Odor source identification method, device, storage medium and electronic device

Abstract

The present specification provides an odor source identification method, device, storage medium and electronic device, which detects the gas to be identified at a designated location through an olfactory sensor to obtain a sensor response feature, and determines each target odor source corresponding to the sensor response feature and the attribute information of each target odor source according to a pre-established odor source database, and constructs a factor graph model with each target odor source as a variable factor according to the sensor response feature and a pre-determined multivariable global function, wherein the multivariable global function uses the relative position of each target odor source to the designated location as a variable, and determines the position of each target odor source according to the factor graph model and the attribute information of each target odor source. Through this method, odor source identification can be accurately performed and the position information of each odor source can be determined.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to an odor source identification method, device, storage medium and electronic device. Background Art

[0002] At present, olfactory navigation technology, with its strong anti-interference, high concealment and excellent robustness, is gradually becoming an important supplement to traditional navigation methods such as GPS. In complex environments where GPS signals are limited, such as urban canyons, indoors and underwater, this technology shows unique advantages. At the same time, it can effectively deal with high-risk areas such as toxic and explosive areas, and provide reliable navigation solutions for small and confined spaces that are difficult for people to enter, showing a wide range of application potential and flexibility.

[0003] However, in practical applications, there are often multiple odor sources in the detection environment, and these odor sources may interfere with each other, resulting in the inability to identify the positions of each odor source in the detection environment and to accurately locate them.

[0004] Therefore, how to accurately locate each odor source in the detection environment has become a problem that needs to be solved urgently. Summary of the invention

[0005] The embodiments of this specification provide an odor source identification method, device, storage medium and electronic device to partially solve the problems existing in the above-mentioned prior art.

[0006] The embodiments of this specification adopt the following technical solutions:

[0007] This specification provides a method for identifying an odor source, the method comprising:

[0008] The gas to be identified at a designated location is detected by an olfactory sensor to obtain a sensor response characteristic;

[0009] Determining each target odor source corresponding to the sensor response characteristic and attribute information of each target odor source according to a pre-established odor source database;

[0010] Constructing a factor graph model with each target odor source as a variable factor according to the sensor response characteristics and a predetermined multivariate global function, wherein the multivariate global function has a relative position of each target odor source to the designated location as a variable;

[0011] The position of each target odor source is determined according to the factor graph model and the attribute information of each target odor source.

[0012] Optionally, the attribute information of each target odor source includes at least one of the category, concentration, distribution, direction and state of each target odor source.

[0013] Optionally, an odor source database is pre-established, specifically including:

[0014] Determine the source of each sample's odor;

[0015] For each sample odor source, the attribute information of the sample gas emitted by the sample odor source is determined by a standard chromatography-mass spectrometry system, and the sample gas is detected by the olfactory sensor to determine the sample response characteristics of the sample odor source;

[0016] According to the attribute information corresponding to each sample odor source and the sample response characteristics, an odor source database is constructed based on a preset database structure.

[0017] Optionally, determining each target odor source corresponding to the sensor response characteristic and attribute information of each target odor source specifically includes:

[0018] Feature matching the sensor response feature with the sample response feature of each sample odor source in the odor source database to determine each target sample response feature;

[0019] According to the target sample response characteristics, the target odor sources corresponding to the sensor response characteristics and the attribute information of the target odor sources are determined.

[0020] Optionally, a multivariable global function is predetermined, specifically including:

[0021] Determining an initial global function with the relative positions of each target odor source and the designated location as a variable;

[0022] A multivariate global function is determined based on predetermined olfactory sensor errors and bias errors, and the initial global function.

[0023] Optionally, constructing a factor graph model with each target odor source as a variable factor specifically includes:

[0024] Determining variable data corresponding to each target odor source according to the sensor response characteristics;

[0025] Determine, according to the variable data, variable nodes with the target odor sources as variables;

[0026] Decomposing the multivariable global function according to the variable data of each variable node to determine a plurality of local functions, wherein the local function takes at least a relative position of a target odor source and the designated location as a variable;

[0027] Determine each function node according to the plurality of local functions;

[0028] According to the variable nodes and the function nodes, a factor graph model with the target odor sources as variable factors is constructed.

[0029] Optionally, determining the position of each target odor source according to the factor graph model and the attribute information of each target odor source specifically includes:

[0030] For each target odor source, according to all function nodes and all variable nodes associated with the target odor source in the factor graph model, determine the edge probability function of the location of the target odor source by using a sum-product algorithm;

[0031] The positions of the target odor sources are determined according to the edge probability functions of the positions of the target odor sources.

[0032] This specification provides an odor source identification device, the device comprising:

[0033] A detection device, used to detect the gas to be identified at a designated location through an olfactory sensor to obtain a sensor response characteristic;

[0034] A determination device, for determining each target odor source corresponding to the sensor response characteristic and attribute information of each target odor source according to a pre-established odor source database;

[0035] A construction device, for constructing a factor graph model with each target odor source as a variable factor according to the sensor response characteristics and a predetermined multivariable global function, wherein the multivariable global function has a relative position of each target odor source to the designated location as a variable;

[0036] A positioning device is used to determine the position of each target odor source according to the factor graph model and the attribute information of each target odor source.

[0037] This specification provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned odor source identification method is implemented.

[0038] The present specification provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned odor source identification method when executing the program.

[0039] At least one of the above technical solutions adopted in the embodiments of this specification can achieve the following beneficial effects:

[0040] The embodiment of this specification provides a method for identifying an odor source, which detects the gas to be identified at a designated location through an olfactory sensor to obtain a sensor response feature, and determines each target odor source corresponding to the sensor response feature and the attribute information of each target odor source according to a pre-established odor source database, and constructs a factor graph model with each target odor source as a variable factor according to the sensor response feature and a pre-determined multivariable global function, wherein the multivariable global function uses the relative position of each target odor source to the designated location as a variable, and determines the position of each target odor source according to the factor graph model and the attribute information of each target odor source. Through this method, odor sources can be accurately identified and the position information of each odor source can be determined. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The drawings described herein are used to provide a further understanding of this specification and constitute a part of this specification. The illustrative embodiments and descriptions of this specification are used to explain this specification and do not constitute an improper limitation on this specification. In the drawings:

[0042] Figure 1 A schematic diagram of a method for identifying an odor source according to an embodiment of this specification;

[0043] Figure 2 is a schematic diagram of the factor graph model structure provided by the embodiments of this specification;

[0044] Figure 3 It is a message passing diagram of variable nodes and function nodes in the factor graph model provided by the embodiment of this specification;

[0045] Figure 4 A schematic diagram of the structure of an odor source identification device provided in an embodiment of this specification;

[0046] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this specification. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solutions and advantages of this specification more clear, the technical solutions of this specification will be clearly and completely described below in combination with the specific embodiments of this specification and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of this specification, not all of them. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this specification.

[0048] The technical solutions provided by the embodiments of this specification are described in detail below in conjunction with the accompanying drawings.

[0049] At present, olfactory navigation is difficult to be interfered by human factors in a large range. It has strong concealment and good robustness. It is an important supplement to the existing navigation methods. It can provide new navigation means for ground mobile equipment, near-Earth space vehicles, underwater mobile equipment, etc. It can play an important role in satellite denial, complex electromagnetic, toxic, explosive and dangerous environments, as well as in small spaces where people cannot enter. According to the characteristics of modern warfare, the application field of active olfactory navigation is also extensive, because olfactory navigation can be used for navigation and positioning of odor source destinations in the air, on the ground, underground and underwater. Therefore, it can be used for helicopters, drones, vehicles, submarines and other carriers to provide navigation support for military operations such as search and rescue, dangerous goods detection, emergency rescue, and security patrols. The key problems of olfactory navigation algorithms mainly include: the divergence of the distribution area of ​​odor source plumes causes the sensor to be unable to provide the target odor source concentration in time, the odor source positioning method in a dynamic environment has low adaptability and poor robustness for diverse plumes, and the target odor source destination positioning accuracy is low under the interference of multiple odor sources.

[0050] Based on this, an embodiment of this specification provides a method for identifying an odor source.

[0051] Figure 1 The following is a flow chart of an odor source identification method in this specification, which may specifically include the following steps:

[0052] S100: Detecting the gas to be identified at a designated location through an olfactory sensor to obtain a sensor response characteristic.

[0053] In the embodiments of this specification, a user may detect and locate the odor source in a specific area through an odor source identification system. First, the gas at the designated location is used as the gas to be identified. Then, the olfactory sensor is mounted on a movable platform such as a drone or a self-guided robot and moved to the designated location. The olfactory sensor is used to detect the gas to be identified to obtain the sensor response characteristics.

[0054] Among them, the olfactory sensor is a device that simulates the human sense of smell, can detect, identify and quantify gas components, and is usually composed of a sensor array and a data analysis system. Furthermore, the gas to be identified can be detected by the sensor array of the olfactory sensor to obtain an array response spectrum, and then the array response spectrum can be analyzed by the analysis system of the olfactory sensor to determine the sensor response characteristics.

[0055] Specifically, the sensor array of the olfactory sensor is the core part of the olfactory sensor. It is composed of multiple different types of chemical sensor elements, each of which is sensitive to specific gases or volatile organic compounds (VOCs). When the gas to be identified comes into contact with the sensor array, each sensor responds to the gas according to its sensitivity and selectivity. This response is usually manifested as a change in electrical signals, such as changes in resistance, capacitance or voltage.

[0056] Since the sensor array contains multiple types of sensors, they can simultaneously detect multiple components in the gas to be identified and generate a set of unique response signals. This set of signals is called an array response spectrum, which can characterize information such as the gas type, concentration, and component ratio corresponding to the gas to be identified.

[0057] Furthermore, after determining the array response spectrum, the analysis system of the olfactory sensor will pre-process the array response spectrum to remove noise, correct baseline drift, etc., to ensure the accuracy and reliability of the data. Then, the analysis system will extract key information that can characterize the gas characteristics from the pre-processed data, such as peak values, valley values, frequency, etc., as sensor response characteristics, which are used to characterize the category, concentration, proportion, distribution and direction of the odor source corresponding to each odor source.

[0058] Furthermore, the terminal device that detects the gas to be identified at the designated location through the olfactory sensor and obtains the sensor response characteristics may be a laptop computer, a tablet computer, etc. Of course, it may also be a server connected to the odor source identification system. For the sake of ease of description, this specification only takes the server as an example of the execution subject to illustrate the odor source identification method provided in this specification.

[0059] S102: Determine, according to a pre-established odor source database, each target odor source corresponding to the sensor response characteristic and attribute information of each target odor source.

[0060] Furthermore, after determining the sensor response characteristics, the target odor sources corresponding to the sensor response characteristics, i.e., the target odor sources contained in the gas to be identified, can be determined according to the pre-established odor source database. The odor source database mentioned here is a database containing sample response characteristics corresponding to the gas emitted by each known sample odor source. For each sample odor source in the odor source database, the database stores the sample response characteristics corresponding to the sample odor source, as well as the concentration, composition, proportion and other attribute information of the gas molecules of the sample gas corresponding to the sample odor source.

[0061] In other words, each record in the odor source database represents a known sample odor source, as well as the sample response characteristics and attribute information generated when the sample odor source is detected by the olfactory sensor under specific conditions. It should be noted that when the odor source database is pre-established, the olfactory sensor used to determine the sample response characteristics of the sample odor source is the same as the specification model of the olfactory sensor used to detect the gas to be identified and determine the sensor response characteristics.

[0062] Then, the sensor response features can be compared with the records in the odor source database by pattern recognition, machine learning or statistics to find the odor source record that best matches the sensor response features. The basis for matching can be the similarity of feature data, distance measurement (such as Euclidean distance, Manhattan distance, etc.) or probability model, etc. The specific data matching method is not limited in the embodiments of this specification. Through feature data matching, one or more target odor sources contained in the gas to be identified, as well as the attribute information of each target odor source, can be identified.

[0063] S104: Constructing a factor graph model with the target odor sources as variable factors according to the sensor response characteristics and a predetermined multivariable global function, wherein the multivariable global function has the relative positions of the target odor sources and the designated location as variables.

[0064] Specifically, the multivariable global function of the set of multiple target odor sources corresponding to the gas to be identified can be locally decomposed according to the multivariable global function and the sensor response characteristics, and expressed in the form of a product of related functions of multiple local variables, so that the factor graph model can be used to describe the joint probability distribution of the global function with multiple target odor sources as variable factors. The multivariable global function is a predetermined multivariable global function with the relative position of each target odor source and the designated location as a variable.

[0065] Furthermore, the factor graph model is a graph model used to represent and process the relationship between variables in a complex system. It consists of two types of nodes: variable nodes and function nodes. Variable nodes represent variables in the system, while function nodes represent the functional relationship between these variables. Specifically, in the factor graph model constructed with each target odor source as a variable factor, the variable data on the variable node is the different characteristics of each target odor source determined according to the sensor response characteristics, characterizing the category, concentration, state, direction and distribution of each target odor source. Each variable node corresponds to one or more attribute information of each target odor source. These attributes are used to distinguish different odor sources in the factor graph model, and the function node represents the interaction relationship between these sensor response characteristics.

[0066] Based on this, when constructing a factor graph model, the variable nodes in the model are defined according to the sensor response characteristics, and the function nodes in the model are defined according to the multivariable global function. The variable nodes and function nodes are connected according to the relationship between them to form a factor graph model. Each target odor source corresponds to a number of function nodes and variable nodes, where the function nodes represent the relationship between the odor source features and the variable nodes represent the specific feature data. By connecting these nodes, a complete factor graph model can be formed.

[0067] S106: Determine the position of each target odor source according to the factor graph model and the attribute information of each target odor source.

[0068] Furthermore, after constructing the factor graph model, the sum-product algorithm can be used to calculate each marginal probability function of the joint probability distribution with the relative position of each target odor source and the designated location as a variable, so as to determine the marginal probability distribution of the location of each target odor source, and thus determine the location information of each target odor source based on the marginal probability distribution.

[0069] Specifically, the sensor response characteristics can be input into the model as the initial variable data of the variable node. The function nodes in the factor graph are used for inference calculation to update the value of the variable node until a stable state is reached. According to the variable node value in the final stable state, the position of each target odor source contained in the gas to be identified is determined.

[0070] In other words, in the iterative calculation of the odor source identification model, when the value of either a variable node or a function node changes, it affects the values ​​of other nodes connected to the node. To reflect this impact, the changed node generates new messages that contain information about the current state of the node, such as updated values, probability distributions, etc., and passes this information to adjacent nodes.

[0071] Then, after receiving these messages, the adjacent nodes will update their values ​​according to their own functional relationships or constraints and the received messages. This process will be repeated until the values ​​of all nodes reach a stable state. That is, in the factor graph model, all the transmitted messages will be multiplied by their local functions and then complemented. Finally, after the root node receives all the transmitted messages, the edge probability function of the location of each target odor source is calculated.

[0072] In addition, the olfactory sensor can obtain the gas to be identified multiple times to improve the accuracy of the positioning information corresponding to each target odor source in the gas to be identified. Specifically, the nearby gas can be re-detected at the same position after a period of time to obtain the sensor response characteristics, or the nearby gas can be detected after the position of the olfactory sensor is moved to obtain the sensor response characteristics. When the factor model receives new measurement information from the olfactory sensor, that is, the new sensor response characteristics, it will update each variable factor in the factor graph model, that is, each target odor source, and the variable nodes and function nodes associated with each updated target odor source, it will iteratively self-evolve to update the odor source location information, and finally achieve accurate identification of the odor source.

[0073] based on Figure 1 A method for identifying an odor source is provided, which can construct a factor graph model according to sensor response characteristics and a pre-constructed multivariate global function, so as to accurately identify the odor source according to the factor graph model and determine the location information of each odor source.

[0074] Furthermore, the attribute information of each odor source includes at least one of the category, concentration, distribution, direction and state of each target odor source, and the specific attribute information is not limited in this embodiment of the specification.

[0075] Furthermore, pre-establishing an odor source database can specifically include determining each sample odor source, and for each sample odor source, determining the attribute information of the sample gas emitted by the sample odor source through a standard chromatography-mass spectrometry system, and detecting the sample gas through the olfactory sensor to determine the sample response characteristics of the sample odor source; and constructing an odor source database based on a preset database structure according to the attribute information and the sample response characteristics corresponding to each sample odor source.

[0076] Specifically, in order to ensure the integrity of the odor source database data, the objects emitting characteristic odors in each application scenario can be determined as sample odor sources, and after each sample odor source is determined, the sample gas emitted by the sample odor source is obtained. Specifically, the sample gas can be divided into two categories, one is a mixed sample gas that can be directly collected in each application scenario, and the other is a single sample gas that cannot be directly collected but can be prepared by a standard gas generator.

[0077] Furthermore, to obtain mixed sample gas, multi-point on-site sampling can be performed on the target application scenario where the sample odor source is located according to different application scenarios. The two gas sample collection methods of polyvinyl fluoride (PVF) air bag and polymer resin adsorption material can be used to effectively collect inorganic and organic odor components in the air to obtain mixed sample gas.

[0078] In addition, urban air testing usually requires the determination of the concentrations of sulfur dioxide (SO2), nitrogen oxides (NOx), carbon monoxide (CO), non-methane hydrocarbons, aromatic hydrocarbons, styrene, phenol, formaldehyde, methyl parathion, methyl isocyanate, carbon disulfide, chlorine, hydrogen chloride, sulfuric acid mist, hydrogen cyanide, hydrogen sulfide, ammonia and other gases. The production of important equipment is prone to produce gases such as hydrogen sulfide (H2S), nitrogen oxides (NO, NO2, N2O3), sulfur dioxide (SO2), carbon monoxide (CO), dichlorodiethyl sulfide (C4H8Cl2S), hydrazine (N2H4), etc. In disaster relief, the location of trapped people can be determined by detecting the concentration of carbon dioxide (CO2), acetone, ammonia, isoprene, methylbenzene, and n-hexane in human exhaled breath. Chemical parks such as petrochemicals, sulfuric acid production, and fertilizer industries often release sulfur dioxide (SO2), sulfur trioxide (SO3), hydrogen sulfide (H2S), dimethyl sulfate, methanol, propylene, butene and other gases. Heavy oil used as fuel for large military weapons such as aircraft carriers and fighter jets releases gases such as carbon monoxide (CO), carbon dioxide (CO2), nitrogen oxides (NOx), cycloalkanes, olefins, alkynes, aromatic compounds, etc. When burning, the above-determined sample gas can be prepared by using a standard gas generator to prepare a single gas commonly found in the target environment to obtain a single sample gas.

[0079] Next, for the sample gases emitted by each odor source, including mixed sample gases and single sample gases, a standard chromatography-mass spectrometry system (standard GC / MS system) can be used to separate the components, analyze the ingredients, and quantitatively analyze the concentration of the mixed odor samples to obtain the precise component information and concentration information of the sample odor, and simulate the complex odor distribution in different application scenarios to form odor component attribute information covering urban governance environment, chemical park environment, equipment processing plant environment, search and rescue environment, aircraft carrier environment, underwater cruise environment, emergency rescue environment, and military area environment. At the same time, the standard GC / MS system is used to detect pure single sample gases to determine the attribute information of the gas molecules of each single sample gas.

[0080] At the same time, for each sample odor emitted by each sample odor source, the sample gas is detected by the sensor array of the olfactory sensor to obtain a sample response spectrum, and the sample response characteristics of the sample gas are determined according to the sample response spectrum by the analysis system of the olfactory sensor, and the pattern response characteristics of the olfactory sensor for each single sample gas and the pattern response characteristics of the olfactory sensor for the mixed sample gas emitted by the sample odor source in each application scenario are obtained. Finally, the odor source database can be determined according to the attribute information and sample response characteristics of the gas molecules of the sample gas corresponding to each odor source, as well as the preset database structure.

[0081] Furthermore, determining each target odor source corresponding to the sensor response feature and the attribute information of each target odor source can be specifically performed by feature matching the sensor response feature with the sample response feature of each sample odor source in the odor source database to determine each target sample response feature, and determining each target odor source corresponding to the sensor response feature and the attribute information of each target odor source based on the target sample response features.

[0082] Specifically, after the odor source database is pre-built, the sensor response characteristics corresponding to the gas to be identified can be feature matched with the sample response characteristics stored in the odor source database. During the matching process, the similarity or distance between the sensor response characteristics and each sample response characteristic can be calculated. This similarity or distance may be calculated based on indicators such as the Euclidean distance, Manhattan distance, and cosine similarity of the feature vector. By comparing these indicators, the system can screen out the sample response characteristics that are closest to the sensor response characteristics, and use them as the target sample response characteristics.

[0083] Then, the sample odor sources corresponding to the target sample response characteristics in the odor source database are determined as the target odor sources corresponding to the sensor response characteristics, that is, the target odor sources corresponding to the gas to be identified. At the same time, the attribute information of each target odor source can also be determined according to the odor source database.

[0084] Furthermore, predetermining a multivariable global function can be specifically determining an initial global function with the relative positions of each target odor source and the designated location as variables, and determining the multivariable global function based on the predetermined olfactory sensor error and bias error, and the initial global function.

[0085] Specifically, the initial multivariable global function of each target odor source corresponding to the gas to be identified can be determined based on a pre-established odor source database by calculating the joint distribution function corresponding to each odor source of the system state over a period of time, wherein the initial multivariable global function is an initial global function with the relative positions of each target odor source and the designated location as variables.

[0086] In addition, in order to improve the accuracy of odor source identification, the error and bias error of the olfactory sensor also need to be considered. These errors may come from the manufacturing process, operating conditions or environmental factors of the sensor. By predetermining these errors, the response characteristic data of the sensor can be corrected, thereby improving the reliability of odor source identification. Specifically, the error and bias error of the olfactory sensor can be determined by using the olfactory sensor to perform multiple tests on the same gas at the same detection location and based on the results of multiple tests. Of course, it can also be determined by comparing the data obtained from the on-site test with the sample response characteristics in the pre-created odor source database. The specific error determination method is not limited in the embodiments of this specification.

[0087] Furthermore, the measurement information of the current variable node and the adjacent variable nodes of the olfactory sensor can be defined as and , multi-sensor error variables , which is related to the sensor error variable of the adjacent nodes The time update equation described is discretized, and the olfactory sensor error and bias error are defined to correct the multivariable global function:

[0088]

[0089]

[0090] in, (.) represents the cost function.

[0091] Then, according to the predetermined olfactory sensor error and bias error, and the initial multivariable global function, the multivariable global function corresponding to each odor source can be determined, and the specific formula is as follows:

[0092]

[0093] in represents the initial node, which can be determined by the sensor response characteristics of the olfactory sensor corresponding to the gas to be identified at the specified detection location, or the initial node can be randomly initialized. represents the olfactory sensor error function node and, represents the bias error function node, represents the initial multivariable global function node, [.] indicates a node cluster.

[0094] Furthermore, constructing a factor graph model with the target odor sources as variable factors can be specifically as follows: according to the sensor response characteristics, determining the variable data corresponding to the target odor sources, determining the variable nodes with the target odor sources as variables according to the variable data, decomposing the multivariable global function according to the variable data of the variable nodes to determine a number of local functions, wherein the local function takes at least the relative position of one target odor source and the designated location as a variable, determining the function nodes according to the several local functions, and constructing a factor graph model with the target odor sources as variable factors according to the variable nodes and the function nodes.

[0095] Specifically, for each target odor source corresponding to the gas to be identified, the target odor source can be used as a variable, and the variable data corresponding to the target odor source and the sub-variable data contained in each variable data can be determined according to the sensor response characteristics related to the target odor source in the sensor response characteristics, wherein each sub-variable data can be used to characterize the attribute information such as the category, concentration, direction, distribution and state corresponding to the odor source. Then, according to the variable data corresponding to the target odor source, the variable node corresponding to the target odor source and the sub-variable nodes contained in the variable node are determined.

[0096] Furthermore, after determining each variable node, the multivariable global function can be factorized according to the variable data of each variable node, that is, under the constraint of the sensor response characteristics, to obtain multiple local functions. Among them, these local functions are functions with at least one target odor source and the relative position of the designated location as variables, which may only involve some variables. For each local function, a corresponding function node is created in the factor graph model. The function node represents a certain constraint or relationship between the variable nodes, that is, the law described by the local function. The function node is connected to the variable nodes involved through edges, indicating that these variables participate in the function.

[0097] Specifically, the multivariate global function of the target odor source set is locally decomposed and expressed in the form of the product of the related functions of multiple local variables. Then, the factor graph model can be used to describe the joint probability distribution of the global function of multiple odor source variables. Can be decomposed into multiple local functions The form of the product of .

[0098]

[0099]

[0100] in, n variable nodes representing multiple target odor sources; , ,..., express The child variable node of represents the i-th local function, which corresponds to multiple function nodes associated with each target odor source, Represents a combination of different variable nodes. Variable nodes and function nodes are connected by edges. Only variable nodes related to local functions are connected to corresponding function nodes by edges.

[0101] Finally, by combining all the variable nodes, function nodes and connections related to each target odor source, a factor graph model with each target odor source as a variable factor is constructed. Figure 2 As shown, Figure 2 It is a structural diagram of the factor graph model. The small balls represent variable nodes, and the variable data represent attribute information such as category, concentration, distribution, direction and state. The small ball represents the function node, which is used to describe the connection relationship between variable nodes. represents the olfactory sensor error function node and, Represents the bias error function node.

[0102] Furthermore, based on the factor graph model and the attribute information of each target odor source, the position of each target odor source is determined. Specifically, for each target odor source, based on all function nodes and all variable nodes associated with the target odor source in the factor graph model, the edge probability function of the position of the target odor source is determined by the sum-product algorithm, and the position of each target odor source is determined based on the edge probability function of the position of the target odor source.

[0103] Specifically, in the factor graph model, for each target odor source, all function nodes and all variable nodes associated with the target odor source can be message-passed and calculated through the Sum-Product Algorithm to determine the edge probability function of the location of the target odor source, thereby determining the location information of the target odor source.

[0104] When solving the edge probability function of the target odor source, the entire solution process can be seen as a message passing process. For each child node, it will pass its own message to the associated parent node, and the parent node will continue to pass its own message to the next associated parent node. During the transmission process, different nodes will process the received messages in different ways. In the variable node, all the passed messages will be multiplied. In the function node, all the passed messages will be multiplied with their local functions and then complemented. Finally, after the root node receives all the passed messages, the calculation of the edge probability function is completed.

[0105] Specifically, Figure 3 As shown, in the global function and local functions During the sum-product operation, the function expression of the variable node passing the message to the function node is as follows:

[0106]

[0107] In the formula, For variable nodes Passed to function node news; For all nodes with variables Associated child function nodes A collection of; For child function nodes Pass to variable node news.

[0108] In the global function and local functions During the sum-product operation, the function expression of the function node passing the message to the variable node is as follows:

[0109]

[0110] In the formula, For function nodes Pass to variable node Message; For all nodes with function Associated child variable nodes A collection of; For child variable nodes Passed to function node news.

[0111] Through the above message passing process, each node of the factor graph model will update its own state according to the received messages. After the root node receives all the passed messages, the edge probability function of the location of each target odor source can be calculated. According to the edge probability function corresponding to each target odor source, the location information of each target odor source can be further determined and output. This information is the key output result of the odor source identification task and can be used to generate a regional odor source distribution map to help users understand and deal with odor source problems.

[0112] Furthermore, when new measurement information is received from the olfactory sensor, the variable factors corresponding to the new odor source will be added, and the corresponding variable nodes and function nodes will be updated. The odor source parameter information will be updated through iterative self-evolution, and ultimately the odor source will be accurately identified, and a regional odor source distribution map will be generated based on the accurate odor source information.

[0113]

[0114] in, It represents the variable factors updated after several iterations, realizing the self-evolution optimization of the model.

[0115] Since the model iteratively evolves by adding new variable factors and updating existing nodes when receiving new olfactory sensor measurement information, the factor graph model used for odor source identification can adapt to changes in the environment and continuously optimize the identification accuracy of odor sources. Through multiple iterations, the parameters in the model will gradually converge to more accurate values, thereby achieving model self-evolution optimization.

[0116] The above is the odor source identification method provided in the embodiments of this specification. Based on the same idea, this specification also provides corresponding devices, storage media and electronic devices.

[0117] Figure 4 A schematic diagram of a structure of an odor source identification provided in an embodiment of this specification, the device includes:

[0118] The detection device 400 is used to detect the gas to be identified at a designated location through an olfactory sensor to obtain a sensor response characteristic;

[0119] The determining means 402 is used to determine each target odor source corresponding to the sensor response characteristic and the attribute information of each target odor source according to a pre-established odor source database;

[0120] A construction device 404 is used to construct a factor graph model with each target odor source as a variable factor according to the sensor response characteristics and a predetermined multivariable global function, wherein the multivariable global function has a relative position of each target odor source and the designated location as a variable;

[0121] The positioning device 406 is used to determine the position of each target odor source according to the factor graph model and the attribute information of each target odor source.

[0122] Optionally, the attribute information of each target odor source includes at least one of the category, concentration, distribution, direction and state of each target odor source.

[0123] Optionally, the determination device 402 is specifically used to determine each sample odor source; for each sample odor source, determine the attribute information of the sample gas emitted by the sample odor source through a standard chromatography-mass spectrometry system, and detect the sample gas through the olfactory sensor to determine the sample response characteristics of the sample odor source; and construct an odor source database based on a preset database structure according to the attribute information and the sample response characteristics corresponding to each sample odor source.

[0124] Optionally, the determination device 402 is specifically used to perform feature matching on the sensor response characteristics with the sample response characteristics of each sample odor source in the odor source database to determine each target sample response characteristic; and determine each target odor source corresponding to the sensor response characteristic and attribute information of each target odor source based on the each target sample response characteristic.

[0125] Optionally, the construction device 404 is specifically used to determine an initial global function with the relative positions of each target odor source and the designated location as variables; and determine a multivariable global function based on a predetermined olfactory sensor error and bias error, and the initial global function.

[0126] Optionally, the construction device 404 is specifically used to determine the variable data corresponding to each target odor source according to the sensor response characteristics; determine the variable nodes with the target odor sources as variables according to the variable data; decompose the multivariable global function according to the variable data of the variable nodes to determine a number of local functions, wherein the local function has at least a relative position of one target odor source and the designated location as a variable; determine the function nodes according to the several local functions; and construct a factor graph model with the target odor sources as variable factors according to the variable nodes and the function nodes.

[0127] Optionally, the positioning device 406 is specifically used to determine, for each target odor source, the edge probability function of the location of the target odor source based on all function nodes and all variable nodes associated with the target odor source in the factor graph model by using a sum-product algorithm; and determine the location of each target odor source based on the edge probability function of the location of the target odor source.

[0128] This specification also provides a computer-readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, it can be used to perform the above Figure 1 A method for identifying an odor source is provided.

[0129] based on Figure 1 The odor source identification method shown in the present specification also provides Figure 5 The structural diagram of the electronic device shown in FIG. Figure 5 At the hardware level, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, and may also include other hardware required for the business. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to achieve the above Figure 1 The odor source identification method.

[0130] Of course, in addition to software implementation, this specification does not exclude other implementation methods, such as logic devices or a combination of software and hardware, etc., that is to say, the executor of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0131] In the 1990s, it was very clear whether the improvement of a technology was hardware improvement (for example, improvement of the circuit structure of diodes, transistors, switches, etc.) or software improvement (improvement of the method flow). However, with the development of technology, many improvements of the method flow today can be regarded as direct improvements of the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that the improvement of a method flow cannot be implemented with hardware entity modules. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system on a PLD by programming it themselves, without having to ask chip manufacturers to design and make dedicated integrated circuit chips. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one kind of HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages ​​and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.

[0132] The controller may be implemented in any suitable manner, for example, the controller may take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (e.g., software or firmware) executable by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller, examples of which include but are not limited to the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320, and the memory controller may also be implemented as part of the control logic of the memory. It is also known to those skilled in the art that, in addition to implementing the controller in a purely computer-readable program code manner, the controller may be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, such a controller may be considered as a hardware component, and the devices for implementing various functions included therein may also be considered as structures within the hardware component. Or even, the devices for implementing various functions may be considered as both software modules for implementing the method and structures within the hardware component.

[0133] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0134] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0135] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0136] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0137] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0138] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0139] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0140] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0141] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0142] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0143] It should be understood by those skilled in the art that the embodiments of this specification may be provided as methods, systems or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0144] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0145] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0146] The above description is only an embodiment of the present specification and is not intended to limit the present specification. For those skilled in the art, the present specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present specification shall be included in the scope of the claims of the present specification.

Claims

1. A method for identifying an odor source, characterized in that: The method comprises: The gas to be identified at a designated location is detected by an olfactory sensor to obtain a sensor response characteristic; Determining each target odor source corresponding to the sensor response characteristic and attribute information of each target odor source according to a pre-established odor source database; Constructing a factor graph model with each target odor source as a variable factor according to the sensor response characteristics and a predetermined multivariate global function, wherein the multivariate global function has a relative position of each target odor source to the designated location as a variable; Determining the location of each target odor source according to the factor graph model and the attribute information of each target odor source; Predetermined multivariable global functions, including: Determining an initial global function with the relative positions of each target odor source and the designated location as a variable; Determining a multivariate global function based on predetermined olfactory sensor errors and bias errors and the initial global function; Constructing a factor graph model with each target odor source as a variable factor, specifically including: Determining variable data corresponding to each target odor source according to the sensor response characteristics; Determine, according to the variable data, variable nodes with the target odor sources as variables; Decomposing the multivariable global function according to the variable data of each variable node to determine a plurality of local functions, wherein the local function takes at least a relative position of a target odor source and the designated location as a variable; Determine each function node according to the plurality of local functions; According to the variable nodes and the function nodes, a factor graph model is constructed with the target odor sources as variable factors; Determining the position of each target odor source according to the factor graph model and the attribute information of each target odor source specifically includes: For each target odor source, according to all function nodes and all variable nodes associated with the target odor source in the factor graph model, determine the edge probability function of the location of the target odor source by using a sum-product algorithm; The positions of the target odor sources are determined according to the edge probability functions of the positions of the target odor sources.

2. The method according to claim 1, characterized in that The attribute information of each target odor source includes at least one of the category, concentration, distribution, direction and state of each target odor source.

3. The method according to claim 1, characterized in that The odor source database is pre-established, including: Determine the source of each sample's odor; For each sample odor source, the attribute information of the sample gas emitted by the sample odor source is determined by a standard chromatography-mass spectrometry system, and the sample gas is detected by the olfactory sensor to determine the sample response characteristics of the sample odor source; According to the attribute information corresponding to each sample odor source and the sample response characteristics, an odor source database is constructed based on a preset database structure.

4. The method according to claim 3, characterized in that Determining each target odor source corresponding to the sensor response characteristic and attribute information of each target odor source specifically includes: Feature matching the sensor response feature with the sample response feature of each sample odor source in the odor source database to determine each target sample response feature; According to the target sample response characteristics, the target odor sources corresponding to the sensor response characteristics and the attribute information of the target odor sources are determined.

5. An odor source identification device, characterized in that: The device comprises: A detection device, used to detect the gas to be identified at a designated location through an olfactory sensor to obtain a sensor response characteristic; A determination device, for determining each target odor source corresponding to the sensor response characteristic and attribute information of each target odor source according to a pre-established odor source database; A construction device, for constructing a factor graph model with each target odor source as a variable factor according to the sensor response characteristics and a predetermined multivariable global function, wherein the multivariable global function has a relative position of each target odor source to the designated location as a variable; A positioning device, used to determine the position of each target odor source according to the factor graph model and the attribute information of each target odor source; The construction device is specifically used to determine an initial global function with the relative position of each target odor source and the designated location as a variable; determine a multivariable global function based on a predetermined olfactory sensor error and bias error, and the initial global function; The construction device is specifically used to determine the variable data corresponding to each target odor source according to the sensor response characteristics; determine the variable nodes with the target odor source as a variable according to the variable data; decompose the multivariable global function according to the variable data of each variable node to determine a plurality of local functions, wherein the local function has at least a relative position of a target odor source and the designated location as a variable; determine the function nodes according to the plurality of local functions; and construct a factor graph model with the target odor source as a variable factor according to the variable nodes and the function nodes; The positioning device is specifically used to determine, for each target odor source, the edge probability function of the location of the target odor source based on all function nodes and all variable nodes associated with the target odor source in the factor graph model through a sum-product algorithm; and determine the location of each target odor source based on the edge probability function of the location of the target odor source.

6. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 4 is implemented.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 4 is implemented.

Images