Odor removal control method and device based on intelligent olfaction chip

Abstract

The invention discloses a peculiar smell removal control method and device based on an intelligent olfactory chip, and relates to the technical field of peculiar smell removal, the peculiar smell removal control method based on the intelligent olfactory chip adopts multi-point sensing and a deep neural network to perform smell recognition, and combines closed-loop feedback to adaptively adjust a preset threshold value, so that the peculiar smell removal efficiency is improved. Therefore, dynamic regulation and control are realized; a plurality of sensing units are arranged in a public place, data of various volatile components in air can be collected in real time, and the defect of response lag of a traditional timing ventilation mode is overcome; according to the method, a convolutional neural network is used for carrying out feature extraction and judgment on segmented data, a continuous odor signal is converted into a binary indication signal, then according to historical data and preset threshold values of environmental parameters, local data are subjected to weighted summation to generate a regulation and control instruction, and more accurate control over fresh air, local ventilation and deodorant spraying devices is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of odor removal, and in particular to an odor removal control method and device based on an intelligent olfactory chip. Background Art

[0002] In some densely populated areas with complex environments, odor problems will greatly affect comfort and safety. Current odor removal methods mostly rely on forced ventilation at fixed times, preset wind speeds, and regular spraying of deodorants. They are difficult to cope with instantaneous changes in human flow and are also difficult to handle a variety of odor types. When the flow of people suddenly increases or local odors are concentrated in a certain area, the established ventilation frequency and wind speed cannot capture the sudden change in time, resulting in the local environment being unable to reach an ideal state in a short period of time.

[0003] To solve this problem, some traditional solutions introduce sensors for monitoring and set fixed thresholds to trigger fans or spray systems. However, such methods have problems with insufficient sensitivity and limited recognition capabilities. Sensors can only reflect the concentration of volatile substances in the overall air and it is difficult to identify specific harmful components or judge the differences in odor changes. At the same time, the control strategies of traditional systems are mostly preset modes and lack the ability to dynamically adjust. Therefore, they are not effective in dealing with diverse and sudden odors.

[0004] In addition, deodorant spraying is mostly fixed, which results in a certain waste of resources. Moreover, after continuous use, it is easy to cause environmental adaptability problems, and the odor control effect will gradually decrease. Therefore, there is an urgent need for an odor control method and device based on an intelligent olfactory chip to solve such problems. Summary of the invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] The present invention provides an odor removal control method and device based on an intelligent olfactory chip to solve the problems of crowded public spaces, sudden odors with various sources, and single detection equipment being prone to missed detection and insufficient response.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In a first aspect, an embodiment of the present invention provides a method for removing odors based on an intelligent olfactory chip, which comprises:

[0009] Step S1, arranging a plurality of sensor units at different locations in a public place, wherein the sensor units include an odor collection device based on an intelligent olfactory chip, wherein the intelligent olfactory chip adopts an electrode film material and is used to collect data on volatile components in the air;

[0010] Step S2, inputting the data collected in step S1 into a data preprocessing module, wherein the data preprocessing module performs noise filtering, normalization processing and time window division on the data to obtain segmented data;

[0011] Step S3, using the segmented data to establish an odor recognition module through a deep learning algorithm, the odor recognition module extracts and distinguishes features of different odor signals to form recognition data;

[0012] Step S4, comparing the identification data with a threshold value preset based on historical data and environmental parameters, and generating a control instruction;

[0013] Step S5, transmitting the control instruction to a deodorization device control unit, wherein the deodorization device control unit includes a fresh air system, a local ventilation device, and a directional deodorant spraying device, and each device adjusts operating parameters according to the control instruction;

[0014] Step S6: establishing a closed-loop feedback module, wherein the closed-loop feedback module includes a data storage unit and an adaptive update unit.

[0015] As a preferred solution of the odor removal control method based on the intelligent olfactory chip described in the present invention, the odor collection device based on the intelligent olfactory chip in the sensing unit includes multiple chip sets, and the chip sets correspond to different types of volatile components respectively.

[0016] As a preferred solution of the odor removal control method based on the intelligent olfactory chip described in the present invention, the data preprocessing module divides the data into time series and uses a statistical method to correct abnormal data.

[0017] As a preferred solution of the odor removal control method based on the intelligent olfactory chip described in the present invention, in step S3, the deep learning algorithm uses a convolutional neural network to extract and distinguish features of odor data.

[0018] As a preferred solution of the odor removal control method based on the intelligent olfactory chip described in the present invention, in step S3, the step of extracting and distinguishing the features of the odor data using a convolutional neural network is as follows:

[0019] For the input segmented data set Each data segment x i Perform primary feature extraction, the feature extraction formula is:

[0020] f i =φ(W (1) *x i +β (1) ),

[0021] Among them, fi represents the feature matrix of the i-th data segment after the first convolutional layer, W (1) Represents the first convolution kernel, dimension is k 1 ×k 1 , x i represents the i-th data segment, with dimensions of p×q, β (1) represents the bias of the first convolutional layer, φ(·) represents the nonlinear activation function, and * represents the convolution operation;

[0022] The primary features are further abstracted through nested convolution and pooling structures. The formula is:

[0023] z i =σ(V·flatten(max{0,W (2) *(max{0,W (1) *x i +β (1)})+β (2)})+γ),

[0024] Among them, z i represents the output of the i-th data segment after passing through the entire convolutional neural network, W (2) Represents the second convolution kernel, dimension is k 2 ×k 2 , β (2) represents the bias of the second convolutional layer, V represents the weight matrix of the fully connected layer, γ represents the bias of the fully connected layer, flatten(·) represents the flattening operation, max{0,·} represents the ReLU activation function, σ(·) represents the normalization function, and the embedded structure realizes the continuous operation of convolution, activation, pooling and fully connected mapping.

[0025] As a preferred solution of the odor removal control method based on the intelligent olfactory chip described in the present invention, in step S4, the identification data is compared with the threshold value preset based on historical data and environmental parameters, and the step of generating the control instruction is as follows:

[0026] Define the indicator function for comparing the odor recognition output with the preset threshold, defined as:

[0027] If i ≥θ, then δ i =1,

[0028] If i <θ, then δ i =0,

[0029] Among them, δ i represents the comparison result of the i-th data segment, z irepresents the output of the i-th data segment in step S3, θ represents a preset threshold parameter, which is determined based on historical data and environmental parameters;

[0030] The control instructions are generated using the indication results of each data segment. The formula is:

[0031]

[0032] Among them, c represents the generated control instruction, ω i represents the weight corresponding to the i-th data segment, N represents the total number of data segments, and g(·) represents a mapping function, which is used to convert the weighted sum into a specific control instruction.

[0033] As a preferred solution of the deodorization control method based on the intelligent olfactory chip described in the present invention, the deodorization device control unit adjusts the ventilation frequency of the fresh air system, the wind speed of the local ventilation equipment and the deodorant spraying amount according to the control instructions.

[0034] As a preferred solution of the odor removal control method based on the intelligent olfactory chip described in the present invention, wherein: in step S6, the data storage unit is used to store the collected data and the recognition results, and the adaptive update unit adjusts the preset threshold and the recognition model parameters according to the stored data;

[0035] In step S6, the adaptive updating unit updates the preset threshold and the recognition model parameters according to the data in the data storage unit.

[0036] As a preferred solution of the odor removal control method based on the intelligent olfactory chip described in the present invention, in step S6, the step of the adaptive updating unit adjusting the preset threshold and the recognition model parameters according to the stored data is as follows:

[0037] The gradient descent method is used to update the preset threshold, and the update formula is:

[0038]

[0039] Among them, θ new represents the updated threshold, θ represents the original preset threshold, λ represents the learning rate, N represents the total number of data segments, and z i Represents the output of the i-th data segment, represents the actual feedback value of the i-th data segment;

[0040] Update the weight parameters in the convolutional neural network, and the update formula is:

[0041]

[0042] Among them, W newrepresents the updated network weight, W represents the current network weight, including W (1) and W (2) , represents the gradient with respect to the weight W, represents the loss function;

[0043] Update the bias parameters, and the update formula is:

[0044]

[0045] Among them, β new represents the updated bias, β represents the current bias parameter, represents the gradient with respect to the bias β.

[0046] In a second aspect, the present invention provides an odor removal control device based on an intelligent olfactory chip, comprising:

[0047] A sensing unit, the sensing unit includes an odor collection device based on an intelligent olfactory chip, the collection device uses an electrode film material and is used to collect data on various volatile components in the air in public places;

[0048] A data preprocessing module, which receives the data output by the sensor unit, performs noise filtering, normalization and time series division on the data, and outputs segmented data;

[0049] An odor recognition module, wherein the odor recognition module uses a deep learning algorithm to extract and distinguish features of the segmented data to form recognition data, and outputs a comparison result after comparing it with a preset threshold;

[0050] A control module, wherein the control module generates a control instruction according to the comparison result and transmits the instruction to a control unit of the deodorization device;

[0051] A control unit for the deodorization device, including a fresh air system, a local ventilation device, and a directional deodorant spraying device, each of which adjusts operating parameters according to the control instructions;

[0052] A closed-loop feedback module includes a data storage unit and an adaptive update unit. The data storage unit is used to store data output by the sensing unit, the data preprocessing module and the odor recognition module. The adaptive update unit updates the preset threshold and recognition model parameters according to the stored data.

[0053] The beneficial effects of the present invention are as follows: the present invention provides an odor removal control method based on an intelligent olfactory chip, which adopts multi-point sensing and deep neural networks for odor recognition, and combines closed-loop feedback to adaptively adjust preset thresholds, thereby achieving dynamic regulation; multiple sensing units are arranged in public places, and the present invention can collect data on various volatile components in the air in real time, overcoming the response lag of traditional timed ventilation methods.

[0054] The present invention utilizes a convolutional neural network to extract and identify features of segmented data, converts continuous odor signals into binary indication signals, and then generates control instructions after weighted summation of each local data based on historical data and preset thresholds of environmental parameters, thereby achieving more accurate control of fresh air, local ventilation and deodorant spraying devices; this method addresses the problems of uneven odor concentration and local odor aggregation, and breaks through the limitations of fixed parameter control in traditional timing control.

[0055] In addition, the present invention introduces a closed-loop feedback mechanism. By storing the collected data and actual feedback, the gradient descent method is used to adaptively update the preset threshold and the internal parameters of the neural network. During the continuous operation, the control strategy can be automatically adjusted according to the actual situation on site. The modules of the whole process cooperate with each other to form a complete closed loop from data collection, feature extraction, comparison generation to control execution, realizing real-time monitoring and dynamic adjustment.

[0056] In summary, the present invention can effectively deal with the problem of odor in public places, improve the pertinence and flexibility of odor removal control, and provide a technically reasonable and operationally efficient solution for solving odor management in crowded environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0058] Figure 1 The figure is a flow chart of the odor removal control method based on the intelligent olfactory chip of the present invention.

[0059] Figure 2 The figure is a schematic diagram of the framework of the odor removal control device based on the intelligent olfactory chip of the present invention. DETAILED DESCRIPTION

[0060] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0061] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0062] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0063] Example 1, reference Figure 1 and Figure 2 This embodiment provides a method for removing odor based on an intelligent olfactory chip, comprising the following steps:

[0064] Step S1, arranging a plurality of sensor units at different locations in a public place, wherein the sensor units include an odor collection device based on an intelligent olfactory chip, wherein the intelligent olfactory chip adopts an electrode film material and is used to collect data on volatile components in the air;

[0065] The odor collection device based on the intelligent olfactory chip in the sensor unit includes multiple chip sets, and the chip sets correspond to different types of volatile components;

[0066] Step S2, inputting the data collected in step S1 into a data preprocessing module, which performs noise filtering, normalization processing and time window division on the data to obtain segmented data;

[0067] The data preprocessing module divides the data into time series and uses statistical methods to correct abnormal data;

[0068] Step S3, using the segmented data, establishing an odor recognition module through a deep learning algorithm, the odor recognition module extracts and distinguishes features of different odor signals to form recognition data;

[0069] In step S3, the deep learning algorithm uses a convolutional neural network to extract and distinguish features from the odor data;

[0070] In step S3, the steps of using a convolutional neural network to extract and distinguish features of odor data are as follows:

[0071] For the input segmented data set Each data segment x i Perform primary feature extraction, the feature extraction formula is:

[0072] f i =φ(W(1) *x i +β (1) ),

[0073] Among them, f i represents the feature matrix of the i-th data segment after the first convolutional layer, W (1) Represents the first convolution kernel, dimension is k 1 ×k 1 , x i represents the i-th data segment, with dimensions of p×q, β (1) represents the bias of the first convolutional layer, φ(·) represents the nonlinear activation function, and * represents the convolution operation;

[0074] The primary features are further abstracted through nested convolution and pooling structures. The formula is:

[0075] z i =σ(V·flatten(max{0,W (2) *(max{0,W (1) *x i +β (1)})+β (2)})+γ),

[0076] Among them, z i represents the output of the i-th data segment after passing through the entire convolutional neural network, W (2) Represents the second convolution kernel, dimension is k 2 ×k 2 , β (2) represents the bias of the second convolutional layer, V represents the weight matrix of the fully connected layer, γ represents the bias of the fully connected layer, flatten(·) represents the flattening operation, max{0,·} represents the ReLU activation function, σ(·) represents the normalization function, and the embedded structure realizes the continuous operation of convolution, activation, pooling and fully connected mapping;

[0077] Specifically, the segmented data is used as input, local features are extracted through two layers of convolution operations, and nonlinear mapping is achieved through ReLU activation. The nested convolution layer and pooling operation make the features more abstract, and finally the network output is obtained by full connection mapping. This process constitutes the odor recognition module;

[0078] The features extracted by the odor recognition module are used for subsequent comparison;

[0079] Step S4, comparing the identification data with a preset threshold value based on historical data and environmental parameters, and generating a control instruction;

[0080] In step S4, the identification data is compared with the threshold value preset based on historical data and environmental parameters, and the step of generating the control instruction is as follows:

[0081] Define the indicator function for comparing the odor recognition output with the preset threshold, defined as:

[0082] If i ≥θ, then δ i =1,

[0083] If i <θ, then δ i =0,

[0084] Among them, δ i represents the comparison result of the i-th data segment, z i represents the output of the i-th data segment in step S3, θ represents a preset threshold parameter, which is determined based on historical data and environmental parameters;

[0085] The control instructions are generated using the indication results of each data segment. The formula is:

[0086]

[0087] Among them, c represents the generated control instruction, ω i represents the weight corresponding to the i-th data segment, N represents the total number of data segments, and g(·) represents a mapping function used to convert the weighted sum into a specific control instruction;

[0088] Specifically, the output of the recognition module is threshold-matched, the continuous output is converted into a binary indication signal, and an indication function is used to indicate whether each data segment exceeds a preset threshold. The weight accumulation of each data segment is then combined to convert it into a control instruction. This method uses a weighted summation method to integrate each local data, and a mapping function converts the result into an actual control signal to drive the deodorization device.

[0089] Step S5, transmitting the control instruction to the deodorization device control unit, the deodorization device control unit includes a fresh air system, a local ventilation device and a directional deodorant spraying device, and each device adjusts the operating parameters according to the control instruction;

[0090] The deodorizing device control unit adjusts the ventilation frequency of the fresh air system, the wind speed of the local ventilation equipment, and the amount of deodorant sprayed according to the control instructions;

[0091] Step S6, establishing a closed-loop feedback module, the closed-loop feedback module includes a data storage unit and an adaptive update unit;

[0092] In step S6, the data storage unit is used to store the collected data and the recognition results, and the adaptive update unit adjusts the preset threshold and the recognition model parameters according to the stored data;

[0093] In step S6, the adaptive updating unit updates the preset threshold and the recognition model parameters according to the data in the data storage unit;

[0094] In step S6, the adaptive updating unit adjusts the preset threshold and the recognition model parameters according to the stored data.

[0095] The gradient descent method is used to update the preset threshold, and the update formula is:

[0096]

[0097] Among them, θ new represents the updated threshold, θ represents the original preset threshold, λ represents the learning rate, N represents the total number of data segments, and z i Represents the output of the i-th data segment, represents the actual feedback value of the i-th data segment;

[0098] Update the weight parameters in the convolutional neural network, and the update formula is:

[0099]

[0100] Among them, W new represents the updated network weight, W represents the current network weight, including W (1) and W (2) , represents the gradient with respect to the weight W, represents the loss function;

[0101] Update the bias parameters, and the update formula is:

[0102]

[0103] Among them, β new represents the updated bias, β represents the current bias parameter, represents the gradient with respect to the bias β;

[0104] Specifically, a closed-loop feedback mechanism is established here. According to the error between the stored recognition output and the actual feedback, the preset threshold and the internal parameters of the neural network are updated by the gradient descent method. The threshold update formula averages the errors of all data segments and adjusts the threshold. The weight and bias update formula corrects the internal parameters of the network, uses the global data to calculate the average, and then multiplies it by the learning rate to update the parameters. The recognition and control strategies can be adjusted according to historical data.

[0105] This embodiment also provides an odor removal control device based on an intelligent olfactory chip, comprising:

[0106] The sensing unit includes an odor collection device based on an intelligent olfactory chip. The collection device uses an electrode membrane material and is used to collect data on various volatile components in the air in public places;

[0107] A data preprocessing module, which receives the data output by the sensor unit, performs noise filtering, normalization and time series division on the data, and outputs segmented data;

[0108] The odor recognition module uses a deep learning algorithm to extract and distinguish features from segmented data, forms recognition data, and outputs the comparison results after comparing with the preset threshold;

[0109] A control module, which generates a control instruction according to the comparison result and transmits the instruction to the control unit of the deodorization device;

[0110] The deodorizing device control unit includes a fresh air system, a local ventilation device, and a directional deodorant spraying device. Each device adjusts the operating parameters according to the control instructions;

[0111] The closed-loop feedback module includes a data storage unit and an adaptive update unit. The data storage unit is used to store data output by the sensing unit, the data preprocessing module and the odor recognition module. The adaptive update unit updates the preset threshold and recognition model parameters according to the stored data.

[0112] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for removing odor based on an intelligent olfactory chip, characterized in that: include, Step S1, arranging a plurality of sensor units at different locations in a public place, wherein the sensor units include an odor collection device based on an intelligent olfactory chip, wherein the intelligent olfactory chip adopts an electrode film material and is used to collect data on volatile components in the air; Step S2, inputting the data collected in step S1 into a data preprocessing module, wherein the data preprocessing module performs noise filtering, normalization processing and time window division on the data to obtain segmented data; Step S3, using the segmented data to establish an odor recognition module through a deep learning algorithm, the odor recognition module extracts and distinguishes features of different odor signals to form recognition data; Step S4, comparing the identification data with a threshold value preset based on historical data and environmental parameters, and generating a control instruction; Step S5, transmitting the control instruction to a deodorization device control unit, wherein the deodorization device control unit includes a fresh air system, a local ventilation device, and a directional deodorant spraying device, and each device adjusts operating parameters according to the control instruction; Step S6: establishing a closed-loop feedback module, wherein the closed-loop feedback module includes a data storage unit and an adaptive update unit.

2. The method for removing odor based on an intelligent olfactory chip according to claim 1, characterized in that: The odor collection device based on the intelligent olfactory chip in the sensor unit includes a plurality of chip groups, and the chip groups correspond to different types of volatile components respectively.

3. The method for removing odor based on an intelligent olfactory chip according to claim 2, characterized in that: The data preprocessing module divides the data into time series and uses statistical methods to correct abnormal data.

4. The method for removing odor based on an intelligent olfactory chip according to claim 3, characterized in that: In step S3, the deep learning algorithm uses a convolutional neural network to extract and identify features of the odor data.

5. The method for removing odor based on an intelligent olfactory chip as claimed in claim 4, characterized in that: In step S3, the step of using a convolutional neural network to extract and distinguish features of odor data is as follows: For the input segmented data set Each data segment x i Perform primary feature extraction, the feature extraction formula is: f i =φ(W (1) *x i +b (1) ), Among them, f i represents the feature matrix of the i-th data segment after the first convolutional layer, W (1) Represents the first convolution kernel, with dimension k1×k1, x i represents the i-th data segment, with dimensions of p×q, β (1) represents the bias of the first convolutional layer, φ(·) represents the nonlinear activation function, and * represents the convolution operation; The primary features are further abstracted through nested convolution and pooling structures. The formula is: z i =σ(V·flatten(max{0,W (2) *(max{0,W (1) *x i +b (1) })+b (2) })+c), Among them, z i represents the output of the i-th data segment after passing through the entire convolutional neural network, W (2) represents the second convolution kernel, with dimension k2×k2, β (2) represents the bias of the second convolutional layer, V represents the weight matrix of the fully connected layer, γ represents the bias of the fully connected layer, flatten(·) represents the flattening operation, max{0,·} represents the ReLU activation function, σ(·) represents the normalization function, and the embedded structure realizes the continuous operation of convolution, activation, pooling and fully connected mapping.

6. The method for removing odor based on an intelligent olfactory chip as claimed in claim 5, characterized in that: In step S4, the identification data is compared with a threshold value preset based on historical data and environmental parameters, and the step of generating a control instruction is as follows: Define the indicator function for comparing the odor recognition output with the preset threshold, defined as: If i ≥θ, then δ i =1, If i <θ, then δ i =0, Among them, δ i represents the comparison result of the i-th data segment, z i represents the output of the i-th data segment in step S3, θ represents a preset threshold parameter, which is determined based on historical data and environmental parameters; The control instructions are generated using the indication results of each data segment. The formula is: Among them, c represents the generated control instruction, ω i represents the weight corresponding to the i-th data segment, N represents the total number of data segments, and g(·) represents a mapping function, which is used to convert the weighted sum into a specific control instruction.

7. The method for removing odor based on an intelligent olfactory chip according to claim 6, characterized in that: The deodorizing device control unit adjusts the ventilation frequency of the fresh air system, the wind speed of the local ventilation equipment, and the deodorant spraying amount according to the control instructions.

8. The method for removing odor based on an intelligent olfactory chip as claimed in claim 7, characterized in that: In step S6, the data storage unit is used to store the collected data and the recognition results, and the adaptive updating unit adjusts the preset threshold and the recognition model parameters according to the stored data; In step S6, the adaptive updating unit updates the preset threshold and the recognition model parameters according to the data in the data storage unit.

9. The method for removing odor based on an intelligent olfactory chip as claimed in claim 8, characterized in that: In step S6, the adaptive updating unit adjusts the preset threshold and the recognition model parameters according to the stored data. The gradient descent method is used to update the preset threshold, and the update formula is: Among them, θ new represents the updated threshold, θ represents the original preset threshold, λ represents the learning rate, N represents the total number of data segments, and z i Represents the output of the i-th data segment, represents the actual feedback value of the i-th data segment; Update the weight parameters in the convolutional neural network, and the update formula is: Among them, W new represents the updated network weight, W represents the current network weight, including W (1) and W (2) , represents the gradient with respect to the weight W, represents the loss function; Update the bias parameters, and the update formula is: Among them, β new represents the updated bias, β represents the current bias parameter, represents the gradient with respect to the bias β.

10. A deodorization control device based on an intelligent olfactory chip, based on the deodorization control method based on an intelligent olfactory chip according to any one of claims 1 to 9, characterized in that: include: A sensing unit, the sensing unit includes an odor collection device based on an intelligent olfactory chip, the collection device uses an electrode film material and is used to collect data on various volatile components in the air in public places; A data preprocessing module, which receives the data output by the sensor unit, performs noise filtering, normalization and time series division on the data, and outputs segmented data; An odor recognition module, wherein the odor recognition module uses a deep learning algorithm to extract and distinguish features of the segmented data to form recognition data, and outputs a comparison result after comparing it with a preset threshold; A control module, wherein the control module generates a control instruction according to the comparison result and transmits the instruction to a control unit of the deodorization device; A control unit for the deodorization device, including a fresh air system, a local ventilation device, and a directional deodorant spraying device, each of which adjusts operating parameters according to the control instructions; A closed-loop feedback module includes a data storage unit and an adaptive update unit. The data storage unit is used to store data output by the sensing unit, the data preprocessing module and the odor recognition module. The adaptive update unit updates the preset threshold and recognition model parameters according to the stored data.

Images