Gas identification methods, devices, equipment, media and products supporting online learning

Abstract

The present application relates to a gas identification method, device, equipment, medium and product supporting online learning. The method comprises: based on the network parameters of multiple pre-stored sample gases, identifying the gas type of the collected target gas, and obtaining the gas type identification result of the target gas; wherein each network parameter is generated when the gas types of multiple sample gases are pre-learned based on the bionic olfactory neural network. The method improves the recognition accuracy of the target gas by obtaining network parameters by learning multiple sample gases in advance.

Description

Technical Field

[0001] The present application relates to the field of gas detection technology, and in particular to a gas identification method, device, equipment, medium and product supporting online learning. Background Art

[0002] With the development of the chemical industry, the types of flammable, explosive and toxic gases are increasing, and the scope of application is also expanding. Once these gases leak during production, transportation and use, they will cause poisoning, fire and even explosion accidents, seriously endangering people's lives and property safety. Therefore, it is very important to detect gases quickly and accurately.

[0003] Usually, gas detection is achieved through an electronic nose, which is a bionic detection technology that simulates the olfactory system of mammals. The electronic nose includes a gas sensor, an analog front-end chip, and a gas recognition processor, among which the gas recognition processor plays an important role in determining the final gas recognition result of the electronic nose.

[0004] However, the gas identification processor in the related art has low identification accuracy when identifying gas. Summary of the invention

[0005] Based on this, it is necessary to provide a gas identification method, device, equipment, medium and product that supports online learning and can improve the accuracy of gas identification in response to the above technical problems.

[0006] In a first aspect, the present application provides a gas identification method. The method comprises:

[0007] Based on the pre-stored network parameters of the multiple sample gases, the gas type of the collected target gas is identified to obtain a gas type identification result of the target gas;

[0008] Among them, each network parameter is generated when the bionic olfactory neural network pre-learns the gas types of multiple sample gases.

[0009] In one embodiment, the bionic olfactory neural network includes an excitation layer and an inhibition layer, and the network parameters of each sample gas include a standard pulse emission time at the excitation layer and weight information of neuron synapses between the excitation layer and the inhibition layer. Based on the pre-stored network parameters of the multiple sample gases, the gas type of the collected target gas is identified to obtain a gas type identification result of the target gas, including:

[0010] When the gas identification processor identifies the target gas using the weight information of the neuron synapses corresponding to each sample gas, the target pulse emission time of the target gas at the excitation layer is obtained;

[0011] Compare the target pulse emission time with each standard pulse emission time;

[0012] The type of the sample gas corresponding to the standard pulse emission time having the highest similarity to the target pulse emission time is used as the gas type of the target gas.

[0013] In one embodiment, the method further comprises:

[0014] Quantify and normalize the target gas to obtain a gas standard digital signal;

[0015] Convert gas specification digital signal into pulse signal;

[0016] In one embodiment, the method further comprises:

[0017] The gas identification processor is controlled by an asynchronous logic control circuit to execute a process of identifying the gas type of the target gas.

[0018] In one embodiment, the asynchronous logic control circuit includes a signal control unit and a signal trigger; the gas identification processor includes a plurality of different functional modules, each functional module is connected to a signal control unit and a signal trigger;

[0019] The process of controlling the gas identification processor to identify the gas type of the target gas by an asynchronous logic control circuit includes:

[0020] Converting the start signal of the current functional module into a control signal through the signal control unit;

[0021] The control signal is input into the current functional module through the signal trigger, instructing the current functional module to start executing the corresponding function.

[0022] In one embodiment, the process of generating network parameters of multiple gases includes:

[0023] Obtain pulse signals of multiple sample gases;

[0024] The network parameters of the bionic olfactory neural network are updated through each pulse signal until the bionic olfactory neural network converges, thereby obtaining the network parameters of a variety of sample gases.

[0025] In one embodiment, the network parameters include weight information of neuron synapses between an excitatory layer and an inhibitory layer in a bionic olfactory neural network, and a pulse emission time of each sample gas at the excitatory layer, and the network parameters of the bionic olfactory neural network are updated by each pulse signal until the bionic olfactory neural network converges, including:

[0026] For each sample gas, obtaining a first pulse emission time of a pulse signal of the sample gas at an excitation layer and a second pulse emission time at an inhibition layer;

[0027] Obtaining a weight change according to a time difference between a first pulse emission time and a second pulse emission time;

[0028] Update the weight information of neuron synapses based on the weight change;

[0029] The iteration is performed multiple times, and if the weight change is zero, it is determined that the bionic olfactory neural network has converged.

[0030] In one embodiment, the method further comprises:

[0031] The bionic olfactory neural network is controlled by an asynchronous logic control circuit to execute a process of updating the network parameters of the bionic olfactory neural network through each pulse signal until the bionic olfactory neural network converges.

[0032] In one embodiment, the asynchronous logic control circuit includes a signal control unit and an enable control unit, and if the weight change is zero, determining that the bionic olfactory neural network converges includes:

[0033] If the weight change is zero, an enable signal is sent to a signal control unit connected to the excitation layer through the enable control unit to instruct the bionic olfactory neural network to converge.

[0034] In one embodiment, the method further comprises:

[0035] After the bionic olfactory neural network converges, the network parameters of multiple sample gases are stored.

[0036] In a second aspect, the present application also provides a gas identification device. The device comprises:

[0037] The gas identification processor is used to identify the gas type of the collected target gas based on the network parameters of multiple pre-stored sample gases, and obtain the gas type identification result of the target gas; wherein each network parameter is generated when the bionic olfactory neural network pre-learns the gas types of multiple sample gases.

[0038] In a third aspect, the present application further provides a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps in any one of the gas identification method embodiments in the first aspect are implemented.

[0039] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which implements the steps in any one of the gas identification method embodiments in the first aspect when executed by a processor.

[0040] In a fifth aspect, the present application further provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the steps in any one of the gas identification method embodiments in the first aspect are implemented:

[0041] The gas identification method, device, equipment, medium and product supporting online learning provided in the embodiments of the present application identify the gas type of the collected target gas based on the network parameters of multiple pre-stored sample gases to obtain the gas type identification result of the target gas. Among them, each network parameter is generated when the bionic olfactory neural network pre-learns the gas types of multiple sample gases. This method learns multiple sample gases in advance, pre-stores the network parameters of multiple sample gases, calls the pre-stored network parameters when identifying the target gas, and by identifying the network parameters of the target gas, the sample gas can continue to be learned after the target gas is identified, thereby improving the gas identification accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 A diagram showing an application environment of a gas identification method in an embodiment;

[0043] Figure 2 A schematic diagram of a gas identification method according to an embodiment;

[0044] Figure 3 A schematic flow chart of a gas identification method in another embodiment;

[0045] Figure 4 A schematic flow chart of a gas identification method in another embodiment;

[0046] Figure 5 A schematic flow chart of a gas identification method in another embodiment;

[0047] Figure 6 A circuit logic diagram of a gas identification method in one embodiment;

[0048] Figure 7 A schematic flow chart of a gas identification method according to another embodiment;

[0049] Figure 8 A schematic flow chart of a gas identification method in another embodiment;

[0050] Fig. 9 A circuit logic diagram of a gas identification method in another embodiment;

[0051] Fig.10 A circuit diagram of a signal control unit of a gas identification method in one embodiment;

[0052] Fig.11 A circuit diagram of a signal control unit of a gas identification method in another embodiment;

[0053] Fig.12 A circuit diagram of a signal control unit of a gas identification method in another embodiment;

[0054] Fig.13 A schematic flow chart of a gas identification method in another embodiment;

[0055] Fig.14 A schematic flow chart of a gas identification method in another embodiment;

[0056] Fig.15 is a structural block diagram of a gas identification device in one embodiment;

[0057] Fig.16 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0059] The gas identification method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the collection device 102 communicates with the gas identification device 104 through the network. The collection device 102 can be a sensor array composed of multiple sensors, the gas identification device 104 is used to identify the gas, and the collection device 102 transmits the collected data to the gas identification device 104. Optionally, the gas identification device 104 is set in the electronic nose, and the collection device 102 interacts with the electronic nose and transmits the collected data to the electronic nose. Optionally, the collection device 102 and the gas identification device can be set in the electronic nose at the same time, and the data collection and gas identification are completed electronically.

[0060] With the increase of toxic gases, it is very important to detect the gas type quickly and accurately, so that once a gas leak accident occurs, remedial measures can be taken as soon as possible through real-time and rapid detection to minimize the loss of the accident.

[0061] As a bionic detection technology that simulates the olfactory system of mammals, the electronic nose has a long history of development and is widely used in food safety, environmental monitoring, agriculture, industry, public safety, etc. However, early electronic noses were bulky, had single functions, and had limited usage scenarios, making it difficult to achieve real-time on-site monitoring of gases. As a result, integrated, intelligent, and portable electronic noses emerged.

[0062] The portable electronic nose includes a gas sensor, an analog front-end chip, and a gas identification processor. Among them, the gas identification methods used by the commonly used gas identification processor include but are not limited to support vector machines, K nearest neighbors, and decision trees. However, these commonly used gas identification methods have the disadvantages of low recognition accuracy and few types of recognition, and can only recognize up to 5 mixed gases at the same time. More importantly, the commonly used gas identification processor has no learning ability, and does not have the function of dynamically adjusting parameters and structures. Under the influence of gas sensor drift (sensor sensitivity and selectivity will gradually change with the passage of time), the parameters stored in the gas identification processor often do not match the actual situation, resulting in a significant decrease in actual recognition accuracy. In addition, the commonly used gas identification processor has a large computing power requirement, and the power consumption is often in the milliwatt level. In the absence of an independent power supply, a portable electronic nose with this as a built-in chip can only work for a few hours, and it is difficult to apply it to long-term standby tasks such as post-disaster search and rescue.

[0063] Based on this, the embodiments of the present application provide a gas identification method, device, equipment, medium and product supporting online learning, which can identify the gas type of the target gas and adjust the network parameters in real time before each identification of the gas type of the target gas. It can identify multiple gas types, improve the identification accuracy, and use asynchronous logic circuits to overcome the disadvantage of high power consumption.

[0064] The gas identification method provided in the embodiment of the present application is described below through specific examples.

[0065] In one embodiment, Figure 2 As shown, a gas identification method is provided, comprising the following steps:

[0066] S201, based on pre-stored network parameters of multiple sample gases, identifying the gas type of the collected target gas to obtain a gas type identification result of the target gas.

[0067] The target gas is a gas whose gas type needs to be identified by a gas identifier. The target gas may be a flammable, explosive, or toxic gas. For example, ammonia, nitrogen dioxide, sulfur dioxide, or hydrogen sulfide may be identified. The target gas may be collected by a sensor. For example, it may be collected by a metal oxide semiconductor conductivity gas sensor, a surface acoustic wave gas sensor, or a quartz crystal microbalance gas sensor. The gas identification processor is a module in the electronic nose. By inputting the collected target gas into the gas identification processor, the gas type of the target gas may be obtained.

[0068] Among them, the gas identification processor includes a bionic olfactory neural network, and each network parameter is generated when the bionic olfactory neural network is pre-learned for the gas types of multiple sample gases. The bionic olfactory neural network is a mathematical model that simulates certain mechanisms of biological olfaction, that is, it imitates the process in which a neuron is activated in a biological neuron and a signal is transmitted to other neurons. The sample gas is the sample gas used when learning the bionic olfactory neural network, and the network parameters of the multiple sample gases pre-stored in the gas identification processor are obtained in the learning mode of the gas identifier. In the process of identifying the target gas, based on the network parameters of the multiple sample gases pre-stored, the identification result of the target gas will be output, and this target result is one of the sample gases. The identification of the gas type of the target gas is performed in the working mode of the gas identifier. Exemplarily, two sample gases, ammonia and nitrogen dioxide, are learned, and the gas type of the target gas is identified based on the network parameters of the two sample gases. Specifically, the identification result can be that the target gas is nitrogen dioxide.

[0069] In the gas identification method provided in the embodiment of the present application, the gas type of the collected target gas is identified based on the network parameters of the pre-stored multiple sample gases to obtain the gas type identification result of the target gas. Among them, each network parameter is generated when the bionic olfactory neural network pre-learns the gas types of multiple sample gases. This method learns multiple sample gases in advance, pre-stores the network parameters of multiple sample gases, and calls the pre-stored network parameters when identifying the target gas. By identifying the network parameters of the target gas, the sample gas can continue to be learned after the target gas is identified, thereby improving the gas identification accuracy.

[0070] In the embodiment of the present application, the gas type of the target gas is identified based on the network parameters of the pre-stored multiple sample gases. First, the network parameters of the pre-stored multiple sample gases need to be obtained. Based on this, in one embodiment, the bionic olfactory neural network includes an excitation layer and an inhibition layer. The network parameters of each sample gas include a standard pulse emission time at the excitation layer and weight information of the neuron synapses between the excitation layer and the inhibition layer. Then, based on the network parameters of the pre-stored multiple sample gases, the gas type of the collected target gas is identified to obtain a gas type identification result of the target gas, such as Figure 3 As shown, including:

[0071] S301, obtaining the target pulse emission time of the target gas at the excitation layer when the gas identification processor identifies the target gas using the weight information of the neuron synapses corresponding to each sample gas.

[0072] Neurons are the basic information processing units in the bionic olfactory neural network. The bionic olfactory neural network is composed of all neurons in the excitatory layer and the inhibitory layer, and is used to simulate the transmission characteristics of biological neuron information, that is, to transmit information under pulse triggering with the help of synapses connecting neurons. Specifically, the excitatory layer neurons in the present application use an integral firing neuron circuit (Integrate and Fire, IF), and the excitatory layer neurons use a leaky integral triggering neuron circuit (Leaky Integrate-and-Fire, LIF). The synaptic connection between the two is an excitatory synapse, and each excitatory layer IF neuron corresponds to five inhibitory layer LIF neurons. Specifically, these six neurons form a group, and each group of five inhibitory layer LIF neurons are connected to different groups of excitatory layer IF neurons, and the connection between the two is considered an inhibitory synapse. Specifically, the number of neuron groups in the present application is at least twelve groups, that is, at least twelve excitatory layer IF neurons are required.

[0073] The target pulse emission time is the time it takes for the pulse information of the target gas to pass through the IF neuron pulse of the excitatory layer. After learning multiple sample gases, the gas identifier using the bionic olfactory neural network will obtain the weight information of the corresponding neuron synapses of multiple sample gases, which is represented by w here. n For example, if two sample gases are learned, the weight information of the corresponding neuron synapses of the multiple sample gases is expressed as w1 and w2. Specifically, if there are twelve groups of neurons at this time, each w n It is a set of weight information corresponding to twelve groups of neurons.

[0074] When studying the sample gas, we will also get the standard pulse emission time of the sample gas at the excited layer, which is expressed as S LnFor example, if two sample gases are learned, the weight information w1 and w2 of the corresponding neuron synapses of the two sample gases are used to obtain the standard pulse emission time S of the two sample gases at the excitation layer. L1 and S L2 , each sample gas corresponds to a neuron synaptic weight information and a standard pulse emission time. Specifically, if there are twelve groups of neurons, each neuron has its own pulse emission time. At this time, each S Ln It is a set of twelve groups of standard pulse firing times of neurons in the excitatory layer.

[0075] When identifying the target gas, the weight information w of the corresponding neuron synapses of multiple sample gases is called respectively. n , obtain the target pulse emission time of multiple target gases in the excited layer, here S tn Indicates that, for example, two sample gases are learned, and S corresponding to w1 and w2 are obtained t1 and S t2 , specifically, each S obtained tn They are all sets of multiple target pulse emission times.

[0076] S302, comparing the target pulse emission time with each standard pulse emission time.

[0077] The target pulse emission time S obtained when identifying the gas tn and the standard pulse emission time S obtained when studying the sample gas Ln For comparison, for example, the standard pulse emission time S obtained by a sample gas under the weight information w1 of the neuron synapse L1 and the target pulse emission time S obtained by using the weight information w1 in the target gas t1 Compare, specifically, the standard pulse emission time S L1 and the target pulse emission time S t1 Both are sets with the same number, and the pulse emission times of corresponding positions in the two sets are compared to see if they are the same.

[0078] S303 , taking the type of the sample gas corresponding to the standard pulse emission time having the highest similarity to the target pulse emission time as the gas type of the target gas.

[0079] Comparison with standard pulse emission time S L1 and the target pulse emission time S t1, mainly compares the number of pulse emission times at corresponding positions in the two sets, and takes the type of sample gas corresponding to the standard pulse emission time with the highest similarity to the target pulse emission time as the gas type of the target gas. The number of comparisons is determined by the number of samples learned. For example, after learning two sample gases, the standard pulse emission time S L1 , S L2 and the target pulse emission time S t1 and S t2 , S L1 and S t1 For comparison, S L2 and S t2 For comparison, if S L1 and S t1 When making a comparison, if the number of pulse emission times at corresponding positions in the two sets is the same is the largest, the corresponding learned sample gas type is used as the gas type of the target gas.

[0080] In the embodiment of the present application, when the gas identification processor identifies the target gas with the weight information of the neuron synapses corresponding to each sample gas, the target pulse emission time of the target gas at the excitation layer is obtained; the target pulse emission time is compared with each standard pulse emission time; the type of sample gas corresponding to the standard pulse emission time with the highest similarity to the target pulse emission time is used as the gas type of the target gas. Before identifying the target gas, multiple sample gases are learned in advance in the learning mode and compared with the network parameters of the target gas in the working mode, thereby improving the gas identification accuracy.

[0081] In the embodiment of the present application, the network parameters of each sample gas are obtained, and the gas type of the target gas can be identified by the pre-stored network parameters of multiple sample gases. Before learning the sample gas or identifying the target gas, it is necessary to perform data preprocessing on the digital signals of the sample gas and the target gas output by the analog front end input into the bionic olfactory neural network. Based on this, in one embodiment, Figure 4 As shown, the method also includes:

[0082] S401, quantifying and normalizing the target gas to obtain a gas normalized digital signal.

[0083] S402, converting the gas standard digital signal into a pulse signal.

[0084] Data quantization and normalization processing refers to the necessary preprocessing such as screening and sorting before the collected data is classified or sorted. In the embodiment of the present application, data preprocessing is performed on the digital signal of the target gas output by the analog front end. Specifically, the digital signal of the target gas can be normalized to the minimum-maximum value, the original digital signal can be linearly transformed, the data can be mapped to a 4-bit digital signal interval of 0 to 15, and the data can be sorted from large to small to obtain a gas standard digital signal.

[0085] Pulse coding is to convert the gas standard digital signal into a random pulse. Specifically, pulse coding adopts an event trigger mechanism expressed by an address event. When data is transmitted, the gas standard digital signal is first digitally encoded. This application does not limit the digital encoding method. Then, the encoded digital signal is pulse decoded to generate a random pulse, which refers to a sudden change in the level state in the electronic circuit, which can be either a sudden increase (the rising edge of the pulse) or a sudden decrease (the falling edge of the pulse). Generally, after the level mutation, the pulse will return to its original level state in a very short time.

[0086] In the real world, data is generally incomplete and inconsistent, lacks unified standards and definitions, and has large differences in data structure, making it impossible to directly mine data. Data preprocessing of digital signals can improve the quality of data mining, reduce the time required for actual mining, and obtain valuable information resources.

[0087] When realizing the function of identifying the gas type of the target gas, it needs to be controlled by an asynchronous logic control circuit. Based on this, in one embodiment, the method also includes: controlling the gas identification processor to execute the process of identifying the gas type of the target gas by an asynchronous logic control circuit.

[0088] Asynchronous logic circuits are circuits whose input and output are not driven by the same clock or causally related clocks. Asynchronous circuits are sometimes also called clockless or self-timing circuits, that is, circuits without global clocks. The Request and Acknowledge signals sent by the previous and next stages inform the other party whether the data can be accepted. Specifically, in asynchronous logic circuits, a control signal is input to trigger the gas identification processor to identify the gas type of the target gas.

[0089] Asynchronous logic circuits are used to replace synchronous logic circuits. Synchronous circuits work under the same global clock, and the clock operating frequency must meet the maximum load requirements, resulting in power consumption waste. Asynchronous logic circuits are driven by data, removing the clock signal. At the same time, the gas identification processor is directly activated only by the input control signal, reducing dynamic power consumption and achieving high-speed, low-power application requirements.

[0090] In the embodiment of the present application, the asynchronous logic control circuit controls the gas identification processor to perform the identification of the gas type of the target gas, which needs to be implemented through multiple different functional modules included in the gas identification processor. In one embodiment, the asynchronous logic control circuit includes a signal control unit and a signal trigger; the gas identification processor includes multiple different functional modules, each of which is connected to a signal control unit and a signal trigger; the asynchronous logic control circuit controls the gas identification processor to perform the process of identifying the gas type of the target gas, such as Figure 5 As shown, including:

[0091] S501, converting the start signal of the current functional module into a control signal through a signal control unit.

[0092] The signal control unit is used to convert the start signal into a control signal. Specifically, when the start signal is flipped, an output pulse, that is, a control signal, is generated. The rising edge of the control signal drives the trigger, thereby completing the logic of the circuit. Among them, the signal control unit is a Click control unit.

[0093] S502, inputting a control signal into the current functional module through a signal trigger, instructing the current functional module to start executing a corresponding function.

[0094] In the embodiment of the present application, the signal trigger is a trigger without a clock, which is not triggered by a clock signal to transmit and process data. The signal trigger is controlled by a control signal generated by an asynchronous logic circuit signal control unit.

[0095] Specifically, the digital signal of the target gas output by the analog front end is first temporarily stored in the trigger, and after receiving the control signal, the trigger is triggered to transmit and process the temporarily stored data. Figure 6 FIG. 1 is a flowchart of a process in which an asynchronous logic control circuit controls a gas identification processor to perform identification of the gas type of a target gas. Figure 6 In the process, the digital signal of the target gas output by the analog front end is first temporarily stored in the trigger, a start signal 1 is input into the signal control unit 1, the start signal 1 is converted into a control signal 1, and then the control signal is sent to the trigger, so that the trigger transfers and preprocesses the temporarily stored data, and the preprocessed data is further temporarily stored in the next trigger. Then the start signal 1 of the previous step is transferred to the next signal control unit 2 as the start signal 2, and so on, until all signal control units generate control signals, triggering the completion of data transmission and processing.

[0096] In actual applications, the trigger can be a D type flip-flop (D type Flip-Flop, DFF), the start signal can be expressed as Reqin, the control signal can be expressed as fire, and the control unit can be expressed as Click. The embodiment of the present application does not limit the specific reference to each part.

[0097] In the process of controlling the gas identification processor to perform gas type identification of the target gas through an asynchronous logic control circuit, it is mainly completed through the functional modules therein. Data transmission and processing can only be triggered after receiving the control signal from the signal control unit. That is to say, one control signal is allowed to be input at a time to avoid the risk of competition between input signals.

[0098] Based on the network parameters of multiple sample gases pre-stored in the gas identification processor, the gas type of the target gas is identified. Based on this, in one embodiment, the generation process of the network parameters of multiple gases is as follows: Figure 7 As shown, including:

[0099] S701, acquiring pulse signals of multiple sample gases.

[0100] S702, updating the network parameters of the bionic olfactory neural network through each pulse signal until the bionic olfactory neural network converges, and obtaining the network parameters of a plurality of sample gases.

[0101] It should be noted that the pulse signal of the sample gas corresponds to the type of the sample gas. Since the embodiment of the present application identifies different types of gases through a bionic olfactory neural network, it is necessary to obtain pulse signals of multiple sample gases.

[0102] After obtaining pulse signals of multiple sample gases, the pulse signals of multiple sample umbilical stickers are used as training sets to train the bionic olfactory neural network. The network parameters of the bionic olfactory neural network will be updated as the training process progresses. At the same time, the training process of the bionic olfactory neural network can be visualized, and the degree of convergence of the training of the bionic olfactory neural network can be judged from the display results. When the training process of the bionic olfactory neural network tends to converge, it can be judged that the training of the bionic olfactory neural network is completed, and the network parameters of the trained bionic olfactory neural network are used as the network parameters of multiple sample gases.

[0103] In the embodiment of the present application, the pulse signals of multiple sample gases are used to update the network parameters, which can improve the training accuracy of the model while improving the generalization ability of the olfactory neural network. In addition, the network parameters of the sample gases obtained in the embodiment of the present application are obtained through pre-training, which means that in the actual gas identification process, there is no need to retrain the network parameters of multiple gases, thereby making the gas identification process faster.

[0104] The gas type of the target gas is identified based on the network parameters of the multiple sample gases pre-stored in the gas identification processor. The network parameters of each sample gas need to be obtained through the calculation of the neural network. Based on this, in one embodiment, the network parameters include the weight information of the neuron synapses between the excitation layer and the inhibition layer in the bionic olfactory neural network, and the pulse emission time of each sample gas at the excitation layer. The network parameters of the bionic olfactory neural network are updated through each pulse signal until the bionic olfactory neural network converges, such as Figure 8 As shown, including:

[0105] S801 , for each sample gas, obtaining a first pulse emission time of a pulse signal of the sample gas at an excitation layer and a second pulse emission time at an inhibition layer.

[0106] The pulse signal is the signal transmitted to the excitation layer neurons, which is obtained through the target signal. When the bionic olfactory neural network learns each sample gas, it must wait until the bionic olfactory neural network converges, that is, the learning ends after the weights of the synapses between neurons stop updating.

[0107] In the bionic olfactory neural network, the pulse time emitted by the excitatory layer neurons is the first pulse emission time, denoted by t1, and the pulse time emitted by the inhibitory layer neurons is the second pulse emission time, denoted by t2.

[0108] S802, obtaining a weight change according to a time difference between a first pulse emission time and a second pulse emission time.

[0109] The time difference can be obtained according to the emission time of the first pulse and the emission time of the second pulse, expressed as Δt=t2-t1, and the obtained time difference Δt is a set composed of multiple time differences. The weight change Δw can be calculated according to the time difference. The weight change Δw is a set composed of weight changes corresponding to multiple groups of neuron synapses. The corresponding weight change will be calculated for each synapse between neurons. For example, if there are twelve groups of neurons, there will be twelve sets of time differences Δt and twelve weight changes Δw.

[0110] Specifically, an exponential operation is performed according to the time difference Δt between the first pulse emission time and the second pulse emission time and the synaptic coefficient between the corresponding neurons to obtain the synaptic weight change Δw, and the weight change is output according to the required accuracy. The weight coefficient of each synapse is given autonomously. Exemplarily, an exponential operation can also be performed in advance according to the time difference between the first pulse emission time and the second pulse emission time and the synaptic coefficient to obtain the corresponding weight change, and the time difference between the first pulse emission time and the second pulse emission time, the synaptic coefficient and the corresponding weight change are stored in a lookup table. After obtaining the time difference, the corresponding weight change is found in the lookup table according to the segmented LUT method to update the weight information.

[0111] S803, updating the weight information of the neuron synapse based on the weight change amount.

[0112] The weight information is updated through the corresponding synaptic weight change Δw between neurons. Specifically, each synapse between neurons will have some corresponding weight change in the weight change Δw. Each synapse between neurons is adjusted up or down according to the corresponding weight change to complete the update of the weight information of the neuron synapse.

[0113] Specifically, when the weight information of the neuron synapse is updated for the first time, the weight change of the weight information of the neuron synapse is calculated by the time difference Δt obtained by the first pulse emission time t1 and the second pulse emission time t2. When the weight information of the neuron synapse is updated for the second time, the weight information of the neuron synapse updated for the first time is used. At this time, the emission time of the first pulse emitted by the neuron in the excitation layer is t3. The time difference used in this round of update is the time difference Δt between the second pulse emission time t2 obtained in the previous round and the first pulse emission time t3 in this round.

[0114] S804, iteratively executing multiple times, if the weight change is zero, it is determined that the bionic olfactory neural network has converged.

[0115] The weight information of the neuron synapse is updated according to the weight change obtained after each pulse transmission based on the time difference between the first pulse emission time and the second pulse emission time, until the obtained weight change is zero, then the cycle ends, indicating that the sample gas learning is completed.

[0116] When learning a sample gas, the network parameters of the bionic olfactory neural network are updated through each pulse signal until the neural network converges, indicating that the learning of the sample gas is completed. At this time, the network parameters obtained are accurate, which is compared with the network parameters of the target gas in the working mode, thereby improving the gas recognition accuracy.

[0117] The generation process of the network parameters of the above-mentioned multiple gases also needs to be controlled by an asynchronous logic control circuit to complete the training of the bionic olfactory neural network. Based on this, in one embodiment, the method also includes: controlling the bionic olfactory neural network through the asynchronous logic control circuit to execute the process of updating the network parameters of the bionic olfactory neural network through each pulse signal until the bionic olfactory neural network converges.

[0118] Before the gas identification processor executes the process of identifying the gas type of the target gas, it is also necessary to control the bionic olfactory neural network through an asynchronous logic control circuit for the sample gas to update the network parameters of the bionic olfactory neural network through each pulse signal until the bionic olfactory neural network converges.

[0119] like Fig. 9 As shown, the signal control units 3 to 8 are used to control the bionic olfactory neural network through the asynchronous logic control circuit to update the network parameters of the bionic olfactory neural network through each pulse signal until the bionic olfactory neural network converges. Fig.10 and Fig.11 As shown, the signal control unit 3 and the signal control unit 3 are both multi-input and multi-output signal control units. Specifically, the signal control unit 3 transmits the start signal 3 to the signal control unit 4 and the signal control unit 6 respectively, which represents the transmission of the pulse signal to the neurons of the inhibitory layer and the first pulse emission time of the pulse signal of the sample gas at the excitation layer to the weight update module. The signal control unit 6 will convert the two start signals into the control signal 6 to trigger the trigger to update the weight. After receiving the start signal 3 after the updated weight, the signal control unit 3 continues to transmit the pulse signal until the weight update is completed. Similarly, the signal control unit can be a Click control unit, the start signal can be expressed as Reqin, and the control signal can be expressed as fire. The embodiments of the present application do not limit the specific references to each part.

[0120] When learning a sample gas, in the process of each pulse signal updating the network parameters of the bionic olfactory neural network, it is also necessary to control the gas recognition processor through an asynchronous logic control circuit to execute the update of the network parameters. Only after receiving the control signal from the signal control unit can the data be triggered for transmission and processing. In other words, one control signal is allowed to be input at a time to avoid the risk of competition between input signals.

[0121] In one embodiment, the asynchronous logic control circuit includes a signal control unit and an enable control unit. If the weight change is zero, determining that the bionic olfactory neural network converges includes:

[0122] If the weight change is zero, an enable signal is sent to a signal control unit connected to the excitation layer through the enable control unit to instruct the bionic olfactory neural network to converge.

[0123] The signal control unit 3 is an enabling control unit with an enabling signal, such as Fig.12 As shown, if the calculated weight change is zero during the weight update, the enable control unit sends an enable signal to the signal control unit connected to the excitation layer, turning off the rising edge, so it will no longer be converted into a control signal. The network parameters stored in the weight storage module are the final results.

[0124] When learning a sample gas, in the process of updating the network parameters of the bionic olfactory neural network with each pulse signal, it is also necessary to control the signal control unit with an enable signal in the gas identification processor through an asynchronous logic control circuit to instruct the bionic olfactory neural network to converge. The sample gas learning can be judged to be completed in time, and the network parameters of the sample gas can be stored.

[0125] Before identifying the gas type of the target gas, it is necessary to call the pre-stored network parameters of multiple sample gases. These identification parameters need to be stored in advance for subsequent calling. Based on this, in one embodiment, the method further includes:

[0126] After the bionic olfactory neural network converges, the network parameters of multiple sample gases are stored.

[0127] After the bionic olfactory neural network converges, that is, after learning multiple sample gases, the network parameters of each sample gas will be obtained. The network parameters of each sample gas include the standard pulse emission time at the excitation layer, and the weight information of the neuron synapses between the excitation layer and the inhibitory layer. Specifically, when the weight information of the neuron synapses between the excitation layer and the inhibitory layer has not been updated, the network parameters at this time are temporarily stored in the buffer group until the weight information is updated, and the final network parameters are stored. When identifying the target gas, it is necessary to obtain the target pulse emission time of the target gas based on the weight information of the neuron synapses between the excitation layer and the inhibitory layer, and then compare it with the standard pulse emission time of each sample gas at the excitation layer. Therefore, the network parameters of multiple sample gases are stored.

[0128] When learning sample gases, there is no need to discard previously learned data. Instead, the new sample gas data needs to be re-input into the processor and the network parameters of multiple sample gases can be stored. This allows for a learn-work-relearn model, overcoming the algorithmic challenge of catastrophic forgetting.

[0129] In one embodiment, Fig.13 As shown in the figure, it is another application environment diagram of the gas identification method. The corresponding specific steps are as follows Fig.14As shown, the above-mentioned gas identification method comprises the following steps:

[0130] S901, quantifying and normalizing the sample gas through a preprocessing module to obtain a gas normalized digital signal;

[0131] S902, converting the gas specification digital signal into a pulse signal through a pulse coding module, and inputting the pulse signals of the multiple sample gases into a gas identification processor;

[0132] S903, for each sample gas, obtaining the first pulse emission time of the pulse signal of the sample gas at the excitation layer;

[0133] S904, for each sample gas, obtaining a second pulse emission time of a pulse signal of the sample gas at the suppression layer;

[0134] S905, obtaining a weight change amount according to a time difference between a first pulse emission time and a second pulse emission time;

[0135] Specifically, the connection between the excitatory layer and the inhibitory layer is described by a synaptic address generator, which is composed of a buffer and a lookup table.

[0136] S906, updating the weight information of the neuron synapse based on the weight change;

[0137] S907, iteratively executing multiple times, if the weight change is zero, it is determined that the bionic olfactory neural network has converged, and the sample gas learning is completed;

[0138] S908, storing network parameters of multiple sample gases;

[0139] S909, obtaining the target pulse emission time of the target gas at the excitation layer when the gas identification processor identifies the target gas using the weight information of the neuron synapse corresponding to each sample gas;

[0140] Specifically, the classifier is used to call the stored weight information of the neuron synapse. The above S901-S907 are the work performed by the gas identifier in the learning mode.

[0141] S910, calling the pulse emission time recorded when learning the sample gas;

[0142] Specifically, the classifier is used to call the stored standard pulse emission time.

[0143] S911, comparing the target pulse emission time with each standard pulse emission time;

[0144] Specifically, a comparator is used to compare the target pulse emission time with each standard pulse emission time.

[0145] S912, taking the type of the sample gas corresponding to the standard pulse emission time having the highest similarity to the target pulse emission time as the gas type of the target gas.

[0146] Specifically, the above S910-S912 as a whole are the work performed by the gas identifier in the working mode.

[0147] Aiming at the problems of few identification types, low identification accuracy and high power consumption in portable electronic noses, this application realizes online learning of gas identifiers based on asynchronous circuits of multiple Click control units and designs low-power gas identification processors, which can solve the problem of performance degradation or even failure caused by sensor drift, improve identification accuracy and increase service life, and meet the actual application requirements of multiple identification types and low power consumption, thus broadening the application field of gas identification.

[0148] The above steps all input the control signal to the function module through the signal trigger, instructing the current function module to start executing the corresponding function.

[0149] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0150] Based on the same inventive concept, the embodiment of the present application also provides a gas identification device for implementing the gas identification method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more gas identification device embodiments provided below can refer to the limitations of the gas identification method above, and will not be repeated here.

[0151] In one embodiment, Fig.15 As shown, a gas identification device is provided, comprising: an acquisition module 10, wherein:

[0152] The acquisition module 10 is used to identify the gas type of the collected target gas based on the network parameters of the pre-stored multiple sample gases, and obtain the gas type identification result of the target gas; wherein each network parameter is generated based on the bionic olfactory neural network that pre-learns the gas types of multiple sample gases.

[0153] In one embodiment, the acquisition module 10 includes: a first acquisition unit, a comparison unit and a determination unit, wherein:

[0154] The first acquisition unit is used to acquire the target pulse emission time of the target gas at the excitation layer when the gas identification processor identifies the target gas using the weight information of the neuron synapse corresponding to each sample gas.

[0155] The comparison unit is used to compare the target pulse emission time with each standard pulse emission time.

[0156] The determination unit is used to use the type of the sample gas corresponding to the standard pulse emission time having the highest similarity to the target pulse emission time as the gas type of the target gas.

[0157] In one embodiment, the above device further includes: a processing module and a conversion module, wherein:

[0158] A processing module, used for quantifying and normalizing the target gas to obtain a gas normalized digital signal;

[0159] The conversion module is used to convert the gas specification digital signal into a pulse signal.

[0160] In one embodiment, the above-mentioned device further includes: an identification module, which is used to control the gas identification processor to execute a process of identifying the gas type of the target gas through an asynchronous logic control circuit.

[0161] In one embodiment, the identification module further includes: a conversion unit and an indication unit, wherein:

[0162] A conversion unit, used to convert the start signal of the current functional module into a control signal through a signal control unit;

[0163] The indication unit is used to input a control signal into the current functional module through a signal trigger, so as to indicate the current functional module to start executing a corresponding function.

[0164] In one embodiment, the above device further includes a learning module, the learning module includes: a second acquisition unit and an update unit, wherein:

[0165] The second acquisition unit is used to acquire pulse signals of multiple sample gases.

[0166] The updating unit is used to update the network parameters of the bionic olfactory neural network through each pulse signal until the bionic olfactory neural network converges to obtain the network parameters of multiple sample gases.

[0167] In one embodiment, the second acquisition unit of the device includes a first acquisition subunit and a second acquisition subunit, and the updating unit of the device includes an updating subunit and a determining subunit.

[0168] The first acquisition subunit is used to acquire, for each sample gas, a first pulse emission time of a pulse signal of the sample gas at the excitation layer and a second pulse emission time at the inhibition layer.

[0169] The second acquisition subunit is used to acquire the weight change amount according to the time difference between the first pulse emission time and the second pulse emission time.

[0170] The updating subunit is used to update the weight information of the neuron synapse based on the weight change.

[0171] A subunit is determined, which is used for iterative execution for multiple times. If the weight change is zero, it is determined that the bionic olfactory neural network converges.

[0172] In one embodiment, the above device further includes: a storage module, wherein:

[0173] The storage module is used to store the network parameters of multiple sample gases after the bionic olfactory neural network converges.

[0174] Each module in the above-mentioned gas identification device can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0175] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Fig.16As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store gas identification data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a gas identification method is implemented.

[0176] Those skilled in the art will understand that Fig.16 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0177] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0178] Based on the pre-stored network parameters of the multiple sample gases, the gas type of the collected target gas is identified to obtain a gas type identification result of the target gas;

[0179] Among them, each network parameter is generated based on that the bionic olfactory neural network is pre-learned about the gas types of multiple sample gases.

[0180] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0181] When the gas identification processor identifies the target gas using the weight information of the neuron synapses corresponding to each sample gas, the target pulse emission time of the target gas at the excitation layer is obtained;

[0182] Compare the target pulse emission time with each standard pulse emission time;

[0183] The type of the sample gas corresponding to the standard pulse emission time having the highest similarity to the target pulse emission time is used as the gas type of the target gas.

[0184] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0185] Quantify and normalize the target gas to obtain a gas standard digital signal;

[0186] The block converts the gas regulation digital signal into a pulse signal.

[0187] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0188] The gas identification processor is controlled by an asynchronous logic control circuit to execute a process of identifying the gas type of the target gas.

[0189] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0190] Converting the start signal of the current functional module into a control signal through the signal control unit;

[0191] The control signal is input into the current functional module through the signal trigger, instructing the current functional module to start executing the corresponding function.

[0192] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0193] Obtain pulse signals of multiple sample gases;

[0194] The network parameters of the bionic olfactory neural network are updated through each pulse signal until the bionic olfactory neural network converges, thereby obtaining the network parameters of a variety of sample gases.

[0195] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0196] For each sample gas, obtaining a first pulse emission time of a pulse signal of the sample gas at an excitation layer and a second pulse emission time at an inhibition layer;

[0197] Obtaining a weight change according to a time difference between a first pulse emission time and a second pulse emission time;

[0198] Update the weight information of neuron synapses based on the weight change;

[0199] The iteration is performed multiple times, and if the weight change is zero, it is determined that the bionic olfactory neural network has converged.

[0200] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0201] The bionic olfactory neural network is controlled by an asynchronous logic control circuit to execute a process of updating the network parameters of the bionic olfactory neural network through each pulse signal until the bionic olfactory neural network converges.

[0202] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0203] If the weight change is zero, an enable signal is sent to a signal control unit connected to the excitation layer through the enable control unit to instruct the bionic olfactory neural network to converge.

[0204] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0205] After the bionic olfactory neural network converges, the network parameters of multiple sample gases are stored.

[0206] The implementation solution for solving the problem provided by the processor in the computer device provided in the embodiment of the present application is similar to the implementation solution recorded in the above method. Therefore, the specific limitations in one or more computer device embodiments provided below can refer to the limitations on the gas identification method above and will not be repeated here.

[0207] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0208] Based on the pre-stored network parameters of the multiple sample gases, the gas type of the collected target gas is identified to obtain a gas type identification result of the target gas;

[0209] Among them, each network parameter is generated when the bionic olfactory neural network pre-learns the gas types of multiple sample gases.

[0210] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0211] When the gas identification processor identifies the target gas using the weight information of the neuron synapses corresponding to each sample gas, the target pulse emission time of the target gas at the excitation layer is obtained;

[0212] Compare the target pulse emission time with each standard pulse emission time;

[0213] The type of the sample gas corresponding to the standard pulse emission time having the highest similarity to the target pulse emission time is used as the gas type of the target gas.

[0214] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0215] Quantify and normalize the target gas to obtain a gas standard digital signal;

[0216] Converts the gas regulation digital signal into a pulse signal.

[0217] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0218] The gas identification processor is controlled by an asynchronous logic control circuit to execute a process of identifying the gas type of the target gas.

[0219] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0220] Converting the start signal of the current functional module into a control signal through the signal control unit;

[0221] The control signal is input into the current functional module through the signal trigger, instructing the current functional module to start executing the corresponding function.

[0222] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0223] Obtain pulse signals of multiple sample gases;

[0224] The network parameters of the bionic olfactory neural network are updated through each pulse signal until the bionic olfactory neural network converges, thereby obtaining the network parameters of a variety of sample gases.

[0225] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0226] For each sample gas, obtaining a first pulse emission time of a pulse signal of the sample gas at an excitation layer and a second pulse emission time at an inhibition layer;

[0227] Obtaining a weight change according to a time difference between a first pulse emission time and a second pulse emission time;

[0228] Update the weight information of neuron synapses based on the weight change;

[0229] The iteration is performed multiple times, and if the weight change is zero, it is determined that the bionic olfactory neural network has converged.

[0230] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0231] The bionic olfactory neural network is controlled by an asynchronous logic control circuit to execute a process of updating the network parameters of the bionic olfactory neural network through each pulse signal until the bionic olfactory neural network converges.

[0232] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0233] If the weight change is zero, an enable signal is sent to a signal control unit connected to the excitation layer through the enable control unit to instruct the bionic olfactory neural network to converge.

[0234] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0235] After the bionic olfactory neural network converges, the network parameters of multiple sample gases are stored.

[0236] The implementation solution for solving the problem provided by the processor in the computer-readable storage medium provided in the embodiment of the present application is similar to the implementation solution recorded in the above method. Therefore, the specific limitations in the one or more computer-readable storage medium embodiments provided below can be referred to the limitations on the gas identification method above and will not be repeated here.

[0237] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0238] Based on the pre-stored network parameters of the multiple sample gases, the gas type of the collected target gas is identified to obtain a gas type identification result of the target gas;

[0239] Among them, each network parameter is generated when the bionic olfactory neural network pre-learns the gas types of multiple sample gases.

[0240] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0241] When the gas identification processor identifies the target gas using the weight information of the neuron synapses corresponding to each sample gas, the target pulse emission time of the target gas at the excitation layer is obtained;

[0242] Compare the target pulse emission time with each standard pulse emission time;

[0243] The type of the sample gas corresponding to the standard pulse emission time having the highest similarity to the target pulse emission time is used as the gas type of the target gas.

[0244] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0245] Quantify and normalize the target gas to obtain a gas standard digital signal;

[0246] Converts the gas regulation digital signal into a pulse signal.

[0247] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0248] The gas identification processor is controlled by an asynchronous logic control circuit to execute a process of identifying the gas type of the target gas.

[0249] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0250] Converting the start signal of the current functional module into a control signal through the signal control unit;

[0251] The control signal is input into the current functional module through the signal trigger, instructing the current functional module to start executing the corresponding function.

[0252] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0253] Obtain pulse signals of multiple sample gases;

[0254] The network parameters of the bionic olfactory neural network are updated through each pulse signal until the bionic olfactory neural network converges, thereby obtaining the network parameters of a variety of sample gases.

[0255] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0256] For each sample gas, obtaining a first pulse emission time of a pulse signal of the sample gas at an excitation layer and a second pulse emission time at an inhibition layer;

[0257] Obtaining a weight change according to a time difference between a first pulse emission time and a second pulse emission time;

[0258] Update the weight information of neuron synapses based on the weight change;

[0259] The iteration is performed multiple times, and if the weight change is zero, it is determined that the bionic olfactory neural network has converged.

[0260] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0261] The bionic olfactory neural network is controlled by an asynchronous logic control circuit to execute a process of updating the network parameters of the bionic olfactory neural network through each pulse signal until the bionic olfactory neural network converges.

[0262] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0263] If the weight change is zero, an enable signal is sent to a signal control unit connected to the excitation layer through the enable control unit to instruct the bionic olfactory neural network to converge.

[0264] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0265] After the bionic olfactory neural network converges, the network parameters of multiple sample gases are stored.

[0266] The implementation solution for solving the problem provided by the processor in the computer program product provided in the embodiment of the present application is similar to the implementation solution recorded in the above method. Therefore, the specific limitations in one or more computer program product embodiments provided below can refer to the limitations on the gas identification method above and will not be repeated here.

[0267] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0268] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0269] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0270] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A gas identification method, characterized in that: The method comprises: Based on the pre-stored network parameters of the plurality of sample gases, the gas type of the collected target gas is identified to obtain a gas type identification result of the target gas; Wherein, each of the network parameters is generated when the bionic olfactory neural network pre-learns the gas types of multiple sample gases, the bionic olfactory neural network includes an excitation layer and an inhibition layer, and the network parameters of each sample gas include a standard pulse emission time at the excitation layer, and weight information of neuron synapses between the excitation layer and the inhibition layer; The method of identifying the gas type of the collected target gas based on the pre-stored network parameters of the plurality of sample gases to obtain the gas type identification result of the target gas includes: Acquire the target pulse emission time of the target gas at the excitation layer when the gas identification processor identifies the target gas using the weight information of the neuron synapse corresponding to each sample gas; comparing the target pulse emission time with each of the standard pulse emission times; The type of the sample gas corresponding to the standard pulse emission time having the highest similarity to the target pulse emission time is used as the gas type of the target gas.

2. The method according to claim 1, characterized in that The method further comprises: quantifying and normalizing the target gas to obtain a gas normalized digital signal; The gas specification digital signal is converted into a pulse signal.

3. The method according to claim 1 or 2, characterized in that: The method further comprises: The gas identification processor is controlled by an asynchronous logic control circuit to execute a process of identifying the gas type of the target gas.

4. The method according to claim 3, characterized in that The asynchronous logic control circuit includes a signal control unit and a signal trigger; the gas identification processor includes a plurality of different functional modules, each functional module is connected to a signal control unit and a signal trigger; The process of controlling the gas identification processor to identify the gas type of the target gas through the asynchronous logic control circuit includes: Converting the start signal of the current functional module into a control signal through the signal control unit; The control signal is input into the current functional module through the signal trigger to instruct the current functional module to start executing the corresponding function.

5. The method according to claim 1 or 2, characterized in that: The generation process of the network parameters of the multiple gases includes: Obtain pulse signals of multiple sample gases; The network parameters of the bionic olfactory neural network are updated by each of the pulse signals until the bionic olfactory neural network converges, thereby obtaining the network parameters of a plurality of sample gases.

6. The method according to claim 5, characterized in that The network parameters include weight information of neuron synapses between the excitation layer and the inhibition layer in the bionic olfactory neural network, and the pulse emission time of each sample gas at the excitation layer; Then, updating the network parameters of the bionic olfactory neural network by each of the pulse signals until the bionic olfactory neural network converges comprises: For each sample gas, obtaining a first pulse emission time of a pulse signal of the sample gas at the excitation layer and a second pulse emission time at the inhibition layer; Obtaining a weight change according to a time difference between the first pulse emission time and the second pulse emission time; Updating the weight information of the neuron synapse based on the weight change; The iteration is performed multiple times, and if the weight change is zero, it is determined that the bionic olfactory neural network has converged.

7. The method according to claim 6, characterized in that The method further comprises: The bionic olfactory neural network is controlled by an asynchronous logic control circuit to execute a process of updating the network parameters of the bionic olfactory neural network by each pulse signal until the bionic olfactory neural network converges.

8. The method according to claim 7, characterized in that The asynchronous logic control circuit includes a signal control unit and an enable control unit, and if the weight change amount is zero, determining that the bionic olfactory neural network converges includes: If the weight change is zero, an enable signal is sent to a signal control unit connected to the excitation layer through the enable control unit to instruct the bionic olfactory neural network to converge.

9. The method according to claim 5, characterized in that The method further comprises: After the bionic olfactory neural network converges, the network parameters of the multiple sample gases are stored.

10. A gas identification device, characterized in that: The device comprises: A gas identification processor, used to identify the gas type of the collected target gas based on the pre-stored network parameters of the multiple sample gases, and obtain a gas type identification result of the target gas; Wherein, each of the network parameters is generated when the bionic olfactory neural network pre-learns the gas types of multiple sample gases, the bionic olfactory neural network includes an excitation layer and an inhibition layer, and the network parameters of each sample gas include a standard pulse emission time at the excitation layer, and weight information of neuron synapses between the excitation layer and the inhibition layer; The method of identifying the gas type of the collected target gas based on the pre-stored network parameters of the plurality of sample gases to obtain the gas type identification result of the target gas includes: Acquire the target pulse emission time of the target gas at the excitation layer when the gas identification processor identifies the target gas using the weight information of the neuron synapse corresponding to each sample gas; comparing the target pulse emission time with each of the standard pulse emission times; The type of the sample gas corresponding to the standard pulse emission time having the highest similarity to the target pulse emission time is used as the gas type of the target gas.

11. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 9 are implemented.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 9 are implemented.

13. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 9 are implemented.

Images