), and value (V) of the i-th attention head, respectively. is the input feature matrix, d o is the dimension of the multi-head output, Represents the matrix shape.

[0136] The neural network training process is the process of "capturing" the hidden related fluid and physical field related characteristics in the entire space. During the training process, the parameters of the MLP module, the attention mechanism module, and the Transformer module are mixed together to minimize the gap between the predicted results and the actual results. The absolute root mean square RMSE and the mean absolute relative error MARE are used to verify the error between the prediction and the actual results:

[0137]

[0138] Among them, y i represents the ith actual result, represents the i-th prediction result, and N represents the total number of samples.

[0139] In addition, the correlation coefficient is used to evaluate the similarity RE between the predicted results and the actual results. The formula is:

[0140]

[0141] in, Represents y i The mean of express The mean of .

[0142] In addition, the structure of the intelligent washing and disinfection gas network is introduced, such as Figure 2As shown in part (e) of the figure. The input of the intelligent gas decontamination network is the perceived distribution of the gas in the entire space. The inference network is a simple MLP network, which builds a connection between the concentration distribution of the gas in the entire space and the decontamination strategy. For example, the angle of the ventilation pipe, the velocity at the inlet, and the mass fraction ratio between gaseous oxides and chemical warfare agents. The intelligent decontamination model was previously trained on a large number of data sets. After that, for actual applications, the concentration distribution of the gas in the entire space is input into the trained model, and the model will efficiently output the optimal parameters for removing the gas.

[0143] S3. Determine the gas concentration distribution data in the predetermined space according to the spatial parameters and the correlation relationship of the predetermined space.

[0144] Specifically, S3 also includes:

[0145] S31, determining a spatial feature vector corresponding to the spatial parameter;

[0146] S32, inputting the spatial feature vector and the attention weighted feature vector corresponding to the spatial parameter into a linear layer to obtain a linear transformation vector, wherein the linear layer includes a predetermined weight matrix and a predetermined bias vector;

[0147] S33, inputting the linear transformation vector into an activation function to obtain a nonlinear vector, wherein the nonlinear vector performs weighted processing on the key feature vector and down-weights the non-key feature vector, the key feature vector is a feature vector whose corresponding importance index is greater than a predetermined threshold, and the non-key feature vector is a feature vector whose corresponding importance index is less than or equal to the predetermined threshold;

[0148] S34, combining the nonlinear vector with the attention weighted feature vector to obtain a combined vector;

[0149] S35. Determine the gas concentration distribution data in the predetermined space based on the combined vector.

[0150] Specifically, after S3, step S4 is also included, wherein S4 includes:

[0151] S411, when the gas concentration distribution data is greater than a predetermined concentration distribution threshold, determining setting parameters corresponding to safety equipment in a predetermined space;

[0152] S412, controlling the safety device to be set to set parameters.

[0153] Specifically, S4 also includes:

[0154] S421, determining whether the gas is a toxic gas;

[0155] S422, when the toxicity result indicates that the gas is toxic, determining a toxicity distribution threshold of the toxic gas;

[0156] S423. When the gas concentration distribution data is greater than the toxicity distribution threshold, determine a planned route corresponding to the predetermined space based on the gas concentration distribution data.

[0157] For example, the specific implementation sequence is as follows:

[0158] (1) First, the flow field data set (same as the gas concentration data mentioned above) is constructed using computational fluid dynamics methods;

[0159] (2) Then, the credibility of the dataset is verified through PIV experiments and concentration experiments (same as the target credibility index mentioned above);

[0160] (3) If the flow field and concentration data sets are not reliable, reconstruct the data sets;

[0161] (4) If the data set is credible (same as the case where the credibility indexes of multiple targets are greater than the predetermined credibility threshold), the internal coupling relationship between the flow field and the concentration is learned through the sparse sampling perception network (same as the above association relationship);

[0162] (5) In addition, based on the learned coupling characteristics of the fluid and concentration field, the distribution of chemical agents in the entire space is sensed (same as the above-mentioned gas concentration distribution data);

[0163] (6) Then, based on the distribution of chemical agents in the entire space, the strategies for constructing cleaning and disinfection gases and safety zones are dynamically and intelligently adjusted;

[0164] (7) Finally, based on the perceived distribution of chemical agents in the entire space, the escape route (same as the planned route mentioned above) is optimized to improve the efficiency of rescuing the masses.

[0165] Through the above optional implementation, at least the following beneficial effects can be achieved:

[0166] (1) Compared with the related art, the present invention can effectively collect key gas concentration information while keeping costs under control by selecting multiple local points for collecting gas concentration data in a predetermined space, thereby helping to provide an accurate data basis for the subsequent determination of gas concentration distribution data. By determining the correlation between gas concentrations of multiple predetermined points based on multiple gas concentration data, wherein the predetermined space includes multiple predetermined points, and the multiple predetermined points include multiple local points; by determining the correlation between gas concentrations of multiple predetermined points based on multiple gas concentration data, the diffusion and flow patterns of gas in space, as well as the mutual influence between gas concentrations at different predetermined points can be accurately reflected. Moreover, the predetermined space includes multiple predetermined points, and the multiple predetermined points include multiple local points, so that there is no need to deploy gas concentration measurement devices at all locations, so that global perception of gas concentration can be achieved under conditions of limited resources, thereby improving resource utilization efficiency. The gas concentration distribution data in the predetermined space is determined based on the spatial parameters and correlation of the predetermined space, which fully considers the geometric characteristics of the predetermined space and the possible flow path of the gas, and their influence on the gas concentration distribution. By analyzing the correlation, the gas distribution of the entire predetermined space can be inferred, even in areas not directly measured, thereby helping to accurately determine the global distribution of the gas in the space, and thus solving the technical problem in related technologies that there are few sensors distributed and it is difficult to accurately determine the gas distribution concentration in the space.

[0167] (2) Compared with the related art, the present invention takes the spatial parameters of the predetermined space as the analysis basis, fully considers the influence of the geometric characteristics of the predetermined space and the possible flow path of the gas on the gas concentration distribution, and contributes to the comprehensiveness and accuracy of the subsequent gas concentration distribution. By combining the spatial parameters of the predetermined space and the correlation between the gas concentrations collected at multiple predetermined points, the gas distribution of the entire predetermined space can be inferred, even in areas that are not directly measured, which helps to accurately determine the global distribution of the gas in the space, thereby effectively solving the technical problem that there are few sensors and it is difficult to accurately determine the gas distribution concentration in the space.

[0168] (3) Compared with the related art, the present invention realizes accurate capture of more features and information of multiple predetermined points and corresponding gas concentrations by converting a one-dimensional concentration vector into a high-dimensional concentration vector, and the high-dimensional concentration vector includes point feature vectors corresponding to multiple predetermined points, which can accurately reflect the spatial position information of each predetermined point. By determining the weight corresponding to each point feature vector in the high-dimensional concentration vector according to the spatial position relationship between the multiple predetermined points and the gas concentration correlation relationship between the multiple local points, not only the gas concentration of the predetermined point itself but also the mutual influence between the predetermined points are considered. According to the weight corresponding to the feature vector of each point in the high-dimensional concentration vector, an attention-weighted feature vector representing the correlation between gas concentrations of multiple predetermined points is obtained, which helps to accurately analyze which predetermined points' gas concentration data are most critical to the gas concentration distribution in the predetermined space, ensuring that subsequent analysis can accurately focus on the truly critical features, thereby helping to infer the gas concentration distribution of the entire predetermined space as accurately as possible when the deployment of gas concentration detection sensors is limited, helping to improve the accuracy and efficiency of determining the gas distribution concentration in the space, and further helping to solve the technical problem of having fewer sensors and difficulty in accurately determining the gas distribution concentration in the space.

[0169] (4) Compared with the related art, the present invention inputs the spatial feature vector and the attention weighted feature vector into the linear layer to obtain a linear transformation vector, which helps to reduce the redundancy of the input features and improve the efficiency of subsequent analysis. The linear transformation vector is input into the activation function to introduce nonlinear characteristics and obtain a nonlinear vector to accurately capture the complex feature relationship, so that the gas concentration distribution in the predetermined space can be accurately determined. In this process, the nonlinear vector weights the key feature vectors (i.e., the feature vectors whose corresponding importance index is greater than the predetermined threshold), thereby enhancing their importance in the subsequent analysis; while the nonlinear vector down-weights the non-key feature vectors (i.e., the feature vectors whose corresponding importance index is less than or equal to the predetermined threshold), thereby reducing their influence. Through weighting and down-weighting processing, it helps to focus on those feature vectors that are closely related to the gas concentration distribution, while reducing the influence of irrelevant or less contributing feature vectors, thereby helping to optimize the allocation of data analysis resources and improve the efficiency and accuracy of processing complex feature relationships. The nonlinear vector is combined with the attention-weighted eigenvector to obtain a combined vector, which provides more comprehensive information for the subsequent determination of the gas concentration distribution data in the predetermined space based on the combined vector, helps to improve the accuracy and efficiency of determining the gas distribution concentration in the space, and further helps to solve the technical problem that there are few sensors distributed and it is difficult to accurately determine the gas distribution concentration in the space.

[0170] (5) Compared with the related art, the present invention helps to understand the physical environment information of how the gas flows and diffuses in the predetermined space by obtaining the flow field data in the predetermined space, and the gas concentration data of multiple local points can directly reflect the real-time distribution of the gas at different locations. Through the flow field data and multiple gas concentration data, the physical information of the gas flow in the space and the chemical information of the actual concentration can be integrated to form a more comprehensive data set, so that the correlation between the gas concentrations of multiple predetermined points determined based on the flow field data and the multiple gas concentration data can better reflect the changes in gas concentration and the actual situation of diffusion, so that the distribution of gas concentration in the space can be accurately determined, which helps to solve the technical problem that there are few sensors and it is difficult to accurately determine the gas distribution concentration in the space.

[0171] (6) By using the particle image velocity method to determine the first credible index corresponding to multiple gas concentration data, the quality of the data is quantified, thereby achieving an accurate assessment of the credibility of the gas flow data (which can be flow field data), which helps to determine whether the data is suitable for subsequent analysis. By using the concentration experiment method to determine the second credible index corresponding to multiple gas concentration data, the credibility of the gas concentration data is accurately assessed, which helps to determine whether the data is suitable for subsequent analysis. By combining multiple first credible indices and multiple second credible indices, target credible indexes corresponding to multiple gas concentration data are obtained, and the gas flow and gas concentration aspects are comprehensively considered to achieve a dual reliability measurement of the credibility of the gas concentration data at a specific point. On this basis, when multiple target credible indices are all greater than a predetermined credible threshold, the correlation between the gas concentrations of multiple predetermined points is determined based on the corresponding multiple gas concentration data, reducing erroneous analysis caused by data quality problems, and providing a comprehensive and reliable data basis for the accuracy and reliability of the subsequent analysis of the gas distribution concentration in the predetermined space, thereby helping to accurately determine the gas distribution concentration in the space, and further helping to solve the technical problem of less sensor distribution and difficulty in accurately determining the gas distribution concentration in the space.

[0172] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0173] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods of various embodiments of the present invention.

[0174] Example 2

[0175] According to an embodiment of the present invention, there is also provided a device for implementing the above-mentioned method for determining gas concentration distribution data. Figure 3 is a structural block diagram of a device for determining gas concentration distribution data according to an embodiment of the present invention. Figure 3 As shown, the device includes: an acquisition module 302, a first determination module 304 and a second determination module 306. The device is described in detail below.

[0176] An acquisition module 302 is used to acquire gas concentration data corresponding to multiple local points in a predetermined space; a first determination module 304 is connected to the above-mentioned acquisition module 302, and is used to determine the correlation relationship between the gas concentrations of multiple predetermined points based on the multiple gas concentration data, wherein the predetermined space includes multiple predetermined points, and the multiple predetermined points include multiple local points; a second determination module 306 is connected to the above-mentioned first determination module 304, and is used to determine the gas concentration distribution data in the predetermined space based on the spatial parameters and the correlation relationship of the predetermined space.

[0177] It should be noted here that the above-mentioned acquisition module 302, the first determination module 304 and the second determination module 306 correspond to steps S102 to S106 in the method for implementing the gas concentration distribution data determination method, and the instances and application scenarios implemented by the multiple modules are the same as the corresponding steps, but are not limited to the contents disclosed in the above-mentioned embodiment 1.

[0178] Example 3

[0179] According to another aspect of an embodiment of the present invention, there is also provided an electronic device, comprising: a processor; and a memory for storing instructions executable by the processor, wherein the processor is configured to execute instructions to implement any of the above-mentioned methods for determining gas concentration distribution data.

[0180] Example 4

[0181] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided. When instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device can execute any of the above-mentioned methods for determining gas concentration distribution data.

[0182] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0183] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0184] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0185] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0186] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0187] 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 storage medium. Based on this understanding, the technical solution of the present invention, in essence, 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, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium 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 codes.

[0188] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for determining gas concentration distribution data, characterized in that: include: Obtaining gas concentration data corresponding to multiple local points in a predetermined space; Determining the correlation relationship between the gas concentrations of a plurality of predetermined points according to the plurality of gas concentration data, wherein the predetermined space includes the plurality of predetermined points, and the plurality of predetermined points include the plurality of local points; The gas concentration distribution data in the predetermined space is determined according to the spatial parameters of the predetermined space and the association relationship.

2. The method according to claim 1, characterized in that The step of determining the correlation between gas concentrations at a plurality of predetermined points based on a plurality of gas concentration data includes: Determine one-dimensional concentration vectors corresponding to the plurality of gas concentration data respectively; According to the multiple one-dimensional concentration vectors, a high-dimensional concentration vector is obtained, wherein the dimension of the high-dimensional concentration vector is higher than the dimension corresponding to the multiple one-dimensional concentration vectors, and the high-dimensional concentration vector includes point feature vectors corresponding to the multiple predetermined points respectively; Determining a weight corresponding to each point feature vector in the high-dimensional concentration vector according to a spatial position relationship between the plurality of predetermined points and a gas concentration correlation relationship between the plurality of local points; An attention weighted feature vector is obtained according to the weight corresponding to each point feature vector in the high-dimensional concentration vector, wherein the attention weighted feature vector represents the correlation relationship between the gas concentrations of the plurality of predetermined points.

3. The method according to claim 2, characterized in that The step of determining the gas concentration distribution data in the predetermined space according to the spatial parameter of the predetermined space and the association relationship comprises: Determine a spatial eigenvector corresponding to the spatial parameter; Inputting the spatial feature vector corresponding to the spatial parameter and the attention weighted feature vector into a linear layer to obtain a linear transformation vector, wherein the linear layer includes a predetermined weight matrix and a predetermined bias vector; Inputting the linear transformation vector into an activation function to obtain a nonlinear vector, wherein the nonlinear vector performs weighted processing on key feature vectors and down-weighted processing on non-key feature vectors, wherein the key feature vector is a feature vector whose corresponding importance index is greater than a predetermined threshold, and the non-key feature vector is a feature vector whose corresponding importance index is less than or equal to the predetermined threshold; Combining the nonlinear vector with the attention weighted feature vector to obtain a combined vector; The gas concentration distribution data in the predetermined space is determined according to the combined vector.

4. The method according to claim 1, characterized in that: Based on multiple gas concentration data, determine the correlation relationship between gas concentrations at multiple predetermined points, including: Acquiring flow field data in the predetermined space; The correlation relationship between the gas concentrations of the plurality of predetermined points is determined according to the flow field data and the plurality of gas concentration data.

5. The method according to claim 1, characterized in that Based on multiple gas concentration data, determine the correlation relationship between gas concentrations at multiple predetermined points, including: Determine the first credibility indexes corresponding to the plurality of gas concentration data respectively according to the particle image velocity method, and determine the second credibility indexes corresponding to the plurality of gas concentration data respectively according to the concentration experiment method; Determining target credibility indexes corresponding to the plurality of gas concentration data respectively according to the plurality of first credibility indexes and the plurality of second credibility indexes; When the plurality of target credibility indexes are all greater than a predetermined credibility threshold, a correlation relationship between the gas concentrations of the plurality of predetermined points is determined based on the plurality of gas concentration data.

6. The method according to claim 1, characterized in that After determining the gas concentration distribution data in the predetermined space according to the spatial parameters of the predetermined space and the association relationship, the method further includes: In the case where the gas concentration distribution data is greater than a predetermined concentration distribution threshold, determining setting parameters corresponding to the safety equipment in the predetermined space; The safety device is controlled to be set to the setting parameters.

7. The method according to any one of claims 1 to 6, characterized in that: After determining the gas concentration distribution data in the predetermined space according to the spatial parameters of the predetermined space and the association relationship, the method further includes: a toxicity result to determine whether the gas is a toxic gas; In a case where the toxicity result indicates that the gas is a toxic gas, determining a toxicity distribution threshold of the toxic gas; In a case where the gas concentration distribution data is greater than the toxicity distribution threshold, a planned route corresponding to the predetermined space is determined based on the gas concentration distribution data.

8. A device for determining gas concentration distribution data, characterized in that: include: An acquisition module, used to acquire gas concentration data corresponding to multiple local points in a predetermined space; A first determination module is used to determine the correlation relationship between the gas concentrations of a plurality of predetermined points according to a plurality of gas concentration data, wherein the predetermined space includes the plurality of predetermined points, and the plurality of predetermined points include the plurality of local points; The second determination module is used to determine the gas concentration distribution data in the predetermined space according to the spatial parameters of the predetermined space and the association relationship.

9. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the method for determining gas concentration distribution data according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the gas concentration distribution data determination method according to any one of claims 1 to 7.

Images